Predicting Anti-cancer therapy related adverse event

Abstract

Methods of predicting development of an anticancer therapy-related adverse event (AE) in a subject to be treated with an anticancer therapy, comprising calculating an AE probability score for factors expressed by the subject, and combining the AE probability score to produce a total AE probability score, wherein a total AE probability score is proportional to the subject's probability of developing an AE in response to the anticancer therapy, are provided.

Description

PREDICTING ANTI-CANCER THERAPY RELATED ADVERSE EVENTCROSS REFERENCE TO RELATED APPLICATIONS

[0001] This application claims the benefit of priority of U.S. Provisional Patent Application No. 63 / 546,547, filed October 31, 2023, U.S. Provisional Patent Application No. 63 / 574,324, filed April 4, 2024, and U.S. Provisional Patent Application No. 63 / 692,748, filed September 10, 2024, the contents of which are all incorporated herein by reference in their entirety.FIELD OF THE INVENTION

[0002] The present invention is in the field of patient- specific diagnostics and adverse events (AEs) prediction.BACKGROUND OF THE INVENTION

[0003] Adverse events are common in patients receiving most types of cancer therapy, affecting patients’ life quality during treatment, and may require intervention, dose adjustments, or even discontinuation of treatment. Adverse events can range from mild symptoms to severe, life-threatening conditions. Successfully managing and predicting these adverse effects is important for improving quality of life of these patients.

[0004] Immune checkpoint inhibitors (ICIs) are therapeutic agents used to treat a variety of solid and hematological cancers. ICIs augment an anti-tumor immune development of an immune-related adverse events (irAE) by targeting checkpoint proteins such as CTLA-4, PD-1 and PD-L1 expressed on tumor and immune cells. Clinically approved ICIs include CTLA-4 inhibitors (ipilimumab), PD-1 inhibitors (pembrolizumab and nivolumab), and PD- L1 inhibitors (atezolizumab, durvalumab, and avelumab). Currently, these drugs are widely used as standard-of-care in first- and advanced-line settings for lung cancer, melanoma, head and neck cancer, kidney cancer and others.

[0005] Activation of the immune system by ICIs can result in irAEs. irAEs commonly affect the skin, liver, gastrointestinal tract, and endocrine organs, ranging in severity from mild (in the majority of the cases) to life threatening. Other relatively rare but potentially fatal irAEsinclude neurotoxicity, cardiotoxicity, and pulmonary toxicity. irAEs are graded by the Common Terminology Criteria for Adverse Events (CTCAE) clinical grading system which classifies the severity of toxicities on a 5-point scale: asymptomatic / mild (grade 1), moderate (grade 2), severe (grade 3), life-threatening (grade 4), and death (grade 5) (“Common Terminology Criteria for Adverse Events (CTCAE) Version 5.0”, U.S. Dept, of Health and Human Services, Nov. 27, 2017, the contents of which are hereby incorporated by reference in its entirety ). A meta-analysis of Phase II and III clinical trials reported a 14%-28% incidence rate of grade 3-4 irAEs in patients treated with CTLA-4, PD-1 and PD-L1 inhibitors (Xu et al., 2018 “Comparative safety of immune checkpoint inhibitors in cancer: systematic review and network meta-analysis”, BMJ 363, k4226).

[0006] irAEs can occur at any time during a patient’s treatment course, most commonly in the first 3 months of treatment. While most irAEs resolve with the appropriate intervention, some become chronic and may require lifelong treatment. Management of irAEs requires a tailored approach primarily involving the use of corticosteroids, immunosuppressants or cytokine antagonists. Hospitalization may be necessary in some cases. The CTCAE grading system guides clinical decisions such as whether to continue or suspend ICI therapy, or to manage irAEs with immunosuppression. Thus, irAEs potentially affect patient morbidity and mortality due to the adverse event itself, the need to discontinue ICI therapy, or possible blunting of ICI- induced immune activation through systemic immunosuppression.

[0007] Chemotherapy is an effective treatment for many cancer types, however is often associated with a range of adverse events (AEs) due to its impact on both cancerous and healthy cells. Common chemo-associated adverse events include nausea, vomiting, fatigue, and hair loss, as well as more serious complications like immunosuppression, which increases the risk of infections. Bone marrow suppression can lead to anemia, neutropenia, and thrombocytopenia, affecting the body's ability to fight infections and control bleeding. Chemotherapy can also cause peripheral neuropathy or gastrointestinal-related adverse events. The frequency and severity of these events vary depending on the type and dosage of the chemotherapy agent, as well as individual patient factors. Currently, there are different kinds of models for prediction of chemo-associated adverse events, such as clinical risk score models (e.g., CARG toxicity score in geriatric oncology), however, no such test is available using plasma-based sample obtained prior to treatment initiation.

[0008] Identifying patients who are at high risk of developing AEs would facilitate the early diagnosis and timely management of AEs, as well as lower the risk of AEs by guidingtreatment choices prior to initiation of therapy. Currently, there are no reliable biomarkers for predicting AEs in anticancer therapy -treated patients, and thus there is a clear clinical need for such biomarkers.SUMMARY OF THE INVENTION

[0009] The present invention provides methods of predicting development of an anticancer therapy-related adverse event (AE) in a subject to be treated with an anticancer therapy, comprising calculating an AE probability score for factors expressed by the subject, and combining the AE probability score to produce a total AE probability score, wherein a total AE probability score is proportional to the subject’s probability of developing an AE in response to the anticancer therapy.

[0010] According to a first aspect, there is provided a method of predicting development of an anticancer therapy -related adverse event (AE) in response to an anticancer therapy in a subject, the method comprising: a. receiving factor expression levels for a plurality of factors i. in a population of subjects known to not develop an AE in response to the anticancer therapy (AE negative); ii. in a population of subjects known to develop an AE in response to the anticancer therapy (AE positive); and iii. in the subject; b. calculate for factors of the plurality of factors an AE probability score, wherein the calculating comprises applying a machine learning algorithm trained on a training set comprising the received factor expression levels in the AE negative population and the AE positive population to individual received factor expression levels from the subject and wherein the machine learning algorithm outputs the AE probability score; and c. combine the calculated AE probability scores to produce a total AE probability score wherein the total AE probability score is proportional to the subject’s probability of developing an AE in response to the anticancer therapy; thereby predicting development of an AE in response to an anticancer therapy in a subject.[Oi l] According to another aspect, there is provided a method of predicting development of an anticancer therapy -related adverse event (AE) in response to an anticancer therapy in a subject, the method comprising: a. receiving factor expression levels for a plurality of factors i. in a population of subjects known to not develop an AE in response to the anticancer therapy (AE negative); ii. in a population of subjects known to develop an AE in response to the anticancer therapy (AE positive); and iii. in the subject; b. calculate for at least one factor of the plurality of factors an AE probability score, wherein the calculating comprises applying a machine learning algorithm and wherein the machine learning algorithm outputs the AE probability score; c. classify a factor with an AE probability score beyond a predetermined threshold as a toxicity associated factor; and d. sum the number of toxicity associated factors expressed in the subject, wherein the number of toxicity associated factors is a total AE probability score and wherein the total AE probability score is proportional to the subject’s probability of developing an AE in response to the anticancer therapy; thereby predicting the development of an AE in response to an anticancer therapy in a subject.

[0012] According to some embodiments, the anticancer therapy is a chemotherapy and the adverse events are chemo-associated adverse events (caAE).

[0013] According to some embodiments, the chemotherapy is selected from: Carboplatin, Abraxane, gemcitabine, paclitaxel, nab-paclitaxel, pemetrexed, vinorelbine, cisplatin, etoposide and combinations thereof.

[0014] According to some embodiments, the caAEs are selected from the group consisting of: fatigue, hair loss, easy bruising and bleeding, infection, anemia (low red blood cell counts), nausea and vomiting, appetite changes / loss of appetite, constipation, diarrhea, mouth / tongue / throat problems such as sores and pain with swallowing, peripheral neuropathy, numbness, tingling, neuropathic pain, skin / nail changes, dry skin, skin colorchange, urine and bladder changes, kidney problems, weight changes, loss of concentration and focus, mood changes, changes in libido and sexual function, fertility problems, leukemia, and neutropenia.

[0015] According to some embodiments, the plurality of factors comprises at least two factors selected from the factors provided in Table 9.

[0016] According to some embodiments, the plurality of factors consists of factors selected from Table 9.

[0017] According to some embodiments, the plurality of factors comprises all factors provided in Table 9.

[0018] According to some embodiments, the anticancer therapy is an immunotherapy and the adverse events are immune-related adverse events (irAE).

[0019] According to some embodiments, the immunotherapy comprises administration of an immune checkpoint protein inhibitor.

[0020] According to some embodiments, the immune checkpoint protein is selected from Programmed cell death protein 1 (PD-1), Programmed death-ligand 1 (PD-L1), cytotoxic T- lymphocyte-associated protein 4 (CTLA-4) and Lymphocyte-activation gene 3 (LAG-3).

[0021] According to some embodiments, the immunotherapy is a monotherapy.

[0022] According to some embodiments, the immunotherapy is a combination therapy and the combination therapy further comprises a chemotherapy or targeted therapy.

[0023] According to some embodiments, the chemotherapy is selected from Carboplatin, Pemetrexed, Paclitaxel, Cisplatin, and Gemcitabine.

[0024] According to some embodiments, the irAE is selected from: itching, rash, pruritus, skin blisters, loss of skin pigmentation, hand-foot syndrome, Lichen planus, psopriasis, mucositis, dry mouth, mouth pain, dermatitis, diarrhea, colitis, hepatitis, vomiting, hypothyroidism, hyperthyroidism, pituitary inflammation (hypophysitis), adrenal insufficiency, thyroiditis, diabetic ketoacidosis, mild joint pain, muscle pain, inflammatory arthritis, muscle weakness, polymyalgia rheumatica, giant cell arteritis, dry eye syndrome, uveitis, episcleritis, blepharitis, scleritis, keratitis, maculopathy, meningitis, encephalitis, myasthenia gravis, Guillain-Barre syndrome, inflamed spinal cord, inflamed brain, myocarditis, pericarditis, myocardial infarction, arrhythmia, vasculitis, Venous thromboembolism, hypertension, anemia, neumonia, neurotoxicity, myositis, nephritis,hematological toxicity, acute kidney injury, hepatotoxicity, nephrotoxicity, pancreatitis, neurotoxicity, cardiovascular toxicity, arthritis, adrenal insufficiency (Al), type 1 diabetes mellitus (T1DM), nausea, autoimmune haemolytic anaemia (AIHA), acquired thrombotic thrombocytopenic purpura (TTP), aplastic anaemia (AA), immune thrombocytopenia (ITP), acquired haemophilia (AH), haemolytic uraemic syndrome (HUS), lymphopenia, polymyalgia-like syndrome, myositis, , neuropathy, increase in serum creatinine levels, oliguria, haematuria or proteinuria, weight gain, oedema, shortness of breath, anorexia, high blood pressure, dark urine, abdominal or pelvic pain, tinnitus, fever, sepsis, leukopenia, stomatitis, pneumonia, and pneumonitis.

[0025] According to some embodiments, the plurality of factors comprises at least two factors selected from the factors provided in Table 1 or Table 3.

[0026] According to some embodiments, the plurality of factors consists of factors selected from Table 1 or Table 3.

[0027] According to some embodiments, the plurality of factors comprises all factors provided in Table 1 or all factors provided in Table 3.

[0028] According to some embodiments, the irAE is a rash.

[0029] According to some embodiments, the plurality of factors comprises at least two factors selected from the factors provided in Table 7.

[0030] According to some embodiments, the plurality of factors consists of factors selected from Table 7.

[0031] According to some embodiments, the plurality of factors comprises all factors provided in Table 7.

[0032] According to some embodiments, a subject with a total AE probability score beyond a predetermined threshold is predicted to develop an AE in response to the anticancer therapy and a subject with a total AE probability score within the predetermined threshold is predicted to not develop an AE in response to the anticancer therapy.

[0033] According to some embodiments, the AE is an early onset AE occurring within 100 days of starting the anticancer therapy, is an AE with a grade of 3 or greater on the CTCAE clinical grading system, is an AE that results in treatment delay or hold, change or termination within 100 days of starting the anticancer therapy, is an AE that caused subject hospitalization, or a combination thereof.

[0034] According to some embodiments, the AE is an early onset AE with a grade of 3 or greater on the CTCAE clinical grading system or which results in treatment delay or hold, change or termination within 100 days of starting the anticancer therapy.

[0035] According to some embodiments, the AE is an AE occurring within 180 days of starting the anticancer therapy.

[0036] According to some embodiments, the AE is an AE of any grade on the CTCAE clinical grading system.

[0037] According to some embodiments, the AE is a systemic AE effecting more than one biological system of the subject.

[0038] According to some embodiments, the AE is a plurality of AEs, the method is a method of predicting development of at least 2 AEs in response to the anticancer therapy, the AE negative population is known not to develop at least two AEs and the AE positive population is known to develop at least two AEs.

[0039] According to some embodiments, the method is a method of predicting development of at least 3 AEs in response to the anticancer therapy, the AE negative population is known not to develop at least three AEs and the AE positive population is known to develop at least three AEs.

[0040] According to some embodiments, the method comprises before (b) selecting a subset of the plurality of factors, wherein the subset comprises factors that best differentiate between the AE positive population and AE negative population, and wherein the calculating is for each factor of the subset.

[0041] According to some embodiments, the selecting comprises applying a statistical test to the received factor expression levels, optionally wherein the statistical test is a Kolmogorov-Smirnov test.

[0042] According to some embodiments, the subset consists of at least 50 factors.

[0043] According to some embodiments, the factor expression level is from a time point before administration of the anticancer therapy to the subject.

[0044] According to some embodiments, the combining is summing.

[0045] According to some embodiments, the subject suffers from cancer.

[0046] According to some embodiments, the cancer is selected from hepato-biliary cancer, cervical cancer, urogenital cancer, anogenital cancer, prostate cancer, thyroid cancer, ovarian cancer, nervous system cancer, ocular cancer, lung cancer, soft tissue cancer, bone cancer, pancreatic cancer, bladder cancer, skin cancer, intestinal cancer, hepatic cancer, rectal cancer, colorectal cancer, esophageal cancer, gastric cancer, gastroesophageal cancer, breast cancer, renal cancer, skin cancer, head and neck cancer, leukemia and lymphoma.

[0047] According to some embodiments, the cancer is non-small cell lung cancer (NSCLC).

[0048] According to some embodiments, the predetermined threshold is determined by performing a cross-validation within the training set or is the median score of the training set.

[0049] According to some embodiments, the factors expression levels are factors expression levels in a biological sample provided by the subjects.

[0050] According to some embodiments, the biological sample is selected from blood plasma, whole blood, blood serum or peripheral blood mononuclear cells.

[0051] According to some embodiments, the biological sample is blood plasma.

[0052] According to some embodiments, the training set further comprises at least one clinical parameter from each subject in the AE negative population and the AE positive population and the machine learning algorithm is applied to individual received factor expression levels from the subject and the at least one clinical parameter from the subject.

[0053] According to some embodiments, the clinical parameter is the type of anticancer therapy received, optionally wherein the type of anticancer therapy is selected from the group consisting of: immunotherapy as a monotherapy, chemotherapy as a monotherapy and immunotherapy in combination with chemotherapy.

[0054] According to some embodiments, the method further comprises administering the anticancer therapy to a subject predicted to not develop an AE in response to the anticancer therapy, stopping or delaying administering the anticancer therapy to the subject predicted to develop an AE in response to the anticancer therapy or administering a reduced dose of the anticancer therapy or an alternative therapy to the subject predicted to develop an AE in response to the anticancer therapy.

[0055] According to some embodiments, the alternative therapy is selected from the immunotherapy in combination with an immunosuppressant, an alternative immunotherapy,an alternative anticancer therapy, a non-chemotherapy anticancer therapy and a nonimmunotherapy anticancer therapy.

[0056] According to some embodiments, the trained machine learning algorithm is trained by a method comprising: at a training stage, training a machine learning algorithm on a training set comprising:(i) factor expression levels of a plurality of factors in samples from subjects known to not develop an AE in response to an anticancer therapy (AE negative) and factor expression levels of a plurality of factors in samples from subjects known to develop an AE in response to the anticancer therapy (AE positive); and(ii) labels associated with the development of an AE in the subjects in response to the anticancer therapy; to produce a trained machine learning algorithm, wherein the trained machine learning algorithm is trained to output the AE probability score.

[0057] According to some embodiments, the training set further comprises at least one clinical parameter from the subjects known to not develop an AE in response to an anticancer therapy and at least one clinical parameter from the subjects known to develop an AE in response to the anticancer therapy.

[0058] According to some embodiments, the clinical parameter is the type of anticancer therapy received, optionally wherein the type of anticancer therapy is selected from the group consisting of: immunotherapy as a monotherapy, chemotherapy as a monotherapy and immunotherapy in combination with a chemotherapy.

[0059] According to some embodiments, the expression levels of the plurality of factors are labeled with the labels, the at least one clinical parameter is labeled with the labels or both.

[0060] Further embodiments and the full scope of applicability of the present invention will become apparent from the detailed description given hereinafter. However, it should be understood that the detailed description and specific examples, while indicating preferred embodiments of the invention, are given by way of illustration only, since various changes and modifications within the spirit and scope of the invention will become apparent to those skilled in the art from this detailed description.BRIEF DESCRIPTION OF THE DRAWINGS

[0061] Figure 1: Cohort description. Overall, data from 1069 patients were explored. Following several filtration steps, 426 patients remained in the analysis. Patients were excluded in the following cases: (i) no available irAE data; (ii) protocol deviations (such as: ECOG performance status >2; targeted therapies; blood collected after therapy; mental health incompatibility; the did not receive therapy; blood collected 2 months before / after therapy); (iii) patients did not receive standard of care (SoC) treatment; (iv) no available plasma samples.

[0062] Figure 2: Patient characteristics. Basic clinical characteristics of the 426 patients participating in the analysis are described in the pie charts.

[0063] Figure 3: Performance of the irAE predictive model. Predicted irAE probability as a function of observed irAE rate. Each dot represents a patient. The observed irAE rate for each predicted irAE probability datapoint refers to the proportion of patients experiencing irAEs within a patient group assigned the predicted irAE probability ±0.15. X=Y is indicated by a black line. The goodness of fit is indicated (R2).

[0064] Figure 4: Venn diagrams describing the overlap between the toxicity -associated proteins (TAPs) and either healthy-NSCLC patients differentially expressed proteins (DEPs) (top) or resistance-associated proteins (RAPs) (bottom). Results of enrichment analysis for the overlap between each two lists of proteins are indicated on the right of each figure. Some proteins measured by the assay are redundant due to different aptamers binding the same protein. The numbers of overall and non-redundant TAPs are indicated.

[0065] Figure 5: Functional interactions between TAPs. An enrichment analysis was performed on a group of strong TAPs (i.e., TAPs that were selected repeatedly, in at least 10 iterations, during model development). The interactions, indicated with edges between nodes, were determined using the STRING database. The TAPs are colored based on different significantly enriched biological processes, as determined using Fisher exact test.

[0066] Figure 6: Voronoi plots showing the main biological functions of TAPs expressed in patients with or without irAEs. TAPs were classified into two groups based on their plasma levels in patients exhibiting or not exhibiting significant irAEs. irAE -positive TAPs refer to TAPs found at higher plasma levels among patients who experienced irAEs; irAE- negative TAPs refer to TAPs found at higher plasma levels among patients who did notexperience irAEs. Shown are Voronoi plots indicating irAE-positive and irAE-negative TAPs. The analysis was performed on a group of strong irAE-positive and irAE-negative TAPs (i.e., proteins selected repeatedly, in at least 10 iterations, during model development). Each polygon represents a TAP, and the size correlates with the TAP strength. The TAPs are grouped based on their KEGG functional group. The plots were generated using Proteomaps tool.

[0067] Figure 7: Cellular localization and potential cellular origin of TAPs and RAPs. TAPs were classified into two groups based on their plasma levels in patients exhibiting or not exhibiting significant irAEs. Similarly, RAPs (disclosed in International Patent Applications W02024033930 and PCT / IL2024 / 050456, the contents of which are hereby incorporated by reference in their entirety) were classified into two groups based on their plasma levels in patients with or without clinical benefit (CB or NCB, respectively). The proteins were then categorized according to cellular localization and origin based on the Human Protein Atlas database (proteinatlas.org).

[0068] Figure 8: Potential immune cell type origin of TAPs. TAPs were classified into two groups based on their plasma levels in patients exhibiting or not exhibiting significant irAEs. irAE-positive TAPs refer to TAPs found at higher plasma levels among patients who experienced irAEs; irAE-negative TAPs refer to TAPs found at higher plasma levels among patients who did not experience irAEs. The proteins were then categorized according to expression in immune cell types based on Human Protein Atlas database (proteinatlas.org). The fraction of irAE-positive and irAE-negative TAPs associated with each immune cell type was calculated. Many irAE-positive TAPs are myeloblast-origin proteins.

[0069] Figure 9: Enrichment analysis of TAPs. Significantly enriched categories in patients experiencing irAEs and patients not experiencing irAEs are indicated in the heatmap on the left and right columns, respectively.

[0070] Figures 10A-10D: (10A) ROC curve showing significant predictive capabilities of the predictor. (10B) Scatter plot showing a high correlation between predicted severe irAE probabilities and incidence of those irAEs (i.e., the rate of patients who experience severe irAEs among a window of ±0.1 severe irAE probability). (IOC) Scatter plot showing an inverse relationship between irAE probability and clinical benefit. (10D) Dot-plot of the distribution of the predicted probability for the no irAE population and the severe / significant irAE population.

[0071] Figures 11A-11B: (11A) Significantly enriched categories in patients experiencing irAEs and patients not experiencing irAEs are indicated in the heatmap, in the left and right columns, respectively. (11B) The fraction of irAE-positive and irAE-negative TAPs associated with each immune cell is shown in the heatmap on the right.

[0072] Figures 12A-12B: Main clinical parameters of the NSCLC cohort that was used for immune-related rash prediction. (12A) Main clinical parameters are presented in a table. (12B) The fraction of patients who experienced immune-related (ir)-rash (Rash). The rest of the patients, who did not experience irAEs, served as a control group (No irAE).

[0073] Figures 13A-13B: Performance of the irAE-rash prediction model. (13A) Receiver Operating Characteristics (ROC) plot for the irAE-rash prediction model. The area under the curve (AUC) is indicated, along with the p-value. (13B) The fit between the predicted rash irAE probability and the observed irAE rash rate (i.e., the rate of patients who experience rash irAEs among a window of ±0.1 rash irAE probability).

[0074] Figures 14A-14D: Main clinical parameters of the NSCLC cohort that was used for chemo-associated adverse events (caAEs) prediction. (14A) Main clinical parameters presented in a table. (14B) The fraction of patients who experienced caAEs. (14C) The different chemo regimens that the patients received. (14D) The effects of caAEs on patient treatment (data is available on some of the patients).

[0075] Figures 15A-15B: Performance of the caAEs prediction model. (15 A) Receiver Operating Characteristics (ROC) plot for the caAEs prediction model. The area under the curve (AUC) is indicated, along with the p-value. (15B) The fit between the predicted caAE probability and the observed caAE rate (i.e., the rate of patients who experience caAE among a window of ±0.1 caAE probability).

[0076] Figures 16A-16B: The differences between the three AE models. (16A) Venn diagram displaying the overlap and the unique TAPs in each model, focusing on TAPs that repeated in at least 10 iterations during model development. (16B) Correlations between the predictions of the 3 models. The AE prediction correlations are based on 441 NSCLC patients that were not included in any AE model development.

[0077] Figures 17A-17B: Bioinformatic analysis of the Toxicity- Associated Proteins (TAPs) of three different models. (17A) Selected significantly enriched biological processes among the TAPs of the three models. (17B) Number of TAPs in different immune cells in each model (severe irAEs, rash irAEs and caAEs). The TAPs in each model weredivided into two groups- those that are higher in patients who experienced the examined AE and TAPs that are higher in patients in the control group who did not experience the examined AE.

[0078] Figures 18A-C: (18A) A bar graph showing the distribution of patients by the number of adverse events (AEs) experienced, with the majority of patients having between 0 and 3 AEs. (18B) A bar graph displaying the number of patients across different categories of adverse events (AEs), with systemic AEs being the most prevalent. (18C) A Bar graph showing the proportion mild or severe (grade 3 or higher or AE resulting in a change or termination of treatment or hospitalization within 180 days) across all categories of AEs.

[0079] Figures 19A-19E: Performance of the prediction models based on AE number. Receiver Operating Characteristics (ROC) plot for five different prediction models based on subjects with (19A) at least 1 AE, (19B) at least 2 AEs, (19C) at least 3 AEs, (19D) at least 4 AEs and (19E) at least 5 AEs. The area under the curve (AUC) is indicated, along with the p-value.

[0080] Figure 20: Performance of the systemic AEs prediction model. Receiver Operating Characteristics (ROC) plot for the systemic AEs prediction model. The area under the curve (AUC) is indicated, along with the p-value.DETAILED DESCRIPTION OF THE INVENTION

[0081] The present invention, in some embodiments, provides methods of predicting development of an anticancer therapy -related adverse event (AE) in a subject treated or to be treated with an anticancer therapy, comprising calculating an AE probability score for factors expressed by the subject, and combining the AE probability score to produce a total AE probability score, wherein a total AE probability score is proportional to the subject’s probability of developing an AE in response to the anticancer therapy. In further embodiments, the invention provides methods of predicting development of an immune- related adverse event (irAE) in a subject treated or to be treated with an immunotherapy, comprising calculating an irAE probability score for factors expressed by the subject, and combining the irAE probability score to produce a total irAE probability score, wherein a total irAE probability score is proportional to the subject’s probability of developing an irAE in response to the immunotherapy. In further embodiments, the invention provides methodsof predicting development of a chemotherapy-associated adverse event (caAE) in a subject treated or to be treated with chemotherapy, comprising calculating a caAE probability score for factors expressed by the subject, and combining the caAE probability score to produce a total caAE probability score, wherein a total caAE probability score is proportional to the subject’s probability of developing an caAE in response to the chemotherapy.

[0082] By a first aspect, there is provided a method of predicting development of an adverse event (AE) in response to a therapy in a subject, the method comprising: a. receiving expression levels for a plurality of factors i. in a population of subjects known to not develop an AE in response to the therapy (AE negative); ii. in a population of subjects known to develop an AE in response to the therapy (AE positive); and iii. in the subject; b. calculate for factors of the plurality of factors an AE probability score, wherein the calculating comprises applying a machine learning algorithm and wherein the machine learning algorithm outputs the AE probability score; and c. combine the calculated AE probability scores to produce a total AE probability score; thereby predicting the development of an AE in response to a therapy in a subject.

[0083] By another aspect, there is provided a method of predicting development of an irAE in response to an immunotherapy in a subject, the method comprising: a. receiving expression levels for a plurality of factors i. in a population of subjects known to not develop an irAE in response to the immunotherapy (irAE negative); ii. in a population of subjects known to develop an irAE in response to the immunotherapy (irAE positive); and iii. in the subject;b. calculate for factors of the plurality of factors an irAE probability score, wherein the calculating comprises applying a machine learning algorithm and wherein the machine learning algorithm outputs the irAE probability score; and c. combine the calculated irAE probability scores to produce a total irAE probability score; thereby predicting the development of an irAE in response to an immunotherapy in a subject.

[0084] By another aspect, there is provided a method of predicting development of a caAE in response to a chemotherapy in a subject, the method comprising: a. receiving expression levels for a plurality of factors i. in a population of subjects known to not develop an caAE in response to the chemotherapy (caAE negative); ii. in a population of subjects known to develop an caAE in response to the chemotherapy (caAE positive); and iii. in the subject; b. calculate for factors of the plurality of factors an caAE probability score, wherein the calculating comprises applying a machine learning algorithm and wherein the machine learning algorithm outputs the caAE probability score; and c. combine the calculated caAE probability scores to produce a total caAE probability score; thereby predicting the development of a caAE in response to chemotherapy in a subject.

[0085] By another aspect, there is provided a method comprising: at a training stage, training a machine learning algorithm on a training set comprising:(i) factor expression levels of toxicity-associated factors (TAPs) in samples from AE negative subjects and AE positive subjects; and(ii) labels associated with AE development in the subjects;to produce a trained machine learning algorithm.

[0086] By another aspect, there is provided a method comprising: at a training stage, training a machine learning algorithm on a training set comprising:(i) factor expression levels of toxicity-associated factors (TAPs) in samples from irAE negative subjects and irAE positive subjects; and(ii) labels associated with irAE development in the subjects; to produce a trained machine learning algorithm.

[0087] By another aspect, there is provided a method comprising: at a training stage, training a machine learning algorithm on a training set comprising:(iii) factor expression levels of toxicity-associated factors (TAPs) in samples from caAE negative subjects and caAE positive subjects; and(iv) labels associated with caAE development in the subjects; to produce a trained machine learning algorithm.

[0088] By another aspect, there is provided a method of predicting development of an AE in response to a therapy in a subject, the method comprising: a. receiving expression levels for a plurality of factors i. in a population of subjects known to not develop an AE in response to the therapy (AE negative); ii. in a population of subjects known to develop an AE in response to the therapy (AE positive); and iii. in the subject; b. calculate for at least one factor of the plurality of factors an AE probability score, wherein the calculating comprises applying a machine learning algorithm and wherein the machine learning algorithm outputs the AE probability score; c. classify a factor with an AE probability score beyond a predetermined threshold as a toxicity associated factor (TAP); andd. combine the number of TAPs expressed in the subject, wherein the number of TAPs is proportional to the probability of developing an AE in response to the therapy; thereby predicting the development of an AE in response to a therapy in a subject.

[0089] By another aspect, there is provided a method of predicting development of an irAE in response to an immunotherapy in a subject, the method comprising: a. receiving expression levels for a plurality of factors i. in a population of subjects known to not develop an irAE in response to the immunotherapy (irAE negative); ii. in a population of subjects known to develop an irAE in response to the immunotherapy (irAE positive); and iii. in the subject; b. calculate for at least one factor of the plurality of factors an irAE probability score, wherein the calculating comprises applying a machine learning algorithm and wherein the machine learning algorithm outputs the irAE probability score; c. classify a factor with an irAE probability score beyond a predetermined threshold as a toxicity associated factor (TAP); and d. combine the number of TAPs expressed in the subject, wherein the number of TAPs is proportional to the probability of developing an irAE in response to the immunotherapy; thereby predicting the development of an irAE in response to an immunotherapy in a subject.

[0090] By another aspect, there is provided a method of predicting development of a caAE in response to a chemotherapy in a subject, the method comprising: a. receiving expression levels for a plurality of factors i. in a population of subjects known to not develop a caAE in response to the chemotherapy (caAE negative); ii. in a population of subjects known to develop a caAE in response to the chemotherapy (caAE positive); andiii. in the subject; b. calculate for at least one factor of the plurality of factors a caAE probability score, wherein the calculating comprises applying a machine learning algorithm and wherein the machine learning algorithm outputs the caAE probability score; c. classify a factor with a caAE probability score beyond a predetermined threshold as a toxicity associated factor (TAP); and d. combine the number of TAPs expressed in the subject, wherein the number of TAPs is proportional to the probability of developing a caAE in response to the chemotherapy; thereby predicting the development of a caAE in response to a chemotherapy in a subject.

[0091] In some embodiments, the method is a diagnostic method. In some embodiments, the method is an in vitro method. In some embodiments, the method is an ex vivo method. In some embodiments, the method is a computer-implemented method. In some embodiments, the method is a statistical method. In some embodiments, the method is a method that cannot be performed in a human mind. In some embodiments, the method is a computerized method. In some embodiments, the processor is a computer processor. In some embodiments, the processor is a computer.

[0092] In some embodiments, the method is for predicting development of an adverse event (AE). In some embodiments, the AE is a specific AE. In some embodiments, the AE is any AE. In some embodiments, the AE is a therapy related AE. In some embodiments, the AE is a treatment related AE. In some embodiments, the therapy or treatment is an anticancer therapy or treatment. In some embodiments, the AE is an immune related AE (irAE). In some embodiments, an immune related AE is an immunotherapy related AE. In some embodiments, the anticancer therapy or treatment is immunotherapy. In some embodiments, the irAE is a severe irAE. In some embodiments, the AE is a severe AE. In some embodiments, severe is a grade of 3 or greater on the CTCAE clinical grading system. In some embodiments, the AE is a chemotherapy-associated AE (caAE). In some embodiments, the anticancer treatment or therapy is chemotherapy. In some embodiments, the method is for predicting development of an AE in response to therapy. In some embodiments, the method is for predicting development of an AE in response to anticancertherapy. In some embodiments, the method is for predicting development of an AE in response to chemotherapy. In some embodiments, the method is for predicting not developing an AE in response to therapy. In some embodiments, the method is for determining development of an AE in response to therapy. In some embodiments, the method is for determining lack of development of an AE in response to therapy. In some embodiments, the method is for determining the risk of developing AE in response to therapy. In some embodiments, the method is for determining the risk of developing irAE in response to immunotherapy. In some embodiments, the method is for determining the risk of developing caAE in response to chemotherapy. In some embodiments, therapy is anticancer therapy. In some embodiments, therapy is chemotherapy. In some embodiments, therapy is radiation therapy. In some embodiments, therapy is targeted therapy. In some embodiments, targeted therapy is tyrosine kinase inhibitor (TKI) therapy.

[0093] In some embodiments, the method is for determining development of an AE probability score. In some embodiments, the method is for determining the probability of developing AE in response to therapy. In some embodiments, the method is for determining probability of development of an AE in response to an immunotherapy. In some embodiments, the method is for determining probability of development of an AE in response to a chemotherapy. In some embodiments, the method is for determining the probability of developing caAE in response to chemotherapy. In some embodiments, the method is for determining development of an caAE probability score. In some embodiments, probability of development of an AE is an AE score. In some embodiments, determining is predicting. According to some embodiments, AE probability score is determined. According to other embodiments, prediction of AE development probability is determined. According to some other embodiments, AE probability below 80%, 75%, 70%, 65%, 60%, 55%, 50%, 45%, 40%, 35%, 30%, 25%, 20%, 15%, 10% or 5% indicates the subject will not develop an AE in response to the therapy. Each possibility represents a separate embodiment of the invention. According to some other embodiments, AE probability below 20% indicates the subject will not develop an AE in response to the therapy. According to some other embodiments, AE probability below 26.2% indicates the subject will not develop an AE in response to the therapy. According to some other embodiments, AE probability below 50% indicates the subject will not develop an AE in response to the therapy. According to some embodiments, development of an AE is determined. According to other embodiments, prediction of development of an AE is determined. According to some other embodiments,an AE probability score beyond 95%, 90%, 85%, 80%, 75%, 70%, 65%, 60%, 55%, 50%, 45%, 40%, 35%, 30%, 20%, 10%, or 5% indicates the subject will develop an AE. Each possibility represents a separate embodiment of the invention. According to some other embodiments, an AE probability score beyond 80% indicates the subject will develop an AE. According to some other embodiments, an AE probability score beyond 50% indicates the subject will develop an AE. According to some other embodiments, an AE probability score beyond 26.2% indicates the subject will develop an AE. In some embodiments, beyond is above. In some embodiments, beyond is below. It will be understood by a skilled artisan that a scale can be designed to be measured in either direction and so above / below depends on the construction of the scale. It will be understood that the AE prediction can also be a relative or qualitative value or can be converted to such. For example, a low score may convert to “low risk of AE”, a medium score to “medium risk of AE” and a high score “high risk of AE”. Other qualitative levels of risk are also envisioned and would be derived from the score.

[0094] In some embodiments, the method is for predicting development of an irAE in response to immunotherapy. In some embodiments, the method is for predicting not developing an irAE in response to immunotherapy. In some embodiments, the method is for determining development of an irAE in response to immunotherapy. In some embodiments, the method is for determining lack of development of an irAE in response to immunotherapy. In some embodiments, the method is for determining development of an irAE probability score. In some embodiments, the method is for determining probability of development of an irAE in response to an immunotherapy. In some embodiments, probability of development of an irAE is an irAE score. In some embodiments, determining is predicting. According to some embodiments, irAE probability score is determined. According to other embodiments, prediction of irAE development probability is determined. According to some other embodiments, irAE probability below 80%, 75%, 70%, 65%, 60%, 55%, 50%, 45%, 40%, 35%, 30%, 25%, 20%, 15%, 10%, or 5% indicates the subject will not develop an irAE in response to the immunotherapy. Each possibility represents a separate embodiment of the invention. According to some other embodiments, irAE probability below 20% indicates the subject will not develop an irAE in response to the immunotherapy. According to some other embodiments, irAE probability below 50% indicates the subject will not develop an irAE in response to the immunotherapy. According to some other embodiments, AE probability below 26.2% indicates the subject will not develop an AE in response to the therapy.According to some embodiments, development of an irAE is determined. According to other embodiments, prediction of development of an irAE is determined. According to some other embodiments, an irAE probability score beyond 95%, 90%, 85%, 80%, 75%, 70%, 65%, 60%, 55%, 50%, 45%, 40%, 35%, 30%, 20%, 10%, or 5% indicates the subject will develop an irAE. Each possibility represents a separate embodiment of the invention. According to some other embodiments, an irAE probability score beyond 80% indicates the subject will develop an irAE. According to some other embodiments, an irAE probability score beyond 50% indicates the subject will develop an irAE. According to some other embodiments, an irAE probability score beyond 26.2% indicates the subject will develop an irAE. In some embodiments, beyond is above. In some embodiments, beyond is below. It will be understood by a skilled artisan that a scale can be designed to be measured in either direction and so above / below depends on the construction of the scale.

[0095] In some embodiments, the method is for determining development of an caAE probability score. In some embodiments, the method is for determining the probability of developing caAE in response to chemotherapy. In some embodiments, the method is for determining probability of development of a caAE in response to a chemotherapy. In some embodiments, the method is for determining probability of development of an caAE in response to a chemotherapy. In some embodiments, the method is for determining the probability of developing caAE in response to chemotherapy. In some embodiments, the method is for determining development of a caAE probability score. In some embodiments, probability of development of a caAE is a caAE score. In some embodiments, determining is predicting. According to some embodiments, caAE probability score is determined. According to other embodiments, prediction of cacAE development probability is determined. According to some other embodiments, caAE probability below 80%, 75%, 70%, 65%, 60%, 55%, 50%, 45%, 40%, 35%, 30%, 25%, 20%, 15%, 10%, or 5% indicates the subject will not develop a caAE in response to the therapy. Each possibility represents a separate embodiment of the invention. According to some other embodiments, caAE probability below 20% indicates the subject will not develop a caAE in response to the chemotherapy. According to some other embodiments, caAE probability below 26.2% indicates the subject will not develop a caAE in response to the chemotherapy. According to some other embodiments, caAE probability below 50% indicates the subject will not develop a caAE in response to the chemotherapy. According to some embodiments, development of a caAE is determined. According to other embodiments, prediction of development of acaAE is determined. According to some other embodiments, a caAE probability score beyond 95%, 90%, 85%, 80%, 75%, 70%, 65%, 60%, 55%, 50%, 45%, 40%, 35%, 30%, 20%, 10% or 5% indicates the subject will develop a caAE. Each possibility represents a separate embodiment of the invention. According to some other embodiments, a caAE probability score beyond 80% indicates the subject will develop a caAE. According to some other embodiments, a caAE probability score beyond 50% indicates the subject will develop a caAE. According to some other embodiments, a caAE probability score beyond 26.2% indicates the subject will develop a caAE. In some embodiments, beyond is above. In some embodiments, beyond is below. It will be understood by a skilled artisan that a scale can be designed to be measured in either direction and so above / below depends on the construction of the scale. It will be understood that the caAE prediction can also be a relative or qualitative value or can be converted to such. For example, a low score may convert to “low risk of caAE”, a medium score to “medium risk of caAE” and a high score “high risk of caAE”. Other qualitative levels of risk are also envisioned and would be derived from the score.

[0096] Examples of adverse events in response to therapy are well known in the art and the term “adverse events” is intended to cover all such adverse events unless specified otherwise. Well known classes of AEs include, for example, skin-related AEs, gastrointestinal toxicity AEs, thyroid dysfunction AEs, musculoskeletal toxicity AEs, ocular toxicity AEs, neurotoxicity AEs, cardiac toxicity AEs among others. Examples of chemotherapy adverse events include, but are not limited to: fatigue, hair loss, easy bruising and bleeding, infection, anemia (low red blood cell counts), nausea and vomiting, appetite changes / loss of appetite, constipation, diarrhea, mouth / tongue / throat problems such as sores and pain with swallowing, peripheral neuropathy, numbness, tingling, neuropathic pain, skin / nail changes, dry skin, skin color change, urine and bladder changes, kidney problems, weight changes, loss of concentration and focus, mood changes, changes in libido and sexual function, fertility problems, leukemia, anemia, and neutropenia. Examples of immune related adverse events include, but are not limited to: itching, rash, pruritus, skin blisters, loss of skin pigmentation, hand-foot syndrome, Lichen planus, psoriasis, mucositis, dry mouth, mouth pain, dermatitis, diarrhea, colitis, hepatitis, vomiting, hypothyroidism, hyperthyroidism, pituitary inflammation (hypophysitis), adrenal insufficiency, thyroiditis, diabetic ketoacidosis, mild joint pain, muscle pain, inflammatory arthritis, muscle weakness, polymyalgia rheumatica, giant cell arteritis, dry eye syndrome, uveitis, episcleritis, blepharitis, scleritis, keratitis, maculopathy, meningitis, encephalitis, myasthenia gravis,Guillain-Barre syndrome, inflamed spinal cord, inflamed brain, Myocarditis, pericarditis, myocardial infarction, arrhythmia, vasculitis, Venous thromboembolism, hypertension, anemia, neumonia, neurotoxicity, myositis, nephritis, hematological toxicity, acute kidney injury, hepatotoxicity, nephrotoxicity, pancreatitis, neurotoxicity, cardiovascular toxicity, arthritis, adrenal insufficiency (Al), Type 1 diabetes mellitus (T1DM), nausea, Autoimmune haemolytic anaemia (AIHA), Acquired thrombotic thrombocytopenic purpura (TTP), Aplastic anaemia (AA), Immune thrombocytopenia (ITP), Acquired haemophilia (AH), Haemolytic uraemic syndrome (HUS), lymphopenia, Polymyalgia-like syndrome, myositis, Guillain-Barre syndrome (GBS), neuropathy, increase in serum creatinine levels, oliguria, haematuria or proteinuria, weight gain, oedema, shortness of breath, anorexia, high blood pressure, dark urine, abdominal or pelvic pain, tinnitus, fever, sepsis, leukopenia, stomatitis, pneumonia, and pneumonitis. In some embodiments, the irAE is rash. In some embodiments, rash is skin rash. A rash can be identified by any skilled medical practitioner and generally is a skin condition in which there is a change in color, appearance and / or texture. Examples of adverse events include, for example, fatigue, skin reaction, dyspnea, constipation, vomiting, pneumonitis, colitis, hepatitis, nausea, neuropathy, anorexia, kidney injury, constipation, alopecia, anemia, pain, mucositis / stomatitis, cough, fever, neutropenia, pneumonia, hypothyroidism, edema, weakness, pulmonary embolism, hyperthyroidism, myocardial infarction, dysphagia, headache, general, gastroesophageal reflux, bleeding, hypokalemia, pruritic, thrush, chest infection, dry mouth, weight loss, rhinitis, hemoptysis, bronchitis, chest pain, lethargy, myopathy, eye disorder, hemoptysis, nephritis, arthralgia, ileitis, myocarditis, insomnia, neurotoxicity, neutropenic fever, sepsis, neurological deterioration, thrombocytopenia, renal function decline, thrombosis, edema, shaking, seizures, rheumatic symptoms, renal disfunction, tinnitus, infection, pleural effusion, dyspnea, vaginal dryness, peripheral vascular disease, periodontitis, tantrums, confusion, lung infection, leukopenia, dysgeusia, drug intolerance, drop in white blood count, drop in kidney function, diabetes, dehydration, conjunctivitis, cold sores, cold hands, cardiogenic shock, bruising, bronchopneumonia, bowel perforation, aphasia, alkaline phosphate increase, abdominal spasms, dyspepsia, arrhythmia, hepatotoxicity, leg swelling, hepatitis, hypophyseal disorder, hypophosphatemia, hyponatremia, hypomagnesemia, thyroiditis, hypertension, hydronephrosis, gastritis, ear infection, eye discharge, embolism, elevated creatine, elevated alanine transaminase and wheezing. Further, adverse events generally speaking an include a combination of all general, immune related and chemotherapy related adverse events.

[0097] In some embodiments, the method is for determining if a subject is AE positive. In some embodiments, the method is for determining if a subject is AE negative. In some embodiments, the method is for determining a subject’s degree of risk of developing an AE. In some embodiments, the score is proportional to the degree of risk. In some embodiments, the method is for predicting a subject’s development of an AE in response to therapy. In some embodiments, the method is for monitoring development of an AE in response to the therapy. In some embodiments, the method is for determining if the therapy should continue, be adjusted (e.g., by further treating the subject with an additional therapy including but not limited to an agent determined by the TAP analysis provided hereinbelow or an immunosuppressant), be delayed, stopped or changed. In some embodiments, the method is for determining a subject as being AE positive or AE negative. In some embodiments, the method is for determining if the subject will benefit or not benefit from the therapy.

[0098] In some embodiments, the method is for determining if a subject is irAE positive. In some embodiments, the method is for determining if a subject is irAE negative. In some embodiments, the method is for predicting a subject’s development of an irAE in response to therapy. In some embodiments, the method is for monitoring development of an irAE in response to the therapy. In some embodiments, the method is for determining if the therapy should continue, be adjusted (e.g., by further treating the subject with an additional therapy including but not limited to an agent determined by the TAP analysis provided hereinbelow or an immunosuppressant), be delayed, stopped or changed. In some embodiments, the method is for determining a subject as being irAE positive or irAE negative. In some embodiments, the method is for determining if the subject will benefit or not benefit from the immunotherapy.

[0099] In some embodiments, the method is for determining if a subject is caAE positive. In some embodiments, the method is for determining if a subject is caAE negative. In some embodiments, the method is for predicting a subject’s development of a caAE in response to therapy. In some embodiments, the method is for monitoring development of a caAE in response to the therapy. In some embodiments, the method is for determining if the therapy should continue, be adjusted (e.g., by further treating the subject with an additional therapy including but not limited to an agent determined by the TAP analysis provided hereinbelow or an immunosuppressant), be delayed, stopped or changed. In some embodiments, the method is for determining a subject as being caAE positive or caAE negative. In someembodiments, the method is for determining if the subject will benefit or not benefit from the chemotherapy.

[0100] In some embodiments, the subject is a mammal. In some embodiments, the subject is a human. In some embodiments, the subject suffers from a disease. In some embodiments, the disease is treatable by the therapy. In some embodiments, the disease is treatable by the immunotherapy. In some embodiments, the disease is treatable by the chemotherapy. In some embodiments, the disease is treatable by target therapy. In some embodiments, the disease is cancer. In some embodiments, the disease is treatable by an immune checkpoint inhibitor (ICI). In some embodiments, the cancer is a PD-L1 positive cancer. In some embodiments, the cancer is a PD-L1 high cancer. In some embodiments, the cancer is a PD- L1 low cancer. In some embodiments, the cancer is a PD-L1 negative cancer. In some embodiments, the cancer is a PD-L1 low or negative cancer. In some embodiments, the cancer is a Lymphocyte-activation gene 3 (LAG-3) positive cancer. In some embodiments, the cancer is solid cancer. In some embodiments, the cancer is a tumor. In some embodiments, the cancer is selected from hepato-biliary cancer, cervical cancer, urogenital cancer (e.g., urothelial cancer), anogenital, testicular cancer, prostate cancer, thyroid cancer, ovarian cancer, nervous system cancer, ocular cancer, lung cancer, soft tissue cancer, bone cancer, pancreatic cancer, bladder cancer, skin cancer, intestinal cancer, hepatic cancer, rectal cancer, colorectal cancer, esophageal cancer, gastric cancer, gastroesophageal cancer, breast cancer (e.g., triple negative breast cancer), renal cancer (e.g., renal carcinoma), head and neck cancer, leukemia and lymphoma. In some embodiments, the cancer is selected from skin cancer, and lung cancer. In some embodiments, the cancer is skin cancer. In some embodiments, the cancer is lung cancer. In some embodiments, the skin cancer is melanoma. In some embodiments, the lung cancer is small cell lung cancer. In some embodiments, the lung cancer is non-small cell lung cancer (NSCLC). In some embodiments, the lung cancer is small cell lung cancer (SCLC). In some embodiments, the melanoma is non-resectable melanoma. In some embodiments, the melanoma is metastatic melanoma. In some embodiments, the renal cancer is kidney cancer. In some embodiments, the renal cancer is renal cell carcinoma. In some embodiments, the cancer is a carcinoma.

[0101] In some embodiments, the subject is naive to the therapy before the determining. In some embodiments, the subject is naive to the immunotherapy before the determining. In some embodiments, the subject is naive to the chemotherapy before the determining. In some embodiments, the subject has not received the therapy before the determining. In someembodiments, the subject has not received the immunotherapy before the determining. In some embodiments, the subject has not received the chemotherapy before the determining. In some embodiments, the subject has received the therapy previously. In some embodiments, the subject has received the immunotherapy previously. In some embodiments, the subject has received the chemotherapy previously.

[0102] In some embodiments, the therapy is a monotherapy. In some embodiments, the therapy is a combination therapy. In some embodiments, the chemotherapy is a monotherapy. In some embodiments, the chemotherapy is administered in combination with one or more conventional cancer therapy including immunotherapy, targeted therapy, and radiotherapy. Combinations of chemotherapy and immunotherapy / radiotherapy / targeted therapy have been studied in multiple clinical trials. It will be understood by a skilled artisan that the predictive proteins disclosed herein are predictive of AE development in response to therapy as a monotherapy, as well as part of a combination therapy. In some embodiments, a monotherapy consists of chemotherapy. In some embodiments, the monotherapy does not comprise immunotherapy. In some embodiments, the combination therapy comprises a chemotherapy and another therapy. In some embodiments, the combination therapy comprises a chemotherapy and an immunotherapy. In some embodiments, the combination therapy comprises a chemotherapy and a targeted therapy. In some embodiments, the combination therapy further comprises radiation.

[0103] In some embodiments, the immunotherapy is a monotherapy. In some embodiments, the immunotherapy is a combination therapy. In some embodiments, an immunotherapy is administered in combination with one or more conventional cancer therapy including chemotherapy, targeted therapy, and radiotherapy. Combinations of ICI and chemotherapy / radiotherapy / targeted therapy have been studied in multiple clinical trials. It will be understood by a skilled artisan that the predictive proteins disclosed herein are predictive of irAE development in response to immunotherapy as a monotherapy, as well as part of a combination therapy. In some embodiments, a monotherapy consists of immunotherapy. In some embodiments, the monotherapy does not comprise chemotherapy. In some embodiments, the combination therapy comprises an immunotherapy and another therapy. In some embodiments, the combination therapy comprises an immunotherapy and a chemotherapy. In some embodiments, the combination therapy comprises at least two immunotherapy drugs. In some embodiments, the combination therapy comprises an immunotherapy and a targeted therapy. In some embodiments, the combination therapyfurther comprises radiation. In some embodiments, the immunotherapy is an antibody therapy. In some embodiments, the immunotherapy is Adoptive cellular therapy (ACT). In some embodiments, the antibody is a monoclonal antibody. In some embodiments, the antibody is a blocking antibody. In some embodiments, the antibody is an inhibitory antibody.

[0104] In some embodiments, the immunotherapy is an anti-PD-l / PD-Ll immunotherapy. In some embodiments, the immunotherapy is an anti-PD-1 immunotherapy. In some embodiments, the immunotherapy is an anti-PD-Ll immunotherapy. In some embodiments, the immunotherapy is PD-1 blockade. In some embodiments, the immunotherapy is PD-1 inhibition. In some embodiments, the immunotherapy is PD-1 modulation. In some embodiments, the immunotherapy is PD-L1 blockade. In some embodiments, the immunotherapy is PD-L1 inhibition. In some embodiments, the immunotherapy is PD-L1 modulation. In some embodiments, the immunotherapy is an anti-PD-1 antibody. In some embodiments, the immunotherapy is an anti-PD-Ll antibody. In some embodiments, the immunotherapy is a PD-1 inhibitor. In some embodiments, the immunotherapy is PD-L1 inhibitor. In some embodiments, the combination therapy further comprises a non-anti-PD- 1 / PD-L1 immunotherapy. In some embodiments, the anti-PD-l / PD-Ll immunotherapy is selected from Pembrolizumab, Nivolumab, Durvalumab and Atezolizumab. In some embodiments, the anti-PD-l / PD-Ll immunotherapy is selected from Pembrolizumab, Nivolumab, Durvalumab, Atezolizumab, Retifanlimab, Dostarlimab, Pidizilumab, and Avelumab. In some embodiments, the anti-PD-l / PD-Ll immunotherapy is selected from Pembrolizumab, Nivolumab, Durvalumab, Atezolizumab, and Cemiplimab. In some embodiments, the immunotherapy comprises Pembrolizumab. In some embodiments, the immunotherapy comprises Nivolumab. In some embodiments, the immunotherapy comprises Durvalumab. In some embodiments, the immunotherapy comprises Atezolizumab.

[0105] The term “chemotherapy” is well known in the art and general refers to a chemical agent that is cytotoxic to dividing cells (e.g., cancer cells). Examples of chemotherapy agents include, but are not limited to alkylating agents, antimetabolites, topoisomerase inhibitors, mitotic inhibitors, antitumor antibiotics among many others. In some embodiments, the chemotherapy is a platinum-based chemotherapy. Examples of chemotherapies include Altretamine , Bendamustine , Busulfan, Carboplatin, Chlorambucil, Cisplatin, Cyclophosphamide, Dacarbazine, Ifosfamide, Mechlorethamine, Melphalan, Oxaliplatin,Procarbazine, Temozolomide, Thiotepa, Trabectedin, ifosfamide, Carmustine, Lomustine, Streptozocin, 5-fluorouracil, 6-mercaptopurine, Azacitidine, Capecitabine, Cladribine, Clofarabine, Cytarabine, Decitabine, Floxuridine, Fludarabine, Gemcitabine, Hydroxyurea, Methotrexate, Nelarabine, Pemetrexed, Pentostatin,Pralatrexate, Thioguanine, Trifluridine / tipiracil combination, Etoposide, Irinotecan, Irinotecan liposomal, Mitoxantrone, Teniposide, Topotecan, Cabazitaxel, Docetaxel, Nab-paclitaxel, Paclitaxel, Vinblastine, Vinrcristine, Vincristine liposomal, Vinorelbine, Daunorubicin, Doxorubicin, liposomal, Epirubicin, Idarubicin, Mitoxantrone, Valrubicin, Bleomycin, Dactinomycin, Mitomycin-C, All-trans-retinoic acid, Arsenic trioxide, Asparaginase, Eribulin, Ixabepilone, Mitotane, Omacetaxine, Pegaspargase, Procarbazine, Romidepsin, Vorinostat, Taxol, prednisolone, bleomycin, dacabazine, mustine, vincristine, methotrexate, folinic acid, oxiplatin, and Abraxane.

[0106] In some embodiments, the chemotherapy is a combination of chemotherapies comprising at least two chemotherapies. In some embodiments, the chemotherapy is selected from: Carboplatin, Abraxane, gemcitabine, paclitaxel, nab-paclitaxel, pemetrexed, vinorelbine, cisplatin, etoposide and combinations thereof. In some embodiments, the chemotherapy is selected from those provided in Figure 14C. In some embodiments, the chemotherapy is selected from, vinorelbine, pemetrexed, paclitaxel, gemcitabine, cisplatin / vinorelbine, cisplatin / pemetrexed, cisplatin / gemcitabine, cisplatin / etoposide, carboplatin / vinorelbine, carboplatin / pemetrexed, carboplatin / paclitaxel, carboplatin / nab- paclitaxel, carboplatin / gemcitabine, and carboplatin / abraxane. In some embodiments, the chemotherapy is selected from Carboplatin, Paclitaxel, Pemetrexed, Gemcitabine, and Cisplatin. In some embodiments, the chemotherapy is selected from Carboplatin, Paclitaxel, Nab-Paclitaxel, Pemetrexed, Vinorelbine, Cisplatin, Gemcitabine, dacarbazine, temozolomide, albumin-bound paclitaxel, and vinblastine. In some embodiments, the chemotherapy is Carboplatin. In some embodiments, the chemotherapy is Paclitaxel. In some embodiments, the chemotherapy is Gemcitabine. In some embodiments, the chemotherapy is Pemetrexed. In some embodiments, the chemotherapy is Cisplatin. In some embodiments, the chemotherapy is vinorelbine. In some embodiments, the combination therapy comprises cisplatin / vinorelbine. In some embodiments, the combination therapy comprises cisplatin / pemetrexed. In some embodiments, the combination therapy comprises cisplatin / gemcitabine. In some embodiments, the combination therapy comprises cisplatin / etoposide. In some embodiments, the combination therapy comprisescarboplatin / vinorelbine. In some embodiments, the combination therapy comprises carboplatin / pemetrexed. In some embodiments, the combination therapy comprises carboplatin / paclitaxel. In some embodiments, the combination therapy comprises carboplatin / nab-paclitaxel. In some embodiments, the combination therapy comprises carboplatin / gemcitabine. In some embodiments, the combination therapy comprises carboplatin / abraxane. In some embodiments, the combination therapy comprises Carboplatin, Durvalumab, and Paclitaxel. In some embodiments, the combination therapy comprises Atezolizumab, Bevacizumab, Carboplatin, and Paclitaxel. In some embodiments, the combination therapy comprises Atezolizumab, Carboplatin, and Paclitaxel. In some embodiments, the combination therapy comprises Atezolizumab, Carboplatin, and Nab- Paclitaxel. In some embodiments, the combination therapy comprises Carboplatin, Nab- Paclitaxel, and Pembrolizumab. In some embodiments, the combination therapy comprises Carboplatin, Nivolumab, and Paclitaxel. In some embodiments, the combination therapy comprises Carboplatin, Paclitaxel, Pembrolizumab. In some embodiments, the combination therapy comprises Carboplatin, Nivolumab, Pemetrexed. In some embodiments, the combination therapy comprises Carboplatin, Paclitaxel, Pembrolizumab, and radiation. In some embodiments, the combination therapy comprises Carboplatin, and Pembrolizumab. In some embodiments, the combination therapy comprises Pemetrexed, and Pembrolizumab. In some embodiments, the combination therapy comprises Carboplatin, Pembrolizumab, and Pemetrexed. In some embodiments, the combination therapy comprises Carboplatin, Pembrolizumab, and Vinorelbine. In some embodiments, the combination therapy comprises Carboplatin, Pembrolizumab, and Gemcitabine. In some embodiments, the combination therapy comprises Cisplatin, Pembrolizumab, and Pemetrexed.

[0107] In some embodiments, the targeted therapy is selected from Bevacizumab (Avastin), Ramucirumab (Cyramza), Sotorasib (Lumakras), adagrasib (Krazati), Erlotinib (Tarceva), Afatinib (Gilotrif), Gefitinib (Iressa), Osimertinib (Tagrisso), Dacomitinib (Vizimpro), Amivantamab (Rybrevant), Mobocertinib (Exkivity), Necitumumab (Portrazza), Crizotinib (Xalkori), Ceritinib (Zykadia), Alectinib (Alecensa), Brigatinib (Alunbrig), Lorlatinib (Lorbrena), Crizotinib (Xalkori), Ceritinib (Zykadia), Lorlatinib (Lorbrena), Entrectinib (Rozlytrek), Dabrafenib (Tafinlar), Trametinib (Mekinist), Selpercatinib (Retevmo), Pralsetinib (Gavreto), Capmatinib (Tabrecta), Tepotinib (Tepmetko), Trastuzumab deruxtecan (Enhertu), Larotrectinib (Vitrakvi), Entrectinib (Rozlytrek), Vemurafenib(Zelboraf), Dabrafenib (Tafinlar), Encorafenib (Braftovi), Binimetinib (Mektovi), Imatinib (Gleevec), Dasatinib (Sprycel) and Nilotinib (Tasigna).

[0108] In some embodiments, the immunotherapy comprises CTLA-4 blockade. In some embodiments, the immunotherapy comprises a CTLA-4 inhibitor. In some embodiments, the immunotherapy comprises an anti-CTLA-4 antibody. In some embodiments, the combination therapy comprises an anti-CTLA-4 antibody. In some embodiments, the CTLA-4 antibody is Ipilimumab. In some embodiments, the CTLA-4 antibody is selected from Ipilimumab and Tremelimumab. In some embodiments, the combination therapy comprises an anti-LAG-3 antibody. In some embodiments, the LAG-3 antibody is Relatlimab.

[0109] In some embodiments, the immunotherapy is a plurality of immunotherapies. In some embodiments, the immunotherapy is immune checkpoint blockade. In some embodiments, the immunotherapy is immune checkpoint protein inhibition. In some embodiments, the immunotherapy is immune checkpoint protein modulation. In some embodiments, the immunotherapy is immune checkpoint protein regulation. In some embodiments, the immunotherapy comprises immune checkpoint inhibition. In some embodiments, the immunotherapy comprises immune checkpoint modulation. In some embodiments, the immunotherapy comprises immune checkpoint regulation. In some embodiments, immune checkpoint blockade and / or immune checkpoint inhibition comprises administering to the subject an immune checkpoint inhibitor. In some embodiments, inhibition comprises administering an immune checkpoint inhibitor. In some embodiments, the inhibitor is a blocking antibody. In some embodiments, the immunotherapy comprises immune checkpoint blockade. In some embodiments, modulation comprises administering an immune checkpoint modulator. In some embodiments, immune checkpoint modulation comprises administering to the subject an immune checkpoint modulator.

[0110] As used herein, the term “an immune checkpoint inhibitor (ICI)” refers to a single ICI, a combination of ICIs and a combination of an ICI with another cancer therapy. The ICI may be a monoclonal antibody, a dual-specific antibody, a humanized antibody, a fully human antibody, a fusion protein, or a combination thereof directed to blocking, inhibition or modulation of immune checkpoint proteins. In some embodiments, an immune checkpoint inhibitor is an immune checkpoint modulator. In some embodiments, an immune checkpoint inhibitor is an immune checkpoint blocker. In some embodiments, an immune checkpoint inhibitor is an immune checkpoint regulator. In some embodiments, the immune checkpointprotein is selected from PD-1 (Programmed Death-1); PD-L1; PD-L2; CTLA-4 (Cytotoxic T-Lymphocyte-Associated protein 4); and LAG-3 (Lymphocyte Activation Gene-3). In some embodiments, the immune checkpoint protein is selected from PD-1, PD-L1 and PD- L2. In some embodiments, the immune checkpoint protein is selected from PD-1 and PD- Ll. In some embodiments, the immune checkpoint protein is PD-LL In some embodiments, the immune checkpoint protein is CTLA-4. In some embodiments, the immune checkpoint protein is LAG-3. In some embodiments, the immune checkpoint protein is PD-1. In some embodiments, immune checkpoint blockade comprises an anti-PD-l / PD-Ll / PD-L2 immunotherapy. In some embodiments, immune checkpoint blockade comprises an anti-PD- 1 immunotherapy. In some embodiments, immune checkpoint blockade comprises an anti- PD-1 and / or anti-PD-Ll immunotherapy. In some embodiments, immune checkpoint blockade comprises an anti-CTLA-4 immunotherapy. In some embodiments, immune checkpoint blockade comprises an anti-LAG-3 immunotherapy. In some embodiments, immune checkpoint blockade comprises an anti-PD-1 and / or anti-PD-Ll immunotherapy and an anti-CTLA-4 immunotherapy. In some embodiments, the immunotherapy is anti-PD- 1 / PD-L1 immunotherapy. In some embodiments, the immunotherapy is anti-PD-l / PD-Ll axis immunotherapy. In some embodiments, immune checkpoint blockade comprises an anti-LAG-3. In some embodiments, immune checkpoint blockade comprises an anti-PD-1 and / or anti-PD-Ll immunotherapy and an anti-LAG-3 immunotherapy.

[0111] In some embodiments, the toxicity-associated factor is determined by a method comprising: a. receiving expression levels for a plurality of factors i. in a population of subjects that do not develop an AE in response to the therapy (AE negative); ii. in a population of subjects that develop an AE in response to the therapy (AE positive); and iii. in the subject; b. calculate for at least one factor of the plurality of factors an AE probability score; and c. classify a factor with an AE probability score beyond a threshold as a toxicity-associated factor (TAP).

[0112] In some embodiments, the toxicity-associated factor is determined by a method comprising: a. receiving expression levels for a plurality of factors i. in a population of subjects that do not develop an irAE in response to the immunotherapy (irAE negative); ii. in a population of subjects that develop an irAE in response to the immunotherapy (irAE positive); and iii. in the subject; b. calculate for at least one factor of the plurality of factors an irAE probability score; and c. classify a factor with an irAE probability score beyond a threshold as a toxicity-associated factor (TAP).

[0113] In some embodiments, the toxicity-associated factor is determined by a method comprising: a. receiving expression levels for a plurality of factors i. in a population of subjects that do not develop a caAE in response to the chemotherapy (caAE negative); ii. in a population of subjects that develop a caAE in response to the chemotherapy (caAE positive); and iii. in the subject; b. calculate for at least one factor of the plurality of factors a caAE probability score; and c. classify a factor with a caAE probability score beyond a threshold as a toxicity-associated factor (TAP).

[0114] In some embodiments, an AE probability score is a toxicity score. In some embodiments, an irAE probability score is a toxicity score. In some embodiments, a caAE probability score is a toxicity score. In some embodiments, toxicity-associated factors are in each subject. In some embodiments, toxicity-associated factors are determined in all subjects. In some embodiments, toxicity-associated factors are in the AE positive subjects. In some embodiments, toxicity-associated factors are in the irAE positive subjects. In someembodiments, toxicity-associated factors are in the caAE positive subjects. In some embodiments, subjects are a population. In some embodiments, toxicity-associated factors are in the AE negative subjects. In some embodiments, toxicity-associated factors are in the irAE negative subjects. In some embodiments, toxicity-associated factors are in the caAE negative subjects. In some embodiments, toxicity-associated factors are in the subject. In some embodiments, the toxicity-associated factors are labeled with the labels. In some embodiments, the expression levels of the toxicity-associated factors are labeled with the labels. In some embodiments, the toxicity-associated factors are toxicity-associated proteins.

[0115] In some embodiments, the immunotherapy is a blocking antibody. In some embodiments, the immunotherapy is administration of a blocking antibody to the subject. In some embodiments, the antibody is a monoclonal antibody.

[0116] In some embodiments, the ICI is a monoclonal antibody (mAb) against PD-1 or PD- Ll. In some embodiments, the ICI is a mAb that neutralizes / blocks / inhibits / modulates / regulates the PD-1 pathway. In some embodiments, the ICI is a mAb against PD-1. In some embodiments, the anti-PD-1 mAb is Pembrolizumab (Keytruda; formerly called lambrolizumab). In some embodiments, the anti-PD-1 mAb is Nivolumab (Opdivo). In some embodiments, the anti-PD-1 mAb is Pidilizumab (CT0011). In some embodiments, the anti-PD-1 mAb is Cemiplimab (Libtayo, REGN2810). In some embodiments, the anti-PD-1 mAb is Dostarlimab (Jemperli). In some embodiments, the anti- PD-1 mAb is Retifanlimab (Zynyz). In some embodiments, the anti-PD-1 mAb is any one of AMP-224, MEDI0680, or PDR001. In some embodiments, the ICI is a mAb against PD- Ll. In some embodiments, the anti-PD-Ll mAb is selected from Atezolizumab (Tecentriq), Avelumab (Bavencio), and Durvalumab (Imfinzi). In some embodiments, the anti-PD-Ll mAb is Atezolizumab. In some embodiments, the anti-PD-Ll mAb is Durvalumab. In some embodiments, the ICI is a mAb against CTLA-4. In some embodiments, the anti-CTLA-4 mAb is ipilimumab. In some embodiments, the anti-CTLA-4 mAb is Tremelimumab (Imjudo). In some embodiments, the ICI is a mAb against LAG-3. In some embodiments, the anti-LAG-3 mAb is Relatlimab.

[0117] As used herein, the term “factor” refers to any measurable biological molecule produced by the subject. In some embodiments, the factor is a protein. In some embodiments, the factor is an RNA. In some embodiments, the factor is a protein. In some embodiments, the factor is a protein. In some embodiments, the factor is a gene. In some embodiments, the factor is a secreted factor. In some embodiments, the secreted factors are selected fromcytokines, chemokines, growth factors, soluble receptors and enzymes. In some embodiments, the factor is a soluble factor. In some embodiment, the factor is cellular factor. In some embodiments, the factor is membranal factor. In some embodiments, the factor is a cell adhesion molecule. In some embodiments, the factor is a factor found in blood. In some embodiments, the factor is a host-generated factor. In some embodiments, the factor is a toxicity factor.

[0118] In some embodiments, expression for a plurality of factors is expression of a plurality of factors. In some embodiments, the expression is protein expression. In some embodiments, the expression is secreted protein expression. In some embodiments, protein expression is soluble protein expression. In some embodiment, the expression is cellular protein expression. In some embodiments, the expression is membranal protein expression. In some embodiments, the expression is mRNA expression. In some embodiments, the expression is protein expression or mRNA expression. In some embodiments, expression level is concentration. In some embodiments, concentration is concentration level. It will be understood by a skilled artisan that when the presence of factor is measured in a liquid sample the expression can be provided as a concentration such as mg / ml or in arbitrary units according to the method of determining the factor’s expression. Arbitrary units can be selected from relative fluorescence unit (RFU) or Normalized Protein expression (NPX), or any other arbitrary units used as measurement of expression. The terms “expression” and “expression levels” are used herein interchangeably and refer to the amount of a gene product present in the sample. In some embodiments, gene product includes polynucleotide, e.g., tumor DNA, circulating tumor DNA, or circulating DNA. In some embodiments, the DNA is cell-free DNA. In some embodiments, determining comprises quantification of expression levels. In some embodiments, determining comprises normalization of expression levels. Determining of the expression level of the factor can be performed by any method known in the art. Methods of determining protein expression include, for example, antibody arrays, immunoblotting, immunohistochemistry, flow cytometry (FACS), ELISA, proximity extension assay (PEA), aptamer-based assays, proteomics arrays, proteome sequencing, flow cytometry (CyTOF), multiplex assays, mass spectrometry and chromatography. In some embodiments, determining protein expression levels comprises ELISA. In some embodiments, determining protein expression levels comprises protein array hybridization. In some embodiments, determining protein expression levels comprises mass-spectrometry quantification. In some embodiments, determining protein expression levels comprises PEA.In some embodiments, determining protein expression levels comprises aptamers. Methods of determining mRNA expression include, for example, RT-PCR, quantitative PCR, realtime PCR, microarrays, northern blotting, in situ hybridization, next generation sequencing, and massively parallel sequencing.

[0119] In some embodiments, the receiving factor expression levels is providing factor expression levels. In some embodiments, the receiving factor expression levels is determining factor expression levels. In some embodiments, determining is measuring. In some embodiments, the measuring is in a sample. In some embodiments, the expression levels were detected in a sample. In some embodiments, the sample is a biological sample. In some embodiments, the sample is provided by the subjects. In some embodiments, the sample is provided by the subject. In some embodiments, the sample is provided by a subject known to not develop an AE in response to the therapy. In some embodiments, the sample is provided by a subject known to develop an AE in response to the therapy. In some embodiments, the sample is provided by an AE negative subject. In some embodiments, the sample is provided by an AE positive subject. In some embodiments, each subject of the AE negative population provided a sample. In some embodiments, each subject of the AE positive population provided a sample. In some embodiments, the sample is provided by a subject known to not develop an AE in response to chemotherapy. In some embodiments, the sample is provided by a subject known to develop an AE in response to chemotherapy. In some embodiments, the sample is provided by a subject before receiving the chemotherapy. In some embodiments, the factor expression level is from a time point before administration of the chemotherapy. In some embodiments, the chemotherapy is a monotherapy. In some embodiments, the chemotherapy is a combination therapy. In some embodiments, the sample is provided by a subject after receiving the chemotherapy. In some embodiments, the sample is provided by a subject known to not develop an irAE in response to the immunotherapy. In some embodiments, the sample is provided by a subject known to develop an irAE in response to the immunotherapy. In some embodiments, the sample is provided by an irAE negative subject. In some embodiments, the sample is provided by an irAE positive subject. In some embodiments, each subject of the irAE negative population provided a sample. In some embodiments, each subject of the irAE positive population provided a sample. In some embodiments, the sample is provided by a subject before receiving the immunotherapy. In some embodiments, the factor expression level is from a time point before administration of the immunotherapy. In some embodiments, theimmunotherapy is a monotherapy. In some embodiments, the therapy is an anti-PD-l / PD- L1 immunotherapy. In some embodiments, the therapy is a combination therapy. In some embodiments, the therapy is an anti-PD-l / PD-Ll immunotherapy and chemotherapy. In some embodiments, the sample is provided by a subject after receiving the immunotherapy. In some embodiments, the determining is directly in the sample. In some embodiments, the determining is in the unprocessed sample. In some embodiments, the determining is in a processed sample. In some embodiments, the method further comprises processing the sample. In some embodiments, processing comprises isolating proteins from the sample. In some embodiments, processing comprises isolating nucleic acids from the sample. In some embodiments, the nucleic acid is RNA. In some embodiments, the RNA is mRNA. In some embodiments, the processing comprises lysing cells in the sample. In some embodiments, the nucleic acid is cell free DNA. In some embodiments, the nucleic acid is tumor cell DNA.

[0120] As used herein, the terms “peptide”, “polypeptide” and “protein” are used interchangeably to refer to a polymer of amino acid residues. In another embodiment, the terms “peptide”, “polypeptide” and “protein” as used herein encompass native peptides, peptidomimetics (typically including non-peptide bonds or other synthetic modifications) and the peptide analogues peptoids and semipeptoids or any combination thereof. In another embodiment, the peptides polypeptides and proteins described have modifications rendering them more stable while in the body or more capable of penetrating into cells. In one embodiment, the terms “peptide”, “polypeptide” and “protein” apply to naturally occurring amino acid polymers. In another embodiment, the terms “peptide”, “polypeptide” and “protein” apply to amino acid polymers in which one or more amino acid residue is an artificial chemical analogue of a corresponding naturally occurring amino acid.

[0121] In some embodiments, the sample is a biological sample. In some embodiments, the sample is tissue. In some embodiments, the tissue sample is tumor sample. In some embodiments, the sample is a fluid. In some embodiments, the fluid is a biological fluid. In some embodiments, the sample is from the subject. In some embodiments, the sample is not a tumor sample. In some embodiments, the sample is a tumor sample. In some embodiments, the sample is not a hematopoietic cancer and the sample is a blood sample. In some embodiments, the sample is a sample that does not comprise cancer cells. In some embodiments, a blood sample comprises a peripheral blood sample, serum sample and a plasma sample. In some embodiments, the sample is a plasma sample. In some embodiments, the sample is a serum sample. In some embodiments, processing comprises isolating plasma.In some embodiments, processing comprises isolating serum. In some embodiments, the biological fluid is selected from blood, plasma, serum, lymph, cerebral spinal fluid, urine, feces, semen, tumor fluid and gastric fluid.

[0122] In some embodiments, the sample obtained from the subject and the AE / irAE / caAE negative population are the same type of sample. In some embodiments, the sample obtained from the subject and the AE / irAE / caAE negative population are different types of samples. In some embodiments, the sample obtained from the subject and the AE / irAE / caAE positive population are the same type of sample. In some embodiments, the sample obtained from the subject and the AE / irAE / caAE positive population are different types of samples. In some embodiments, the sample obtained from the AE / irAE / caAE positive population and the AE / irAE / caAE negative population are the same type of sample. In some embodiments, the sample obtained from the AE / irAE / caAE positive population and the AE / irAE / caAE negative population are different types of samples. In some embodiments, the sample obtained from the subject, the AE / irAE / caAE positive population and the AE / irAE / caAE negative population are the same type of sample. In some embodiments, the sample obtained from the subject, the AE / irAE / caAE positive population and the AE / irAE / caAE negative population are blood samples. In some embodiments, the sample obtained from the subject, the AE / irAE / caAE positive population and the AE / irAE / caAE negative population are plasma samples. In some embodiments, the sample obtained from the subject, the AE / irAE / caAE positive population and the AE / irAE / caAE negative population are serum samples. In some embodiments, the sample obtained from the subject, the AE / irAE / caAE positive population and the AE / irAE / caAE negative population are different types of samples.

[0123] In some embodiments, a factor is a factor of the plurality of factors. In some embodiments, expression levels of a plurality of factors are received. In some embodiments, expression levels of at least 2, 3, 4, 5, 6 ,7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 30, 35, 40, 45, 50, 100, 150, 200, 250, 300, 400, 500, 600, 700, 800, 900, 1000, 1100, 1200, 1300, 1400, 1500, 1600, 1700, 1800, 1900, 2000, 3000, 4000, 5000, 6000, 7000, 8000, 9000, 10000, 12000, 15000, 20000, 25000, 30000, 35000, or 40000 factors is received. Each possibility represents a separate embodiment of the invention. In some embodiments, expression levels of at least 50 factors are received. In some embodiments, expression levels of at least 100 factors are received. In some embodiments, expression levels of at least 200 factors are received. In some embodiments, expression levels of at least300 factors are received. In some embodiments, expression levels of at least 350 factors are received. In some embodiments, expression levels of at least 400 factors are received. In some embodiments, expression levels of at least 425 factors are received. In some embodiments, expression levels of at least 430 factors are received. In some embodiments, expression levels of at least 440 factors are received. In some embodiments, a plurality is at least 2, 3, 4, 5, 6 ,7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 30, 35, 40, 45, 50, 100, 150, 200, 250, 300, 400, 500, 600, 700, 800, 900, 1000, 1100, 1200, 1300, 1400, 1500, 1600, 1700, 1800, 1900, 2000, 3000, 4000, 5000, 6000, 7000, 8000, 9000, 10000, 12000, 15000, 20000, 25000, 30000, 35000, or 40000. Each possibility represents a separate embodiment of the invention. In some embodiments, a plurality is at least 50 factors. In some embodiments, a plurality is at least 100 factors. In some embodiments, a plurality is at least 200 factors. In some embodiments, a plurality is at least 300 factors. In some embodiments, a plurality is at least 350 factors. In some embodiments, a plurality is at least 400 factors. In some embodiments, a plurality is at least 425 factors. In some embodiments, a plurality is at least 430 factors. In some embodiments, a plurality is at least 440 factors. In some embodiments, expression levels of at least 50 factors are received. In some embodiments, expression levels of at least 100 factors are received. In some embodiments, expression levels of at least 200 factors are received. In some embodiments, expression levels of at least 300 factors are received. In some embodiments, expression levels of at least 350 factors are received. In some embodiments, expression levels of at least 400 factors are received. In some embodiments, expression levels of at least 425 factors are received. In some embodiments, expression levels of at least 430 factors are received. In some embodiments, expression levels of at least 440 factors are received. In some embodiments, expression levels of at least 450 factors are received. In some embodiments, expression levels of at least 500 factors are received. In some embodiments, expression levels of at least 550 factors are received. In some embodiments, expression levels of at least 600 factors are received. In some embodiments, expression levels of at least 650 factors are received. In some embodiments, expression levels of at least 681 factors are received. In some embodiments, expression levels of at least 1000 factors are received. In some embodiments, expression levels of at least 5000 factors are received. In some embodiments, expression levels of at least 6000 factors are received. In some embodiments, expression levels of at least 7000 factors are received. In some embodiments, expression levels of at least 8000 factors are received.

[0124] In some embodiments, the factor is selected from a factor provided in Table 1. In some embodiments, the plurality of factors is selected from the factors provided in Table 1. In some embodiments, the plurality of factors comprises at least two factors selected from those provided in Table 1. In some embodiments, the plurality of factors consists of factors selected from Table 1. In some embodiments, the factors provided in Table 1 are: ELK1, IL- 23, Mcl-1, SCAR5, CaCC, LAT, B2M, PUF60, LRP1, SLIK1, PROK2, HABP4, HHIP, IL27B, P4R3A, TMM52, C1QR1, ITGB7, TXND5, FGFP3, ALDH1A1, SELPL:ECD, COL11A2, PHS, CLK2, DGC14, ADH7, RBM24, KRT7, RN146, DTX1, RBM9, RBM23, MET, PTPS, HS71L, TADBP, MSTN, EGFLA, APOF, TRA2B, CAN3, RL30, OTUD5, MCTS1, LTOR3, EFS, PURA2, GNAI3, CSNK1G2, KNTC2, DRBP76, PPR3B, SF01, GSKIP, TEAD3, TM237, RDH16, SAP30, ENOXI, RAB 18, SRSF7, EWS, P5I11, PSD1, FSTL1, SPP1, DB 113, IL-20 Rb, CI167, CHAD, PKD2:CD 1, HSPA9B, TRADD, NP1L2, NH2L1, MD1L1, RBBP5, ELANE, CFD, NAB2, BLC, HMCS2, PRTN3, IL-10, SPOP, RPGP1, sFRP-3, IGFBP-1, CA6, UGPA1, S26A7, CYTF, IL-1 R AcP, TNFSF15, FCN3, IL-18 Ra, ADM, C1QR1, UCRP, HSP 70, SNX7, MBD1, VPS29, TPPC3, F210A, CA I 30, SERPINB5, OTOR, VNN2, LL-37, AGRN, ATRN, LEFTY2, BCHE, CLSTN1, GAPR1, EphB4, PRSS27, HTRA1, Keratin 19, KPYR, OBCAM, HSPG2, BTD, PACAP, GSN, IL- 26, LECT2, ENASE, KI2L1, SYJ2B, ADH1A, CD46, SEPTIN6, WWOX, THIK, AFM, CB032, GGE2D, ZWINT, ASRGL1, CPLX1, SERC, LAMA2, ALB, LRIG1, CANB2, CEAM8, HDGR3, H-ras (WT), SPB8, UB2D1, MLRA, Holo-TC I, FGL2, ADA, CSTB, UB2R2, BASP, DPYS, GBP5, MME, PDE4C, CIB 1, MED20, NRG3, RET1L, SCRB1, SIGI LECD, HNP-3, NEP1, NHP2, RCAN2, RPB3, CIA30, ZC4H2, GFRP, FA84B, SEN15, METK1, DHRS4, KB RSI, MRM1, T2EB, PHOP2, PRTG, VAP-1, AL7A1, BMF, LSP1, KBRS2, FOPNL, PDE1B, LRFN4, FUT11, GALT4, H0ME3, MGT5A, CLC1A, Fl l, AB1IP, BCAP, RBFOX1, TIMP-1, K1C14, KC1AL, KRT6A, RIP, FZD9, TMA16, ZBP1, FCN1, MCP-3, MDC, DAG1, TRI62, REEP2, PMF1, HOP2, RAB38, F118A, XRP2, TM221, STAR7, HSH2D, ACO12, T11L1, MALD2, VMAC, CARME, XPP3, CK049, S100A8, NELFA, KDM8, EGLN2, SAP, MPPA, FUBP2, MALT1, SCPDL, NT5D1, MACD2, CP2CJ, IGFBP-3, IL-6, MPO, PRL, CRTC3, PRAME, CD30, MRC1, M-CSF R, TrkC, TECK, EMAP-2, NovH, BMP-14, IL-22, LD78-beta, MMP-3, Lipocalin 2, Hatl, MPIF-1, CYCS, IGF-I, NKp30, FSH, MIP-la, PARC, PTN, RETN, C9, CTSA, HPLN1, RET, SARP-2, FCN2, NET4, TFPI, ACY1, ECM1, PCI, TBK1, CMA1, LTA / LTB, SERPINA1, AHSG, ARSA, OLR1, PROC, CYTD, ZAP70, BPI, IgE, PAPP-A, PLG, SERPINA3, CFH, IL-1 R4, TK1, BOC, PCSK7, C5b, 6 Complex, IL-11, BSSP4, KYNU,SFN, F9, LYZ, STC1, ARTS1, PI3, ERP29, GRN, sCD163, SIRT2, CLM6, IL-17B R, TIMD3, F9, S100A9, IL-IRa, SET, PKC-G, SP0CK2, GREM2, PGRP-L, AMD, CM35H, MFAP4, CTRB2, S0D3, VASN, CBLN4, DB 112, CRLD2, CLEC3B, FCRL1, SEM4D, IFN-g Rl, NG36, GRP, PTH, EPHAA, FGRL1, TP53 R175H, PLTP, CYR61, CLSTN3, 0IT3, GGT2, DLK1, CEI, ATL2, GP116, CPN2, MMP19, PCYOX, NPTX2, FAIM3, PZP, KISSI, REL2, SELPL:CD, HAAH, LIRB5, CETP, IFN-w, LRFN2, CD3-zeta, B2L10, NPPB, FMRI, GBP6, SCO1, CNDP1, ADA22, FCRL2, HBAT, Cathepsin B, SURF1, NRG2, ITM2B, JAML1, ITIH5, PTPRJ, NOTUM, Apo D, DLL1, PEAREECD, CMPK1, VSIG1, CTNNB1, PSA, IGFBP-2, FBLN3, G3ST1, A4GALT, F171B:ECD, CILP2, FAK2:FAT, CHL1, LRAP, G0LM1, PRG3, PCD10:ECD, TIM-1, ZNT3, NT5D3, POMC, QORL1, CATF, BTC, CBPA4, SDF-1, MFAP2, CREG1, MPZL2, Apo LI, PLA1A, GZMK, B4GT2, CDHI LECD, LCP1, IL-10 Ra:CD, ARI1A, AT1B4, and AMGOLCD. The amino acid sequences of these factors can be found in the Uniprot database, for example, and each factor’s Uniprot accession number is provided in Table 1. Further, methods, reagents, and assays for measuring expression levels of these factors are well known in the art and are commercially available.

[0125] In some embodiments, the factor is selected from a factor provided in Table 2. In some embodiments, the plurality of factors is selected from the factors provided in Table 2. In some embodiments, the plurality of factors comprises at least two factors selected from those provided in Table 2. In some embodiments, the plurality of factors consists of factors selected from Table 2. In some embodiments, the factors provided in Table 2 are: RL30, NH2L1, TEAD3, F9, S100A8, CYR61, B2M, GP116, RET1L, OIT3, BTD, TFPI, VASN, DTX1, REL2, KYNU, FGRL1, OTOR, OTUD5, BOC, C1QR1, C1QR1, NELFA, MMP-3, SRSF7, RIP, DRBP76, C9, ENOXI, CMPK1, MED20, EPHAA, AL7A1, CA6, TRADD, KI2L1, DLL1, MALD2, PLTP, ZAP70, G3ST1, ENASE, GSKIP, EphB4, SELPL:CD, FAIM3, SURF1, ERP29, B2L10, CETP, MMP19, TM221, CEAM8, QORL1, BASP, PARC, CTSB, FLANE, CSTB, NEP1, TMM52, ITGB7, K1C14, GOLM1, IL27B, MCTS1, KRT7, LCN2, Apo F, OLR1, NRG3, KRT6A, IGFBP-1, PROK2, CYTF, IL-22, CRLD2, IgE, THIK, ADA22, KPYR, MDC, DB113, BTC, SOD3, CLEC3B, PPR3B, SAP30, CBLN4, NT5D1, NG36, SCPDL, MFAP4, GAPR1, SPOCK2, FOPNL, Asparaginase-like protein 1, PCI, AHSG, Proteinase-3, SARP-2, ECM1, NT5D3, AGRN, TNFSF15, CK049, RCAN2, PGRP-L, SET, ARSA, LTOR3, PKD2:CD 1, CSNK1G2, ADM, MLRA, and VSIGL

[0126] In some embodiments, the factor is selected from a factor provided in Table 3. In some embodiments, the plurality of factors is selected from the factors provided in Table 3. In some embodiments, the plurality of factors comprises at least two factors selected from those provided in Table 3. In some embodiments, the plurality of factors consists of factors selected from Table 3. In some embodiments, the factors provided in Table 3 are LCN2, PRTN3, S100A8, BTD, AHSG, CEACAM8, ADAM22, ERP29, ADM, CLEC3B, KRT7, MARVELD2, FCMR, CD93, OLR1, ARSA, KRT6A, BTC, ELANE, SCCPDH, NELFA, ENOXI, ACAA1, ENGASE, ASRGL1, CA6, AGRN, RAET1L, CST7, TMEM221, TFPI, SPOCK2, BCL2L10, SAP30, SNUB, C9, EMG1, ADGRF5, EBI3, PGLYRP2, TMEM52, APOF, PPP1R3B, PROK2, TRADD, CSNK1G2, KRT14, SELPLG, MCTS1, MZB1, SFRP1, EHMT2, OTOR, CCN1, TEAD3, GLIPR2, LRIG1, B2M, Cllorf49, CBLN4, CSTB, MYL7, PKLR, RLN2, CMPK1, DLL1, GAL3ST1, G0LM1, NT5DC3, PRTN3, MMP19, CTSB, AFM, ITIH5, MRM1, OPCML, SET, VSIG1, EPHB4, SOD3, BOC, KYNU, NOTUM, CD93, CRISPLD2, MFAP4, S100A9, SERPINA3| KLK3, DEFBI 13, ITGB7, IL1RN, SERPINA5, ALDH7A1, ECM1, NTRK3, PLTP, ZAP70, LECT2, PSAT1, RCAN2, RNF146, ALB, CFD, FGFRL1, IGFBP1, RPL30, ALDH1A1, FAM171B, SPP1, TNFSF15, VASN, CCL23, F9, LAMTOR3, MED20, MMP3, ZBP1, CCL22, IGHE, ILF3, OIT3, RABIS, RIPK1, BCHE, NAB2, NDUFAF1, AOC3, CA6, DTX1, IL1RL1, JAML, KIR2DL1, RAP1GAP, CARNMT1, CYCS, GSKIP, IL22, SPOP, TCP11L1, CXCL13, GGT2, OTUD5, PKLR, COL11A2, MBD1, NKIRAS1, NT5DC1, PKD2, SNX7, TIMP1, APBB1IP, CCL18, CD163, GCHFR, GGH, HHIP, LILRA5, LRFN2, MAT1A, TCN1, AFM, CEP20, CETP, FMRI, GRP, IL17RB, IL26, SURF1, B2M, CDH11, CYP2C19, GOLM1, SLC3OA3, APCS, ASPH, CRYZL1, DEFA3, HAT1, IL10, ITM2B, KISSI, MRC1, PCYOX1, RBM24, SERPINA1, TRA2B, TRIM62, TSEN15, A4GALT, BGN, CCL22, CHAD, CNRIP1, CNTN1, CTRB2, EWSR1, F9, IGHG1, PHOSPHO2, PSMC3IP, TNFRSF8, CTNNB1, EPHA10, FZD9, GALNT4, GTF2E2, HRAS, IGFBP2, IL6, INA, NPTX2, PAPPA, PDE4C, PIK3AP1, PMF1, PRTG, VNN2, CLK2, CNDP1, FUT11, IGF2, IGFBP3, LEFTY2, NCR3, PCBD1, PRKCG, PSD, RBFOX2, RETN, SAA2, TK1, ACOT12, BTD, CST5, Fll, FAM210A, FCN1, IGF1, NAP1L2, NRG3, PMPCA, PUF60, SERPINA4, B4GALT2, C6| C5, CFHR5, CGA| FSHB, FGL1, GNAI3, HGF, KRT19, LBP, MUC16, PDE1B, PLCD1, PTPRJ, SCO1, SRSF7, TOR4A, TXNDC5, BMP10, CILP2, CNTN6, EGFLAM, EGLN2, GAGE2D, HAVCR2, HDGFL3, LYZ, MCL1, MGAT5, PLG, PPP4R3A, RDH16, REEP2, RPS20, SFRP1, TARDBP, ADA, BASP1, C9, CAMP, CAMP, CCL3L1, CCL7, CD300C, CGA| LHB, CLSTN3, CPLX1, CTSA, DAG1, DPYS, FCRL1,FCRL2, FRZB, GDF5, GZMA, LILRB5, NCMAP, NUCB1, PCDH10, RBM23, STARD7, TRAPPC3, UBE2D1, ZC4H2, ADH1A, ADSS2, AMIGO1, ARL8B, ATP13A1, BTN2A2, CCL3, CD46, CDH11, CHST11, CIB1, CLCA1, CLUH, CSF1R, CTSB, CTSF, DEFBI 12, FAM189A2, FCN3, FGA| FGB| FGG, GBP6, GRN, HAPLN1, HAUS1, HSH2D, HSPA1A, IFNGR1, INO80E, KDM8, LCP1, MACROD2, MALT1, MPO, NTRK2, PAM, PCSK7, PLA1A, PTK2B, RBBP5, SERPINA10, SERPINB5, SF3B4, TP53, TP53I11, ANTXR2, CCNE1, CD247, CDON, CETN2, CLEC1A, CLSTN1, CX3CL1, DHRS4, FAM151A, GBP5, GPC3, GREM2, HAVCR1, HPCAL1, HSPA1L, HSPA9, IL10RA, INHBA, KIAA1143, KLK10, LAT, LRP1, LSP1, MMP13, NCF1, NHP2, NKIRAS2, NOTCH3, PI3, POLR2C, PRAME, RAB17, RP2, SAR1B, SERPINB8, SLC26A7, TBK1, TIMP1, TMA16, ZWINT, ACY1, ADAMI 1, BACH1, BST1, C1GALT1C1, CABP2, CD300A, CD A, CFH, CHL1, CLUL1, CPA4, DCBLD1, EFEMP1, ERAP2, ESS2, FCN2, FGFR2, FOXO4, GALNT16, GGT5, GZMK, HBQ1, HSPG2, IFNE, IGFBP2, IHH, IL12B| IL23A, IL20RB, IL6, ISG15, LAMP1, LIPN, LRATD2, PCDH10, PGM5, PROS1, PRSS22, PTS, RAB2B, RAB38, RBFOX1, SAA1, SIGLEC11, SIRT2, SYNJ2BP, UBE2R2, ADH7, AGA, AGGF1, AIMP1, AKR7A3, APBB2, ASNS, ATF6, B3GNT2, BCDIN3D, BCHE, BMF, C1QTNF3, C4B| C4A, CAPN3, CCDC69, CCN3, CETN1, CFI, CHST5, CMA1, CRTC3, CSTF1, CTSV, DCC, DIABLO, DLG3, DLK1, DMKN, EGFR, ENTPD1, ENTPD3, FGF19, FGFBP3, FTL| FTH1, GSN, HABP4, HTRA1, IGFBP2, IGHA1| IGHA2, IL1RAP, IL1RAP, IL22, ITIH1, KIR2DL3, LMNB2, M1AP, MFAP2, MGAT5, MME, MMP17, NAT14, NRG1, NRG2, NRG3, NXT1, PDCD1LG2, PEAR1, PENK, PHEX, PKN2, PPP3R2, PROC, PZP, RAB26, RAB3A, RBL2, RBM18, SCARB1, SEPTIN6, SERPINA9, SF1, SFN, STC1, THRA, TNFRSF4, TNFSF14, TPPP3, UBC, UMOD, WAS, ZNF526, ACHE, ACP6, ADA2, ADI1, AKR1C4, ANGPTL3, ANKRD45, APOM, ARHGAP30, ARHGEF10, ARRDC5, ART3, ASF1A, ATRN, AZGP1, B3GALT2, BGN, BPI, C1QL2, CAPG, CAPS, CARD9, CASP1O, CBLN2, CCL25, CCL25, CD22, CD244, CD27, CDNF, CEMIP2, CFHR2, CGREF1, CHEK1, CKAP2, CLEC4C, COA4, COQ6, CPN2, CR2, CREG1, CSAG1, CSF1R, CSNK1A1L, CXCL12, CXCL8, DBNDD1, DNAJB9, ERLEC1, FAM20B, FAM50A, FCGR3B, FGFBP1, FGFR1, FGL2, FTL, GPX3, GSS, H2AW, HAMP, HECTD3, HK2, HMGB2, HPGD, HPX, ICAM1, IGDCC4, IGFBP1, IGHG4, IL11, IL11RA, IL15RA, IL3RA, IL4, INHBA, INHBC, INIP, LDLRAD3, LRFN4, LRRC15, MAGEA3, MAGOHB, MDGA2, MEIG1, MFAP2, MFAP5, MLN, MMACHC, MRC2, MSTN, MSTN| GDF11, NAGPA, NDC80, NDRG4, NDUFB4, NDUFV2, NPPB, NPTN, NT5E, NTN4, PDGFRL, PIGR, PON1, POP7, PRCP, PRG3, PRPF6, PRRT2, PRSS27,PTGR2, PTPRU, PYCARD, QPCT, RBBP4, RBP4, RCSD1, REN, RNASE3, R0B02, RP9, RPN1, RRAS, S100A11, S100A12, SAA4, SCUBE3, SDCBP, SELE, SEMA4A, SERPINA10, SERPINA4, SFTPD, SIGLEC7, SIRPA, SIRT5, SLITRK1, TBCA, TCEA2, TCL1A, TCP10L2, TGFBI, THAP11, TIMP1, TMEM59L, TMX3, TNC, TNC, TPMT, TRIM5, TSC2, TSR2, TXLNA, TYRP1, UGP2, USP21, VEGFA, VMAC, VPS29, XPNPEP3, YBX1, and YBX2. The amino acid sequences of these factors can be found in the Uniprot database, for example, and each factor’s Uniprot accession number is provided in Table 3. Further, methods, reagents, and assays for measuring expression levels of these factors are well known in the art and are commercially available.

[0127] In some embodiments, the factor is selected from a factor provided in Table 4. In some embodiments, the plurality of factors is selected from the factors provided in Table 4. In some embodiments, the plurality of factors comprises at least two factors selected from those provided in Table 4. In some embodiments, the plurality of factors consists of factors selected from Table 4. In some embodiments, the factors provided in Table 4 are: LCN2, PRTN3, S100A8, BTD, AHSG, CEACAM8, ADAM22, ERP29, ADM, CLEC3B, KRT7, MARVELD2, FCMR, CD93, OLR1, ARSA, KRT6A, BTC, ELANE, SCCPDH, NELFA, ENOXI, ACAA1, ENGASE, ASRGL1, CA6, AGRN, RAET1L, CST7, TMEM221, TFPI, SPOCK2, BCL2L10, SAP30, SNUB, C9, EMG1, ADGRF5, EBI3, PGLYRP2, TMEM52, APOF, PPP1R3B, PROK2, TRADD, CSNK1G2, KRT14, SELPLG, MCTS1, MZB 1, SFRP1, EHMT2, OTOR, CCN1, TEAD3, GLIPR2, LRIG1, B2M, Cl lorf49, CBLN4, CSTB, MYL7, PKLR, RLN2, CMPK1, DLL1, GAL3ST1, G0LM1, NT5D’3, PRTN3, MMP19, CTSB, AFM, ITIH5, MRM1, OPCML, SET, VSIG1, EPHB4, SOD3, BOC, KYNU, NOTUM, CD93, CRISPLD2, MFAP4, S100A9, SERPINA3| KLK3, DEFBI 13, ITGB7, IL1RN, SERPINA5, ALDH7A1, ECM1, NTRK3, PLTP, ZAP70, LECT2, PSAT1, RCAN2, RNF146, ALB, CFD, FGFRL1, IGFBP1, RPL30, ALDH1A1, FAM171B, SPP1, TNFSF15, VASN, CCL23, F9, LAMTOR3, MED20, MMP3, ZBP1, CCL22, IGHE, ILF3, OIT3, RABIS, RIPK1, BCHE, NAB2, NDUFAF1, AOC3, CA6, DTX1, IL1RL1, JAML, KIR2DL1, RAP1GAP, CARNMT1, CYCS, GSKIP, IL22, SPOP, TCP11L1, CXCL13, GGT2, OTI, PKLR, COL11A2, MBD1, NKIRAS1, NT5D’l, PKD2, SNX7, TIMP1, APBB 1IP, CCL18, CD163, GCHFR, GGH, HHIP, LILRA5, LRFN2, MAT1A, TCN1, AFM, CEP20, CETP, FMRI, GRP, IL17RB, IL26, SURF1, B2M, CDH11, CYP2C19, GOLM1, SLC3OA3, APCS, ASPH, CRYZL1, DEFA3, HAT1, IL10, ITM2B, KISSI, MRC1, PCYOX1, RBM24, SERPINA1, TRA2B, TRIM62, TSEN15, A4GALT, BGN,CCL22, CHAD, CNRIP1, CNTN1, CTRB2, EWSR1, F9, IGHG1, PH0SPH02, PSMC3IP, TNFRSF8, CTNNB1, EPHA10, FZD9, GALNT4, GTF2E2, HRAS, IGFBP2, IL6, INA, NPTX2, PAPPA, PDE4C, PIK3AP1, PMF1, PRTG, VNN2, CLK2, CNDP1, FUT11, IGF2, IGFBP3, LEFTY2, NCR3, PCBD1, PRKCG, PSD, RBFOX2, RETN, SAA2, and TK1.

[0128] In some embodiments, the factor is selected from a factor provided in Table 5. In some embodiments, the plurality of factors is selected from the factors provided in Table 5. In some embodiments, the plurality of factors comprises at least two factors selected from those provided in Table 5. In some embodiments, the plurality of factors consists of factors selected from Table 5. In some embodiments, the factors provided in Table 5 are IL12B|IL23A, MCL1, CLCA1, LAT, EPHA10, PUF60, LRP1, SLITRK1, CRYZL1, HABP4, HHIP, CEP20, PPP4R3A, CETP, SURF1, CCL18, PKD2, TXNDC5, FGFBP3, ALDH1A1, COL11A2, PCBD1, CLK2, ESS2, ADH7, RBM24, NT5DC1, RNF146, OTUD5, RBFOX2, RBM23, PTS, HSPA1L, TARDBP, MSTN, EGFLAM, GSKIP, TRA2B, CAPN3, IL22, DTX1, KIR2DL1, ILF3, ADSS2, GNAI3, RIPK1, NDC80, CCL22, IGHE, SF1, OIT3, LAMTOR3, RDH16, MED20, MMP3, RAB18, SRSF7, EWSR1, TP53I11, PSD, SPP1, F9, IL20RB, TOR4A, CHAD, TNFSF15, HSPA9, VASN, NAP1L2, RPL30, RBBP5, IGFBP1, CFD, NAB2, CXCL13, FGFRL1, C9, IL10, SPOP, RAP1GAP, FRZB, RCAN2, IL22, CA6, UGP2, SLC26A7, ALDH7A1, SFRP1, IL1RAP, ECM1, FCN3, ZAP70, PLTP, ISG15, HSPA1A, SNX7, MBD1, VPS29, TRAPPC3, FAM210A, NCMAP, SERPINB5, SERPINA5, VNN2, CAMP, ITGB7, ATRN, LEFTY2, BCHE, CLSTN1, DEFBI 13, CD93, PRSS27, HTRA1, KRT19, PKLR, OPCML, HSPG2, BTD, CDH11, MZB1, GSN, IL26, LECT2, MFAP4, CRISPLD2, SYNJ2BP, ADH1A, GOLM1, CD46, SEPTIN6, BOC, AFM, CNRIP1, GAGE2D, ZWINT, KYNU, CPLX1, PSAT1, ALB, LRIG1, PPP3R2, EPHB4, HDGFL3, HRAS, SERPINB8, UBE2D1, SOD3, TCN1, FGL2, ADA, SET, UBE2R2, BASP1, DPYS, GBP5, MME, PDE4C, CIB1, VSIG1, NRG3, CTSB, SCARB1, SIGLEC11, DEFA3, MMP19, NHP2, DLL1, POLR2C, NDUFAF1, ZC4H2, GCHFR, LRATD2, TSEN15, MAT1A, DHRS4, NKIRAS1, MRM1, GTF2E2, PHOSPHO2, PRTG, AOC3, CMPK1, BMF, LSP1, NKIRAS2, GAL3ST1, PDE1B, LRFN4, FUT11, GALNT4, MGAT5, MGAT5, CLEC1A, Fll, APBB1IP, PIK3AP1, RBFOX1, TIMP1, GOLM1, CSNK1A1L, NT5DC3, PKLR, FZD9, TMA16, ZBP1, FCN1, CCL7, CCL22, DAG1, TRIM62, REEP2, PMF1, PSMC3IP, RAB38, RP2, RLN2, STARD7, HSH2D, ACOT12, TCP11L1, B2M, VMAC, CARNMT1, XPNPEP3, MYL7, CSTB, Cllorf49, KDM8, EGLN2, APCS, PMPCA, MALT1, CBLN4, GLIPR2,MACR0D2, CYP2C19, IGFBP2, IGFBP3, IL6, MPO, CRTC3, PRAME, TIMP1, TNFRSF8, MRC1, CSF1R, NTRK3, CCL25, AIMP1, CCN3, GDF5, TEAD3, CCL3L1, CCN1, OTOR, HAT1, CCL23, CYCS, IGF1, NCR3, CGA|FSHB, CCL3, EHMT2, RETN, MCTS1, SFRP1, CTSA, HAPLN1, CSNK1G2, FCN2, NTN4, KRT14, ACY1, SELPLG, PROK2, APOF, TBK1, CMA1, B2M, PPP1R3B, PRTN3, SERPINA1, TRADD, EBI3, TMEM52, PROC, CST5, PGLYRP2, BPI, ADGRF5, PAPPA, PLG, SERPINA3|KLK3, CFH, IL1RL1, TK1, EMG1, PCSK7, C6|C5, IL11, PRSS22, C9, IL6, AFM, SFN, F9, LYZ, STC1, PI3, SAP30, GRN, CD163, SIRT2, CD300C, IL17RB, HAVCR2, SNUB, S100A9, IL1RN, BCL2L10, PRKCG, SPOCK2, GREM2, TFPI, PAM, CD300A, CST7, CTRB2, TMEM221, AGRN, RAET1L, DEFBI 12, ASRGL1, CA6, FCRL1, IFNGR1, ENGASE, GRP, ACAA1, ENOXI, TP53, NELFA, SCCPDH, CLSTN3, ELANE, GGT2, DLK1, BTC, CPN2, KRT6A, PCYOX1, NPTX2, ARSA, PZP, KISSI, CD93, OLR1, ASPH, LILRB5, FCMR, LRFN2, CD247, KRT7, NPPB, FMRI, GBP6, SCO1, CNDP1, MARVELD2, FCRL2, HBQ1, CTSB, CLEC3B, NRG2, ITM2B, JAML, ITIH5, PTPRJ, NOTUM, ADM, PEAR1, ERP29, ADAM22, CTNNB 1, IGFBP2, EFEMP1, CEACAM8, A4GALT, FAM171B, CILP2, PTK2B, CHL1, ERAP2, AHSG, PRG3, PCDH10, HAVCR1, SLC3OA3, BTD, PRTN3, CTSF, S100A8, CPA4, LCN2, CXCL12, MFAP2, CREG1, PLA1A, GZMK, B4GALT2, CDH11, LCP1, IL10RA, and AMIG01. The amino acid sequences of these factors can be found in the Uniprot database, for example, and each factor’s Uniprot accession number is provided in Tables 1 and 3. Further, methods, reagents, and assays for measuring expression levels of these factors are well known in the art and are commercially available.

[0129] In some embodiments, the factor is selected from a factor provided in Table 6. In some embodiments, the plurality of factors is selected from the factors provided in Table 6. In some embodiments, the plurality of factors comprises at least two factors selected from those provided in Table 6. In some embodiments, the plurality of factors consists of factors selected from Table 6. In some embodiments, the factors provided in Table 6 are: EPHA10, CRYZL1, CEP20, CETP, SURF1, CCL18, PKD2, NT5DC1, OTUD5, GSKIP, IL22, DTX1, KIR2DL1, ILF3, RIPK1, CCL22, IGHE, OIT3, LAMTOR3, MED20, MMP3, F9, TNFSF15, VASN, RPL30, IGFBP1, FGFRL1, RCAN2, ALDH7A1, ECM1, ZAP70, PLTP, SERPINA5, ITGB7, DEFB 113, CD93, MFAP4, CRISPLD2, BOC, KYNU, EPHB4, SOD3, SET, VSIG1, CTSB, MMP19, DLL1, CMPK1, GAL3ST1, GOLM1, NT5DC3, PKLR, RLN2, B2M, MYL7, CSTB, Cl lorf49, CBLN4, GLIPR2, TEAD3, CCN1, OTOR, EHMT2,MCTS1, SFRP1, CSNK1G2, KRT14, SELPLG, PR0K2, APOF, PPP1R3B, TRADD, EBI3, TMEM52, PGLYRP2, ADGRF5, EMG1, C9, SAP30, SNUB, BCL2L10, SPOCK2, TFPI, CST7, TMEM221, AGRN, RAET1L, ASRGL1, CA6, ENGASE, ACAA1, ENOXI, NELFA, SCCPDH, ELANE, BTC, KRT6A, ARSA, CD93, OLR1, FCMR, KRT7, MARVELD2, CLEC3B, ADM, ERP29, ADAM22, CEACAM8, AHSG, BTD, PRTN3, S100A8, and LCN2.

[0130] In some embodiments, the factor is a factor that appeared in at least 10 iterations. In some embodiments, the factor that appeared in at least 10 iterations is a factor provided in Table 2. In some embodiments, the factor that appeared in at least 10 iterations is a factor provided in Table 4. In some embodiments, the factor that appeared in at least 10 iterations is a factor provided in Table 6. In some embodiments, a factor that appeared in at least 10 iterations is a factor that appeared in at least 20 iterations. In some embodiments, a factor that appeared in at least 20 iterations is selected from: LCN2, PRTN3, S100A8, BTD, AHSG, CEACAM8, ADAM22, ERP29, ADM, CLEC3B, KRT7, MARVELD2, FCMR, CD93, OLR1, ARSA, KRT6A, BTC, ELANE, SCCPDH, NELFA, ENOXI, ACAA1, ENGASE, ASRGL1, CA6, AGRN, RAET1L, CST7, TMEM221, TFPI, SPOCK2, BCL2L10, SAP30, SNUB, C9, EMG1, ADGRF5, EBI3, PGLYRP2, TMEM52, APOF, PPP1R3B, PROK2, TRADD, CSNK1G2, KRT14, SELPLG, MCTS1, MZB1, SFRP1, EHMT2, OTOR, CCN1, TEAD3, GLIPR2, LRIG1, B2M, Cl lorf49, CBLN4, CSTB, MYL7, PKLR, RLN2, CMPK1, DLL1, GAL3ST1, G0LM1, NT5DC3, PRTN3, MMP19, CTSB, AFM, ITIH5, MRM1, OPCML, SET, VSIG1, EPHB4, SOD3, BOC, KYNU, NOTUM, CD93, CRISPLD2, MFAP4, S100A9, SERPINA3| KLK3, DEFBI 13, ITGB7, IL1RN, SERPINA5, ALDH7A1, ECM1, NTRK3, PLTP, ZAP70, LECT2, PSAT1, RCAN2, RNF146, ALB, CFD, FGFRL1, IGFBP1, RPL30, ALDH1A1, FAM171B, SPP1, TNFSF15, VASN, CCL23, F9, LAMTOR3, MED20, MMP3, ZBP1, CCL22, IGHE, ILF3, OIT3, RAB IS, RIPK1, BCHE, NAB2, NDUFAF1, AOC3, CA6, DTX1, IL1RL1, JAML, KIR2DL1, and RAP1GAP. In some embodiments, 10 iterations is 10 out of 80. In some embodiments, 20 iterations is 20 out of 80. In some embodiments, 80 iterations of the machine learning algorithm are run to determine the toxicity associated factors.

[0131] In some embodiments, the factors provided in Tables 1-6 are relevant for irAEs from immunotherapy. In some embodiments, the factors provided in Tables 1-6 are modulated in response to immunotherapy. In some embodiments, the factors provided in Tables 1-6 are relevant for severe irAEs from immunotherapy.

[0132] In some embodiments, the factor is selected from a factor provided in Table 7. In some embodiments, the plurality of factors is selected from the factors provided in Table 7. In some embodiments, the plurality of factors comprises at least two factors selected from those provided in Table 7. In some embodiments, the plurality of factors consists of factors selected from Table 7. In some embodiments, the factors provided in Table 7 are: IGFBP1, , SFRP1, , FGFBP1, ADA2, TREM2, KRT5, SIRPB1, TREM2, MFAP4, PTN, SSBP1, CXCL10, AOC2, PKN2, SAP30, TREM2, IGHG4, IGFBP4, AGRN, NT5DC3, LECT2, EEF2KMT, SRA1, CDCP1, RPN2, COA7, CRYZL1, SERPINA10, CXCL9, BTC, SUMF1, KRT19, KLK10, IP6K2, TNFRSF11B, RGS7, TMUB1, RRAS, TNFSF13B, GLIPR2, POMC, VAC14, COA4, CLTA, GGH, GGT2, SPRN, ITGB1; ITGA4, DCUN1D5, RDH16, MSMB, NHLRC2, FRZB, S100A11, DTNA, ELK1, NUCB1, CETN2, MAGE, LAMTOR3, ADAMI 1, BGN, GNAQ, DDOST, CLUL1, RHOB, PKD2, FCGR3B, EFEMP1, CD300A, UBXN4, CTNNB1, FDPS, SRSF6, HSPA1A, NXT1, B2M, SMOC2, NTN1, SERPINA10, NFKBID, UBE2D3, HAX1, MAP3K10, CKAP2, PHOSPHO2, NAA20, VSIG1, BMPER, PSAPL1, PRKCSH, DUSP26, SPARC, PMF1, BCL2L2, TRIM54, SLITRK1, ANKRA2, FGF7, SERPINA11, RGS21, DDHD2, MPP7, M1AP, RSPO4, STC1, NELFE, ZWINT, GNAI3, HCAR2, EFNB3, ACBD4, EFCAB1, CDCP1, TYRP1, TMEM59L, Cllorf49, HAVCR2, NDC80, FZD10, PLAU, RNASET2, RNF122, PMPCA, GPNMB, AKT2, SEPTIN6, CST2, FGB, ERLEC1, CES1, ADAM15, COPS2, UBE2R2, LRFN3, EFS, SDSL, FAM118A, ARSA, CHST5, CES3, ADM, TAF12, CFHR1, PRKCG, COMTD1, HOMER3, CCL25, PDIA4, FLRT3, SPA17, MPZL2, NPIPB3, GDF5, MAPT, FCGR2A, DEPPI, PRKG1, RNF146, NDUFA2, IFNGR1, SPHK2, ACAN, HMGB2, BTG1, DYRK1A, AOC3, S100A8, KLK11, TNFSF15, DLX3, KRT18, LCP1, LOXL2, KLKB1, CCL7, TRPT1, ZNF34, MSR1, RETN, MAGEA4, SERPINA9, NRG2, DAZAP1, ESPN, RFXAP, TNFRSF12A, PDGFC, NPFF, CD28, NANP, CXCL13, FMOD, WFDC8, CTSB, INHBB; INHBA, HPX, THAP4, GSKIP, IFNW1, COL8A1, FGFR1, ATF5, WAS, OMD, C16orf72, PEAR1, PRTN3, IGHG1, TNC, EIF4EBP3, CFHR2, NTM, VEGFB, IL26, MAN1A2, OGGI, SFRP4, CD86, RPL30, SULF2, CFTR, RCAN2, CFD, TNC, GRP, EGFR, ZSWIM7, ERP44, SGK1, PSMD7, DLL1, FGF19, MAPKAPK5, PRR15, GRN, GAS2, AOC1, MRPL58, BMF, TNFSF14, NOVAI, MEIG1, PHEX, IL18R1, NRG3, SMPDL3A, NTMT1, TEAD3, ADIPOQ, DDC, LYG2, CDH5, TSC2, NEIL2, FSTL1, CKB, CD5, ZAP70, ST6GAL2, AKR1D1, CAPS, XPNPEP3, FGFR2, C1QL3, PLA2G12B, CHST11, RLN2, HSPA9, C4B; C4A, LGI3, INIP, STX3, SFRP5, SF3B4, CLSTN2, PBRM1, EMG1, MORF4L1, CCL23, CTSV, CD247, ADH1A,RNASE3, TRAPPC3, MSANTD2, CCN4, UBE2C, RAB26, DEFBI, KEAP1, LILRA4, FCHSD1, CEACAM8, MSTN; GDF11, KNG1, WWOX, ITGB3, MASP1, TFPI, PCSK7, LRP12, CD163, ACHE, TRIM62, PC, TAFA3, RNASE3, FLT4, IBSP, PRG3, GSTA1, CRTAC1, CARHSP1, S1OOA8; S100A9, IHH, NRG3, CST7, CTSF, HMGCS2, IDUA, ALDOC, KPNA2, LIPN, CILP2, NGF, COX7A2L, TCEA3, MPO, ST3GAL6, S100A9, HSD11B1, SERPINA3; KLK3, DNALI1, ERP29, GPD1, FAM151A, SPARCL1, MXI1, NDUFAF1, FCMR, FDCSP, SEMA5B, IGHE, TCEA2, AKR7A3, HGF, AKR1C4, ICAM1, ELANE, LEPR, ADGRF5, RET, ALDOB, DAP, AKR1C3, IZUMO4, MMP19, TMEM52, ALDH1A2, LEAP2, CKM, CYP2C19, BGLAP, Fl 1R, ALPG, PRTN3, ALDH2, LILRB3, APOA5, VNN2, LAT, TCN1, MAT1A, OLR1, ALDH1A1, PZP, OTC, LEP, LEP, CNRIP1, PKLR, IL1RL1, BPI, PKLR, CST7, CKM; CKB, PROK2, and CRISPLD2. The amino acid sequences of these factors can be found in the Uniprot database, for example, and each factor’s Uniprot accession number is provided in Table 7. Further, methods, reagents, and assays for measuring expression levels of these factors are well known in the art and are commercially available.

[0133] In some embodiments, the factor is selected from a factor provided in Table 8. In some embodiments, the plurality of factors is selected from the factors provided in Table 8. In some embodiments, the plurality of factors comprises at least two factors selected from those provided in Table 8. In some embodiments, the plurality of factors consists of factors selected from Table 8. In some embodiments, the factors provided in Table 8 are: CRISPLD2, OLR1, PROK2, ALPG, VNN2, MORF4L1, CST7, HMGCS2, GRP, CETN2, OGGI, FSTL1, PKLR, CKAP2, CNRIP1, DNALI1, BPI, DTNA, DAP, ICAM1, RLN2, MSANTD2, IHH, RPN2, VEGFB, IFNW1, BGLAP, RDH16, OTC, TRAPPC3, PKLR, CEACAM8, CKM|CKB, TNFSF13B, ERP29, ZWINT, WAS, BMF, RNASE3, TYRP1, PRG3, NDC80, CAPS, UBE2D3, POMC, SEMA5B, PHEX, KRT19, TFPI, ADA2, PC, PBRM1, IGFBP4, TNFRSF12A, PRTN3, GNAQ, MXH, IGFBP1, ELANE, SAP30, HGF, SULF2, GGH, FAM118A, CRTAC1, MFAP4, MAT1A, IP6K2, EFCAB 1, DLL1, LAMTOR3, S100A11, ALDOC, FGFR2, CCN4, ERLEC1, HAX1, CKM, ST3GAL6, SERPINA10, IZUM04, CLUL1, CYP2C19, IL1RL1, KNG1, LIPN, HSD11B1, NOVAI, RCAN2, IL18R1, ADGRF5, GLIPR2, PSMD7, LEPR, AGRN, CTSV, SPARC, C4B|C4A, MASP1, DDC, DDHD2, MMP19, RET, SRA1, LECT2, RNF122, WWOX, CTSF, and INIP.

[0134] In some embodiments, the factor is selected from a factor provided in Table 9. In some embodiments, the plurality of factors is selected from the factors provided in Table 9. In some embodiments, the plurality of factors comprises at least two factors selected from those provided in Table 9. In some embodiments, the plurality of factors consists of factors selected from Table 9. In some embodiments, the factors provided in Table 9 are: POP7, FKBP1A, GNAQ, RGS7, MVK, MZB1, TRIM54, MCL1, HBQ1, RSPO1, CRYZL1, PAK5, CD300A, ADPRH, EFCAB14, AFP, NPFF, DCPS, LRP4, UBL3, PLXNC1, TNFSF15, PDE7A, RHOB, CADM1, PPP4R3A, ARHGEF10, APOD, UBE2D3, GOLM1, GGT2, AIF1L, FMOD, DLG3, ENTPD1, CDH15, MSANTD2, PTN, COPS2, TLL1, IL3RA, TP53I11, SRP14, ERCC1, PSMD7, SELPLG, RBM24, AKR1D1, TMEM221, NDRG4, FH, RGS21, UBB; UBE2L3, DDR1, CAMK1, ADGRF1, KLK13, OTUD5, GRP, PON2, BMX, CDA, MFAP3L, NOTCH2, IGFBP7, GTF2E2, ASAP2, NCR1, CEACAM8, SFRP1, IDS, HAT1, WWOX, NDST1, GBP1, LGALS3BP, KLK10, CXCL5, F2, SHANK1, EFEMP1, INHBB, C9, ICAM3, TPST1, RFX5, EPHB4, FDCSP, GULP1, FSTL1, KRT14, CDK5RAP3, LY86, SERPINB8, POMGNT2, CES3, ACTN4, GEMIN7, KCNRG, RNF122, BRK1, TACSTD2, PTGR2, PLXDC1, NEO1, TNFRSF11B, HAVCR2, PLCD1, MAK16, ITPKA, GREM2, BTN2A2, SFRP4, KIR2DL3, TTC33, CMA1, TP63, CRELD1, APOA2, F13B; F13A1, CD86, MET, BMP5, OBP2B, KISSI, ENGASE, TRADD, XCL2, INO80E, RBM18, SIRT2, PLA2G5, LRIG3, SIRPA, SIGLEC11, APOL1, APOA5, AUH, SPARC, DNAJB9, MAP3K10, TNFAIP6, TPPP3, CHRDL2, PLXND1, AOC3, HGFAC, NAAA, INSR, PZP, POLR2C, ARFGAP1, GPT, PSPH, CD22, STC1, VBP1, VEGFA, HPGDS, TRIM62, LILRB3, DDX58, LAMTOR3, NCR3, TPPP2, ADAM15, PRTG, TIMP1, PI3, CLCA1, IL22, NGFR, ACHE, NSF, EXOSC8, HECTD3, FGB, STAB2, FZD10, FGF8, TP53, FDPS, LAMC1; LAMB1; LAMA2, NT5E, IDO1, GOLM1, CD28, ATRN, CLUL1, ESMI, NOS3, CDNF, ACE, ESPN, B2M, APBB1IP, INHBB, SPOCK2, PLA2G2A, PRSS22, SIRPB 1, ADA2, CRH, PNLIP, FAM210A, ZFP36, CSDE1, COMTD1, NRG2, B3GALT2, DUSP26, APOL1, PIN1, HK2, C5orf38, PHYH, C8G; C8A; C8B, AURKB, THRA, MSR1, C4B; C4A, AMIG02, CAMKID, PON1, GID8, ETV2, DCBLD1, GLIPR2, SCLY, HSPA1A, CXCL12, KYNU, CD46, NTM, DCTPP1, RNASET2, B3GAT3, NGF, SEMA4A, S100A11, OAT, MMP13, PUF60, CXCL9, NRG3, THBS3, FCGR3B, S100A8, MTRF1L, SURF1, TMUB1, MMACHC, BLOC1S2, ARHGEF2, MEIG1, KRT19, MPST, SERPINA3, HPX, FMRI, GPR26, SNX7, PFKFB4, ZSWIM7, VLDLR, GK5, PRKCG, MST1R, FGF23, CDH1, DEFBI, CABP2, OTOR, PCOLCE, SERPINA11, SLPI, ADAM23, ST3GAL1, CCN1, TNFSF13B, NHLRC2, OMD,IDUA, ATF6, CETP, REG3A, MACR0D2, R0B02, CDH11, ITLN1, NTRK3, GUSB, VAC14, IGHA1; IGHA2, PDE4C, NTRK2, CRABP2, SFRP1, RAET1L, PKLR, LILRA2, NRG3, EPB41L1, ANKRD45, NOVAI, FOXO1, HAAG, ANKRD63, SPOCK3, PBRM1, TRAPPC3, APOF, RARRES1, IL18RAP, FCGR1A, S100A9, EYA2, PTPN9, CD247, HEY1, NCBP1, COTL1, MATN4, ECM1, RAP1GAP, TOM1L1, DHRS4, SLITRK1, SEPTIN6, NDC80, OTULIN, HPX, KIR2DS2, BACH1, APOCI, AKT2, SERPINC1, FAM171B, CMPK1, DSG1, DAG1, HSF1, CAPN3, PIK3AP1, CTRB2, GZMK, C8G, PCBD1, CCL7, LY6D, PCDH10, RSPO3, FCN1, CFD, GSKIP, MRPL58, C1QC, IGF1, CEMIP2, RAB38, B3GNT2, SLITRK5, CNTFR, A1BG, GSR, NTM, STX3, DLL4, ACP5, MGAT5, EGFLAM, MET, CD93, FLRT2, H6PD, CFHR2, CSTB, SCGB3A1, NTM, FLT4, CKM; CKB, NAGPA, BCAM, PCDH9, SERPINB2, KHSRP, PIGR, IL16, TFPI2, MRC2, COQ6, TREM2, SELPLG, OMD, LRRC15, SCARB 1, ICAM1, FUT5, IL12B; IL23A, FGF19, DYTN, ITGB1; ITGA4, LAMP1, MSMB, and CPA4. The amino acid sequences of these factors can be found in the Uniprot database, for example, and each factor’s Uniprot accession number is provided in Table 9. Further, methods, reagents, and assays for measuring expression levels of these factors are well known in the art and are commercially available.

[0135] In some embodiments, the factor is selected from a factor provided in Table 10. In some embodiments, the plurality of factors is selected from the factors provided in Table 10. In some embodiments, the plurality of factors comprises at least two factors selected from those provided in Table 10. In some embodiments, the plurality of factors consists of factors selected from Table 10. In some embodiments, the factors provided in Table 10 are:POP7, FKBP1A, GNAQ, RGS7, MVK, MZB1, TRIM54, MCL1, HBQ1, RSPO1, CD300A, PAK5, CRYZL1, ADPRH, AFP, EFCAB14, NPFF, DCPS, ERP4, UBE3, PEXNC1, PPP4R3A, TNFSF15, CADM1, PDE7A, RHOB, ARHGEF10, GOEM1, UBE2D3, APOD, AIF1E, GGT2, FMOD, DEG3, ENTPD1, MSANTD2, CDH15, PTN, IE3RA, TEE1, COPS2, TP53I11, SRP14, PSMD7, ERCC1, RBM24, SEEPEG, AKR1D1, NDRG4, TMEM221, UBB;UBE2L3, DDR1, RGS21, FH, CAMK1, ADGRF1, KLK13, PON2, OTUD5, GRP, BMX, CDA, MFAP3L, IGFBP7, NOTCH2, GTF2E2, ASAP2, SFRP1, IDS, NCR1, CEACAM8, KLK10, GBP1, LGALS3BP, WWOX, HAT1, NDST1, SHANK1, EFEMP1, F2, CXCL5, EPHB4, C9, ICAM3, FDCSP, RFX5, INHBB, TPST1, FSTL1, GULP1, KRT14, LY86, POMGNT2, CDK5RAP3, SERPINB8, TNFRSF11B,RNF122, PLXDC1, BRK1, KCNRG, NE01, TACSTD2, GEMIN7, PLCD1, HAVCR2, PTGR2, CES3, ACTN4, KIR2DL3, ITPKA, BTN2A2, SFRP4, MAK16, and GREM2.

[0136] In some embodiments, the factor is a factor that appeared in at least 10 iterations. In some embodiments, the factor that appeared in at least 10 iterations is a factor provided in Table 8. In some embodiments, the factor that appeared in at least 10 iterations is a factor provided in Table 10. In some embodiments, 10 iterations is 10 out of 80. In some embodiments, 80 iterations of the machine learning algorithm are run to determine the toxicity associated factors.

[0137] In some embodiments, the factors provided in Tables 7-8 are relevant for rash from immunotherapy. In some embodiments, the factors provided in Tables 7-8 are modulated in response to immunotherapy in subjects that develop rash. In some embodiments, rash is a rash adverse event. In some embodiments, the factors provided in Tables 9-10 are relevant for AE from chemotherapy (caAE). In some embodiments, the factors provided in Table 11 are relevant for irAEs from immunotherapy. In some embodiments, the factors are selected from GFBP1, MMP19, PRTN3, ADGRF5, GGH, DLL1, MFAP4, RLN2, ERP29, CRISPLD2, TFPI, IL1RL1, CST7, OLR1, CYP2C19, SAP30, LAMTOR3, CNRIP1, RCAN2, MAT1A, ELANE, LECT2, PKLR, PROK2, AGRN, VNN2, and GLIPR2 are relevant for irAEs from immunotherapy. In some embodiments, the factors provided in Table 11 are relevant for AEs from therapy. In some embodiments, the factors provided in Table 11 are relevant for AEs from anticancer therapy. In some embodiments, the factors are selected from C9, GGT2, GOLM1, EPHB4, CRYZL1, HAT1, SFRP1, TMEM221, KRT14, OTUD5, GTF2E2, SELPLG, TNFSF15, MZB1, and RBM24 and are relevant for AEs from therapy. In some embodiments, the factors are selected from PSMD7, MSANTD2, GNAQ, UBE2D3, FSTL1, WWOX, and RNF122 and are relevant for AEs from therapy. In some embodiments, the factors are selected from CAECAM8, and GRP and are relevant for AEs from therapy. In some embodiments, therapy is anticancer therapy.

[0138] In some embodiments, the AE / irAE / caAE negative population suffers from the disease. In some embodiments, the AE / irAE / caAE positive population suffers from the disease. In some embodiments, the disease is a proliferative disease. In some embodiments, the disease is cancer. In some embodiments, the AE / irAE / caAE negative subjects all have the same disease. In some embodiments, the AE / irAE / caAE negative population suffers from the disease. In some embodiments, the AE / irAE / caAE positive subjects all suffer from the same disease. In some embodiments, the AE / irAE / caAE positive population suffers fromthe disease. In some embodiments, the AE / irAE / caAE positive population and the AE / irAE / caAE negative population all suffer from the same disease. In some embodiments, the AE / irAE / caAE negative population and the subject suffer from the same disease. In some embodiments, the AE / irAE / caAE positive population and the subject suffer from the same disease. In some embodiments, the AE / irAE / caAE positive population, the AE / irAE / caAE negative population and the subject suffer from the same disease.

[0139] In some embodiments, the expression levels are from the subject before receiving the therapy. In some embodiments, the expression levels are from the subject before receiving the immunotherapy. In some embodiments, the expression levels are from the subject before receiving the chemotherapy. In some embodiments, the expression levels are determined for the subject before receiving the therapy. In some embodiments, the expression levels are determined for the subject before receiving the anticancer therapy. In some embodiments, the expression levels are determined for the subject before receiving the immunotherapy. In some embodiments, the expression levels are determined for the subject before receiving the chemotherapy. In some embodiments, the expression levels are determined for the subject before receiving the therapy. In some embodiments, the expression levels are determined for the subject before receiving the anticancer therapy. In some embodiments, the expression levels are from time TO. In some embodiments, time TO is before receiving the therapy. In some embodiments, the expression levels are baseline expression levels. In some embodiments, the baseline expression levels are before receiving the therapy. In some embodiments, the sample is provided by the subject before receiving the therapy. In some embodiments, the sample is provided by the subject before receiving the anticancer therapy. In some embodiments, the sample is provided by the subject before receiving the immunotherapy. In some embodiments, the sample is provided by the subject before receiving the chemotherapy. In some embodiments, the expression levels are from the subject before receiving a first treatment of the therapy. In some embodiments, the expression levels are from the subject before receiving a first treatment of the anticancer therapy. In some embodiments, the expression levels are from the subject before receiving a first treatment of the immunotherapy. In some embodiments, the expression levels are from the subject before receiving a first treatment of the chemotherapy. In some embodiments, the expression levels are from the subject before receiving a first treatment of the therapy. In some embodiments, the expression levels are from the subject before receiving a first treatment of the anticancer therapy. In some embodiments, the expression levels are fromthe subject before receiving the first cycle of the immunotherapy. In some embodiments, the expression levels are from the subject before receiving the first cycle of the chemotherapy. In some embodiments, the expression levels are from the subject before receiving the first cycle of the therapy. In some embodiments, the expression levels are from the subject before receiving the first cycle of the anticancer therapy. In some embodiments, a treatment is a dose. In some embodiments, a treatment is a regimen. In some embodiments, a treatment is a combination of dose and regimen.

[0140] In some embodiments, before is at least 1 hour, 2 hours, 3 hours, 6 hours, 8 hours, 12 hours, 1 day, 2 days, 3 days, 5 days, 1 week, 2 weeks, 3 weeks, 4 weeks, 1 month, 2 months, 3 months, 4 months, 5 months, or 6 months before the therapy or before administration of the therapy. Each possibility represents a separate embodiment of the invention. In some embodiments, before is at least 1 hour before. In some embodiments, before is just before the therapy or before administration of the therapy. In some embodiments, before is at most 1 hour, 2 hours, 3 hours, 4 hours, 6 hours, 9 hours, 12 hours, 18 hours, 24 hours, 2 days, 3 days, 5 days, 1 week, 2 weeks, 3 weeks, 4 weeks, 1 month, 2 months, 3 months, 4 months, 5 months, or 6 months before the therapy or before administration of the therapy. Each possibility represents a separate embodiment of the invention. In some embodiments, before is at most 24 hours before the therapy or before administration of the therapy. In some embodiments, administration of the therapy is the first administration of the therapy. In some embodiments, administration of the therapy is any administration of the therapy.

[0141] In some embodiments, the expression levels are from the subject after receiving the immunotherapy. In some embodiments, the expression levels are from the subject after receiving the chemotherapy. In some embodiments, the expression levels are from the subject after receiving the therapy. In some embodiments, the expression levels are from the subject after receiving the anticancer therapy. In some embodiments, the expression levels are from time Tl. In some embodiments, time T1 is time after receiving the therapy, anticancer therapy, immunotherapy or chemotherapy. In some embodiments, the sample is provided by the subject after receiving the immunotherapy. In some embodiments, the sample is provided by the subject after receiving the chemotherapy. In some embodiments, the sample is provided by the subject after receiving the therapy. In some embodiments, the sample is provided by the subject after receiving the anticancer therapy. In some embodiments, the expression levels are from the subject after receiving a first treatment ofthe immunotherapy. In some embodiments, the expression levels are from the subject after receiving a first treatment of the chemotherapy. In some embodiments, the expression levels are from the subject after receiving a first treatment of the therapy. In some embodiments, the expression levels are from the subject after receiving a first treatment of the anticancer therapy. In some embodiments, the expression levels are from the subject after receiving any treatment with the immunotherapy. In some embodiments, the expression levels are from the subject after receiving any treatment with the chemotherapy. In some embodiments, the expression levels are from the subject after receiving any treatment with the therapy. In some embodiments, the expression levels are from the subject after receiving any treatment with the anticancer therapy.

[0142] In some embodiments, after is at a time after initiation of the immunotherapy, or after administration of the immunotherapy, sufficient for altered expression of the at least one factor. In some embodiments, after is at a time after initiation of the chemotherapy, or after administration of the chemotherapy, sufficient for altered expression of the at least one factor. In some embodiments, after is at a time after initiation of the therapy, or after administration of the therapy, sufficient for altered expression of the at least one factor. In some embodiments, after is at a time after initiation of the anticancer therapy, or after administration of the anticancer therapy, sufficient for altered expression of the at least one factor. In some embodiments, after is at a time after initiation of the immunotherapy, or after administration of the first treatment of the immunotherapy. In some embodiments, after is at a time after initiation of the chemotherapy, or after administration of the first treatment of the chemotherapy. In some embodiments, after is at a time after initiation of the therapy, or after administration of the first treatment of the therapy. In some embodiments, after is at a time after initiation of the anticancer therapy, or after administration of the first treatment of the anticancer therapy. In some embodiments, after is at least 1 day, 2 days, 3 days, 4 days, 5 days, 6 days, 1 week, 2 weeks, 3 weeks, 4 weeks, 6 weeks, 1 month, 2 months, 3 months,4 months, 5 months, 6 months, or a year after. Each possibility represents a separate embodiment of the invention. In some embodiments, after is at least 24 hours after. In some embodiments, after is at least 2 weeks after. In some embodiments, after is at least 3 weeks after. In some embodiments, after is at least 6 weeks after. In some embodiments, after is at most 1 week, 2 weeks, 3 weeks, 4 weeks, 6 weeks, 1 month, 2 months, 3 months, 4 months,5 months, 6 months or a year after initiation of the immunotherapy, chemotherapy, therapy,or anticancer therapy, or after administration of them. Each possibility represents a separate embodiment of the invention.

[0143] In some embodiments, the receiving expression levels comprises receiving factor expression levels for a group of factors larger than the plurality of factors. In some embodiments, the received expression levels for the larger group are received for AE / irAE / caAE positive subjects and AE / irAE / caAE negative subjects. In some embodiments, a subgroup of proteins is selected from the group. In some embodiments, a subgroup is a subset. In some embodiments, the subgroup is designated the plurality of factors. In some embodiments, the method comprises designating. In some embodiments, the receiving further comprises for each factor of the group applying a machine learning algorithm. In some embodiments, the algorithm classifies factors as from AE / irAE / caAE negative subjects and AE / irAE / caAE positive subjects. In some embodiments, the algorithm outputs if a subject that provided the sample that had the measured factor expression level is AE / irAE / caAE negative or AE / irAE / caAE positive. In some embodiments, the receiving further comprises selecting a subgroup of factors for which the algorithm most evenly divides the subjects into AE / irAE / caAE negative subjects and AE / irAE / caAE positive subjects. In some embodiments, the subjects are all the subjects in the populations of AE / irAE / caAE negative subjects and AE / irAE / caAE positive subjects. In some embodiments, the factors are processed with an algorithm that most evenly divides all subjects, AE / irAE / caAE negative subjects and AE / irAE / caAE positive subjects, into groups of AE / irAE / caAE negative subjects and AE / irAE / caAE positive subjects (even if designations are incorrect) are selected as the subgroup. In some embodiments, the algorithm is trained on the received factor expression levels in AE / irAE / caAE negative subjects and AE / irAE / caAE positive subjects. In some embodiments, the algorithm is trained on a training set. In some embodiments, training is on expression levels and labels indicating if an expression level was from an AE / irAE / caAE negative subject or an AE / irAE / caAE positive subject. In some embodiments, training is on expression levels, clinical information and labels indicating if an expression level was from an AE / irAE / caAE negative subject or an AE / irAE / caAE positive subject.

[0144] In some embodiments, the receiving further comprises for each factor of the group determining the average difference between AE / irAE / caAE negative subjects and AE / irAE / caAE positive subjects. In some embodiments, the receiving further comprises for each factor of the group determining the statistical significance between the levels inAE / irAE / caAE negative subjects and AE / irAE / caAE positive subjects. In some embodiments, the statistical significance is between the averages. In some embodiments, the statistical significance is the p-value. In some embodiments, the receiving further comprises selecting a subgroup of factors with the greatest statistical significance. In some embodiments, a statistical test is applied to determine significance. In some embodiments, the test is a Kolmogorov-Smirnov test. In some embodiments, the subgroup comprises a predetermined number of factors with the greatest significance. In some embodiments, the predetermined number is about 50 factors. In some embodiments, the predetermined number is at least 50 factors. In some embodiments, the determining is whether the factor is highly expressed in AE / irAE / caAE positive subjects. In some embodiments, the determining is whether the factor is highly expressed in AE / irAE / caAE negative subjects. In some embodiments, an AE / irAE / caAE negative factor is a factor whose level is statistically significantly higher in AE / irAE / caAE negative subjects. In some embodiments, an AE / irAE / caAE positive factor is a factor whose level is statistically significantly higher in AE / irAE / caAE positive subjects. In some embodiments, the subgroup comprises the factors whose algorithm most evenly divides the subjects. In some embodiments, evenly divides is into AE / irAE / caAE positive and AE / irAE / caAE negative. In some embodiments, the subgroup is the top 50, 100, 150, 200, 250, 300, 350, 400, 450, 500, 600, 700, 750, 800, 900, 1000, 1100, 1200, 1300, 1400, 1500, 1600, 1700, 1800, 1900, 2000, 3000, 4000, or 5000 factors. Each possibility represents a separate embodiment of the invention. In some embodiments, the subgroup is the top 50. In some embodiments, the subgroup is the top 100. In some embodiments, the subgroup is the top 200. In some embodiments, the subgroup is the top 500.

[0145] In some embodiments, the method further comprises performing a dimensionality reduction step. In some embodiments, the reduction is with respect to the plurality of factors. In some embodiments, the reduction is reducing the number of factors in the plurality. In some embodiments, the dimensionality reduction step identifies a subgroup or a subset of factors. In some embodiments, factors are principal factors. In some embodiments, the training set comprises only the expression levels of the subset / subgroup of factors. In some embodiments, the subgroup or subset of factors are the factors that most evenly balance the predicted number of AE / irAE / caAE negative subjects and AE / irAE / caAE positive subjects. In some embodiments, predicted is predicted by the machine learning algorithm. In some embodiments, the machine learning algorithm is the trained machine learning algorithm. Insome embodiments, the machine learning algorithm is the machine learning algorithm during training.

[0146] In some embodiments, a preprocessing stage may take place to preprocess the received expression levels. In some embodiments, the preprocessing stage may comprise at least one of data cleaning and normalizing, feature selection, feature extraction, dimensionality reduction, and / or any other suitable preprocessing method or technique. Feature selection can be performed by statistical tests, such as the Kolmogorov Smirnov (KS) test, or any other test known in the art.

[0147] In some embodiments, factor selection and / or dimensionality reduction steps may be performed, to reduce the number of factors in each sample and / or to obtain a set of principal factors, e.g., those factors that may have significant predictive power. In some embodiments, factor selection is TAP selection. Accordingly, in some embodiments, a factor selection and / or dimensionality reduction step may result in a reduction of the number of factors in each sample and / or set of values. In some embodiments, dimensionality reduction selects principal factors, e.g., proteins, based on the level of AE / irAE / caAE predictive power a factor generates with respect to the desired prediction. In specific embodiments, the dimensionality reduction involves regarding all or some factors as vector components and calculating their norm.

[0148] In some embodiments, any suitable factor selection and / or dimensionality reduction method or technique may be employed, such as, but not limited to:• ANQVA with So parameter: Analysis of variance with an additional parameter (So) that controls for the relative importance of features based on resulted test p-values and difference between the group means (see, e.g., Tusher, Tibshirani and Chu, PNAS 98, pp5116-21, 2001).• Scalable Empirical Bayes Model Selection (SEMMS): An empirical Bayes feature selection method which applies a parsimonious mixture model to identify significant predictors (see, e.g., Bar, Booth, and Wells. A scalable empirical Bayes approach to variable selection in generalized linear models, 2019).• L2N: A method for differential expression analysis that uses a three-component mixture model. The model consists of two log-normal components (L2) for differentially expressed features, one component for under-expressed features and the other for overexpressed features, and a single normal component (N) for non-differentially expressed features (see,e.g., Bar and Schifano. Differential variation and expression analysis. Stat 8, e237, doi:10.1002 / sta4.237, 2019).• Genetic algorithms: A family of heuristic optimization algorithms that employ organic evolutionary techniques such as random mutations, recombination, and natural selection as methods for achieving optimal configurations (see, e.g., Popovic, Sifrim, Pavlopoulos, Moreau, and Bart De Moor. A Simple Genetic Algorithm for Biomarker Mining. 2012).• Naive classifier: The naive classifier evaluates a development of an AE / irAE score by reducing the dimension to a single score. This is performed by regarding all features (e.g., specific profiles such as protein expression levels) as component of a vector and calculating its norm. The dimension reduction reduces the possible risk of an over-fitting. In some embodiments, the vector components are normalized according to the typical component value among patients that belong to the same development of an AE / irAE group (e.g., responders), such that the normalized norm quantifies the amount of deviation from the typical respective class value. In additional embodiments, the naive classifier enables training using data of subjects that belong only to part of the development of an AE / irAE groups.

[0149] Kolmogorov Smirnov: The Kolmogorov-Smirnov (K-S) test is a nonparametric statistical test used to compare a sample with a reference probability distribution or to compare two samples to determine if they come from the same distribution. It evaluates the goodness of fit between distributions and is particularly useful for continuous, univariate data.

[0150] As used herein, the term “AE negative subject” or a subject “known to not develop AE in response to the therapy” are used interchangeably and refer to a subject that when administered a therapy does not develop an AE. In some embodiments, does not develop an AE is does not develop a severe AE. In some embodiments, does not develop an AE is does not develop an early onset AE. In some embodiments, does not develop an AE is does not develop a systemic AE. In some embodiments, does not develop an AE is does not develop a rash. In some embodiments, does not develop an AE is does not develop a specific AE from the list provided herein. In some embodiments, an AE negative subject is a subject that when administered a therapy does not develop an early onset AE. In some embodiments, an AE negative subject is a subject that when administered a therapy does not develop late onset AE. In some embodiments, an early onset AE is an AE that develops within 30, 50, 60, 75,90, 100, 120, 125, 150, 175, 180, 200, 225, 250, 275 or 300 days after receiving the therapy. Each possibility represents a separate embodiment of the invention. In some embodiments, an early onset AE is an AE that develops within 100 days after receiving the therapy. In some embodiments, an early onset AE is an AE that develops within 180 days after receiving the therapy. In some embodiments, a late onset AE is an AE that develops after the early onset time period. In some embodiments, a late onset AE is an AE that develops after 30, 50, 60, 75, 90, 100, 120, 125, 150, 175, 180, 200, 225, 250, 275 or 300 days after receiving the therapy. Each possibility represents a separate embodiment of the invention. In some embodiments, a late onset AE is an AE that develops after 100 days after receiving the therapy. In some embodiments, a late onset AE is an AE that develops after 180 days after receiving the therapy. In some embodiments, the AE is an early onset AE. In some embodiments, the AE is a later onset AE. In some embodiments, an AE negative subjects is a subject that when administered a therapy does not develop a severe AE. In some embodiments, a severe AE is an AE with grade of 3 or greater on the CTCAE clinical grading system. In some embodiments, a severe AE is an AE that results in treatment change. In some embodiments, treatment change is within 30, 50, 60, 75, 90, 100, 120, 125, 150, 175, 180, 200, 225, 250, 275 or 300 days of receiving said therapy. Each possibility represents a separate embodiment of the invention. In some embodiments, treatment change is within 100 days of receiving said therapy. In some embodiments, treatment change is within 180 days of receiving said therapy. In some embodiments, a severe AE is an AE that results in treatment termination. In some embodiments, treatment termination is within 30, 50, 60, 75, 90, 100, 120, 125, 150, 175, 180, 200, 225, 250, 275 or 300 days of receiving the therapy. Each possibility represents a separate embodiment of the invention. In some embodiments, treatment termination is within 100 days of receiving the therapy. In some embodiments, treatment termination is within 180 days of receiving the therapy. In some embodiments, a severe AE is an AE that results in treatment delay. In some embodiments, the treatment is the therapy. In some embodiments, receiving the therapy is starting the therapy. In some embodiments, a severe AE is an AE that causes hospitalization. In some embodiments, an AE is an early onset AE and a severe AE. In some embodiments, the AE is a systemic AE. In some embodiments, a systemic AE is an AE that affects more than one biological system of the subject. Examples of biological systems include gastroenterological, pulmonary, dermatological, neurological, hematological, cardiac, hepatic, renal, musculoskeletal, ear / nose / throat, endocrine, ocular and vascular. In some embodiments, a systemic AE effects at least two systems selected from gastroenterological, pulmonary, dermatological,neurological, hematological, cardiac, hepatic, renal, musculoskeletal, ear / nose / throat, endocrine, ocular and vascular. In some embodiments, the AE is a plurality of AEs. In some embodiments, an AE negative subject is a subject that does not develop at least 1 AE. In some embodiments, an AE negative subject is a subject that does not develop at least 2 AEs. In some embodiments, an AE negative subject is a subject that does not develop at least 3 AEs. In some embodiments, an AE negative subject is a subject that does not develop at least 4 AEs. In some embodiments, an AE negative subject is a subject that does not develop at least 5 AEs.

[0151] As used herein, the term “irAE negative subject” or a subject “known to not develop irAE in response to the immunotherapy” are used interchangeably and refer to a subject that when administered an immunotherapy does not develop an irAE. In some embodiments, an irAE negative subject is a subject that when administered an immunotherapy does not develop an early onset irAE. In some embodiments, an early onset irAE is an irAE that develops within 30, 50, 60, 75, 90, 100, 120, 125, 150, 175, 180, 200, 225, 250, 275 or 300 days after receiving the immunotherapy. Each possibility represents a separate embodiment of the invention. In some embodiments, an early onset irAE is an irAE that develops within 100 days after receiving the immunotherapy. In some embodiments, an early onset irAE is an irAE that develops within 180 days after receiving the immunotherapy. In some embodiments, an irAE negative subjects is a subject that when administered an immunotherapy does not develop a severe irAE. In some embodiments, a severe irAE is an irAE with grade of 3 or greater on the CTCAE clinical grading system. In some embodiments, a severe irAE is an irAE that results in treatment change. In some embodiments, treatment change is within 30, 50, 60, 75, 90, 100, 120, 125, 150, 175, 180, 200, 225, 250, 275 or 300 days of receiving said immunotherapy. Each possibility represents a separate embodiment of the invention. In some embodiments, treatment change is within 100 days of receiving said immunotherapy. In some embodiments, treatment change is within 180 days of receiving said immunotherapy. In some embodiments, a severe irAE is an irAE that results in treatment delay. In some embodiments, a severe irAE is an irAE that results in treatment termination. In some embodiments, treatment termination is within 30, 50, 60, 75, 90, 100, 120, 125, 150, 175, 180, 200, 225, 250, 275 or 300 days of receiving the immunotherapy. Each possibility represents a separate embodiment of the invention. In some embodiments, treatment termination is within 100 days of receiving the immunotherapy. In some embodiments, treatment termination is within 180 days of receiving theimmunotherapy. In some embodiments, the treatment is the immunotherapy. In some embodiments, receiving the immunotherapy is starting the immunotherapy. In some embodiments, a severe irAE is an irAE that causes hospitalization. In some embodiments, an irAE is an early onset irAE and a severe irAE. In some embodiments, the irAE is a systemic irAE. In some embodiments, a systemic irAE is an irAE that affects more than one biological system of the subject. In some embodiments, a systemic irAE affects at least two systems selected from gastroenterological, pulmonary, dermatological, neurological, hematological, cardiac, hepatic, renal, musculoskeletal, ear / nose / throat, endocrine, ocular and vascular. In some embodiments, the irAE is a plurality of irAEs. In some embodiments, an irAE negative subject is a subject that does not develop at least 1 irAE. In some embodiments, an irAE negative subject is a subject that does not develop at least 2 irAEs. In some embodiments, an irAE negative subject is a subject that does not develop at least 3 irAEs. In some embodiments, an irAE negative subject is a subject that does not develop at least 4 irAEs. In some embodiments, an irAE negative subject is a subject that does not develop at least 5 irAEs.

[0152] As used herein, the term “caAE negative subject” or a subject “known to not develop caAE in response to the chemotherapy” are used interchangeably and refer to a subject that when administered a chemotherapy does not develop a caAE. In some embodiments, a caAE negative subject is a subject that when administered a chemotherapy does not develop an early onset caAE. In some embodiments, an early onset caAE is a caAE that develops within 30, 50, 60, 75, 90, 100, 120, 125, 150, 175, 180, 200, 225, 250, 275 or 300 days after receiving the chemotherapy. Each possibility represents a separate embodiment of the invention. In some embodiments, an early onset caAE is a caAE that develops within 100 days after receiving the chemotherapy. In some embodiments, development is within 180 days after receiving the chemotherapy. In some embodiments, a caAE negative subjects is a subject that when administered a chemotherapy does not develop a severe caAE. In some embodiments, a severe caAE is a caAE with grade of 3 or greater on the CTCAE clinical grading system. In some embodiments, a severe caAE is an caAE that results in treatment change. In some embodiments, treatment change is within 30, 50, 60, 75, 90, 100, 120, 125, 150, 175, 180, 200, 225, 250, 275 or 300 days of receiving said chemotherapy. Each possibility represents a separate embodiment of the invention. In some embodiments, treatment change is within 100 days of receiving said chemotherapy. In some embodiments, treatment change is within 180 days of receiving said chemotherapy. In some embodiments,a severe caAE is a caAE that results in treatment delay. In some embodiments, a severe caAE is a caAE that results in treatment termination. In some embodiments, treatment termination is within 30, 50, 60, 75, 90, 100, 120, 125, 150, 175, 180, 200, 225, 250, 275 or 300 days of receiving the chemotherapy. Each possibility represents a separate embodiment of the invention. In some embodiments, treatment termination is within 100 days of receiving the chemotherapy. In some embodiments, treatment termination is within 180 days of receiving the chemotherapy. In some embodiments, the treatment is the chemotherapy. In some embodiments, receiving the chemotherapy is starting the chemotherapy. In some embodiments, a severe caAE is a caAE that causes hospitalization. In some embodiments, a caAE is an early onset caAE and a severe caAE. In some embodiments, the caAE is a systemic caAE. In some embodiments, a systemic caAE is a caAE that affects more than one biological system of the subject. In some embodiments, a systemic caAE affects at least two systems selected from gastroenterological, pulmonary, dermatological, neurological, hematological, cardiac, hepatic, renal, musculoskeletal, ear / nose / throat, endocrine, ocular and vascular. In some embodiments, the caAE is a plurality of caAEs. In some embodiments, a caAE negative subject is a subject that does not develop at least 1 caAE. In some embodiments, a caAE negative subject is a subject that does not develop at least 2 caAEs. In some embodiments, a aAE negative subject is a subject that does not develop at least 3 caAEs. In some embodiments, a caAE negative subject is a subject that does not develop at least 4 caAEs. In some embodiments, a caAE negative subject is a subject that does not develop at least 5 caAEs.

[0153] As used herein, the term “AE positive subject” or a subject “known to develop AE in response to the therapy” are used interchangeably and refer to a subject that when administered a therapy develops an AE. In some embodiments, an AE positive subject is a subject that when administered a therapy develops an early onset AE. In some embodiments, an early onset AE is an AE that develops within 30, 50, 60, 75, 90, 100, 120, 125, 150, 175, 180, 200, 225, 250, 275 or 300 days after receiving the therapy. Each possibility represents a separate embodiment of the invention. In some embodiments, an early onset AE is an AE that develops within 100 days after receiving the therapy. In some embodiments, an early onset AE is an AE that develops within 180 days after receiving the therapy. In some embodiments, an AE positive subject is a subject that when administered a therapy develops a severe AE. In some embodiments, a severe AE is an AE with a grade of 3 or greater on the CTCAE clinical grading system. In some embodiments, a severe AE is an AE that results intreatment change. In some embodiments, treatment change is within 30, 50, 60, 75, 90, 100, 120, 125, 150, 175, 180, 200, 225, 250, 275 or 300 days of receiving said therapy. Each possibility represents a separate embodiment of the invention. In some embodiments, treatment change is within 100 days of receiving said therapy. In some embodiments, treatment change is within 180 days of receiving said therapy. In some embodiments, a severe AE is an AE that results in treatment delay. In some embodiments, a severe AE is an AE that results in treatment termination. In some embodiments, treatment termination is within 30, 50, 60, 75, 90, 100, 120, 125, 150, 175, 180, 200, 225, 250, 275 or 300 days of receiving the therapy. Each possibility represents a separate embodiment of the invention. In some embodiments, treatment termination is within 100 days of receiving the therapy. In some embodiments, treatment termination is within 180 days of receiving the therapy. In some embodiments, the treatment is the therapy. In some embodiments, receiving the therapy is starting the therapy. In some embodiments, a severe AE is an AE that causes hospitalization. In some embodiments, the AE is a systemic AE. In some embodiments, a systemic AE is an AE that affects more than one biological system of the subject. In some embodiments, a systemic AE affects at least two systems selected from gastroenterological, pulmonary, dermatological, neurological, hematological, cardiac, hepatic, renal, musculoskeletal, ear / nose / throat, endocrine, ocular, and vascular. In some embodiments, the AE is a plurality of AEs. In some embodiments, an AE positive subject is a subject that does develop at least 1 AE. In some embodiments, an AE positive subject is a subject that does develop at least 2 AEs. In some embodiments, an AE positive subject is a subject that does develop at least 3 AEs. In some embodiments, an AE positive subject is a subject that does develop at least 4 AEs. In some embodiments, an AE positive subject is a subject that does develop at least 5 AEs.

[0154] As used herein, the term “irAE positive subject” or a subject “known to develop irAE in response to the immunotherapy” are used interchangeably and refer to a subject that when administered an immunotherapy develops an irAE. In some embodiments, an irAE positive subject is a subject that when administered an immunotherapy develops an early onset irAE. In some embodiments, an early onset irAE is an irAE that develops within 30, 50, 60, 75, 90, 100, 120, 125, 150, 175, 180, 200, 225, 250, 275 or 300 days after receiving the immunotherapy. Each possibility represents a separate embodiment of the invention. In some embodiments, an early onset irAE is an irAE that develops within 100 days after receiving the immunotherapy. In some embodiments, an early onset irAE is an irAE that developswithin 180 days after receiving the immunotherapy. In some embodiments, an irAE positive subjects is a subject that when administered an immunotherapy develops a severe irAE. In some embodiments, a severe irAE is an irAE with grade of 3 or greater on the CTCAE clinical grading system. In some embodiments, a severe irAE is an irAE that results in treatment change. In some embodiments, treatment change is within 30, 50, 60, 75, 90, 100, 120, 125, 150, 175, 180, 200, 225, 250, 275 or 300 days of receiving said immunotherapy. Each possibility represents a separate embodiment of the invention. In some embodiments, treatment change is within 100 days of receiving said immunotherapy. In some embodiments, treatment change is within 180 days of receiving said immunotherapy. In some embodiments, a severe irAE is an irAE that results in treatment delay. In some embodiments, a severe irAE is an irAE that results in treatment termination. In some embodiments, treatment termination is within 30, 50, 60, 75, 90, 100, 120, 125, 150, 175, 180, 200, 225, 250, 275 or 300 days of receiving the immunotherapy. Each possibility represents a separate embodiment of the invention. In some embodiments, treatment termination is within 100 days of receiving the immunotherapy. In some embodiments, treatment termination is within 180 days of receiving the immunotherapy. In some embodiments, the treatment is the immunotherapy. In some embodiments, receiving the immunotherapy is starting the immunotherapy. In some embodiments, a severe irAE is an irAE that causes hospitalization. In some embodiments, the irAE is a systemic irAE. In some embodiments, a systemic irAE is an irAE that effects more than one biological system of the subject. In some embodiments, a systemic irAE effects at least two systems selected from gastroenterological, pulmonary, dermatological, neurological, hematological, cardiac, hepatic, renal, musculoskeletal, ear / nose / throat, endocrine, ocular and vascular. In some embodiments, the irAE is a plurality of irAEs. In some embodiments, an irAE positive subject is a subject that does develop at least 1 irAE. In some embodiments, an irAE positive subject is a subject that does develop at least 2 irAEs. In some embodiments, an irAE positive subject is a subject that does develop at least 3 irAEs. In some embodiments, an irAE positive subject is a subject that does develop at least 4 irAEs. In some embodiments, an irAE positive subject is a subject that does develop at least 5 irAEs.

[0155] As used herein, the term “caAE positive subject” or a subject “known to develop caAE in response to the chemotherapy” are used interchangeably and refer to a subject that when administered a chemotherapy develops a caAE. In some embodiments, a caAE positive subject is a subject that when administered a chemotherapy develops an early onset caAE.In some embodiments, an early onset caAE is a caAE that develops within 30, 50, 60, 75, 90, 100, 120, 125, 150, 175, 180, 200, 225, 250, 275 or 300 days after receiving the chemotherapy. Each possibility represents a separate embodiment of the invention. In some embodiments, an early onset caAE is a caAE that develops within 100 days after receiving the chemotherapy. In some embodiments, an early onset caAE is a caAE that develops within 180 days after receiving the chemotherapy. In some embodiments, a caAE positive subject is a subject that when administered a chemotherapy develops a severe caAE. In some embodiments, a severe caAE is a caAE with grade of 3 or greater on the CTCAE clinical grading system. In some embodiments, a severe caAE is a caAE that results in treatment change. In some embodiments, treatment change is within 30, 50, 60, 75, 90, 100, 120, 125, 150, 175, 180, 200, 225, 250, 275 or 300 days of receiving said chemotherapy. Each possibility represents a separate embodiment of the invention. In some embodiments, treatment change is within 100 days of receiving said chemotherapy. In some embodiments, treatment change is within 180 days of receiving said chemotherapy. In some embodiments, a severe caAE is a caAE that results in treatment delay. In some embodiments, a severe caAE is a caAE that results in treatment termination. In some embodiments, treatment termination is within 30, 50, 60, 75, 90, 100, 120, 125, 150, 175, 180, 200, 225, 250, 275 or 300 days of receiving the chemotherapy. Each possibility represents a separate embodiment of the invention. In some embodiments, treatment termination is within 100 days of receiving the chemotherapy. In some embodiments, treatment termination is within 180 days of receiving the chemotherapy. In some embodiments, the treatment is the chemotherapy. In some embodiments, receiving the chemotherapy is starting the chemotherapy. In some embodiments, a severe caAE is a caAE that causes hospitalization. In some embodiments, the caAE is a systemic caAE. In some embodiments, a systemic caAE is a caAE that affects more than one biological system of the subject. In some embodiments, a systemic caAE affects at least two systems selected from gastroenterological, pulmonary, dermatological, neurological, hematological, cardiac, hepatic, renal, musculoskeletal, ear / nose / throat, endocrine, ocular and vascular. In some embodiments, the caAE is a plurality of caAEs. In some embodiments, a caAE positive subject is a subject that does develop at least 1 caAE. In some embodiments, a caAE positive subject is a subject that does develop at least 2 caAEs. In some embodiments, a caAE positive subject is a subject that does develop at least 3 caAEs. In some embodiments, a caAE positive subject is a subject that does develop at least 4 caAEs. In some embodiments, a caAE positive subject is a subject that does develop at least 5 caAEs.

[0156] In some embodiments, a toxicity score is a TAP score. In some embodiments, a toxicity score is an AE / irAE / caAE probability score. In some embodiments, toxicity score is total toxicity score. In some embodiments, AE / irAE / caAE probability score is a total AE / irAE / caAE probability score. In some embodiments, a TAP score is a total TAP score. In some embodiments, the toxicity score is based on similarity of the factor expression level in the subject to the factor expression level in the AE / irAE / caAE positive subjects. In some embodiments, the toxicity score is based on similarity of the factor expression level in the subject to the factor expression level in the AE / irAE / caAE negative subjects. In some embodiments, based on is calculated based on. In some embodiments, similarity is lack of similarity. In some embodiments, similarity to AE / irAE / caAE positive subjects is lack of similarity to AE / irAE / caAE negative subjects. In some embodiments, similarity to AE / irAE / caAE negative subjects is lack of similarity to AE / irAE / caAE positive subjects. In some embodiments, similarity is measured on a scale. In some embodiments, the toxicity score is produced by a trained machine learning model. In some embodiments, a factor more highly expressed in AE / irAE / caAE positive subjects is an AE / irAE / caAE positive TAP. In some embodiments, a factor more highly expressed in AE / irAE / caAE negative subject is an AE / irAE / caAE negative TAP. In some embodiments, more highly expressed is with higher expression. In some embodiments, the scale is from 0 to 1, wherein 1 is perfectly similar to AE / irAE / caAE positive subjects and 0 is perfectly similar to AE / irAE / caAE negative subjects. In some embodiments, the toxicity score is from 0 to 1, wherein 1 is perfectly similar to AE / irAE / caAE positive subjects and 0 is perfectly similar to AE / irAE / caAE negative subjects. In some embodiments, the toxicity score is based on similarity of the factor expression level in the subject to the factor expression level in the AE / irAE / caAE negative subjects and the factor expression level in the AE / irAE positive subjects. In some embodiments, the scale is from 0 to 10, wherein 10 is perfectly similar to AE / irAE / caAE positive subjects and 0 is perfectly similar to AE / irAE / caAE negative subjects. In some embodiments, the toxicity score is from 0 to 10, wherein 10 is perfectly similar to AE / irAE / caAE positive subjects and 0 is perfectly similar to AE / irAE / caAE negative subjects. In some embodiments, the scale is from 0% to 100%, wherein 100% is perfectly similar to AE / irAE / caAE positive subjects and 0% is perfectly similar to AE / irAE / caAE negative subjects. In some embodiments, the toxicity score is from 0% to 100%, wherein 100% is perfectly similar to AE / irAE / caAE positive subjects and 0% is perfectly similar to AE / irAE / caAE negative subjects. In some embodiments, the toxicity score is based on similarity of the factor expression level in the subject to the factor expression level in theAE / irAE / caAE negative subjects and the factor expression level in the AE / irAE / caAE positive subjects. In some embodiments, an AE / irAE / caAE positive TAP is a factor with expression level similar to AE / irAE / caAE positive subjects. In some embodiments, an AE / irAE / caAE negative TAP is a factor with expression level similar to AE / irAE / caAE negative subjects.

[0157] In some embodiments, the method further comprises administering the therapy to a subject predicted to not develop an AE. In some embodiments, the method further comprises continuing to administering the therapy to a subject predicted to not develop an AE. In some embodiments, the method further comprises not administering the therapy to a subject predicted to develop an AE. In some embodiments, the method further comprises discontinuing the therapy to a subject predicted to develop an AE.

[0158] In some embodiments, the method further comprises administering the immunotherapy to a subject predicted to not develop an irAE. In some embodiments, the method further comprises continuing to administering the immunotherapy to a subject predicted to not develop an irAE. In some embodiments, the method further comprises not administering the immunotherapy to a subject predicted to develop an irAE. In some embodiments, the method further comprises discontinuing the immunotherapy to a subject predicted to develop an irAE.

[0159] In some embodiments, the method further comprises administering the chemotherapy to a subject predicted to not develop a caAE. In some embodiments, the method further comprises continuing to administer the chemotherapy to a subject predicted to not develop a caAE. In some embodiments, the method further comprises not administering the chemotherapy to a subject predicted to develop a caAE. In some embodiments, the method further comprises discontinuing the chemotherapy to a subject predicted to develop a caAE.

[0160] In some embodiments, the method further comprises reducing the dose of therapy administered to the subject predicted to develop an AE. In some embodiments, the method further comprises stopping administering the therapy to the subject predicted to develop an AE. In some embodiments, the method further comprises delaying the administering of the therapy to the subject predicted to develop an AE. In some embodiments, the method further comprises administering an alternative therapy to the subject predicted to develop an AE. In some embodiments, the alternative therapy is an additional therapy. In some embodiments,the additional therapy is immunotherapy. In some embodiments, the additional therapy is chemotherapy. In some embodiments, the additional therapy is targeted therapy. In some embodiments, the additional therapy is a different therapy. In some embodiments, a different therapy is a second therapy. In some embodiments, the additional therapy is a nonchemotherapy anticancer therapy. In some embodiments, the alternative therapy comprises an alternative chemotherapy. In some embodiments, an alternative therapy is a therapy that is not the chemotherapy. In some embodiments, an alternative therapy is an alternative combination of chemotherapies. In some embodiments, the alternative therapy is any anticancer therapy that is not the chemotherapy. In some embodiments, the alternative therapy is a non-chemotherapy anti-cancer therapy. In some embodiments, the nonchemotherapy anti-cancer therapy is selected from radiation, immunotherapy, and targeted therapy. In some embodiments, the non-chemotherapy anti-cancer therapy comprises radiation. In some embodiments, the non-chemotherapy anti-cancer therapy comprises immunotherapy. In some embodiments, the non-chemotherapy anti-cancer therapy comprises targeted therapy. In some embodiments, targeted therapy is targeted anticancer therapy.

[0161] In some embodiments, the method further comprises reducing the dose of immunotherapy administered to the subject predicted to develop an irAE. In some embodiments, the method further comprises administering an alternative therapy to the subject predicted to develop an irAE. In some embodiments, the alternative therapy is an additional therapy. In some embodiments, the additional therapy is chemotherapy. In some embodiments, the additional therapy is an immunosuppressant. In some embodiments, the additional therapy is a different immunotherapy. In some embodiments, a different immunotherapy is a second immunotherapy. In some embodiments, the additional therapy is a non-immunotherapy anticancer therapy. In some embodiments, the alternative therapy comprises the immunotherapy in combination with an immunosuppressant. In some embodiments, the alternative therapy comprises an alternative immunotherapy. In some embodiments, an alternative immunotherapy is an immunotherapy that is not the immunotherapy. For example, if the subject was predicted to develop an irAE in response to an anti-PD-1 immunotherapy then an anti-CTLA4 or anti-LAG-3 immunotherapy (or vice versa) could be used instead. In some embodiments, an alternative therapy is an alternative combination of immunotherapies. For example, if a subject was predicted to develop an irAE in response to anti-PDl+anti-CTLA4 then an anti-PDl+anti-LAG-3 could be used instead.In some embodiments, the alternative therapy is any anticancer therapy that is not the immunotherapy. In some embodiments, the alternative therapy is a non-immunotherapy anticancer therapy. In some embodiments, the non-immunotherapy anti-cancer therapy is selected from radiation, chemotherapy, and targeted therapy. In some embodiments, the nonimmunotherapy anti-cancer therapy comprises radiation. In some embodiments, the nonimmunotherapy anti-cancer therapy comprises chemotherapy. In some embodiments, the non-immunotherapy anti-cancer therapy comprises targeted therapy. In some embodiments, targeted therapy is targeted anticancer therapy.

[0162] In some embodiments, the method further comprises administering the therapy or continuing to administer the therapy in combination with an agent or therapy that blocks or inhibits at least one of the toxicity-associated factors in the subject predicted to develop an AE. In some embodiments, the method further comprises administering the immunotherapy or continuing to administer the immunotherapy in combination with an agent or therapy that blocks or inhibits at least one of the toxicity-associated factors in the subject predicted to develop an irAE. In some embodiments, the method further comprises administering the chemotherapy or continuing to administer the chemotherapy in combination with an agent or therapy that blocks or inhibits at least one of the toxicity-associated factors in the subject predicted to develop a caAE. In some embodiments, an agent or therapy that blocks or inhibits at least one of the toxicity-associated factors is an additional therapy. In some embodiments, an agent or therapy that blocks or inhibits the signaling pathway of at least one of the toxicity-associated factors is an additional therapy.

[0163] In some embodiments, the method further comprises administering to the subject (e.g., an AE / irAE / caAE positive subject) an agent that modulates the at least one factor. In some embodiments, modulates comprises inhibits, blocks and regulates. In some embodiments, modulates is inhibits. In some embodiments, the method further comprises administering to the subject (e.g., an AE / irAE / caAE positive subject) an agent that modulates a pathway that comprises the at least one factor. In some embodiments, modulating the at least one factor is modulating a pathway comprising the at least one factor. In some embodiments, modulating a pathway comprising modulating a driver protein / gene that controls the at least one factor. In some embodiments, modulating a pathway comprising modulating a driver protein / gene that controls the pathway. In some embodiments, modulating a pathway comprising the at least one factor is modulating a receptor of the factor (e.g., using a receptor agonist or antagonists), a ligand or the factor, a paralog of the factor,or a combination thereof. In some embodiments, the modulating is modulating a plurality of factors. In some embodiments, the modulating is modulating a plurality of factors in the subset of factors. In some embodiments, the modulation is modulating each factor in the subset of factors. In some embodiments, the modulation achieves reduced development of an AE / irAE in response to the immunotherapy. In some embodiments, the modulation achieves a reduced chance or risk of development of an AE / irAE in response to the immunotherapy. In some embodiments, the modulation achieves reduced development of an irAE in response to the immunotherapy. In some embodiments, the modulation achieves a reduced chance or risk of development of an irAE in response to the immunotherapy. In some embodiments, the modulation achieves reduced development of an AE in response to the therapy. In some embodiments, the modulation achieves a reduced chance or risk of development of an AE in response to the therapy. In some embodiments, the modulation achieves reduced development of a caAE in response to the chemotherapy. In some embodiments, the modulation achieves a reduced chance or risk of development of a caAE in response to the chemotherapy. In some embodiments the factor is a toxicity-associated factor.

[0164] In some embodiments, an AE / irAE / caAE probability score is a TAP score. In some embodiments, an AE / irAE probability score is an AE / irAE / caAE score. In some embodiments, an AE / irAE / caAE probability score is a no AE / irAE / caAE probability score. In some embodiments, a no AE / irAE / caAE probability score is 1- an AE / irAE / caAE probability score. In some embodiments, a no AE / irAE / caAE probability score is 10- an AE / irAE / caAE probability score. In some embodiments, a no AE / irAE / caAE probability score is 100%- an AE / irAE / caAE probability score. It will be understood by a skilled artisan that the no AE / irAE / caAE probability score and AE / irAE / caAE probability score are inverses. Thus, if the scale of the scores is 0-1 then the conversion of one score to the other is 1-score. Whereas if the scale of the scores is 0-10 then the conversion of one score to the other is 10-score, and if the scale of the scores is 0-100% then the conversion of one score to the other is 100%-score. The same can be used for any scale being used for the two scores. In some embodiments, AE / irAE / caAE probability score is total AE / irAE / caAE probability score. In some embodiments, a TAP score is a total TAP score. In some embodiments, the AE / irAE / caAE probability score is based on similarity of the factor expression level in the subject to the factor expression level in the AE / irAE / caAE positive population. In some embodiments, the AE / irAE / caAE probability score is based on similarity of the factorexpression level in the subject to the factor expression level in AE / irAE / caAE negative population. In some embodiments, based on is calculated based on. In some embodiments, similarity is lack of similarity. In some embodiments, similarity to AE / irAE / caAE negative subjects is lack of similarity to AE / irAE / caAE positive subjects. In some embodiments, similarity to AE / irAE / caAE positive subjects is lack of similarity to AE / irAE / caAE negative subjects. In some embodiments, similarity is measured on a scale.

[0165] In some embodiments, the scale is from 0 to 1, wherein 1 is perfectly similar to AE / irAE / caAE positive subjects and 0 is perfectly similar to AE / irAE / caAE negative subjects. In some embodiments, the AE / irAE / caAE probability score is from 0 to 1, wherein 1 is perfectly similar to AE / irAE / caAE positive subjects and 0 is perfectly similar to AE / irAE / caAE negative subjects. In some embodiments, the AE / irAE / caAE probability score is from 0% to 100%, wherein 100% is perfectly similar to AE / irAE / caAE positive subjects and 0% is perfectly similar to AE / irAE / caAE negative subjects. In some embodiments, the AE / irAE / caAE probability score is based on similarity of the factor expression level in the subject to the factor expression level in the AE / irAE / caAE positive subjects and the factor expression level in the AE / irAE / caAE negative subjects. In some embodiments, a development of an AE / irAE / caAE score from 0.5 to 1 indicates the subject is an AE / irAE / caAE positive subject. In some embodiments, the threshold is 0.5. In some embodiments, the threshold is 50%. In some embodiments, a development of an AE / irAE / caAE score above 0.5 indicates the subject is an AE / irAE / caAE positive subject. In some embodiments, a development of an AE / irAE / caAE score from 0.5 to 0 indicates the subject is an AE / irAE / caAE negative subject. In some embodiments, a development of an AE / irAE / caAE score below 0.5 indicates the subject is an AE / irAE / caAE negative subject. In some embodiments, a development of an AE / irAE / caAE score from 0.26 to 1 indicates the subject is an AE / irAE / caAE positive subject. In some embodiments, the threshold is 0.26. In some embodiments, a development of an AE / irAE / caAE score above 0.26 indicates the subject is an AE / irAE / caAE positive subject. In some embodiments, a development of an AE / irAE / caAE score from 0.26 to 0 indicates the subject is an AE / irAE / caAE negative subject. In some embodiments, a development of an AE / irAE / caAE score below 0.26 indicates the subject is an AE / irAE / caAE negative subject.

[0166] In some embodiments, the scale is from 0 to 10, wherein 10 is perfectly similar to AE / irAE / caAE positive subjects and 0 is perfectly similar to AE / irAE / caAE negative subjects. In some embodiments, the AE / irAE / caAE probability score is based on similarityof the factor expression level in the subject to the factor expression level in the AE / irAE / caAE positive subjects and the factor expression level in the AE / irAE / caAE negative subjects. In some embodiments, a development of an AE / irAE / caAE score from 5 to 10 indicates the subject is an AE / irAE / caAE positive subject. In some embodiments, a development of an AE / irAE / caAE score above 5 indicates the subject is an AE / irAE / caAE positive subject. In some embodiments, a development of an AE / irAE / caAE score from 5 to 0 indicates the subject is an AE / irAE / caAE negative subject. In some embodiments, a development of an AE / irAE / caAE score below 5 indicates the subject is an AE / irAE / caAE negative subject.

[0167] In some embodiments, the scale is from 0% to 100%, wherein 100% is perfectly similar to AE / irAE / caAE positive subjects and 0% is perfectly similar to AE / irAE / caAE negative subjects. In some embodiments, the AE / irAE / caAE probability score is based on similarity of the factor expression level in the subject to the factor expression level in the AE / irAE / caAE positive subjects and the factor expression level in the AE / irAE / caAE negative subjects. In some embodiments, a development of an AE / irAE / caAE score from 50% to 100% indicates the subject is an AE / irAE / caAE positive subject. In some embodiments, a development of an AE / irAE / caAE score above 50% indicates the subject is an AE / irAE / caAE positive subject. In some embodiments, a development of an AE / irAE / caAE score from 50% to 0% indicates the subject is an AE / irAE / caAE negative subject. In some embodiments, a development of an AE / irAE / caAE score below 50% indicates the subject is an AE / irAE / caAE negative subject. In some embodiments, the threshold is 50%. In some embodiments, a development of an AE / irAE / caAE score from 26% to 100% indicates the subject is an AE / irAE / caAE positive subject. In some embodiments, a development of an AE / irAE / caAE score above 26% indicates the subject is an AE / irAE / caAE positive subject. In some embodiments, a development of an AE / irAE / caAE score from 26% to 0% indicates the subject is an AE / irAE / caAE negative subject. In some embodiments, a development of an AE / irAE / caAE score below 26% indicates the subject is an AE / irAE / caAE negative subject. In some embodiments, the threshold is 26%.

[0168] In some embodiments, the method comprises before step (b) selecting a subset of factors. In some embodiments, the subset is a subset of the plurality of factors. In some embodiments, before step (b) is before the calculating. In some embodiments, the subset is a subset of the plurality of factors. In some embodiments, the subset comprises the factorsthat best differentiate between the AE / irAE / caAE negative subjects and AE / irAE / caAE positive subjects. In some embodiments, the factors that best differentiate are the top percentage. In some embodiments, the top percentage is the top 1, 3, 5, 10, 15, 20, 25, 30, 35, 40, 45 or 50% of factors. Each possibility represents a separate embodiment of the invention. In some embodiments, the top percentage is the top 20%. In some embodiments, the top factors are the top 10, 20, 25, 30, 40, 50, 60, 70, 75, 80, 90 or 100 factors. Each possibility represents a separate embodiment of the invention. In some embodiments, the top factors are the top 50 factors. In some embodiments, selection comprises applying a Kolmogorov-Smirnov test. In some embodiments, the Kolmogorov-Smirnov test is applied to the received factor expression levels. In some embodiments, the Kolmogorov-Smirnov test determines the top percentage. In some embodiments, statistical significance in the Kolmogorov-Smirnov test determines the top percentage. In some embodiments, the Kolmogorov-Smirnov test determines how well a factor differentiates between AE / irAE / caAE negative subjects and AE / irAE / caAE positive subjects. In some embodiments, the Kolmogorov-Smirnov test outputs a measure of how well a factor differentiates and the best factors are the factors with the highest scores. In some embodiments, selection comprises applying an XGBoost algorithm. In some embodiments, the subset comprises the factors that repeat in iterations of the ML model. In some embodiments, repeating is at least twice. In some embodiments, repeating is at least 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 25, 30, 40, 50, 60, 70, 75, 80, 90 or 100 times. Each possibility represents a separate embodiment of the invention. In some embodiments, repeating is at least 10 times. In some embodiments, repeating is at least 80 times. In some embodiments, repeating is repeating in at least 10 iterations of the ML model. In some embodiments, repeating is repeating in at least 80 iterations of the ML model. In some embodiments, each iteration comprises selecting a training group and a testing group. In some embodiments, the training group is 75% of the subjects and the testing group is 25% of the subjects. In some embodiments, each iteration comprises determining TAPs for the training group. In some embodiments, each iteration comprises applying the determined TAPs to test predicted response on the test group. In some embodiments, the calculating is for the subset. In some embodiments, the calculating is for each factor of the subset. In some embodiments, each iteration comprises identifying using the Kolmogorov-Smirnov test proteins displaying differential levels between no AE / irAE / caAE (AE / irAE / caAE negative) and AE / irAE / caAE (AE / irAE / caAE positive) patients. In some embodiments, the protein identified from at least one iteration are TAPs. In some embodiments, the protein identifiedin at least ten iterations are TAPs. In some embodiments, the protein identified in at least twenty iterations are TAPs. In some embodiments, the protein identified in all iterations are TAPs.

[0169] In some embodiments, calculating comprises applying a machine learning algorithm. In some embodiments, calculating comprises applying a machine learning model. In some embodiments, the machine learning model is a machine learning algorithm. In some embodiments, the machine learning model implements a machine learning algorithm. In some embodiments, the algorithm is a classifier. In some embodiments, the algorithm is a regression model. In some embodiments, the algorithm is supervised. In some embodiments, the algorithm is unsupervised. In some embodiments, the machine learning algorithm is trained on the expression levels in AE / irAE / caAE negative subjects. In some embodiments, AE / irAE / caAE negative subjects are the AE / irAE / caAE negative population. In some embodiments, the machine learning algorithm is trained on the expression levels in AE / irAE / caAE positive subjects. In some embodiments, AE / irAE / caAE positive subjects are the AE / irAE positive population. In some embodiments, the machine learning algorithm is trained on the expression levels in AE / irAE / caAE negative subjects and AE / irAE / caAE positive subjects. In some embodiments, the machine learning algorithm is trained on a training set. In some embodiments, the training comprises cross-validation. In some embodiments, the cross-validation is in the training set. In some embodiments, the machine learning algorithm is trained by a method of the invention. In some embodiments, a machine learning algorithm is applied to factors of the plurality of factors. In some embodiments, a machine learning algorithm is applied to each factor of the plurality of factors. In some embodiments, a machine learning algorithm is applied to the subset. In some embodiments, a machine learning algorithm is applied to the subset of factors. In some embodiments, a machine learning algorithm is applied to each factor of the subset of factors. In some embodiments, each factor is analyzed and calculated separately, and the machine learning algorithm does not use expression levels of more than one factor as the training set. In some embodiments, a trained machine learning algorithm is applied to individual protein expression levels from the subject. In some embodiments, a machine learning algorithm trained on expression levels of a specific factor in AE / irAE / caAE negative subjects and AE / irAE / caAE positive subjects is applied to the expression level of that specific factor in the subject. It will be understood by a skilled artisan, that for each of the factors of the plurality of factors, a different algorithm will be trained and then applied to each expressionlevel of the subject. Thus, if three algorithms are separately trained on expression in AE / irAE / caAE negative subjects and AE / irAE / caAE positive subjects for Factor A, Factor B and Factor C, then the algorithm trained on Factor A expression levels will be applied to the subject’s expression level of Factor A, the algorithm trained on Factor B expression levels will be applied to the subject’s expression level of Factor B, and the algorithm trained on Factor C expression levels will be applied to the subject’s expression level of Factor C. In some embodiments, during a training phase, the machine learning model is trained on a training set comprising expression data for a single factor from AE / irAE / caAE negative subjects and AE / irAE / caAE positive subjects, using corresponding annotations of “AE / irAE / caAE negative” or “AE / irAE / caAE positive” to predict or classify factor expression data according to classes “AE / irAE / caAE negative subject” and “AE / irAE / caAE positive subject”. In some embodiments, during an inference stage, the machine learning model is applied to expression data of the single factor from a subject to predict classification of the factor as similar to an AE / irAE / caAE negative subject or AE / irAE / caAE positive subject. In some embodiments, the classification is an AE / irAE / caAE probability score. In some embodiments, the classification is a development of an AE / irAE / caAE score. In some embodiments, the classification is a measure of how similar the factor is to AE / irAE / caAE positive subjects and dissimilar to AE / irAE / caAE negative subjects.

[0170] In some embodiments, the trained machine learning algorithm is trained to predict development of an AE / irAE / caAE in subjects suffering from the disease in response to the Therapy. In some embodiments, the trained machine learning algorithm is trained to predict development of an AE / irAE / caAE in subjects suffering from the disease in response to the anticancer therapy. In some embodiments, the trained machine learning algorithm is trained to predict development of an AE / irAE in subjects suffering from the disease in response to the immunotherapy. In some embodiments, the trained machine learning algorithm is trained to predict development of an AEcaAE in subjects suffering from the disease in response to the chemotherapy. In some embodiments, the trained machine learning algorithm is trained to output an AE / irAE / caAE probability score. In some embodiments, the trained machine learning algorithm is trained to output a probability of developing an AE / irAE / caAE. In some embodiments, the trained machine learning algorithm is trained to output a probability of developing a specific type of AE / irAE / caAE. In some embodiments, the trained machine learning algorithm is trained to output probability of not developing an AE / irAE / caAE. In some embodiments, an AE / irAE / caAE probability score is a total AE / irAE / caAE probabilityscore. In some embodiments, the trained machine learning algorithm is trained to output an activity score. In some embodiments, the trained machine learning algorithm is trained to predict activity of a toxicity-associated factor in a subject. In some embodiments, the trained machine learning algorithm is trained to predict if a factor is a toxicity-associated factor in the subject. In some embodiments, the trained machine learning algorithm is trained to predict if a factor of the subject is a toxicity-associated factor in the subject.

[0171] In some embodiments, the trained machine learning algorithm is trained to predict AE / irAE / caAE development in the subjects suffering from the disease to the therapy. In some embodiments, the trained machine learning algorithm is trained to output an AE / irAE / caAE score. In some embodiments, the trained machine learning algorithm is trained to output an AE / irAE / caAE probability. In some embodiments, the trained machine learning algorithm is trained to output an activity score. In some embodiments, the trained machine learning algorithm is trained to predict activity of a toxicity-associated factor in a subject. In some embodiments, the trained machine learning algorithm is trained to predict if a factor is an AE / irAE / caAE development factor in the subject. In some embodiments, the trained machine learning algorithm is trained to predict if a factor of the subject is an AE / irAE / caAE non-development factor in the subject.

[0172] In some embodiments, the training set comprises received factor expression levels. In some embodiments, the training set comprises received factor expression levels in both AE / irAE / caAE negative subjects and AE / irAE / caAE positive subjects. In some embodiments, the training set comprises received factor expression levels for only one factor. In some embodiments, the training set comprises the number of toxicity-associated factors or AE / irAE / caAE development factors expressed in samples. In some embodiments, the sample are from subjects suffering from the disease. In some embodiments, the sample is from AE / irAE / caAE negative subjects. In some embodiments, the sample is from AE / irAE / caAE positive subjects. In some embodiments, the training set comprises at least one clinical parameter. In some embodiments, the clinical parameter is from the subjects. In some embodiments, subjects are AE / irAE / caAE negative subjects and AE / irAE / caAE positive subjects. In some embodiments, the training set comprises labels. In some embodiments, the labels are associated with the AE / irAE / caAE status of the subjects. In some embodiments, the labels are associated with the AE / irAE / caAE development in the subjects following treatment. In some embodiments, the labels are an AE / irAE / caAE negative subject or an AE / irAE / caAE positive subject. In some embodiments, the toxicity-associated factors are labeled with the labels. In some embodiments, the expression levels of the toxicity-associated factors are labeled with the labels. In some embodiments, the at least one clinical parameter is labeled with the label.

[0173] According to some embodiments, the training set further comprises at least one clinical parameter of each AE / irAE / caAE negative subjects and AE / irAEcaAE positive subjecst and the machine learning algorithm is applied to individual received factor expression levels from the subject and the subject’s at least one clinical parameter. In some embodiments, the at least one clinical parameter is the sex of the subjects. In some embodiments, the training set further comprises the sex of the subjects. In some embodiments, the subjects are each subject. In some embodiments, sex is gender. In some embodiments, the at least one clinical parameter is sex. In some embodiments, sex is a subject’s sex. In some embodiments, sex is male or female. In some embodiments, sex is sex at birth. In some embodiments, the training set comprises the sex of each AE / irAE / caAE negative subject. In some embodiments, the training set comprise the sex of each AE / irAE / caAE positive subject. In some embodiments, the clinical parameter is age. In some embodiments, age is a subject’s age.

[0174] In some embodiments, the clinical parameter is the line of treatment. In some embodiments, the line of treatment parameter is whether the therapy was a first line of treatment or an advanced treatment. In some embodiments, the line of treatment parameter is whether the immunotherapy was a first line of treatment or an advanced treatment. In some embodiments, the line of treatment parameter is whether the therapy was a first line of treatment or an advanced treatment. In some embodiments, the line of treatment parameter is whether the chemotherapy was a first line of treatment or an advanced treatment. In some embodiments, a line of treatment is first line treatment. In some embodiments, a line of treatment is a secondary treatment. In some embodiments, secondary treatment is an advanced treatment. It will be understood by a skilled artisan that advanced treatment may be any line of treatment after the first, e.g., second line, third line, fourth line, fifth line, etc. In some embodiments, the clinical parameter is whether the treatment is a first line treatment or an advanced treatment. In some embodiments, the clinical factor is the therapy administered. In some embodiments, the clinical factor is the immunotherapy administered. In some embodiments, the clinical factor is the chemotherapy administered. In some embodiments, the clinical factor is whether the immunotherapy is a monotherapy or a combination therapy. In some embodiments, the clinical factor is whether the chemotherapyis a monotherapy or a combination therapy. In some embodiments, the clinical factor is whether the immunotherapy is just immunotherapy or a combination of immunotherapy with a non-immunotherapy anticancer therapy. In some embodiments, the clinical factor is immunotherapy or immunotherapy in combination with chemotherapy. In some embodiments, the clinical factor is whether the chemotherapy is just chemotherapy or a combination of chemotherapy with a non-chemotherapy anticancer therapy. In some embodiments, the clinical factor is chemotherapy or immunotherapy in combination with chemotherapy.

[0175] In some embodiments, the clinical parameter is PD-L1 status. In some embodiments, PD-L1 status is PD-L1 status of the cancer. Methods of measuring PD-L1 levels in cancer cells (e.g., a tumor) are well known in the art and any such method may be employed. In some embodiments, PD-L1 status comprises high PD-L1 or low PD-L1. In some embodiments, PD-L1 status comprises high PD-L1, low PD-L1 or no PD-L1. In some embodiments, PD-L1 status comprises high PD-L1, medium PD-L1 or low PD-L1. In some embodiments, PD-L1 levels are numeric values between 0 to 100. In some embodiments, PD-L1 levels are percentages between 0 to 100. In some embodiments, PD-L1 status comprises PD-L1 expression in less than 1% of cancer cells, in 1-49% of cancer cells, or in 50% or more of cancer cells. In some embodiments, PD-L1 expression in less than 1% of cancer cells is no PD-L1 expression. In some embodiments, PD-L1 low or negative cancer comprises fewer than 50% of cancer cells being positive for PD-L1 expression. In some embodiments, expression is surface expression. In some embodiments, PD-L1 negative cancer comprises fewer than 1% of cancer cells being positive for PD-L1 expression. In some embodiments, PD-L1 expression in less than 1% of cancer cells is low PD-L1 expression. In some embodiments, PD-L1 expression in 1-49% of cancer cells is low PD-L1 expression. In some embodiments, PD-L1 low cancer comprises fewer than 1-49% of cancer cells being positive for PD-L1 expression. In some embodiments, PD-L1 expression in 1- 49% of cancer cells is medium PD-L1 expression. In some embodiments, PD-L1 expression in 50% or more of cancer cells is high PD-L1 expression. In some embodiments, a high PD- L1 cancer comprises expression in at least 50% of cells. In some embodiments, PD-L1 high cancer comprises at least 50% of cancer cells being positive for PD-L1 expression. In some embodiments, a low PD-L1 cancer comprises expression in 1-49% of cells. In some embodiments, a no PD-L1 cancer comprises expression in 0% of cells. In some embodiments, a no PD-L1 cancer comprises expression in less than 1% of cells. In someembodiments, the PD-L1 low or negative cancer is PD-L1 low cancer. In some embodiments, the PD-L1 low or negative cancer is PD-L1 negative cancer. In some embodiments, a no PD-L1 cancer is a PD-L1 negative cancer.

[0176] In some embodiments, the clinical parameter is a known biomarker of the disease or mutations in known biomarkers of the disease. In some embodiments, the biomarker is selected from MYC, NOTCH, EGFR, HER2, BRAF, KRAS, MAP2K1, MET, NRAS, NTRK1, NTRK2, NTRK3, PIK3CA, RET, ROS1, TP53, ALK, CDKN2A, KIT, NF1, BFAST, FGFR, LDH, PTEN, RB I, KEAP1, STK11, PD-L1, MSI (Microsatelite Instability), TMB (Tumor Mutational Burden), or a combination thereof. In some embodiments, the clinical parameter is expression of the biomarker. In some embodiments, expression is percent expression. In some embodiments, expression is mutational status.

[0177] In some embodiments, the calculating comprises applying a machine learning algorithm trained on a training set comprising the received factor expression levels in AE / irAE / caAE negative subjects and AE / irAE / caAE positive subjects and at least one clinical parameter, to the expression levels from the subject and the subject’s at least one clinical parameter and wherein the machine learning algorithm outputs the AE / irAE / caAE probability score. In some embodiments, the applying comprises cross-validation. In some embodiments, the training comprises the received factor expression levels in AE / irAE / caAE negative subjects and AE / irAE / caAE positive subjects and clinical parameters of each AE / irAE / caAE negative subject and AE / irAE / caAE positive subject and the machine learning algorithm is applied to individual received factor expression levels from the subject and the subject’s clinical parameters and wherein the machine learning algorithm outputs an AE / irAE / caAE probability score.

[0178] In some embodiments, the training set comprises the received factor expression levels in AE / irAE / caAE negative subjects and AE / irAE / caAE positive subjects. In some embodiments, the training set comprises the received factor expression levels in AE / irAE / caAE negative subjects and AE / irAE / caAE positive subjects and a clinical parameter. In some embodiments, the training set comprises the received factor expression levels in AE / irAE / caAE negative subjects and AE / irAE / caAE positive subjects and sex of each of the AE / irAE / caAE negative subjects and AE / irAE / caAE positive subjects. In some embodiments, the trained machine learning algorithm is applied to individual received factor expression levels from the subject. In some embodiments, the trained machine learning algorithm is applied to each received factor expression level from the subject. In someembodiments, the trained machine learning algorithm is applied to individual received factor expression levels from the subject and a clinical parameter from the subject. In some embodiments, the trained machine learning algorithm is applied to individual received factor expression levels from the subject and the subject’s sex.

[0179] In some embodiments, the clinical parameter is the type of treatment. In some embodiments, the type of treatment is selected from monotherapy or combination therapy. In some embodiments, the type of treatment is selected from immunotherapy alone or immunotherapy in combination with a non-immunotherapy anticancer therapy. In some embodiments, the type of treatment is selected from chemotherapy alone or chemotherapy in combination with a non-chemotherapy anticancer therapy. In some embodiments, the nonimmunotherapy anticancer therapy is an alternative therapy. In some embodiments, the nonchemotherapy anticancer therapy is an alternative therapy. In some embodiments, the clinical parameter is expression of a target of the immunotherapy. In some embodiments, the clinical parameter is expression of a protein within a process that is a target of the immunotherapy. In some embodiments, the process is a process comprising the target of the immunotherapy. In some embodiments, the clinical parameter is expression of a target of the chemotherapy. In some embodiments, the clinical parameter is expression of a protein within a process that is a target of the chemotherapy. In some embodiments, the process is a process comprising the target of the chemotherapy. In some embodiments, expression is expression in the subject. In some embodiments, expression is expression in a diseased tissue. In some embodiments, expression is expression in a diseased tissue sample. In some embodiments, expression is expression in the tumor. In some embodiments, expression is expression in a tumor sample. In some embodiments, a tumor sample is a biopsy. In some embodiments, expression is expression not in the tumor. In some embodiments, expression is expression not in a tumor sample. In some embodiments, expression is expression in a liquid biopsy. In some embodiments, expression is expression in the blood plasma. In some embodiments, expression is expression in the blood serum. In some embodiments, expression is expression in either the blood plasma or the blood serum. In some embodiments, expression is expression in either the blood plasma and / or the blood serum. In some embodiments, expression is percent expression. In some embodiments, percent is percent of cells. In some embodiments, the therapy is anti-PD-1 therapy and the protein in the process is PD-L1. In some embodiments, the therapy is anti-PD-Ll therapy, and the target protein is PD-L1. In some embodiments, the clinical parameter is PD-L1 expression.

[0180] Additionally clinical parameters may also be included. A skilled artisan will be able to select relevant clinical parameters for inclusion in the training set. Examples of additional clinical parameters include, but are not limited to, histological type of the sample (e.g., adenocarcinoma, squamous cell carcinoma, etc.), age, metastatic location, tumor location, cancer staging (such as tumor, nodes and metastases, TNM, staging for example), performance status (such as ECOG performance status), genetic mutations, epigenetic status, general medical history (such as background diseases, hereditary disease), vital signs, blood measurements (e.g., LDH levels), renal and liver function, weight, height, pulse, PFS (progression-free survival), clinical benefit (progressive disease, stable disease, complete remission, partial remission), RECIST, blood pressure, and smoking history.

[0181] In some embodiments, at an inference stage the trained machine learning algorithm is applied. In some embodiments, the trained machine learning algorithm is applied to individual received factor expression levels. In some embodiments, the trained machine learning algorithm is applied to individual received factor expression levels and the at least one clinical parameter. In some embodiments, the trained machine learning algorithm is applied to individual received factor expression levels from the subjects and the subject’s sex.

[0182] In some embodiments, at the inference stage an input is received. In some embodiments, the sample is from a subject. In some embodiments, the input comprises at least one clinical parameter. In some embodiments, the subject suffers from the disease. In some embodiments, the subject has unknown AE / irAE / caAE development in response to the therapy. In some embodiments, the parameter is of the subject with unknown AE / irAE / caAE development. In some embodiments, at the inference stage the trained machine learning algorithm is applied. In some embodiments, applied is applied to the input. In some embodiments, the input is the received input. In some embodiments, the inference stage is to predict AE / irAE / caAE development. In some embodiments, AE / irAE / caAE development is AE / irAE / caAE development in response to the therapy in the subject with unknown responsiveness.

[0183] In some embodiments, a positive TAP is a TAP that is upregulated in subjects that develop an AE / irAE / caAE. In some embodiments, a positive TAP is a TAP that is down- regulated in subjects that do not develop an AE / irAE / caAE. In some embodiments, a negative TAP is a TAP that is down-regulated in subjects that develop an AE / irAE / caAE. In some embodiments, a negativfe TAP is a TAP that is upregulated in subjects that do notdevelop an AE / irAE / caAE. Thus, positive TAPs correlate with AE / irAE / caAE development and negative TAPs negatively correlate with AE / irAE / caAE development. In some embodiments, the average or median expression of a positive TAP is higher in subjects that develop AE / irAE / caAEs. In some embodiments, the average or median expression of a negative TAP is higher in subjects that do not develop AE / irAE / caAEs.

[0184] In some embodiments, the machine learning algorithm outputs the AE / irAE / caAE probability score. In some embodiments, the AE / irAE / caAE probability score is the TAP score. In some embodiments, the outputted AE / irAE / caAE probability score is scaled from 0 to 1. In some embodiments, 1 is perfectly similar to AE / irAE / caAE positive subjects and 0 is perfectly similar to AE / irAE / caAE negative subjects. In some embodiments, for a development of an AE / irAE / caAE score 1 is perfectly similar to AE / irAE / caAE negative subjects and 0 is perfectly similar to AE / irAE / caAE positive subjects. In some embodiments, the machine learning algorithm calculates similarity to AE / irAE / caAE negative subjects. In some embodiments, the machine learning algorithm calculates similarity to AE / irAE / caAE positive subjects. In some embodiments, the machine learning algorithm outputs a numeric value of similarity to AE / irAE / caAE negative subjects and AE / irAE / caAE positive subjects. In some embodiments, a protein is considered to be a TAP if it’s expression is significantly higher in AE / irAE / caAE positive subjects. In some embodiments, a protein is considered to be a TAP if it’s expression is significantly lower in AE / irAE / caAE negative subjects. In some embodiments, significance is statistical significance. In some embodiments, significance is determined by a statistical test. In some embodiments, the test is a Kolmogorov-Smirnov test. In some embodiments, a protein is considered to be a toxicity- associated protein (TAP) if its AE / irAE / caAE probability score is beyond a predetermined threshold. In some embodiments, the threshold for the AE / irAE / caAE probability score is calculated on a scale of 0 to 1. In some embodiments, the threshold for the AE / irAE / caAE probability score of a certain protein is between 0.2 and 0.95. In some embodiments, the threshold for the AE / irAE / caAE probability score of a certain protein is about 0.01, 0.05, 0.1, 0.15, 0.2, 0.25, 0.3, 0.35, 0.4, 0.42, 0.45, 0.5, 0.55, 0.6, 0.65, 0.7, 0.75, 0.8, 0.85, 0.9, or 0.95. Each possibility represents a separate embodiment of the invention. In some embodiments, the threshold for the AE / irAE / caAE probability score is 0.25. In some embodiments, the threshold for the AE / irAE / caAE probability score is 0.42. In some embodiments, the threshold for the AE / irAE / caAE probability score is 0.6. In some embodiments, the threshold for the AE / irAE / caAE probability score when calculated by amachine learning algorithm is about 0.2, 0.25, 0.3, 0.35, 0.4, 0.42, 0.45, 0.5, 0.55, 0.6, 0.65, 0.7, 0.75, 0.8, 0.85, 0.9, or 0.95. Each possibility represents a separate embodiment of the invention. In some embodiments, the threshold for the AE / irAE / caAE probability score when calculated with a machine learning algorithm is 0.25. In some embodiments, the threshold for the AE / irAE / caAE probability score when calculated with a machine learning algorithm is 0.42. In some embodiments, the threshold for the AE / irAE / caAE probability score when calculated with a machine learning algorithm is 0.6.

[0185] In some embodiments, the algorithm outputs AE / irAE / caAE probability. In some embodiments, the algorithm outputs AE / irAE / caAE probability score. In some embodiments, probability is probability score. In some embodiments, the algorithm outputs AE / irAE / caAE probability, and the AE / irAE / caAE probability is calculated on a scale of 0 to 1. In some embodiments, the algorithm outputs AE / irAE / caAE probability, and the AE / irAE / caAE probability is calculated on a scale of 0 to 10. In some embodiments, 10 is perfectly similar to AE / irAE / caAE positive subjects and 0 is perfectly similar to AE / irAE / caAE negative subjects. In some embodiments, the algorithm outputs the AE / irAE / caAE probability, and the AE / irAE / caAE probability is calculated on a scale of 0% to 100%, wherein 100% is a perfect AE / irAE / caAE positive subject and 0% is a perfect AE / irAE / caAE negative subject. In some embodiments, an AE / irAE / caAE probability above 50% indicates a subject likely to develop an AE / irAE / caAE. In some embodiments, an AE / irAE / caAE probability below 50% indicates a subject unlikely to develop an AE / irAE / caAE. In some embodiments, an AE / irAE / caAE probability above 30% indicates a subject likely to develop an AE / irAE / caAE. In some embodiments, an AE / irAE / caAE probability below 30% indicates a subject unlikely to develop an AE / irAE / caAE. In some embodiments, the threshold for the AE / irAE / caAE probability when calculated with a machine learning algorithm is 0.75. In some embodiments, a protein with an AE / irAE / caAE probability above 0.75 is active in the subject. In some embodiments, active is is a TAP in the subject. In some embodiments, active is is an active TAP in the subject. In this context active does not refer to biologically active but rather it is a TAP that is predictive in the subject. In some embodiments, the threshold for AE / irAE / caAE probability when calculated with a machine learning algorithm is 0.5. In some embodiments, a protein with an AE / irAE / caAE probability above 0.5 is active in the subject. In some embodiments, the threshold for AE / irAE / caAE probability when calculated with a machine learning algorithm is 0.3. In some embodiments, a protein with an AE / irAE / caAE probability above 0.3 is activein the subject. In some embodiments, an active protein is a TAP. In some embodiments, the threshold for AE / irAE / caAE probability when calculated with a machine learning algorithm is 0.25. In some embodiments, a protein with an AE / irAE / caAE probability above 0.25 is active in the subject. In some embodiments, an active protein is a TAP.

[0186] In some embodiments, development of a no AE / irAE / caAE probability is determined by the calculation (1-AE / irAE / caAE probability score). In some embodiments, 1- AE / irAE / caAE probability score is 1 -total AE / irAE / caAE probability score. In some embodiments, the AE / irAE / caAE probability score is the total AE / irAE / caAE probability score. In some embodiments, the no AE / irAE / caAE probability score is the total no AE / irAE / caAE probability score. In some embodiments, development of an AE / irAE / caAE probability is a development of a no AE / irAE / caAE probability score. In some embodiments, the machine learning algorithm outputs the AE / irAE / caAE probability score. In some embodiments, the machine learning algorithm outputs the no AE / irAE / caAE probability score. In some embodiments, the outputted no AE / irAE / caAE probability score is scaled from 0 to 1. In some embodiments, 1 is perfectly similar to AE / irAE / caAE negative subjects and 0 is perfectly similar to AE / irAE / caAE positive subjects. In some embodiments, the machine learning algorithm calculates similarity to AE / irAE / caAE negative subjects. In some embodiments, the machine learning algorithm calculates similarity to AE / irAE / caAE positive subjects. In some embodiments, the machine learning algorithm outputs a numeric value of similarity to AE / irAE / caAE negative subjects and AE / irAE / caAE positive subjects. In some embodiments, a protein is considered to be a TAP if its no AE / irAE / caAE score is beyond a certain threshold. In some embodiments, a protein is considered to be an active TAP if its no AE / irAE / caAE probability score is beyond a certain threshold. In some embodiments, the threshold for the no AE / irAE / caAE probability score is calculated on a scale of 0 to 1. In some embodiments, the threshold for the no AE / irAE / caAE probability score of a certain protein is about 0.2, 0.25, 0.3, 0.35, 0.4, 0.42, 0.45, 0.5, 0.55, 0.6, 0.65, 0.7, 0.75, 0.8, 0.85, 0.9, or 0.95. Each possibility represents a separate embodiment of the invention. In some embodiments, the threshold for the no AE / irAE / caAE probability score is 0.25. In some embodiments, the threshold for the no AE / irAE / caAE probability score is 0.3. In some embodiments, the threshold for the no AE / irAE / caAE probability score score is 0.5. In some embodiments, the threshold for the no AE / irAE / caAE probability score when calculated by a machine learning algorithm is about 0.2, 0.25, 0.3, 0.35, 0.4, 0.42, 0.45, 0.5, 0.55, 0.6, 0.65, 0.7, 0.75, 0.8, 0.85, 0.9, or 0.95. Each possibility represents a separateembodiment of the invention. In some embodiments, the threshold for the no AE / irAE / caAE probability score when calculated with a machine learning algorithm is 0.25. In some embodiments, the threshold for the no AE / irAE / caAE probability score when calculated with a machine learning algorithm is 0.3. In some embodiments, the threshold for the no AE / irAE / caAE probability score when calculated with a machine learning algorithm is 0.5.

[0187] In some embodiments, the algorithm outputs no AE / irAE / caAE probability, and the no AE / irAE / caAE probability is calculated on a scale of 0 to 1. In some embodiments, the algorithm outputs no AE / irAE / caAE probability, and the no AE / irAE / caAE probability is calculated on a scale of 0 to 10. In some embodiments, 10 is perfectly similar to AE / irAE / caAE negative subjects and 0 is perfectly similar to AE / irAE / caAE positive subjects. In some embodiments, the algorithm outputs the no AE / irAE / caAE probability, and the no AE / irAE / caAE probability is calculated on a scale of 0% to 100%, wherein 100% is a perfect AE / irAE / caAE negative subject and 0% is a perfect AE / irAE / caAE positive subject. In some embodiments, a no AE / irAE / caAE probability above 50% indicates a subject unlikely to develop an AE / irAE / caAE. In some embodiments, a no AE / irAE / caAE probability above 30% indicates a subject unlikely to develop an AE / irAE / caAE. In some embodiments, unlikely to develop an AE / irAE / caAE is likely to not develop an AE / irAE. In some embodiments, a no AE / irAE / caAE probability below 50% indicates a subject likely to develop an AE / irAE / caAE. In some embodiments, a no AE / irAE / caAE probability below 30% indicates a subject likely to develop an AE / irAE / caAE. In some embodiments, the threshold for the no AE / irAE / caAE probability when calculated with a machine learning algorithm is 0.75. In some embodiments, a protein with a no AE / irAE / caAE probability above 0.75 is active in the subject. In some embodiments, the threshold for the no AE / irAE / caAE probability when calculated with a machine learning algorithm is 0.25. In some embodiments, a protein with a no AE / irAE / caAE probability above 0.25 is active in the subject. In some embodiments, the threshold for the no AE / irAE / caAE probability when calculated with a machine learning algorithm is 0.3. In some embodiments, a protein with a no AE / irAE / caAE probability above 0.3 is active in the subject. In some embodiments, active is is a TAP in the subject. In some embodiments, active is an active TAP in the subject. In some embodiments, an active TAP is a TAP that is predictive of AE / irAE / caAE development in the subject. In some embodiments, the threshold for no AE / irAE / caAE probability when calculated with a machine learning algorithm is 0.5. In some embodiments, a protein with a no AE / irAE / caAE probability above 0.5 is active in the subject. In someembodiments, the threshold for no AE / irAE / caAE probability when calculated with a machine learning algorithm is 0.3. In some embodiments, a protein with a no AE / irAE / caAE probability above 0.3 is active in the subject.

[0188] In some embodiments, the score is between zero and 1. In some embodiments, active is active in the cancer. In some embodiments, active is active in the subject. In some embodiments, active is active in promoting AE / irAE / caAE development. In some embodiments, active is active in promoting an AE / irAE / caAE. In some embodiments, beyond a threshold is below a threshold. In some embodiments, beyond a threshold is above a threshold. In some embodiments, within a threshold is below a threshold. In some embodiments, within a threshold is above a threshold. In some embodiments, the predetermined threshold is 0.5, 0.4, 0.3, 0.25, 0.2, 0.15, 0.1, 0.05, 0.01, 0.005, 0.001, 0.0005 or 0.0001. Each possibility represents a separate embodiment of the invention. In some embodiments, the threshold is 0.05. In some embodiments, the threshold is 5%. In some embodiments, the score or total score is linearized to provide a total score between 0 and 1. In some embodiments, the score or total score is linearized to provide a total score between 0 and 10. In some embodiments, the threshold is 0.5. In some embodiments, the threshold is 5. In some embodiments, linearized is linearly scaled. In some embodiments, linearizing comprises a linear regression.

[0189] In some embodiments, the predetermined threshold is determined by performing a cross-validation within the training set. In some embodiments, the cross-validation is performed in multiple iterations. In some embodiments, at least 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 15, 20, 25, 30, 40, 50, 60, 70, 75, 80, 90, or 100 iterations are performed. Each possibility represents a separate embodiment of the invention. In some embodiments, at least 10 iterations are performed. In some embodiments, at least 80 iterations are performed. In some embodiments, a cross-validation comprises applying the model produced from the training set onto the test set. In some embodiments, the threshold is selected based on the best prediction within the test set. In some embodiments, the threshold is selected based on the best prediction within the training set. In some embodiments, the predetermined threshold is the median score in the training set. In some embodiments, the predetermined threshold is the score that best distinguishes between AE / irAE / caAE negative subjects and AE / irAE / caAE positive subjects in the training set. In some embodiments, the predetermined threshold is the ratio between the number of AE / irAE / caAE positive subjects in the development cohort to the total size of the development cohort. In some embodiments, thepredetermined threshold is the ratio between the number of AE / irAE / caAE negative subjects in the development cohort to the total size of the development cohort. In some embodiments, the predetermined threshold is determined by the fraction of AE / irAE / caAE in a population of subjects suffering from the same disease and treated by the same treatment. In some embodiments, the threshold is determined by the prevalence of the AE / irAE / caAE in a population of subjects suffering from the same disease and treated by the same treatment. In some embodiments, the threshold is determined by the prevalence of the AE / irAE / caAE in a population of subjects in the training set. In some embodiments, the threshold is determined by the prevalence of the AE / irAE / caAE in a population of subjects in the development set.

[0190] In some embodiments, the calculated AE / irAE / caAE probability scores are combined to produce a total AE / irAE / caAE probability score. In some embodiments, the calculated no AE / irAE / caAE probability scores are combined to produce a total no AE / irAE / caAE probability score. It will be understood by a skilled artisan that as the development of an AE / irAE / caAE and no AE / irAE / caAE probability scores are just 1 minus the other, they are always interchangeable. The conversion of no AE / irAE / caAE development to AE / irAE / caAE development can be performed on the individual factor level or after the scores are combined and performed on the total level. In some embodiments, combine is sum. In some embodiments, the AE / irAE / caAE probability scores are summed to produce a total AE / irAE / caAE probability score. In some embodiments, combine is average. In some embodiments, the AE / irAE / caAE probability scores are averaged to produce a total AE / irAE / caAE probability score. In some embodiments, the scores are weighted when combined.

[0191] In some embodiments, the method comprises determining the number of factors of the plurality of factors that are active in the subject. In some embodiments, an active factor is a factor with an AE / irAE / caAE probability score above a predetermined threshold. In some embodiments, an active factor is a factor with an AE positive prediction. In some embodiments, the threshold is 0.25. In some embodiments, a factor with an AE / irAE / caAE / caAE probability score above 0.25 is a factor active in the subject. In some embodiments, only the active factors are combined. In some embodiments, combining the calculated AE / irAE / caAE probability scores is combining the active AE / irAE / caAE probability scores. In some embodiments, converted comprises applying a linear regression model. In some embodiments, linearized is linearly scaled. In some embodiments, linearizing comprises a linear regression.

[0192] In some embodiments, an AE / irAE / caAE positive factor with a score above a predetermined threshold is an active TAP. In some embodiments, an AE / irAE / caAE negative factor with a score below a predetermined threshold is an active TAP. In some embodiments, the number of active TAPs in the subject is calculated. In some embodiments, the number of active TAPs in the subject is summed. In some embodiments, the number of TAPs is the AE / irAE / caAE probability score. In some embodiments, the number of TAPs is proportional to the AE / irAE / caAE probability score. In some embodiments, the number of TAPs is linearized. In some embodiments, the number of TAPs is proportional to the probability of developing an AE / irAE / caAE . In some embodiments, the number of TAPs is the total AE / irAE / caAE probability score. In some embodiments, the number of TAPs is proportional to the total AE / irAE / caAE probability score.

[0193] In some embodiments, a subject with a total AE probability score beyond a predetermined threshold is predicted to develop an AE in response to the therapy. In some embodiments, a subject with a total AE probability score beyond a predetermined threshold will develop an AE in response to the therapy. In some embodiments, beyond is above. In some embodiments, a subject with a total AE probability score within a predetermined threshold is predicted to not develop an AE in response to the therapy. In some embodiments, a subject with a total AE probability score within a predetermined threshold will not develop an AE in response to the therapy. In some embodiments, the total AE probability score is proportional to the subject’s probability of developing an AE in response to the therapy. In some embodiments, the total AE probability score is proportional to the probability the subject will develop an AE in response to the therapy. In some embodiments, the total AE probability score is proportional to subject’s risk of developing an AE in response to the therapy.

[0194] In some embodiments, a subject with a total irAE probability score beyond a predetermined threshold is predicted to develop an irAE in response to the immunotherapy. In some embodiments, a subject with a total irAE probability score beyond a predetermined threshold will develop an irAE in response to the immunotherapy. In some embodiments, beyond is above. In some embodiments, a subject with a total irAE probability score within a predetermined threshold is predicted to not develop an irAE in response to the immunotherapy. In some embodiments, a subject with a total irAE probability score within a predetermined threshold will not develop an irAE in response to the immunotherapy. In some embodiments, the total irAE probability score is proportional to the subject’sprobability of developing an irAE in response to the immunotherapy. In some embodiments, the total irAE probability score is proportional to the probability the subject will develop an irAE in response to the immunotherapy. In some embodiments, the total irAE probability score is proportional to subject’s risk of developing an irAE in response to the immunotherapy .

[0195] In some embodiments, a subject with a total caAE probability score beyond a predetermined threshold is predicted to develop a caAE in response to the chemotherapy. In some embodiments, a subject with a total caAE probability score beyond a predetermined threshold will develop a caAE in response to the chemotherapy. In some embodiments, beyond is above. In some embodiments, a subject with a total caAE probability score within a predetermined threshold is predicted to not develop a caAE in response to the chemotherapy. In some embodiments, a subject with a total caAE probability score within a predetermined threshold will not develop an caAE in response to the chemotherapy. In some embodiments, the total caAE probability score is proportional to the subject’s probability of developing a caAE in response to the chemotherapy. In some embodiments, the total caAE probability score is proportional to the probability the subject will develop a caAE in response to the chemotherapy. In some embodiments, the total caAE probability score is proportional to subject’s risk of developing a caAE in response to the chemotherapy.

[0196] In some embodiments, the machine learning model is a machine learning algorithm. In some embodiments, the algorithm is a supervised learning algorithm. In some embodiments, the algorithm is an unsupervised learning algorithm. In some embodiments, the algorithm is a reinforcement learning algorithm. In some embodiments, the machine learning model is a Convolutional Neural Network (CNN). In some embodiments, the machine learning model is XGBoost. In some embodiments, the at least one hardware processor trains a machine learning model. In some embodiments, the model is based, at least in part, on a training set. In some embodiments, the model is based on a training set. In some embodiments, the model is trained on a training set. In some embodiments, the model is trained on a training set and checked on a test set. In some embodiments, the model is trained on a training set and validated on a test set. In some embodiments, the model is trained on a training set, checked on a test set, and validated on a validation set. In some embodiments, the validation is performed on a validation set with unknown AE prevalnce. I think I made a confusion here: In some embodiments, the training set is divided into development and test sets. In some embodiments, the model is developed on thedevelopment set and checked on the test set. In some embodiments, the model is validated on a validation set. In some embodiments, validation is performed on a validation set without information on AE. In some embodiments, the at least one hardware processor applies the machine learning model to a factor expression level from a subject.

[0197] In some embodiments, the calculating comprises calculating a mean expression for each protein in AE / irAE / caAE negative subjects. In some embodiments, the calculating comprises calculating a mean expression for each protein in AE / irAE / caAE positive subjects. In some embodiments, the calculating comprises calculating a mean expression for each protein in AE / irAE / caAE negative subjects and a mean expression for each protein in AE / irAE / caAE positive subjects. In some embodiments, the calculating comprises calculating a distribution of the expression for each protein in AE / irAE / caAE negative subjects and AE / irAE / caAE positive subjects. In some embodiments, the calculating comprises calculating a standard deviation of expression for each protein in AE / irAE / caAE negative subjects and AE / irAE / caAE positive subjects. In some embodiments, in AE / irAE / caAE negative subjects is in the AE / irAE / caAE negative subject population. In some embodiments, in AE / irAE / caAE positive subjects is in the AE / irAE / caAE positive subject population. In some embodiments, the AE / irAE / caAE probability score is based on the ratio of deviation of the factor expression in the subject from the calculated mean in AE / irAE / caAE negative subjects to the deviation of the factor expression in the subject from the calculated mean in AE / irAE / caAE positive subjects. Calculation of deviation is well known to one skilled in the art. It will be understood that the more dissimilar the expression in the subject is from a mean the larger the deviation will be. Thus, factors that are very dissimilar to the mean in AE / irAE / caAE negative subjects will have a large numerator in the calculation of this ratio and factors that are lowly dissimilar to the mean in AE / irAE / caAE positive subjects will have a small denominator. Thus, the more dissimilar to AE / irAE / caAE negative expression and the more similar to AE / irAE / caAE positive expression is expression of a factor in a subject the higher the AE / irAE / caAE probability score will be. In some embodiments, an AE / irAE / caAE probability score beyond a predetermined threshold indicates a factor is a toxicity-associated factor. In some embodiments, a toxicity-associated factor is a toxicity-associated protein (TAP). In some embodiments, toxicity-associated factor is a TAP if its expression in AE / irAE / caAE negative subjects is statistically different from its expression in AE / irAE / caAE positive subjects.

[0198] In some embodiments, the calculating further comprises calculating a distribution for each factor in AE / irAE / caAE negative subjects. In some embodiments, the calculating further comprises calculating a distribution for each factor in AE / irAE / caAE positive subjects. In some embodiments, the calculating further comprises calculating a distribution for each factor in AE / irAE / caAE negative subjects and a distribution for each factor in AE / irAE / caAE positive subjects. In some embodiments, the calculating further comprises calculating a standard deviation for each factor in AE / irAE / caAE negative subjects. In some embodiments, the calculating further comprises calculating a standard deviation for each factor in AE / irAE / caAE positive subjects. In some embodiments, the calculating further comprises calculating a standard deviation for each factor in AE / irAE / caAE negative subjects and a standard deviation for each protein in AE / irAE / caAE positive subjects. In some embodiments, the calculating further comprises calculating a standard deviation for each factor in a mix of AE / irAE / caAE negative subjects and AE / irAE / caAE positive subjects. In some embodiments, the deviation is measured as a multiple of the calculated standard deviation. It will be understood by a skilled artisan that by scaling the deviation to the standard deviation for a group of expression values the deviation can be given in more absolute terms allow for the comparison of factors and populations with very small and very large stand deviations (which may also have very low and very high expression levels).

[0199] In some embodiments, the AE / irAE / caAE probability score is based on a Z-score for the expression level of each factor in the subject. In some embodiments, the AE / irAE / caAE probability score is based on the Z-score relative to AE / irAE / caAE negative subjects. In some embodiments, the AE / irAE / caAE probability score is based on the Z-score relative to AE / irAE / caAE positive subjects. In some embodiments, the AE / irAE / caAE probability score is based on both the Z-score relative to AE / irAE / caAE negative subjects and the Z- score relative to AE / irAE / caAE positive subjects. In some embodiments, the AE / irAE / caAE probability score is based on the ratio of the Z-score relative to AE / irAE / caAE negative subjects to the Z-score relative to AE / irAE / caAE positive subjects. It will be well known to a skilled artisan that a Z-score counts the distance of the individual level from the population mean in units of the population standard deviation. In some embodiments, the Z-score is calculated by Equation 1.

[0200] In some embodiments, the AE / irAE / caAE probability score is calculated by the( ■ — — I. In some embodiments, ZN is the deviation of the factor expression in the jZp l +C / subject from the calculated mean in AE / irAE / caAE negative subjects. In some embodiments,Zp is the deviation of the factor expression in the subject from the calculated mean in AE / irAE / caAE positive subjects. In some embodiments, 1 1 is the Z-score of the deviation. In some embodiments, | | is the standardizing of the deviation to a multiple of the standard deviation. In some embodiments, c is a constant. In some embodiments, constant is a regulation constant that prevents the score from divergence for ZP= 0. In some embodiments, the AE / irAE / caAE probability score is calculated by Equation 2. In some embodiments, monotonoic is an ad-hoc function that prevents the AE / irAE / caAE probability score from decreasing for extreme values within the AE / irAE / caAE positive distributions. In some embodiments, function is the function provided in Algorithm 1.

[0201] In some embodiments, an AE / irAE / caAE probability score beyond a predetermined threshold indicates a factor is a TAP. In some embodiments, beyond is above. In some embodiments, the threshold is a predetermined threshold. In some embodiments, threshold is a threshold value. In some embodiments, the threshold for the AE / irAE / caAE probability score is about 1.0, 1.1, 1.2, 1.3, 1.4, 1.5, 1.6, 1.7, 1.8, 1.9, 2.0, 2.1, 2.2, 2.3, 2.4, 2.5, 2.6, 2.7, 2.8, 2.9, 3.0, 3.1, 3.2, 3.3, 3.4, 3.5, 3.6, 3.7, 3.8, 3.9, 4.0, 4.1, 4.2, 4.3, 4.4, 4.5, 4.6, 4.7, 4.8, 5.0, 5.1, 5.2, 5.3, 5.4, 5.5, 5.6, 5.7, 5.8, 5.9, 6.0. 6.1, 6.2, 6.3, 6.4, 6.5, 6.6, 6.7, 6.8, 6.9, or 7.0. Each possibility represents a separate embodiment of the invention. In some embodiments, the threshold for the AE / irAE / caAE probability score is about 0.26. In some embodiments, the threshold for the AE / irAE / caAE probability score is about 0.3. In some embodiments, the threshold is about 0.05, 0.1, 0.15, 0.2, 0.25, 0.3, 0.35, 0.4, 0.45, 0.5, 0.55, 0.6, 0.67, 0.7, 0.75, 0.8, 0.85 or 0.9. Each possibility represents a separate embodiment of the invention. In some embodiments, the threshold is about 0.26. In some embodiments, the threshold is about 0.3.

[0202] In some embodiments, the method further comprises classifications of the toxicity- associated factors into at least one pathway, process, or network. In some embodiments, the method further comprises performing analysis on toxicity associated factors to determine at least one pathway, process, or network in which the toxicity-associated factors are involved. In some embodiments, the pathway, process, or network causes development of AE / irAE / caAE to the therapy. In some embodiments, the analysis is selected from pathway analysis, process analysis and network analysis. In some embodiments, the method further comprises performing pathway analysis on TAPs. In some embodiments, the method further comprises performing process analysis on TAPs. In some embodiments, the method further comprises performing network analysis on TAPs. In some embodiments, at least onepathway, process or network comprises at least 2, 3, 4, 5, 6, 7, 8, 9 or 10 pathways, processes, or networks. Each possibility represents a separate embodiment of the invention. In some embodiments, at least one pathway, process or network is all the pathways, processes or networks known to include the toxicity-associated factors. In some embodiments, at least one pathway, process or network is all the pathways, processes or networks enriched with toxicity associated factors. In some embodiments, enriched is the most enriched. In some embodiments, enriched comprises contains the most TAPs of any or the pathways, processes or networks.

[0203] In some embodiments, the method further comprises administering to a subject that is an AE / irAE / caAE positive subject, or predicted to be an AE / irAE / caAE positive subject, an agent that modulates at least one toxicity associated factor. In some embodiments, the agent inhibits the at least one toxicity associated factor. In some embodiments, the agent activates the at least one toxicity associated factor. In some embodiments, the agent modulates the at least one toxicity associated factor. In some embodiments, the agent affects a factor that was measured. In some embodiments, the agent affects a factor for which a probability score was calculated. In some embodiments, the agent affects a factor that is active in the subject. In some embodiments, the factor is the most active factor. In some embodiments, the factor is the factor that best differentiates AE / irAE / caAE positive and AE / irAE / caAE negative subjects. In some embodiments, the factor appears in most of the iterations of the ML. In some embodiments, the factor is the most statistically significant. Thus, for example, if the factor is more highly expressed in AE / irAE / caAE positive subjects then the agent would inhibit the factor. Conversely, if the factor, for example, is more lowly expressed in AE / irAE / caAE positive subjects then the agent would activate the factor. In some embodiments, a factor more lowly expressed in AE / irAE / caAE positive subjects than in AE / irAE / caAE negative subject is administered to the AE / irAE / caAE positive subjects. In some embodiments, the agent is the factor or a derivative thereof. Essentially, the agent should result in factor expression or function more similar to what is observed in AE / irAE / caAE negative subjects.

[0204] In some embodiments, the method comprises selecting a pathway, process or network. In some embodiments, the selected pathway, process or network is hypothesized to affect non-development of an AE / irAE / caAE to the therapy. In some embodiments, the selected pathway, process or network is hypothesized to cause non-development of an AE / irAE / caAE to the therapy. In some embodiments, the selected pathway, process ornetwork is hypothesized to affect development of an AE / irAE / caAE to the therapy. In some embodiments, the selected pathway, process or network is hypothesized to cause the development of an AE / irAE / caAE to the therapy. In some embodiments, the selected pathway, process or network is known to be druggable. In some embodiments, known to be druggable comprises a known therapeutic agent that modulates the pathway, process, or network. In some embodiments, the selected pathway, process or network is known to be druggable. In some embodiments, known to be druggable comprises a known therapeutic agent that modulates the pathway, process or network. In some embodiments, the known therapeutic agent is in or has concluded clinical trials. In some embodiments, the known therapeutic agent is approved for human use. In some embodiments, approved for human use is approved for use in treating the disease in a human. In some embodiments, the disease is cancer.

[0205] In some embodiments, the method further comprises administering to a subject that is an AE / irAE / caAE positive subject, or predicted to be an AE / irAE / caAE positive subject, an agent that modulates the at least one pathway, process, or network containing a toxicity associated factor. In some embodiments, the agent inhibits a target in said pathway, process, or network. In some embodiments, the agent activates a target in said pathway, process, or network. In some embodiments, the agent modulates the pathway, process or network. In some embodiments, the target is a gene. In some embodiments, the target is a protein. In some embodiments, the protein is a regulatory RNA. In some embodiments, the target is a no AE / irAE / caAE associated factor. In some embodiments, the target is an AE / irAE / caAE associated factor. In some embodiments, the pathway’s activity induces development of an AE / irAE / caAE, and the agent inhibits the pathway. In some embodiments, the pathway’s activity reduces development of an AE / irAE / caAE, and the agent activates the pathway. It will be understood by a skilled artisan that a no AE / irAE / caAE associated factor is identified by its expression in a subject being more similar to the expression in AE / irAE / caAE negative subjects than AE / irAE / caAE positive subjects. Thus, for example, if the factor is more highly expressed in AE / irAE / caAE positive subjects and increases activity of the pathway / process / network then the agent would inhibit the pathway. If, for example, the factor is more highly expressed in AE / irAE / caAE positive subjects, but decreases activity of the pathway / process / network then the agent would activate the pathway / process / network. Similarly, if the factor, for example, is more lowly expressed in AE / irAE / caAE positive subjects and decreases activity of the pathway / process / network the agent would inhibit thepathway / process / network. And lastly, if, for example, the factor is more lowly expressed in AE / irAE / caAE positive subjects but increases activity of the pathway / process / network the agent would activate the pathway / process / network. Essentially, the agent should induce the pathway / process / network to function more as it does in AE / irAE / caAE negative subjects. In some embodiments, the agent targets a hub target in the pathway. In some embodiments, the agent targets a regulator target in the pathway. In some embodiments, the process activity induces development of an AE / irAE / caAE, and the agent inhibits the process. In some embodiments, the processes’ activity reduces development of an AE / irAE / caAE, and the agent activates the process. In some embodiments, the agent targets a hub target in the process. In some embodiments, the agent targets a regulator target in the process. In some embodiments, the network activity induces development of an AE / irAE / caAE, and the agent inhibits the network. In some embodiments, the network activity reduces development of an AE / irAE / caAE, and the agent activates the network. In some embodiments, the agent targets a hub factor in the network. In some embodiments, the agent targets a regulator factor in the network. In some embodiments, the regulator is a master regulator. The factors can be classified into pathways, protein interaction or signals using any analysis tool known in the art. Examples include, but are not limited to, Fisher’s exact test, ID enrichment analysis, GO analysis, Ingenuity analysis, Metacore analysis (Clarivate Analytics), reactome pathway analysis and functional analysis.

[0206] By another aspect there is provided, a computer program product comprising a non- transitory computer-readable storage medium having program code embodied thereon, the program code executable by at least one hardware processor to perform a method of the invention.

[0207] The present invention may be a system, a method, and / or a computer program product. The computer program product may include a computer readable storage medium (or media) having computer readable program instructions thereon for causing a processor to carry out aspects of the present invention.

[0208] The computer readable storage medium can be a tangible device that can retain and store instructions for use by an instruction execution device. The computer readable storage medium may be, for example, but is not limited to, an electronic storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any suitable combination of the foregoing. A non- exhaustive list of more specific examples of the computer readable storage medium includesthe following: a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), a static random access memory (SRAM), a portable compact disc read-only memory (CD-ROM), a digital versatile disk (DVD), a memory stick, a floppy disk, a mechanically encoded device having instructions recorded thereon, and any suitable combination of the foregoing. A computer readable storage medium, as used herein, is not to be construed as being transitory signals per se, such as radio waves or other freely propagating electromagnetic waves, electromagnetic waves propagating through a waveguide or other transmission media (e.g., light pulses passing through a fiber-optic cable), or electrical signals transmitted through a wire. Rather, the computer readable storage medium is a non-transient (i.e., not-volatile) medium.

[0209] Computer readable program instructions described herein can be downloaded to respective computing / processing devices from a computer readable storage medium or to an external computer or external storage device via a network, for example, the Internet, a local area network, a wide area network and / or a wireless network. The network may comprise copper transmission cables, optical transmission fibers, wireless transmission, routers, firewalls, switches, gateway computers and / or edge servers. A network adapter card or network interface in each computing / processing device receives computer readable program instructions from the network and forwards the computer readable program instructions for storage in a computer readable storage medium within the respective computing / processing device.

[0210] Computer readable program instructions for carrying out operations of the present invention may be assembler instructions, instruction-set-architecture (ISA) instructions, machine instructions, machine dependent instructions, microcode, firmware instructions, state-setting data, or either source code or object code written in any combination of one or more programming languages, including Python, an object oriented programming language such as Java, Smalltalk, C++ or the like, and conventional procedural programming languages, such as the “C” programming language or similar programming languages. The computer readable program instructions may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the latter scenario, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or theconnection may be made to an external computer (for example, through the Internet using an Internet Service Provider). In some embodiments, electronic circuitry including, for example, programmable logic circuitry, field-programmable gate arrays (FPGA), or programmable logic arrays (PLA) may execute the computer readable program instructions by utilizing state information of the computer readable program instructions to personalize the electronic circuitry, in order to perform aspects of the present invention.

[0211] These computer readable program instructions may be provided to a processor of a general-purpose computer, special purpose computer, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions / acts specified in the flowchart and / or block diagram block or blocks. These computer readable program instructions may also be stored in a computer readable storage medium that can direct a computer, a programmable data processing apparatus, and / or other devices to function in a particular manner, such that the computer readable storage medium having instructions stored therein comprises an article of manufacture including instructions which implement aspects of the function / act specified in the flowchart and / or block diagram block or blocks.

[0212] The computer readable program instructions may also be loaded onto a computer, other programmable data processing apparatus, or other device to cause a series of operational steps to be performed on the computer, other programmable apparatus or other device to produce a computer implemented process, such that the instructions which execute on the computer, other programmable apparatus, or other device implement the functions / acts specified in the flowchart and / or block diagram block or blocks. As used herein, the term "about" when combined with a value refers to plus and minus 10% of the reference value. For example, a length of about 1000 nanometers (nm) refers to a length of 1000 nm+- 100 nm.

[0213] It is noted that as used herein and in the appended claims, the singular forms "a," "an," and "the" include plural referents unless the context clearly dictates otherwise. Thus, for example, reference to "a polynucleotide" includes a plurality of such polynucleotides and reference to "the polypeptide" includes reference to one or more polypeptides and equivalents thereof known to those skilled in the art, and so forth. It is further noted that the claims may be drafted to exclude any optional element. As such, this statement is intendedto serve as antecedent basis for use of such exclusive terminology as "solely," "only" and the like in connection with the recitation of claim elements, or use of a "negative" limitation.

[0214] In those instances where a convention analogous to "at least one of A, B, and C, etc." is used, in general such a construction is intended in the sense one having skill in the art would understand the convention (e.g., "a system having at least one of A, B, and C" would include but not be limited to systems that have A alone, B alone, C alone, A and B together, A and C together, B and C together, and / or A, B, and C together, etc.). It will be further understood by those within the art that virtually any disjunctive word and / or phrase presenting two or more alternative terms, whether in the description, claims, or drawings, should be understood to contemplate the possibilities of including one of the terms, either of the terms, or both terms. For example, the phrase "A or B" will be understood to include the possibilities of "A" or "B" or "A and B."

[0215] It is appreciated that certain features of the invention, which are, for clarity, described in the context of separate embodiments, may also be provided in combination in a single embodiment. Conversely, various features of the invention, which are, for brevity, described in the context of a single embodiment, may also be provided separately or in any suitable sub-combination. All combinations of the embodiments pertaining to the invention are specifically embraced by the present invention and are disclosed herein just as if each and every combination was individually and explicitly disclosed. In addition, all subcombinations of the various embodiments and elements thereof are also specifically embraced by the present invention and are disclosed herein just as if each and every such sub-combination was individually and explicitly disclosed herein.

[0216] Additional objects, advantages, and novel features of the present invention will become apparent to one ordinarily skilled in the art upon examination of the following examples, which are not intended to be limiting. Additionally, each of the various embodiments and aspects of the present invention as delineated hereinabove and as claimed in the claims section below finds experimental support in the following examples.

[0217] Various embodiments and aspects of the present invention as delineated hereinabove and as claimed in the claims section below find experimental support in the following examples.EXAMPLES

[0218] Generally, the nomenclature used herein and the laboratory procedures utilized in the present invention include molecular, biochemical, microbiological and recombinant DNA techniques. Such techniques are thoroughly explained in the literature. See, for example, "Molecular Cloning: A laboratory Manual" Sambrook et al., (1989); "Current Protocols in Molecular Biology" Volumes I-III Ausubel, R. M., ed. (1994); Ausubel et al., "Current Protocols in Molecular Biology", John Wiley and Sons, Baltimore, Maryland (1989); Perbal, "A Practical Guide to Molecular Cloning", John Wiley & Sons, New York (1988); Watson et al., "Recombinant DNA", Scientific American Books, New York; Birren et al. (eds) "Genome Analysis: A Laboratory Manual Series", Vols. 1-4, Cold Spring Harbor Laboratory Press, New York (1998); methodologies as set forth in U.S. Pat. Nos. 4,666,828; 4,683,202; 4,801,531; 5,192,659 and 5,272,057; "Cell Biology: A Laboratory Handbook", Volumes I-III Cellis, J. E., ed. (1994); "Culture of Animal Cells - A Manual of Basic Technique" by Freshney, Wiley-Liss, N. Y. (1994), Third Edition; "Current Protocols in Immunology" Volumes I-III Coligan J. E., ed. (1994); Stites et al. (eds), "Basic and Clinical Immunology" (8th Edition), Appleton & Lange, Norwalk, CT (1994); Mishell and Shiigi (eds), "Strategies for Protein Purification and Characterization - A Laboratory Course Manual" CSHL Press (1996); all of which are incorporated by reference. Other general references are provided throughout this document.Materials and Methods

[0219] Patient cohort and specimen collection: Blood plasma samples and clinical data, including the occurrence of adverse events in response treatment, were collected from advanced stage non-small cell lung cancer (NSCLC) patients prior to commencement of ICL based treatment. One cohort included patients who were treated with Id-based regimens. Another cohort included NSCLC patients who received chemotherapy treatment.

[0220] Blood samples were collected from each patient into EDTA-anticoagulated tubes. Following plasma separation, the plasma samples were stored frozen at -80°C and were shipped frozen to the analysis laboratory.

[0221] Inclusion criteria: provision of informed consent; age older than 18 years; stage IIIB-IV NSCLC; ECOG performance status 0-2; normal hematological, renal and liver functions. In addition, exclusion criterion was any concurrent and / or other active malignancy that required systemic treatment within 2 years prior to receiving the first dose of Id-based treatment.

[0222] Clinical benefit data were retrieved from patient medical records and verified by the investigators through a review of radiologic images, i.e., CT chest / abdomen and brain MRI performed every 2-3 months, based on Response Evaluation Criteria In Solid Tumors (RECIST) 1.1. Clinical benefit (CB) was assessed based on Progression Free Survival (PFS) at 12 months after the commencement of treatment.

[0223] AE data were retrieved from patient medical records and from patient reporting as well as patient monitoring.

[0224] Proteomic measurements: Proteomic profiling of plasma samples was performed using SomaScan assay that simultaneously measures a total of 7596 protein targets. The assay is based on chemically-modified single stranded oligonucleotides that fold into molecular structures capable of binding to proteins with high affinity and specificity. The measurement is performed using DNA microarray technology with a readout provided in relative fluorescence units (RFU). Since protein level distributions are roughly log-normal (i.e., the logarithm of the measurement is normally distributed), and given that many statistical methods assume normality, log2 transformation was applied unless stated otherwise. There were no data imputations.

[0225] The proteomic dataset was narrowed down to a set of proteins with high analytical reliability.

[0226] Model development: The model was developed using cross-validation by applying a random sampling approach with multiple iterations. In each iteration, the development set was randomly divided into a train set and a test set (75% and 25% of the cohort, respectively) while maintaining the examined AE rate proportion. In each iteration, proteins displaying differential levels between patients who experienced AE and those who did not were identified using Kolmogorov-Smirnov test. These proteins were termed Toxicity-Associated Proteins (TAPs). A prediction model based on a single protein was constructed for each TAP using XGBoost algorithm. Based on the output of all TAPs, the AE probability was determined.

[0227] Model performance: The performance of all models was evaluated using two metrics: (i) The area under the curve (AUC) of the receiver operating characteristics plot, (ii) Agreement between the predicted AE probability and the observed AE rate in terms of goodness of fit (R2of a linear regression), where the observed AE rate for each AE valuewas defined as the proportion of patients who experience AEs among a group of patients within the range of the AE probability ±0.1 or ±0.15 window.

[0228] Data analysis: All data analyses were conducted using Python, Perseus computational platform and GraphPad Prism (San Diego, California, USA, graphpad.com). A level of 0.05 or lower was considered significant. Enrichment analysis for the TAPs was done using Fisher’s exact test against the overall background of 1578 examined proteins (false discovery rate < 0.1).Example 1: Generation of predictive model

[0229] Pre-treatment plasma samples were collected from 1069 advanced-stage non-small cell lung cancer (NSCLC) patients as part of the PROPHETIC multi-center clinical trial (NCT04056247). Deep proteomic profiling of plasma samples was performed, measuring the levels of approximately 7000 proteins in each sample. Clinical data, including irAE occurrence, were recorded for each patient. Following a detailed curation process, 426 patients with available plasma samples and well- annotated irAE data remained in the study cohort (Fig. 1). The main clinical parameters are presented in Figure 2. Overall, 60 patients experienced significant irAEs at early onset, where significance is defined as high-grade (grade 3 or higher) or irAEs resulting in treatment change or termination within the first 100 days of treatment. A total of 197 patients had low- grade irAEs, irAEs at late onset or adverse events that are not immune -related, and 169 patients did not display any adverse events. The cohort included 177 female and 249 male patients. The patients were treated with either ICI as monotherapy (276 patients, receiving pembrolizumab, nivolumab, atezolizumab, or durvalumab) or ICI with chemotherapy (150 patients, combination with at least one of carboplatin, pemetrexed, cisplatin, paclitaxel, and gemcitabine).

[0230] A predictive model for irAEs was developed as follows: proteins displaying differential plasma levels in patients with significant irAEs versus patients without significant irAEs were identified using Kolmogorov-Smirnov statistical test in an iterative manner. Briefly, 80 iterations were performed, where in each iteration, the cohort was randomly divided into training and test sets, with 75% and 25% of the patients in the training and test sets, respectively. The algorithm also used the type of treatment received (monotherapy or combination therapy) as an input. In each iteration, the Kolmogorov- Smirnov test was performed and the top 50 differentially expressed proteins between the twogroups of patients were selected as proteins for inclusion in the predictive model. Such proteins are collectively termed Toxicity Associated Proteins (TAPs). Overall, 430 TAPs were identified (Table 1), some of which were selected multiple times over different iterations (115 were identified at least 10 times and provided in Table 2). Following the TAP selection process, a predictive model for significant (severe) irAEs was developed per TAP using a machine learning algorithm in an iterative process of cross validation. irAE predictions inferred from each TAP were summed up to yield a probability for irAEs. The predictor displayed good correlation between predicted irAE probability and observed irAE rate, with a goodness of fit (R2) of 0.92 (Fig. 3). The observed irAE rate was determined by calculating the fraction of patients that experienced irAEs within a patient group assigned the irAE probability ±0.15.

[0231] Table 1: TAPs, first analysis

[0232] Table 2: Repeated TAPs, first analysisExample 2: Biologically relevant enrichment in TAPs

[0233] Of the 430 TAPs, approximately 50% were identified as proteins displaying differential plasma levels when comparing between NSCLC patients and healthy subjects; this overlap is significantly enriched, with enrichment factor of 1.12 and false discovery rate (FDR) p-value of 0.002 (Fisher’s exact test; Fig. 4). This may imply a predisposition of NSCLC patients for processes that are related to irAEs. In addition, approximately 30% of the TAPs overlap with the list of resistance associated proteins (RAPs; proteins that are differentially expressed between NSCLC patients who experience clinical benefit from treatment and those who do not; these proteins are predictive for clinical benefit from ICIs), with enrichment factor of 1.23 and FDR p-value of 0.0003 (Fisher’s exact test). This finding is consistent with the known association between development of an irAE to ICIs and irAEs.

[0234] From a biological perspective, multiple TAPs are functionally interconnected, and many participate in different immune -related biological processes (Fig. 5). TAPs were classified into two groups based on their plasma levels in patients exhibiting or not exhibiting irAEs (i.e., significant early onset irAEs). Specifically, irAE-positive TAPs (total of 265 proteins) refer to TAPs found at higher plasma levels among patients who experienced irAEs, whereas irAE-negative TAPs (total of 166 proteins) refer to TAPs found at higher plasma levels among patients who did not experience irAEs. Functional differences between irAE-positive and irAE-negative TAPs were found. Specifically, multiple irAE-negative TAPs were involved in metabolism-related processes, as well as signaling and interaction related processes, while vesicular-related processes, as well as immune-related processes were more evident in irAE-positive TAPs (Fig. 6). Moreover, in comparison to irAE- negative TAPs, a larger percentage of irAE-positive TAPs appear to be membranal, lung tumor-related proteins and immune -related proteins (Fig. 7). Following these results, we examined the association between irAE-positive / negative TAPs with different immune cells. Many irAE-positive TAPs appear to be myeloblast-origin proteins. Specifically, a high fraction of neutrophil- and classical monocyte-related TAPs was identified in patients experiencing irAEs (Fig. 8), which coincides with the induced immune development of an irAE by ICIs. Neutrophils can elicit inflammation, which agrees with these findings. Moreover, irAE-positive TAPs had a lower fraction of lymphocyte -related proteins compared to irAE-negative TAPs. Altogether, this may reflect a high neutrophil-to- lymphocyte ratio (NLR) among patients who experienced irAEs. NLR was previously described in the literature as a biomarker for irAEs, as well as for clinical benefit. Notably, some studies indicate that high NLR is associated with increased risk for irAEs, while others suggest the contrary (see Zhang et al., 2023, “Neutrophil to Lymphocyte ratio as a predictor for immune-related adverse events in cancer patients treated with immune checkpoint inhibitors: a systematic review and meta-analysis”, Front Immunol 14, 1234142). It is possible that neutrophil- and lymphocyte -related proteins would serve as a more robust signal for irAEs in comparison to overall neutrophil and lymphocyte cell count, especially given the dual role of neutrophils in cancer progression and the tumor microenvironment.

[0235] Finally, we ran an enrichment test for irAE-positive and -negative TAPs (Fig. 9; Fisher’s exact test); the results reinforce the findings described here for the involvement of neutrophils and monocytes in irAEs. Specifically, proteins regulated by the SPI1 transcription factor were significantly enriched among irAE-positive TAPs (enrichmentfactor = 5.5; related to 11 irAE TAPs). SPI1 has multiple roles in hematopoiesis. It is a lineage- specifying transcription factor that positively regulates many genes in the macrophage, granulocyte, dendritic -cell and B-cell lineages. This suggests an upregulation of these cells. Neutrophil count signature related proteins were also significantly enriched, as well as proteins involved in inflammation. irAE-negative TAPs, on the other hand, displayed an opposite trend, with de-enrichment of proteins related to neutrophils and monocytes, as well as inflammation related proteins.Example 3: Reanalysis and derivation of new TAPs

[0236] Another model was developed based on the same data from Example 1 but without considering the treatment received as a parameter (unlike the first model). 80 iterations were performed, where in each iteration, the cohort was randomly divided into training and test sets, with 75% and 25% of the patients in the training and test sets, respectively. In each iteration, the Kolmogorov-Smirnov test was performed and the top 100 differentially expressed proteins between the two groups of patients were selected as proteins for inclusion in the predictive model. Overall, 681 TAPs were identified (Table 3), some of which were selected multiple times over different iterations (231 were identified at least 10 times, Table 4). A total of 415 proteins were identified as TAPs in both models (Table 5) and of those 113 proteins were identified at least 10 times in both iterations (Table 6).

[0237] Table 3: TAPs, second analysis

[0238] Table 4: Repeated TAPs, second analysis

[0239] Table 5: TAPs common to the first and second analyses

[0240] Table 6: Repeated TAPs common to the first and second analyses

[0241] A new predictor was designed using the new list of TAPs. In this predictor the total number of TAPs active in a patient (expression above a predetermined threshold) was calculated. That number was then linearized to produce a total severe irAE score that correlates with the likelihood of developing a severe irAE. The higher the score the more likely a subject is to develop irAEs. The new severe irAE model demonstrated statistically significant predictive capabilities (Fig. 10A) and had a high correlation between predicted severe irAE probability and the observed severe irAE rate (Fig. 10B). Further, a negative correlation exists between irAE probability and clinical benefit (Fig. 10C). This coincides with recent findings showing that while moderate irAEs is positively associated with overall survival, severe irAEs is associated with poor prognosis (Jarushka et al., ESMO 2023). Figure 10D shows a comparison of predicted severe irAE probability between patients who experienced severe irAEs and those who did not. There is a statistically significant difference between the predicted probability distributions of patients with and without severe irAEs (Mann-Whitney p-value =0.008).

[0242] Biological analysis on the list of TAPs demonstrated that multiple TAPs are functionally interconnected, and many participate in different immune-related biological processes. Also, many irAE-positive TAPs appear to be myeloblast-origin proteins. Specifically, a high fraction of neutrophil- and classical monocyte-related TAPs was identified in patients experiencing irAEs (Fig. 11), which coincides with the induced immune development of an irAE by ICIs.Example 4: Immune-related rash prediction

[0243] Next, a predictive model was developed for a specific immune-related adverse event. Immune-related rash was chosen as a proof of concept. To this end, a cohort comprised of 162 NSCLC patients who received immune checkpoint inhibitors (ICI) was recruited. The main clinical parameters are presented in Figure 12A. The majority of the cohort was males(59%). 93% of the patients had ECOG PS (performance status) of 0-1. Most of the patients had non-squamous cell carcinoma (74%). Most of the patients received ICI as a monotherapy (67%), while 31% were treated with a combination of ICI and chemotherapy and 2% received a combination of two immunotherapy agents. Out of the 162 patients, 26 experienced rash as an irAE (Fig. 12B).

[0244] A predictive model for rash irAE was developed using cross-validation. During model development, the cohort was divided into training and test sets (75% and 25%, respectively), and Toxicity-Associated Proteins (TAPs) were identified using a statistical test (Kolmogorov-Smirnov test). The 50 proteins with the lowest p-value were selected as TAPs. This process was repeated for 80 iterations, ultimately yielding 391 TAPs (Table 7), some of which were selected in more than one iteration (109 were identified at least 10 times and provided in Table 8). Next, for each patient, each TAP served as an indicator for whether the patient will have rash irAE by applying a single-protein machine learning-based model using the XGBoost algorithm. The integration of all single -protein models provides the rash irAE probability. The rash irAE probability is a value between 0 and 1.

[0245] Overall, the model displays strong predictive capabilities, with area under the curve (AUC) of the Receiver Operating Characteristics (ROC) plot of 0.75 and p-value lower than 0.0001 (Fig.l3A). When examining the correlation between the predicted rash irAE probability and the observed rash irAE rate, the R2is 0.91 and p-value is 0.018 (Fig. 13B).

[0246] Table 7: TAPs, rash analysis

[0247] Table 8: Repeated TAPs, rash analysisExample 5: Chemo-associated toxicity prediction

[0248] The concept of using levels of TAPs in blood to predict adverse events was now expanded to other anticancer treatments. Aiming to develop a predictive model for chemotherapy-associated adverse events (caAEs), a cohort comprised of 138 NSCLC patients who received chemotherapy was recruited. The main clinical parameters are presented in Figures 14A-14C. The majority of the cohort was males (70%). 99% of the patients had ECOG PS (performance status) of 0-1. Most of the patients had non-squamous cell carcinoma (67%). Out of the 138 patients, 31 experienced caAEs (Fig. 14B). The patients received different kinds of chemotherapy regimens (Fig. 14C). The effect of the CAAEs on the patients varied and included dose reduction, change of chemotherapy, postponing treatment cycle and treatment stop (Fig. 14D).

[0249] A predictive model for caAEs was developed using cross-validation. During model development, the cohort was divided into training and test sets (75% and 25%, respectively), and Toxicity- Associated Proteins (TAPs) were identified using a statistical test (Kolmogorov-Smirnov test). The 50 proteins with the lowest p-value were selected as TAPs. This process was repeated for 80 iterations, ultimately yielding 405 TAPs (Table 9), some of which were selected in more than one iteration (114 were identified at least 10 times and provided in Table 10). Next, for each patient, each TAP serves as an indicator for whether the patient will have caAE by applying a single -protein machine learning-based model using the XGBoost algorithm. The integration of all single -protein models provides the rash irAE probability. The rash irAE probability is a value between 0 and 1.

[0250] Overall, the model displays strong predictive capabilities, with area under the curve (AUC) of the Receiver Operating Characteristics (ROC) plot of 0.67 and p-value lower than 0.002 (Fig. 15A). When examining the correlation between the predicted rash irAE probability and the observed rash irAE rate, the R2is 0.80 and p-value is 0.01 (Fig. 15B).

[0251] Table 9: TAPs, caAE analysis

[0252] Table 10: Repeated TAPs, caAE analysisExample 6: Exploration of the 3 AE models

[0253] Examination of the TAPs of each model, namely- severe irAE model (second severe irAE model), rash irAE model, and caAE model- shows a minimal overlap of two proteins between all three models (Fig. 16A). Furthermore, the overlap between the two irAE predictors is larger than the overlap between any of the two irAE models and the caAEs model. When examining the model outputs, there was a positive and significant correlationbetween severe irAEs and Rash irAEs, while a negative correlation between severe irAEs and caAEs was observed (Fig. 16B). There was no significant correlation between the predictions of rash irAE and chemo-associated AEs. The overlapping proteins are summarized in Table 11.

[0254] Table 11: Overlapping TAPs between analyses

[0255] Bioinformatic analysis showed that the three models differ in the underlying biology. A representative significantly enriched biological process from each model showed that the severe irAE model is involved in inflammation, the rash irAE is involved in extracellular matrix-related proteins, and the caAE model is mainly involved in angiogenesis related proteins (Fig. 17A). Exploring the number of cell- specific immune-related proteins in each model showed that many of the TAPs that are higher in patients who experienced severe irAEs are neutrophil- or monocyte-related; this signal is observed also for rash irAE, but to a lesser extent, while it is not observed for the caAE TAPs (Fig. 17B). In addition, the number of immune cell specific caAE TAPs are lower than any of the two immune-related models. It should be noted that Figures 16A and 17A-17B were focused on TAPs that were selected in at least 10 iterations during model development (holds for all 3 models).Example 7: Prediction by number of distinct AEs, systemic AEs, and AEs with potentially severe outcomes

[0256] Predictive models were developed based on the number of adverse events (AEs), taking into consideration all AEs of all grades. To this end, a cohort comprised of 221 advanced stage NSCLC patients who received ICI treatment was used (Table 12). The majority of the cohort was males (54%). 88% of the patients had ECOG PS (performancestatus) of 0-1. Most of the patients had non-squamous cell carcinoma (73%). Most of the patients received ICI in combination with chemotherapy (58%), while 42% were treated with ICI as a monotherapy. Out of the 221 patients, 163 (74%) reported about AEs, out of which 88 (54% of those with AE, and 40% in total) reported experiencing severe AEs. 115 patients (71% of those with AE, and 52% in total) reported multiple occurrences of AEs, and 83 patients (51% of those with AE, and 38% in total) reported a systemic AE. Systemic AEs included fatigue, anorexia, alopecia, pain, fever, weakness, edema and more. 84% of the AEs and 88% of severe AEs occurred within 180 days following the first ICI dose of treatment.

[0257] For the model development, the cohort dataset was divided into a development set of 168 patients and a validation set of 53 patients. The sets were stratified in a manner that preserves the portion of patients having AEs in each set.

[0258] Table 12: Cohort description

[0259] The occurrence of AEs was evaluated during the first 180 days following treatment, with patients assessed for the number of AEs reported (Fig. 18A). Most patients had 0, 1, 2 or 3 AEs but a significant number of patients had a higher number of AEs. The list of reported AEs included: fatigue, skin reaction, dyspnea, constipation, vomiting, pneumonitis, colitis, hepatitis, nausea, neuropathy, anorexia, kidney injury, constipation, alopecia, anemia, pain, mucositis / stomatitis, cough, fever, neutropenia, pneumonia, hypothyroidism, edema, weakness, pulmonary embolism, hyperthyroidism, myocardial infarction, dysphagia, headache, general, gastroesophageal reflux, bleeding, hypokalemia, pruritic, thrush, chest infection, dry mouth, weight loss, rhinitis, hemoptysis, bronchitis, chest pain, lethargy, myopathy, eye disorder, hemoptysis, nephritis, arthralgia, ileitis, myocarditis, insomnia, neurotoxicity, neutropenic fever, sepsis, neurological deterioration, thrombocytopenia, renal function decline, thrombosis, edema, shaking, seizures, rheumatic symptoms, renal disfunction, tinnitus, infection, pleural effusion, dyspnea, vaginal dryness, peripheral vascular disease, periodontitis, tantrums, confusion, lung infection, leukopenia, dysgeusia, drug intolerance, drop in white blood count, drop in kidney function, diabetes, dehydration, conjunctivitis, cold sores, cold hands, cardiogenic shock, bruising, bronchopneumonia, bowel perforation, aphasia, alkaline phosphate increase, abdominal spasms, dyspepsia, arrhythmia, hepatotoxicity, leg swelling, hepatitis, hypophyseal disorder, hypophosphatemia, hyponatremia, hypomagnesemia, thyroiditis, hypertension,hydronephrosis, gastritis, ear infection, eye discharge, embolism, elevated creatine, elevated alanine transaminase and wheezing. Over 80 subjects had what was classified as a systemic AE, while others had AEs in specific systems (Fig. 18B). 182 AEs were categorized as severe AEs (29% of all events) and 443 events were categorized as mild AEs (71% of all events). The severe and mild AEs were spread across the various categories of AEs (Fig. 18C).

[0260] Five predictive models were developed based on the number of AEs. All models were trained and tested on a development set using cross-validation and eventually validated on a validation set. During model development, the development set was divided into train and test sets (75% and 25%, respectively), and Toxicity- Associated Proteins (TAPs) were identified using a statistical test (Kolmogorov-Smirnov test). The 50 proteins with the lowest p-value were selected as TAPs. This process was repeated for 80 iterations, ultimately yielding 1028 TAPs, some of which were selected in more than one iteration (740 were identified in at least 10 iterations). Next, for each patient, each TAP served as an indicator of whether the patient will have AEs by applying a single-protein machine learning-based model using the XGBoost algorithm. The integration of all single-protein models provided the AE probability. The AE probability is a value between 0 and 1. 47 of the 1028 TAPs appear as TAPs for all 7 models.

[0261] When patients with at least 1 AE (of any severity or category) were considered as AE positive, the model displayed an area under the curve (AUC) of the Receiver Operating Characteristics (ROC) plot of 0.63 and a p-value of only 0.06 (Fig.l9A). When patients with at least 2 AEs were considered, the predictive ability increased with an AUC of the ROC plot of 0.81 and p-value of less than 0.001 (Fig. 19B). Increasing the stringency to subjects with at least 3 AEs produced a further improvement with an AUC of the ROC plot of 0.84 and a p-value of less than 0.001 (Fig. 19C). Surprisingly, further increasing to at least 4 AEs or at least 5 AEs did not improve the prediction and indeed made it worse (Fig. 19D-19E, respectively). When examining the correlation between the predicted AE probability in the development set and the observed AE rate in the validation set, the R2 for the at least 1, 2, 3, 4, and 5 AE predictors were 0.77, 0.90, 0.96, 0.84 and 0.85, respectively with p-values of less than 0.0001 for all of them. It was thus determined that the number of AEs could be used to produce a highly accurate classifier and that 2-3 AEs as the cutoff for classifying the subject as AE positive was ideal.

[0262] As systemic AEs were observed to be the most common category of AEs (Figure 18B), a model for systemic AEs was developed using cross-validation based on the presence of systemic AEs. As before, during model development, the cohort was divided into training and test sets (75% and 25%, respectively), and Toxicity- Associated Proteins (TAPs) were identified using a statistical test (Kolmogorov-Smirnov test). The 50 proteins with the lowest p-value were selected as TAPs. This process was repeated for 80 iterations. Next, for each patient, each TAP serves as an indicator for whether the patient will have AEs by applying a single-protein machine learning-based model using the XGBoost algorithm. The integration of all single-protein models provides the systemic AE probability. The systemic AE probability is a value between 0 and 1.

[0263] Overall, the model displayed strong predictive capabilities, with area under the curve (AUC) of the Receiver Operating Characteristics (ROC) plot of 0.79 and p-value lower than 0.001 (Fig. 20). When examining the correlation between the predicted systemic AE probability and the observed systemic AE rate, the R2was 0.70 and p-value less than 0.0001.

[0264] Altogether, herein are described novel computational models for predicting significant AEs in patients with NSCLC based on proteomic profiling of pre-treatment plasma samples. The three models for severe AEs, rash AEs and chemo associated AEs demonstrate that this is a universal concept that can be used to predict any AEs from any type of anticancer treatment. Further, the TAPs provide insights into the biological processes underlying AEs. Early prediction of AEs could enable personalized management plans and mitigation strategies to reduce the risk of AEs during cancer treatment.

[0265] Although the invention has been described in conjunction with specific embodiments thereof, it is evident that many alternatives, modifications, and variations will be apparent to those skilled in the art. Accordingly, it is intended to embrace all such alternatives, modifications and variations that fall within the spirit and broad scope of the appended claims.

Claims

CLAIMS:

1. A method of predicting development of an anticancer therapy-related adverse event (AE) in response to an anticancer therapy in a subject, the method comprising: a. receiving factor expression levels for a plurality of factors i. in a population of subjects known to not develop an AE in response to said anticancer therapy (AE negative); ii. in a population of subjects known to develop an AE in response to said anticancer therapy (AE positive); and iii. in said subject; b. calculate for factors of said plurality of factors an AE probability score, wherein said calculating comprises applying a machine learning algorithm trained on a training set comprising said received factor expression levels in said AE negative population and said AE positive population to individual received factor expression levels from said subject and wherein said machine learning algorithm outputs said AE probability score; and c. combine said calculated AE probability scores to produce a total AE probability score wherein said total AE probability score is proportional to said subject’s probability of developing an AE in response to said anticancer therapy; thereby predicting development of an AE in response to an anticancer therapy in a subject.

2. A method of predicting development of an anticancer therapy-related adverse event (AE) in response to an anticancer therapy in a subject, the method comprising: a. receiving factor expression levels for a plurality of factors i. in a population of subjects known to not develop an AE in response to said anticancer therapy (AE negative); ii. in a population of subjects known to develop an AE in response to said anticancer therapy (AE positive); and iii. in said subject; b. calculate for at least one factor of said plurality of factors an AE probability score, wherein said calculating comprises applying a machine learningalgorithm and wherein the machine learning algorithm outputs said AE probability score; c. classify a factor with an AE probability score beyond a predetermined threshold as a toxicity associated factor; and d. sum the number of toxicity associated factors expressed in said subject, wherein said number of toxicity associated factors is a total AE probability score and wherein said total AE probability score is proportional to said subject’s probability of developing an AE in response to said anticancer therapy; thereby predicting the development of an AE in response to an anticancer therapy in a subject.

3. The method of claim 1 or 2, wherein said anticancer therapy is a chemotherapy and said adverse events are chemo-associated adverse events (caAE).

4. The method of claim 3, wherein said chemotherapy is selected from: Carboplatin, Abraxane, gemcitabine, paclitaxel, nab-paclitaxel, pemetrexed, vinorelbine, cisplatin, etoposide and combinations thereof.

5. The method of claim 3 or 4, wherein said caAEs are selected from the group consisting of: fatigue, hair loss, easy bruising and bleeding, infection, anemia (low red blood cell counts), nausea and vomiting, appetite changes / loss of appetite, constipation, diarrhea, mouth / tongue / throat problems such as sores and pain with swallowing, peripheral neuropathy, numbness, tingling, neuropathic pain, skin / nail changes, dry skin, skin color change, urine and bladder changes, kidney problems, weight changes, loss of concentration and focus, mood changes, changes in libido and sexual function, fertility problems, leukemia, and neutropenia.

6. The method of any one of claims 3 to 5, wherein said plurality of factors comprises at least two factors selected from the factors provided in Table 9.

7. The method of claim 6, wherein said plurality of factors consists of factors selected from Table 9.

8. The method of claim 6 or 7, wherein said plurality of factors comprises all factors provided in Table 9.

9. The method of claim 1 or 2, wherein said anticancer therapy is an immunotherapy and said adverse events are immune -related adverse events (irAE).

10. The method of claim 9, wherein said immunotherapy comprises administration of an immune checkpoint protein inhibitor.

11. The method of claim 10, wherein said immune checkpoint protein is selected from Programmed cell death protein 1 (PD-1), Programmed death-ligand 1 (PD-L1), cytotoxic T-lymphocyte-associated protein 4 (CTLA-4) and Lymphocyte-activation gene 3 (LAG-3).

12. The method of any one of claims 9 to 11, wherein said immunotherapy is a monotherapy.

13. The method of any one of claims 9 to 12, wherein said immunotherapy is a combination therapy and said combination therapy further comprises a chemotherapy or targeted therapy.

14. The method of claim 13, wherein said chemotherapy is selected from Carboplatin, Pemetrexed, Paclitaxel, Cisplatin, and Gemcitabine.

15. The method of any one of claims 9 to 14, wherein said irAE is selected from: itching, rash, pruritus, skin blisters, loss of skin pigmentation, hand-foot syndrome, Lichen planus, psopriasis, mucositis, dry mouth, mouth pain, dermatitis, diarrhea, colitis, hepatitis, vomiting, hypothyroidism, hyperthyroidism, pituitary inflammation (hypophysitis), adrenal insufficiency, thyroiditis, diabetic ketoacidosis, mild joint pain, muscle pain, inflammatory arthritis, muscle weakness, polymyalgia rheumatica, giant cell arteritis, dry eye syndrome, uveitis, episcleritis, blepharitis, scleritis, keratitis, maculopathy, meningitis, encephalitis, myasthenia gravis, Guillain-Barre syndrome, inflamed spinal cord, inflamed brain, myocarditis, pericarditis, myocardial infarction, arrhythmia, vasculitis, Venous thromboembolism, hypertension, anemia, neumonia, neurotoxicity, myositis, nephritis, hematological toxicity, acute kidney injury, hepatotoxicity, nephrotoxicity, pancreatitis, neurotoxicity, cardiovascular toxicity, arthritis, adrenal insufficiency (Al), type 1 diabetes mellitus (T1DM), nausea,autoimmune haemolytic anaemia (AIHA), acquired thrombotic thrombocytopenic purpura (TTP), aplastic anaemia (AA), immune thrombocytopenia (ITP), acquired haemophilia (AH), haemolytic uraemic syndrome (HUS), lymphopenia, polymyalgialike syndrome, myositis, , neuropathy, increase in serum creatinine levels, oliguria, haematuria or proteinuria, weight gain, oedema, shortness of breath, anorexia, high blood pressure, dark urine, abdominal or pelvic pain, tinnitus, fever, sepsis, leukopenia, stomatitis, pneumonia, and pneumonitis.

16. The method of any one of claims 9 to 15, wherein said plurality of factors comprises at least two factors selected from the factors provided in Table 1 or Table 3.

17. The method of claim 16, wherein said plurality of factors consists of factors selected from Table 1 or Table 3.

18. The method of claim 16 or 17, wherein said plurality of factors comprises all factors provided in Table 1 or all factors provided in Table 3.

19. The method of claim 15, wherein said irAE is a rash.

20. The method of claim 19, wherein said plurality of factors comprises at least two factors selected from the factors provided in Table 7.

21. The method of claim 20, wherein said plurality of factors consists of factors selected from Table 7.

22. The method of claim 20 or 21, wherein said plurality of factors comprises all factors provided in Table 7.

23. The method of any one of claims 1 to 22, wherein a subject with a total AE probability score beyond a predetermined threshold is predicted to develop an AE in response to said anticancer therapy and a subject with a total AE probability score within said predetermined threshold is predicted to not develop an AE in response to said anticancer therapy.

24. The method of any one of claims 1 to 23, wherein said AE is an early onset AE occurring within 100 days of starting said anticancer therapy, is an AE with a grade of 3 or greater on the CTCAE clinical grading system, is an AE that results in treatmentdelay or hold, change or termination within 100 days of starting said anticancer therapy, is an AE that caused subject hospitalization, or a combination thereof.

25. The method of claim 24, wherein said AE is an early onset AE with a grade of 3 or greater on the CTCAE clinical grading system or which results in treatment delay or hold, change or termination within 100 days of starting said anticancer therapy.

26. The method of any one of claims 1 to 23, wherein said AE is a systemic AE effecting more than one biological system of the subject.

27. The method of any one of claims 1 to 26, wherein said AE is a plurality of AEs, said method is a method of predicting development of at least 2 AEs in response to said anticancer therapy, said AE negative population is known not to develop at least two AEs and said AE positive population is known to develop at least two AEs.

28. The method of claim 27, wherein said method is a method of predicting development of at least 3 AEs in response to said anticancer therapy, said AE negative population is known not to develop at least three AEs and said AE positive population is known to develop at least three AEs.

29. The method of any one of claims 1 to 28, comprising before (b) selecting a subset of said plurality of factors, wherein said subset comprises factors that best differentiate between said AE positive population and AE negative population, and wherein said calculating is for each factor of said subset.

30. The method of claim 29, wherein said selecting comprises applying a statistical test to said received factor expression levels, optionally wherein said statistical test is a Kolmogorov-Smirnov test.

31. The method of claim 29 or 30, wherein said subset consists of at least 50 factors.

32. The method of any one of claims 1 to 31, wherein said factor expression level is from a time point before administration of said anticancer therapy to said subject.

33. The method of any one of claims 1 and 3 to 32, wherein said combining is summing.

34. The method of any one of claims 1 to 33, wherein said subject suffers from cancer.

35. The method of claim 34, wherein said cancer is selected from hepato-biliary cancer, cervical cancer, urogenital cancer, anogenital cancer, prostate cancer, thyroid cancer, ovarian cancer, nervous system cancer, ocular cancer, lung cancer, soft tissue cancer, bone cancer, pancreatic cancer, bladder cancer, skin cancer, intestinal cancer, hepatic cancer, rectal cancer, colorectal cancer, esophageal cancer, gastric cancer, gastroesophageal cancer, breast cancer, renal cancer, skin cancer, head and neck cancer, leukemia and lymphoma.

36. The method of claim 35, wherein said cancer is non-small cell lung cancer (NSCLC).

37. The method of any one of claims 1 to 36, wherein said predetermined threshold is determined by performing a cross-validation within said training set or is the median score of said training set.

38. The method of any one of claims 1 to 37, wherein said factors expression levels are factors expression levels in a biological sample provided by said subjects.

39. The method of claim 38, wherein said biological sample is selected from blood plasma, whole blood, blood serum or peripheral blood mononuclear cells.

40. The method of claim 39, wherein said biological sample is blood plasma.

41. The method of any one of claims 1 and 3 to 40, wherein said training set further comprises at least one clinical parameter from each subject in said AE negative population and said AE positive population and said machine learning algorithm is applied to individual received factor expression levels from said subject and said at least one clinical parameter from said subject.

42. The method of claim 41, wherein said clinical parameter is the type of anticancer therapy received, optionally wherein said type of anticancer therapy is selected from the group consisting of: immunotherapy as a monotherapy, chemotherapy as a monotherapy and immunotherapy in combination with a chemotherapy.

43. The method of any one of claims 1 to 42, further comprising administering said anticancer therapy to a subject predicted to not develop an AE in response to said anticancer therapy, stopping or delaying administering said anticancer therapy to said subject predicted to develop an AE in response to said anticancer therapy oradministering a reduced dose of said anticancer therapy or an alternative therapy to said subject predicted to develop an AE in response to said anticancer therapy.

44. The method of claim 43, wherein said alternative therapy is selected from said immunotherapy in combination with an immunosuppressant, an alternative immunotherapy, an alternative anticancer therapy, a non-chemotherapy anticancer therapy and a non-immunotherapy anticancer therapy.

45. The method of any one of claims 1 to 44, wherein said trained machine learning algorithm is trained by a method comprising: at a training stage, training a machine learning algorithm on a training set comprising:(i) factor expression levels of a plurality of factors in samples from subjects known to not develop an AE in response to an anticancer therapy (AE negative) and factor expression levels of a plurality of factors in samples from subjects known to develop an AE in response to said anticancer therapy (AE positive); and(ii) labels associated with the development of an AE in said subjects in response to said anticancer therapy; to produce a trained machine learning algorithm, wherein said trained machine learning algorithm is trained to output said AE probability score.

46. The method of claim 45, wherein said training set further comprises at least one clinical parameter from said subjects known to not develop an AE in response to an anticancer therapy and at least one clinical parameter from said subjects known to develop an AE in response to said anticancer therapy.

47. The method of claim 46, wherein said clinical parameter is the type of anticancer therapy received, optionally wherein said type of anticancer therapy is selected from the group consisting of: immunotherapy as a monotherapy, chemotherapy as a monotherapy and immunotherapy in combination with a chemotherapy.

48. The method of any one of claims 45 to 47, wherein said expression levels of said plurality of factors are labeled with said labels, said at least one clinical parameter is labeled with said labels or both.

Images