Methods for predicting prognosis and treatment response

Abstract

The present invention provides a computer-implemented method for predicting the therapeutic response of mammalian cancer patients to anti-cancer immunotherapy, the method comprising: a) providing a set of features including the expression levels of at least eight of the following oncogenic genes previously measured in a sample obtained from a tumor of a mammalian cancer patient: PTGS2, CXCL1, CXCL2, CXCL5, CXCL6, IL1A, HAS2, IL11, IL1B, EREG, IL6, CXCL8 (IL8), MMP12, NRG1, VCAN, and AREG; b) calculating a score using the features provided in a); and c) predicting the therapeutic response of the mammalian cancer patient based on the score calculated in step b). Related methods and systems for predicting therapeutic response are also provided.

Description

[Technical Field]

[0001] The present invention relates to materials and methods for predicting the response to cancer treatment and the overall survival of cancer patients, and is particularly applicable to patients undergoing immunotherapy, but is not limited to these. [Background technology]

[0002] The concept that cancer induces inflammation and that inflammatory cells in the tumor site can promote cancer progression is well established (Coussens et al., 2013; Hanahan and Weinberg, 2011; Mantovani et al., 2008). Several cellular and molecular inflammatory mediators commonly found in clinically apparent tumors are known to have pro-tumorogenic effects and are associated with many features of aggressive and invasive tumors in both preclinical models and cancer patients. These include tumor-infiltrating leukocytes such as macrophages, neutrophils, immature myeloid cells, or regulatory T cells, as well as molecules produced by these and other leukocytes, stromal cells, or molecules directly produced by cancer cells. IL-6, IL-8, CCL2, CXCL, or vascular endothelial growth factor (VEGF) are typical examples of soluble factors with multifaceted effects that can promote cancer growth and spread (Coussens et al., 2013; Mantovani et al., 2008).

[0003] Inflammation at the tumor site also exerts an anti-cancer effect by contributing to the recognition and elimination of cancer cells by the immune system. In particular, cytotoxic T cells (CTLs) are recognized as anti-tumor effectors in preclinical cancer models, and their abundance within tumors is associated with improved patient outcomes and responses to cancer treatment (Binnewies et al., 2018; Fridman et al., 2012; Galon, 2006; Thorsson et al., 2018). Therefore, high intratumoral levels of CTL chemotactic factors such as CXCL9 or CXCL10, or cytokines that promote type I immunity, CTL differentiation, and effector function, such as IL-12 or type I and type II interferons (IFNs), also lead to better prognosis (Gajewski et al., 2013; Spranger and Gajewski, 2018; Vesely et al., 2011). The levels of these factors within the tumor microenvironment (TME) can vary depending on the systemic and / or local inflammatory state, and their combined effects contribute to the strength and extent of the anti-tumor immune response. Conventional helper CD4 + T cells and cytotoxic CD8 +Further evidence indicates that tumor infiltration by other inflammatory cells, in addition to T cells, similarly promotes immune attack and correlates with a favorable prognosis. Natural killer (NK) cells, gamma delta T cells, innate-like lymphocytes, and Batf3-dependent conventional dendritic cell type I (cDC1) constitute a subset of the immune system often associated with improved outcomes (Bottcher et al., 2018; Broz et al., 2014; Gentles et al., 2015; Mittal et al., 2017; Morvan and Lanier, 2016; Ruffell et al., 2014; Sanchez-Paulete et al., 2016; Spranger et al., 2015; 2017). This applies to both spontaneous and treatment-induced antitumor responses, such as those resulting from the administration of immune checkpoint inhibitors, which have revolutionized cancer treatment by inducing beneficial responses, including long-term remission, in a wide range of cancer types (Ribas and Wolchok, 2018). In particular, with immune checkpoint inhibitor (ICB) therapy, the abundance of specific immune cells or inflammatory mediators is positively or negatively correlated with the treatment response (Ayers et al., 2017; De Henau et al., 2016; Herbst et al., 2014; Roh et al., 2017; Rooney et al., 2015; Tumeh et al., 2014).

[0004] However, how antagonistic, cancer-promoting or suppressive inflammatory mesenchymal infiltration (TMEs) are established during cancer development and progression remains unclear. Furthermore, the signals and pathways that control the quality and quantity of various elements of inflammatory infiltration are not well defined. Improving our understanding of these processes is clinically critical, as it can enhance our ability to predict responses to treatment and provide even more compelling therapeutic targets to improve the effectiveness of anti-cancer treatments.

[0005] The concept of immune-based cancer control has historically been a subject of debate, but it has gained undeniable support over the past decade due to the remarkable clinical success of cancer treatments utilizing the immune system's tumor-suppressing function (Alfaro et al., 2016; Van Allen et al., 2015). In particular, so-called immune checkpoint inhibitor (ICB) therapies targeting the CTLA-4 and PD(L)-1 axes have transformed the landscape of cancer treatment (Andrea et al., 2012). Despite their widespread use in clinical practice, including as first-line treatment for multiple cancer types, many patients have received little benefit from ICBs, and some have experienced serious side effects (Auslander et al., 2018). Therefore, there is an urgent need for highly sensitive and accurate biomarkers to predict the response to ICBs and guide treatment selection, and this need is particularly heightened in situations where alternative therapies may benefit patients (Ayers et al., 2017; Azizi et al., 2018).

[0006] Established and emerging immunobiological biomarkers for ICB therapy fall into two categories that broadly assess T cell activity or the (no)antigenicity of cancer cells. Tumors described as “hot” or “T cell inflammation” are characterized by T cell infiltration, particularly CD8 + Tumors exhibiting signs of T cell infiltration and / or increased levels of inflammatory mediators associated with the differentiation and function of cytotoxic T cell effectors. Patients with tumors possessing these features tend to show a better response to ICB. Similarly, measures of cancer cell antigenicity, such as tumor mutagenesis (TMB) and neoantigen clonality, are also associated with the ICB response. Importantly, the correlation between T cell inflammation and TMB is low, but they exhibit co-predictive power when combined (Cristescu et al., 2018). However, even when these biomarkers are combined, they lack sufficient precision, and CD8 +Numerous tumors with high T cell infiltration do not respond to ICB, while conversely, cold, low-antigenic tumors sometimes respond actively. Furthermore, in some cancer types, such as clear cell renal cell carcinoma (ccRCC), one of the malignancies with the highest response rate to ICB, none of the currently available immunobiomarkers have shown meaningful predictive utility (Bhattacharya et al., 2020; Bisawas et al., 2019). A common feature of past and present tumor immunoprofiling efforts, as well as established and emerging immunotherapy biomarkers, is the focus on T-cell inflammation. However, inflammation is a common feature of all solid tumors. Even in so-called "immune desert" tumors, where immune cell infiltration is extensively lacking, oncogene-driven inflammatory signaling is generally upregulated in cancer cells or stromal cells. However, in contrast to T-cell-mediated inflammation, the type of inflammatory response most frequently present in clinically apparent tumors is associated with the growth and treatment resistance of malignant cancer. Therefore, tumors exhibit quantitatively and qualitatively different inflammatory profiles that have cancer-promoting or cancer-suppressing effects, rather than simply being inflamed.

[0007] Multiple studies have highlighted the close association between tumorigenesis and tumor progression and inflammation and immune evasion (van Weverwij Pellyk and de Visser, 2023; Spranger and Gajewski, 2018). Among these studies, recent work has revealed that the cyclooxygenase (COX)-2 / prostaglandin E2 (PGE2) pathway, which is upregulated in many cancers and involved in various aspects of malignant proliferation, plays a central role as a major pro-tumorogenic inflammatory axis that enables cancer immune evasion and drives resistance to immunotherapy (Zelenay et al., 2015; Bonavita et al., 2020; Pelly et al., 2021). Therefore, anti-inflammatory or antagonistic agents that inhibit PGE2 synthesis or signaling enhance the efficacy of ICB in combination with chemoimmunotherapy in preclinical models (Pelly et al., 2021; Bell et al., 2022). Furthermore, by investigating a large dataset of cancer patients, the inventors previously demonstrated the prognostic and predictive power of mouse-derived inflammatory gene signatures integrating both COX-2-related pro-tumorogenesis and T-cell activation aspects. Therefore, activity levels of the COX-2 / PGE2 pathway, which is upregulated in many cancers and involved in multiple aspects of malignant tumor growth, constitute a candidate for molecular stratification of tumor inflammatory phenotypes. [Overview of the project] [Problems that the invention aims to solve]

[0008] While previously reported cancer prediction and prognosis models are promising, there is still a need for more human-centered models that can predict the treatment response and / or survival of cancer patients. This invention aims to meet these needs and provides further related advantages. [Means for solving the problem]

[0009] The inventors conducted a systematic meta-analysis of thousands of patient samples and cancer cell lines to study the value of monitoring pro-tumor inflammation as an immune-related biomarker for prognosis and prediction. Using COX-2 expression as a putative indicator of intratumor pro-tumor inflammatory activity, the inventors defined human-derived pro-tumor inflammatory (PTI) gene signatures that exhibit important and consistent prognostic value across molecularly defined immune subtypes, all cancers, and individual malignancies. Through analysis of genomic, transcriptome, and proteome data, the inventors elucidated the cellular origin, oncogenicity correlates, and impact of PTI on T-cell inflammation. In particular, PTI has been shown to exhibit co-prognostic utility when combined with the most potent current predictors of ICB response. Furthermore, in clear cell renal cell carcinoma (ccRCC), a cancer type in which established immune biomarkers do not show predictive utility, PTI has been shown to selectively predict immunotherapy outcomes. Molecular profiling of archived ccRCC samples and validation through two independent ICB-treated patient cohort analyses indicate that patient stratification based on pre-treatment PTI levels may guide immunotherapy decision-making. PTI is established as an independent prognostic and predictive immunobiomarker that may guide treatment selection in clinical practice. Overall, pro-tumor-induced inflammation is a critical factor in determining outcomes in cancer patients, and the PTI signature is established as a potent prognostic and predictive biomarker for response to immune checkpoint inhibitor therapy.

[0010] A first aspect of the present invention provides a computer-implemented method for predicting the therapeutic response of a mammalian cancer patient to anti-cancer immunotherapy, the method comprising: a) providing a set of features including the gene expression levels of at least eight, at least nine, at least ten, at least eleven, at least twelve (e.g., all) of the following oncogenic genes previously measured in a sample obtained from a tumor of a mammalian cancer patient: PTGS2, CXCL1, CXCL2, CXCL5, CXCL6, IL1A, HAS2, IL11, IL1B, EREG, IL6, CXCL8 (IL8), MMP12, NRG1, VCAN, AREG; b) calculating a score using the features provided in a); and c) predicting the therapeutic response of the mammalian cancer patient based on the score calculated in step b).

[0011] In some embodiments, the features further include Clonal_TMB and CXCL9. In some embodiments, the features include the expression levels of all genes: PTGS2, CXCL1, CXCL2, CXCL5, CXCL6, IL1A, HAS2, IL11, IL1B, EREG, IL6, CXCL8 (IL8), MMP12, NRG1, VCAN, and AREG.

[0012] In some embodiments, the gene expression level is a normalized gene expression level. In some embodiments, the score is calculated by normalizing the individual gene expression values ​​of the features using Z-scores and then calculating the average of the normalized measurements.

[0013] In some embodiments, the score is given by the formula:

[0014]

number

[0015] It is calculated according to the formula, where n p This is the number of the aforementioned gene expression features,

[0016]

number

[0017] (i) are positively correlated genes within the interval of unit values, and (e) represent gene expression values ​​expressed in log2 count per million (CPM); optionally,

[0018]

number

[0019] This represents the average expression level of positive genes in log2-converted counts per million (read per kilobase million (FPKM)). In other embodiments, the score is given by the formula:

[0020]

number

[0021] This is a probability score calculated using the formula, where β0 is the intercept weight, β is a vector of weights for each of the k variables, and x is a vector of variables related to the sample, and the variables include the aforementioned multiple features.

[0022] In some embodiments, the computer implementation method further includes evaluating tumor burden and / or neoantigen frequency in cancer patients. In some embodiments, predicting the treatment response of mammalian cancer patients involves comparing the score calculated from step b) with a reference median score derived from a responder sample cohort and a reference median score derived from a non-responder sample cohort, where the responder and non-responder sample cohorts are i) cancer patients having the same type of cancer as the aforementioned cancer patient, or ii) cancer patients whose cancer is tumor type independent.

[0023] In some embodiments, the score calculated in step b) indicates that the cancer is predicted to respond to anti-cancer immunotherapy, and the method further includes selecting mammalian cancer patients for anti-cancer immunotherapy.

[0024] In some embodiments, the gene expression levels of at least eight oncogenic genes are obtained by measuring the gene expression levels of at least eight, at least nine, at least ten, at least eleven, at least twelve or more (e.g., all) of the following oncogenic genes in samples obtained from tumors of mammalian cancer patients: PTGS2, CXCL1, CXCL2, CXCL5, CXCL6, IL1A, HAS2, IL11, IL1B, EREG, IL6, CXCL8 (IL8), MMP12, NRG1, VCAN, and AREG.

[0025] A second aspect of the present invention provides a method for predicting the therapeutic response of a mammalian cancer patient to anti-cancer immunotherapy, the method comprising: a) measuring a number of features, including measuring the gene expression levels of at least eight, at least nine, at least ten, at least eleven, at least twelve (e.g., all) of the following oncopromoting genes in a sample obtained from the patient's tumor: PTGS2, CXCL1, CXCL2, CXCL5, CXCL6, IL1A, HAS2, IL11, IL1B, EREG, IL6, CXCL8 (IL8), MMP12, NRG1, VCAN, AREG; b) determining a score using the features measured in a); and c) predicting the patient's therapeutic response based on the score determined in step b).

[0026] In some embodiments, the features further include Clonal_TMB and CXCL9. In some embodiments, the step of determining a score for predicting a subject's therapeutic response includes inputting data representing features of the subject, including gene expression levels of at least eight cancer-promoting genes, into a machine learning classifier, wherein the machine learning classifier is trained using a training dataset containing gene expression levels of at least the same cancer-promoting genes that form the subject's score, and the training dataset consists of at least two sample cohorts; one cohort includes multiple samples from multiple mammalian cancer patients (responders) known to have responded to anti-cancer immunotherapy, and the other cohort includes multiple samples from mammalian cancer patients (non-responders) known not to have responded to anti-cancer immunotherapy; and the step of causing the machine learning classifier to classify the subject's score as belonging to the responding group or the non-responding group.

[0027] In some embodiments, the training dataset includes gene expression levels of at least eight cancer-promoting genes from at least 50, at least 100, at least 150, at least 200, at least 250, and at least 300 samples from mammalian cancer patients known to have responded to anti-cancer immunotherapy (responders), and gene expression levels of at least eight cancer-promoting genes from at least 50, at least 100, at least 150, at least 200, at least 250, at least 300, at least 350, at least 400, at least 450, at least 500, and at least 550 samples from mammalian cancer patients known not to have responded to anti-cancer immunotherapy (non-responders).

[0028] In some embodiments, the machine learning classifier is trained with multiple scores from at least three sample cohorts, one cohort comprising multiple samples from multiple mammalian cancer patients known to have fully responded to anti-cancer immunotherapy (full responders), another cohort comprising multiple samples from multiple mammalian cancer patients known to have partially responded to anti-cancer immunotherapy (partial responders), and another cohort comprising multiple samples from mammalian cancer patients known not to have responded to anti-cancer immunotherapy (non-responders).

[0029] In some embodiments, the machine learning classifier is a logistic regression model, an elastic network regression model, or a tree-based model, such as a random forest classifier. A third aspect of the present invention provides a method for classifying a sample obtained from a mammalian cancer subject, wherein the sample comprises nucleic acids and / or proteins, and the method includes a) analyzing the sample to obtain gene expression levels of at least eight cancer-promoting genes selected from PTGS2, CXCL1, CXCL2, CXCL5, CXCL6, IL1A, HAS2, IL11, IL1B, EREG, IL6, CXCL8 (IL8), MMP12, NRG1, VCAN, and AREG, and b) performing a method according to the first or second aspect of the present invention using the gene expression levels obtained in step a).

[0030] In some embodiments, the sample is derived from one or more cancer cells of the subject. In some embodiments, the sample obtained from the patient's tumor is a formalin-fixed paraffin-embedded (FFPE) sample.

[0031] A fourth aspect of the present invention provides a method for treating cancer in a mammalian cancer patient, the method comprising: a) performing a method according to any one of the prior aspects of the present invention to predict the therapeutic response of a mammalian cancer patient to an anti-cancer immunotherapy; b) determining in step a) that the mammalian cancer patient is expected to respond to the anti-cancer immunotherapy; and c) administering a therapeutically effective dose of the anti-cancer immunotherapy to a subject in need thereof.

[0032] A fifth aspect of the present invention provides a method for treating cancer in a mammalian cancer patient, comprising the steps of: a) performing a method according to any one of the first to fourth aspects of the present invention; b) determining that a score of the mammalian cancer patient indicates that the mammalian cancer patient is expected to respond to anti-cancer immunotherapy; and c) administering immunotherapy to the mammalian cancer patient in need of immunotherapy.

[0033] In some embodiments, the method further includes analyzing one or more additional markers of anti-cancer immunotherapy response derived from mammalian cancer patients to supplement and / or support predictions of anti-treatment response or non-response.

[0034] A sixth aspect of the present invention provides a system for predicting the therapeutic response of a mammalian cancer patient to anti-cancer immunotherapy, the system comprising at least one processor and at least one non-transient computer-readable medium containing, if executed by at least one processor, instructions causing at least one processor to perform a method according to the first or second aspect of the present invention.

[0035] In a seventh aspect of the present invention, one or more computer-readable media are provided, which, when executed by one or more processors, include instructions causing one or more processors to perform steps of a method according to a first or second aspect of the present invention.

[0036] In an eighth aspect of the present invention, a pharmaceutical composition is provided for use in the treatment of a mammalian cancer patient, including anti-cancer immunotherapy, wherein the patient is expected to respond to the anti-cancer immunotherapy by any one of the first to third aspects of the present invention.

[0037] In some embodiments of the first to eighth aspects of the present invention, a mammalian cancer patient has, and / or cancers described above, renal cancer, e.g., renal cell carcinoma or renal cell carcinoma of the kidney (KIRC); gastroesophageal tumors; esophageal cancer (ECA); melanoma; bladder cancer (BLCA); gastric cancer; lung cancer, e.g., adenocarcinoma of the lung (LUAD), squamous cell carcinoma of the lung (LUSC), or non-small cell lung cancer (NSCLC); pancreatic cancer (PAAD); liver cancer (LIHC); glioblastoma (GBM); adrenal cortical carcinoma (ACC); cervical cancer (CESC); head and neck cancer, e.g., head and neck squamous cell carcinoma (HNSC); thyroid cancer (THCA); uterine cancer (UCEC); giant cell tumor of the tendon sheath (TGCT); melanoma; BRCA; sarcoma and all cancers. In some embodiments, renal cell carcinoma is clear cell renal cell carcinoma (ccRCC). In some embodiments, the cancer immune subtype is one of the following subtypes: wound healing, IFNγ dominant, TGFβ dominant, inflammatory, and / or lymphocyte depletion.

[0038] In some embodiments, the anti-cancer immunotherapy includes IL-2 immunotherapy or immune checkpoint blockade therapy alone or in combination with anti-VEGF therapy. In some embodiments, the immune checkpoint blockade therapy includes programmed cell death-1 (PD-1) blockade, programmed cell death ligand 1 (PD-L1) blockade and / or cytotoxic T lymphocyte-associated protein 4 (CTLA-4) blockade. In some embodiments, the immune checkpoint blockade therapy includes treatment with nivolumab, pembrolizumab, atezolizumab and / or ipilimumab. In some embodiments, the anti-VEGF therapy is bevacizumab.

[0039] The present invention includes combinations of the described embodiments and preferred features, unless such combinations are clearly unacceptable or explicitly avoided. Examples and experiments illustrating the principle of the present invention will be discussed below with reference to the attached drawings. [Brief explanation of the drawing]

[0040] [Figure 1A] This figure shows the derivation of all cancerous tumor-promoting inflammatory signatures linked to COX-2. (A) COX-2 mRNA (PTGS2) expression in 31 TCGA (Cancer Genome Atlas) datasets. [Figure 1B] (B) A bioinformatics pipeline used to discover the gene signature of COX-2-centered PTI (pro-tumor inflammation). [Figure 1C] (C) COX-2 high tumor: Protein-protein interaction network related to genes upregulated in all cancers. [Figure 1D] (D) Gene ontology of biological processes significantly richer in the COX-2 high protein-protein interaction network. [Figure 1E] (E) Kaplan-Meier (KM) plots and associated risk tables for all solid tumor TCGA patients stratified by quantiles based on PTI signature score. [Figure 2A] Pro-tumoricidal inflammatory mediators are strong indicators of poor prognosis. (A) HR and 95% confidence intervals of continuous PTI and IFNγ signatures from a whole-cancer multivariate CoxPH model stratified by cancer type and stage, and their interaction terms. [Figure 2B] (B) Simulation of hazard ratios at various PTI values ​​in representative patients receiving high and low doses of IFNγ. The lines indicate the median hazard ratio. Dark areas represent the interquartile range, and light areas represent the minimum and maximum hazard ratios. [Figure 2C] (C) HR and 95% confidence interval from univariate Cox regression survival analysis of target biomarkers in six immune subtypes. [Figure 2D](D) KM survival curves (left panel) of patients with the "IFNγ-dominant" subtype, stratified into four groups based on PTI and IFNγ signatures. Asterisks indicate FDR values. Box plots (right panel) comparing CD8+ T cell scores across groups. Unless otherwise noted, all P values ​​are represented by asterisks (*<0.05, **<0.01, ***<0.001). [Figure 2E] (E) HR, 95% confidence intervals, and Wald test p-values ​​from multivariate CoxPH models of OS in ccRCC (n=531), LIHC (n=374), PAAD (n=176), glioblastoma (GBM, n=168), CESC (n=306), adrenocortical carcinoma (ACC, n=79), UCEC (n=277), THCA (n=513), and HNSC (n=240). [Figure 3A] MEK inhibition reduces the secretion of tumorigenic inflammatory mediators by cancer cells. (A) Pairwise Spearman correlation coefficients of the PTI gene in non-hematological, non-fibroblastic cancer cell lines derived from CCLE. Non-significant relationships (P>0.05) are indicated by "X". [Figure 3B] (B) PTI status coefficients from a linear mixed-effects model using PTI status (low / high) to predict drug sensitivity of cancer-type adjusted cell lines. Drug names (black) and targets (light gray / dark gray) are labeled on the "y" axis. All bars shown are based on statistically significant associations (P<0.05). [Figure 3C] (C) Correlation between cell line sensitivity to refametinib, calculated using the natural logarithm of drug IC50 values, and cancer-specific PTI transcription programs in all cell lines (top panel) and NSCLC cell lines (bottom panel). [Figure 3D] (D) Secretion protein concentrations detected by Luminex after treatment of NCIH1339 cells (top) and NCIH3255 cells (bottom) with 1 μM and 10 μM trametinib (6 hours, 8 hours, and 24 hours). [Figure 4A]NFκB links MAPK pathway mutations to cancer cell-specific PTI expression. (A) Box plot showing the relationship between cancer cell-specific PTI transcription programs and KRAS mutations in all cancer cell lines. [Figure 4B] (B) Estimated RELA gene copy number in cancer cell lines with low, moderate, and high cancer cell-specific PTI expression (0 = diploid). [Figure 4C] (C) Levels of p65 / RELA pS536 in RPPA data from KRAS mutant cancer cell lines stratified according to cancer-specific PTI status. [Figure 4D] (D) A table showing the top 10 motifs and predicted transcription factor binders that were significantly enriched in the expanded cancer cell-specific PTI gene list. [Figure 4E] (E) Sequence logos obtained from multiple sequence alignments of NFKB1 and NFKB2 binding motifs from chromatin immunoprecipitation sequencing (ChIPseq) experiments, showing the consensus nucleotide sequences of the associated motifs. [Figure 4F] (F) Odds ratios and 95% confidence intervals from a multivariate linear regression model using mutation status to predict PTI scores in the whole cancer TCGA dataset adjusted for cancer type and TMB. [Figure 4G] (G) The proportion of significant mutation-PTI associations belonging to different oncogenic signaling pathways. Breakdown of significant MAPK pathway mutation-PTI associations in individual cancers. [Figure 4H] (H) Comparison of intratumor PTI signature scores between all TCGA tumors with diploid copies of EGFR and RELA (WT) and tumors with chromosomal amplification. Wilcoxon test p-values ​​are shown. [Figure 5A] Pro-tumoricidal inflammatory mediators originate from heterogeneous cell types and influence the transcriptional phenotype of tumor-infiltrating T cells. Dimensionality reduction of single-cell transcriptomics data from primary HNSC tumors, color-coded by (A) major cell type or (B) PTI signature module score. [Figure 5B]Dimensionality reduction of single-cell transcriptomics data from primary HNSC tumors, color-coded by (A) major cell type or (B) PTI signature module score. [Figure 5C] (C) PTI module scores of individual cancer cells derived from primary HNSCs. Patients defined as having high cancer cell-specific PTI expression are shown in light gray, and patients with low expression are shown in dark gray. Asterisks indicate adjusted p-values ​​from nonparametric t-tests comparing cancer cells in the PTIhigh (light gray) vs. PTIlow (dark gray) groups (*<0.05, **<0.01, ***<0.001). The p-value for the Kruskal-Wallis test was <0.001. [Figure 5D] (D) Heatmap using circle size and color to represent the mean expression levels of individual PTI genes within individual clusters of immune and stromal cells in the HNSC scRNAseq dataset. Gray squares highlight selected co-expression gene groups. [Figure 5E] (E) Mean proportion of CIBERSORT estimated immune cell population in "IFNγ-dominant" subtype tumors with high, moderate, and low PTI levels. [Figure 5F] (F) Correlation between PTI signature score and T cell anergy signature, including contours showing point density. Spearman's correlation coefficient and P-value are shown. [Figure 5G] (G) CD8 ICB response signature scores in individual CD8+ T cells derived from MEL and NSCLC tumors, classified by PTI status. Adjusted P values ​​by multiple nonparametric t-tests are shown (*<0.05, **<0.01, ***<0.001). [Figure 6A] PTI predicts no response to checkpoint inhibitors and is superior to other resistance signatures. (A) Pie chart showing the proportion of samples (n=1032) in the total cancer immunotherapy metadata for five cancer types. [Figure 6B] (B) A table showing the individual cohorts, sample counts, and treatments within the overall cancer immunotherapy metadata set. [Figure 6C](C) Bar graph showing the proportion of responders (PR / CR) in patient groups stratified according to median PTI, TIS, IFNγ signaling, CD8+ T cell, and CD274 expression. [Figure 6D] (D) Proportion of responders in four patient groups simultaneously stratified by PTI and “hot” tumor biomarkers such as left-to-right TIS, IFNγ signaling, CD8+ T cells, and CD274 expression. Selected χ2 test p-values ​​are shown (*<0.05, **<0.01, ***<0.001). [Figure 6E] (E) Overall survival (OS) of patients in each PTI / TIS quadrant of the whole cancer ICB dataset, accompanied by a related risk table. Wald test p-values ​​are shown. [Figure 6F] (F) HR, 95% confidence interval, and -log10P value from a multivariate CoxPH model of randomly generated signatures using the median resistance signature and cutoff value, TIS status, and dataset as covariates. The graph shows only the signatures associated with poor prognosis (HR>1). [Figure 6G] (G) Comparison of PTI scores between patients with PD or PR / CR after ICB in the CPI1000 dataset. The p-value refers to a nonparametric t-test. [Figure 6H] (H) Effect size and standard error from a meta-analysis of PTI signatures and other putative biomarkers across seven separate ICB datasets from the CPI1000 cohort. Error bars are capped at an odds ratio of 5 for visualization. CYT = cell lysis score. [Figure 6I] (I) Overall survival (OS) of CPI1000 patients with high and low levels of clonal TMB, stratified by PTI signature (upper panel). Log-rank p-values ​​are shown. [Figure 7A] Pro-tumoricidal inflammation is a predictive biomarker in clear cell renal cell carcinoma. (A) Heatmap of pairwise Spearman correlation coefficients between individual PTI genes in the in-house Nanostring dataset. [Figure 7B](B) A heatmap showing the immune, stromal, and TME gene signature scores within the sample, arranged from left to right in order of increasing PTI score. The asterisks on the right side of the heatmap indicate the Spearman correlation P-value of the signature to the PTI signature (*<0.05, **<0.01, ***<0.001). [Figure 7C] (C) Correlation between PTI signatures and CD8+ T cell and IFNγ signaling gene signatures. Points are color-coded according to the PTI score. Spearman's Rho (R) and P values ​​are shown. [Figure 7D] (D) Heatmap of individual PTI gene expression levels in the Manchester Nanostring Cohort of patients with evaluable prognoses. High and low PTI groups are indicated by annotations at the top (light gray = high, dark gray = low). [Figure 7E] (E) Table of OS and associated risks by PTI for Manchester cohort patients. P values ​​refer to log-rank tests. [Figure 7F] (F) HR and 95% confidence interval of a univariate Cox regression model predicting OS in patients with selected genetic signatures. Patients were stratified by the median cutoff value. [Figure 7G] (G) KM plots with associated risk tables showing OS for Checkmate 009, 010, and 025 patients treated with nivolumab (left) or everolimus (right), categorized by median PTI expression level. [Figure 7H] (H) Comparison of PTI signature scores in PD patients and PR / CR patients treated with atezolizumab, atezolizumab + bevacizumab, or sunitinib in the IMMotion150 trial. The p-values ​​of the nonparametric t-test comparing PD patients and PR / CR patients are shown. [Figure 8]COX-2 expression and PGE2 signaling are associated with the expression of PTI mediators across multiple cell types in vitro. (A) Normalized enrichment scores from GSEA by Hallmark gene signatures, using a ranked list of genes obtained from a comparison of high and low PTGS2 cancers. (B) Log2FC values ​​and associated FDRs of PTI genes obtained from differential gene expression analyses of high and low PTGS2 human cancer cell lines. (C) Log2FCs of PTI genes from differential gene expression (DGE) analyses of mouse BMDM, human fibroblasts, and CD4+ T cells treated with synthetic PGE2 in vitro. Genes that were significantly upregulated (FDR<0.05, log2FC>1) are shown in light gray, and genes with decreased expression (FDR<0.05, log2FC<-1) are shown in dark gray. [Figure 9A] PTI is a strong, independent prognostic indicator that correlates with tumorigenesis and disease progression. (A) Box plots showing scores for PTI, CD8+ T cells, T-cell inflammation (TIS), and IFNγ signaling signatures across all cancer immune subtypes by Thorsson et al. (2018). Results of nonparametric t-tests are shown. [Figure 9B] (B) OS of "IFNγ-dominant" patients stratified clockwise by PTI, TIS and CD8 signature, TIL fraction and CD274 expression. Individual FDR values ​​are shown for the overall survival analysis using Wald's test (*FDR<0.05, **FDR<0.01, ***FDR<0.001). [Figure 9C] (C) HR, 95% confidence interval, and Wald test p-value from multivariate CoxPH models of OS in KIRK / ccRCC (n=531), LIHC (n=374), PAAD (n=176), glioblastoma (GBM, n=168), CESC (n=306), adrenocortical carcinoma (ACC, n=79), UCEC (n=277), THCA (n=513), and HNSC (n=240). [Figure 9D] (D) Average PTI signature score (±SEM) of samples at various stages of LUSC. [Figure 9E](E) Correlation between disease stage and PTI score (mean ± SEM) in the TCGA datasets for LUSC (n=498), LUAD (n=509), PAAD (n=176), ccRCC (n=531), and TGCT (n=139). [Figure 10A] Cancer cell-specific PTI expression is associated with MEK activity. (A) Heatmap showing stratification of high and low PTI cancer cell lines from Garnett et al. (2012) drug screening dataset, which includes microarray gene expression data from over 600 cell lines. These cell lines partially overlap with those in the CCLE dataset. [Figure 10B] (B) Dose-response curves of 10 NSCLC cell lines treated with rifametinib at PTIhigh and PTIlow. Shapes represent different cell lines. [Figure 10C] (C) Changes in confluence rates from baseline over time in cell lines NCIH1339 (top) and NCIH3255 (bottom) treated with two concentrations of trametinib in 96-well plates. Lines indicate time points for supernatant analysis. [Figure 10D] (D) Representative images of confluence measurements at 8, 24, and 72 hours after 10 μM trametinib treatment. The confluence rate for the entire well is shown in the lower right corner of each image. [Figure 10E] (E) Concentrations of four proteins in the supernatant after treating NCIH1339 lung cancer cells with 1 or 10 μM celemetinib for 6 hours, 8 hours, and 24 hours. [Figure 11] Cancer cell lines with high PTI mediator expression exhibit different copy number profiles and protein expression. (A) Estimated copy numbers of four different genes distinguished by the cancer cell-specific PTI status (0 = diploid). FDR values ​​are shown by nonparametric t-tests (*<0.05, **<0.01, ***<0.001). (B) Volcano plots showing differential expression of proteins and phosphorylated proteins measured by RPPA comparing PTIhigh and PTIlow KRAS mutant cell lines. [Figure 12A]PTI mediators originate from heterogeneous cell types. (A) Heatmap of Spearman correlations of PTI signatures to cell types estimated by the MCPcounter algorithm across the entire TCGA dataset. Significant correlations (p<0.05) are indicated by asterisks. [Figure 12B] (B) Single-cell t-SNE clustering from two independent scRNAseq patient datasets derived from NSCLC and MEL. The upper panel is color-coded by cell type, and the lower panel by PTI module score. [Figure 12C] (C) and (D) are heatmaps (displayed by size and color) showing the mean expression levels of individual PTI genes within individual clusters of immune and stromal cells across the entire scRNAseq datasets for NSCLC and MEL, respectively. Gray squares highlight co-expressed genes. [Figure 12D] (C) and (D) are heatmaps (displayed by size and color) showing the mean expression levels of individual PTI genes within individual clusters of immune and stromal cells across the entire scRNAseq datasets for NSCLC and MEL, respectively. Gray squares highlight co-expressed genes. [Figure 12E] (E) A scatter plot with contour lines showing the point density of the correlation between PTI signature scores and gene set scores for CD8+ T cell-related gene signatures in "IFNγ-dominant" subtype tumors (Singer et al., 2016; Guo et al., 2018). Spearman's correlation coefficient (E) and associated p-values ​​are shown. [Figure 13A] PTI adds predictive and prognostic value to current ICB response biomarkers. (A) AUC values ​​of a multivariate logistic regression model predicting patient ICB response (PD vs. PR / CR) across 16 datasets of TIS combined with resistance signatures. Asterisks indicate p-values ​​(*P<0.05) from nonparametric t-tests. Gray lines show mean and upper / lower confidence intervals from 1000 bootstrap trials. [Figure 13B](B) Heatmap of pairwise Spearman correlation coefficients between sequential immune biomarkers, including the PTI signature. The shape indicates the direction and strength of the correlation. SDI = Shannon Diversity Index. Hierarchical clustering is used to arrange boxes surrounding the four main clusters. [Figure 13C] (C) Box plot showing the association between PTI scores and binary clinical and genomic biomarkers in the CPI1000 dataset. Wilcoxon test p-values ​​are shown. B1501 refers to HLA type, and B44 and B62 refer to HLA supertypes. DDR = DNA damage response pathway mutation. RTK = receptor tyrosine kinase mutation. Max het = maximum HLA heterozygosity. HLA-LOH = loss of heterozygosity at the HLA locus. [Figure 13D] (D) Overall survival (OS) of patients with high, moderate, and low PTI levels in CPI1000 (Mariathasan dataset n=331, Riaz n=38, VanAllen n=34, Hugo n=27, Snyder PLOS Med n=25, Snyder NEJM n=7). Asterisks indicate Wald test p-values ​​(**<0.01). [Figure 13E] (E) Bar graph showing deviation estimates from univariate and multivariate logistic regression models predicting patient response (PD vs. PR / CR) to ICB treatment in the CPI1000 cohort. Green dots below the x-axis indicate variables included in each model. Lines between green dots indicate interaction terms. Statistical significance between nested models is determined by likelihood ratio tests. [Figure 13F] (F) KM plots showing OS in patients in the CPI1000 cohort stratified by PTI status, with high and low CXCL9 expression (upper panel) or CD274 expression (lower panel). Log-rank P-values ​​are shown. [Figure 13G] (G) Overall survival (OS) of patients with clonal TMB high CXCL9 high CPI 1000, stratified by PTI signature status. Log-rank P-values ​​are shown. [Figure 14]Experimental workflow and clinical characteristics of the Manchester RCC cohort. (A) Workflow showing the methodology applied to the autologous ccRCC cohort. (B) Table summarizing patient clinical characteristics by total sample size and percentage. aPD(L)1 refers to treatment with PD1 or PDL1 blocking antibodies. TKI = tyrosine kinase inhibitor. Here, aVEGF refers to bevacizumab. [Figure 15] Table 3 shows the receiver operating characteristic curve for tumor type-specific logistic regression model A, illustrating the false positive rate (x-axis) and the true positive rate (y-axis). [Figure 16A] AUC values ​​of the CPI1000 dataset for training (x-axis) and testing (y-axis) of elastic net classification models trained on the PTI gene, clonal TMB, and CXCL9. Gene expression levels were (A) normalized according to tumor type (tumor type-specific dataset model) or (B) tumor type independent (model B). Points indicate models trained with different values ​​of the α hyperparameter from 0 to 1. [Figure 16B] AUC values ​​of the CPI1000 dataset for training (x-axis) and testing (y-axis) of elastic net classification models trained on the PTI gene, clonal TMB, and CXCL9. Gene expression levels were (A) normalized according to tumor type (tumor type-specific dataset model) or (B) tumor type independent (model B). Points indicate models trained with different values ​​of the α hyperparameter from 0 to 1. [Figure 16C] (C) AUC values ​​for training and testing of a model trained to predict immunotherapy response in the CPI1000 dataset, featuring only the PTI gene. The optimal model had an α hyperparameter of 0.2 and seven key features (AREG, CXCL1, CXCL2, EREG, HAS2, IL1A, and PTGS2). The α value is shown next to the corresponding point (trained model). [Figure 17] Protocol for RNA extraction and gene expression profiling from formalin-fixed, paraffin-embedded (FFPE) surgical excision specimens. [Modes for carrying out the invention]

[0041] Aspects and embodiments of the present invention will be discussed with reference to the accompanying drawings. Those skilled in the art will readily understand further aspects and embodiments. All references mentioned herein are incorporated herein by reference.

[0042] In describing the present invention, the following terms are intended to be used and defined as follows. The features and advantages of this disclosure will be more readily understood by those skilled in the art by reading the following detailed description. For clarity, it should be noted that certain features of this disclosure described above and below in the context of separate embodiments may also be provided in combination in a single element. Conversely, various features of this disclosure described in the context of a single embodiment for the sake of brevity may also be provided individually or in any partial combination. It will be understood below that embodiments referred to in relation to one broad aspect of the invention are equally applicable to each of the other broad aspects of the invention described above. Furthermore, it will be understood that, unless otherwise specified in the context, the embodiments described below may be combined.

[0043] As used herein, “and / or” is interpreted to mean that each of the two specified functions or components is specifically disclosed, with or without the other function or component. For example, “A and / or B” means that (i) A, (ii) B, and (iii) A and B are disclosed separately, as they are described separately in this specification.

[0044] sample As used herein, “test specimen” may be a cell or tissue sample (e.g., a biopsy), a biological fluid, or an extract (e.g., a protein or DNA extract obtained from the subject). In particular, the specimen may be a tumor specimen, such as a solid tumor including a gastroesophageal junction tumor, melanoma, bladder tumor, or kidney tumor. The specimen may be freshly collected from the subject or may have been processed and / or preserved prior to the determination (e.g., frozen, fixed, or subjected to one or more purification, concentration, or extraction steps).

[0045] In some embodiments, the sample is a formalin-fixed paraffin-embedded (FFPE) sample. In some embodiments, the sample is a surgically excised sample. In some embodiments, the sample is H&E (hematoxylin and eosin) stained. In preferred embodiments, the sample is an FFPE surgically excised sample. Preferably, the FFPE sample, for example, an FFPE surgically excised sample, is H&E stained.

[0046] gene expression "Determining the expression level" refers to determining the expression level of a gene's expression product. Expression levels can be determined at the nucleic acid level and / or the protein level.

[0047] The determined gene expression levels can be considered to provide an “expression profile.” As used herein, “expression profile” is understood to mean a dataset of expression levels of one or more relevant genes in an individual, in a format that allows for comparison with comparable expression profiles from individuals with known prognoses, for example, to assist in determining the prognosis and selecting appropriate treatment for individual patients.

[0048] Determining gene expression levels may involve determining the presence or amount of mRNA in a sample of cancer cells. Methods for doing so are well known to those skilled in the art. Gene expression levels can be determined in a sample of cancer cells using any conventional method, for example, by using nucleic acid microarrays or nucleic acid synthesis (such as quantitative PCR). For example, gene expression levels can be determined using the NanoString nCounter analysis system (see, for example, U.S. Patent No. 7,473,767).

[0049] In some embodiments, the gene expression levels of at least eight cancer-promoting genes are obtained by measuring the gene expression levels of at least eight cancer-promoting genes in samples obtained from tumors of mammalian cancer patients.

[0050] Alternatively, the determination of gene expression levels may involve determining the levels of proteins expressed from those genes in samples containing cancer cells obtained from an individual. Protein expression levels can be determined by any available means, including the use of immunological assays. For example, expression levels can be determined by immunohistochemistry (IHC), Western blotting, ELISA, immunoelectrophoresis, immunoprecipitation, flow cytometry, mass cytometry, and immunostaining. By using any of these methods, it is possible to determine the relative expression levels of proteins expressed from the genes listed in Table 1.

[0051] In some embodiments, the gene expression level of each of the cancer-promoting genes is a normalized gene expression level. The gene expression levels and PTI scores derived therefrom, as detailed herein, can be compared to the expression levels and corresponding scores of the same genes in cancers of a patient group with known survival and / or treatment response. The patients being compared may be referred to as the “control group.” Therefore, the determined gene expression levels and PTI scores can be compared to the expression levels in the control group of individuals with cancer. The comparison can be made against the determined expression levels in cancer cells of the control group. The comparison can be made against the determined expression levels in a sample of cancer cells from the control group. The cancer in the control group may be of the same type as the individual. For example, when determining expression in an individual with renal cancer, the expression levels and PTI scores can be compared to the expression levels and PTI scores in cancer cells of patients who also have renal cancer.

[0052] Other factors can also be matched between the control group and the individuals and cancers being tested. For example, the cancer stage may be the same, the immune subtype of the cancer (as determined by Thorsson et al., 2018) may be the same, and the subjects and control group may be matched by age and / or sex.

[0053] Furthermore, the control group may have been treated with the same form of surgery and / or the same anti-cancer treatment, such as chemotherapy.

[0054] Clonal TMB Clonal tumor mutagenesis (TMB) is an indicator of somatic mutations presumed to be present in all tumor cells. Clonal TMB has been proposed as a comprehensive cancer predictor that forecasts the outcome of checkpoint blockade therapy. Clonal TMB measurements were obtained from Litchfield et al., 2021.

[0055] PTI Signature We discovered PTI gene signatures using a bioinformatics pipeline (Figure 1B). Gene expression data (raw count format) from the TCGA (Cancer Genome Atlas) was downloaded from the SAGE synapse database (syn4311114) along with clinical data (syn12026747). Raw counts were TMM normalized using calcNormFactors in edgeR (3.24.3). Genes that did not reach 0.25 count per million (CPM) in less than 10% of the samples were excluded from the analysis. Patients belonging to individual TCGA studies were stratified into high and low PTGS2 groups using the upper and lower quartiles of log2 normalized expression levels. Non-solid tumor types, such as LAML (acute myeloid leukemia) and DLBC (diffuse large B-cell lymphoma), were excluded from subsequent analyses. We designed a model matrix to compare patients with high and low PTGS2 while adjusting for cancer type. This was performed using the top 10 cancer types with the highest median PTGS2 levels (LUAD, BLCA, LUSC, HNSC, PRAD, PAAD, ESCA, STAD, CESC, KICH). DGE analysis was performed using limma-trend (P<0.05, FC>1.2) (Ritchie et al., 2015). 54 upregulated genes were filtered according to a list of ligand:receptor interactions (Ramilowski et al., 2015), and the resulting 16 genes were used as the final PTI signature (Table 1).

[0056] [Table 1]

[0057] PTI score In this specification, the terms “gene expression score,” “score,” and “PTI score” are used interchangeably. The terms “oncopromoting gene,” “protumoric inflammatory gene,” and “PTI gene” as used herein may be used interchangeably.

[0058] The oncogenic genes or pro-tumoricidal inflammation (PTI) genes listed in Table 1 were found to have oncogene expressions that positively correlate with PTGS2 expression, tumor growth, and poor response to immunotherapy. The inventors found that calculating gene expression scores for the oncogenic genes enhanced predictive power compared to previous predictive signatures under specific conditions, including the previously evaluated COX-IS ratio disclosed in International Publication WO2019 / 243567 (see, for example, Figure 6F). As used herein, the term “score,” such as “PTI score,” is intended to have a broader meaning, encompassing not only the sum of individual gene expression values ​​but also any relationships between features, such as the combination of gene expression values ​​of the oncogenic genes.

[0059] In some embodiments, at least eight oncogenic genes include HAS2 and / or EREG. In some embodiments, at least eight oncogenic genes include HAS2, EREG, CXCL8 (IL8), and / or VCAN. In some embodiments, at least eight oncogenic genes include HAS2, EREG, CXCL1, CXCL2, and / or AREG.

[0060] In some embodiments, the gene expression of at least eight, nine, ten, eleven, twelve, thirteen, fourteen, fifteen, and / or all (i.e., all 16) of the cancer-promoting genes listed in Table 1 can be measured in a sample obtained from a patient's tumor. The gene expression levels of at least eight PTI genes listed in Table 1 can be used to calculate a PTI score for gene expression in a sample obtained from a patient's tumor. The PTI score can be used to predict a patient's treatment response and / or prognosis after treatment, based on the gene expression levels of at least eight PTI genes.

[0061] PTI scores calculated using gene expression data from a reduced subset of the PTI gene set may have equivalent or greater predictive power than PTI scores calculated using gene expression data from all 16 PTI genes. The optimal reduced PTI gene set may vary depending on the clinical setting.

[0062] In some embodiments, the PTI score (also known as the “score”) is calculated by Z-score normalizing the individual gene expression values ​​of at least eight PTI genes and then calculating or averaging the normalized measurements.

[0063] In some embodiments, the PTI score is given by the formula:

[0064]

number

[0065] Calculated or calculated according to, In the formula, n p This is the number of the aforementioned cancer-promoting genes,

[0066]

number

[0067] (i) represents positively correlated genes within the unit value interval, and (e) represents gene expression values ​​expressed in log2 count per million (CPM). In some embodiments,

[0068]

number

[0069] This represents the average expression level of the positive gene in log2-converted counts per million (read per kilobase million (FPKM)). In other embodiments, the PTI score is calculated or computed as the weighted sum of Z-score normalized gene expression values of at least 8 PTI genes plus an intercept. Next, the probability of response to immunotherapy can be calculated using a logistic (sigmoid) function according to the following formula, where β0 is the intercept, and β 1-n refers to the coefficients of the at least 8 PTI genes. PTI score = β0 + β1X1 + ... + β n X n Pr(Response to ICB) = 1 / 1+e -PTIスコア The expression values can be represented by, for example, any of RPKM (Reads Per Kilobase per Million), FPKM (Fragments Per Kilobase per Million), CPM (Counts Per Million), and / or NanoString counts. Differentiation Method Based on Gene Expression In some embodiments, the present invention provides methods for classification, prognosis, or monitoring of cancer in a subject. In particular, data obtained from analysis of gene expression can be evaluated using one or more pattern recognition algorithms. Such analytical methods can be used to form predictive models that can be used to classify test data. For example, one convenient and particularly effective classification method employs multivariate statistical analysis modeling, firstly forming a model ("predictive mathematical model") using data ("modeling data") from samples of known subgroups (e.g., subjects known to have specific cancer prognosis subgroups: high risk and low risk), and secondly, classifying unknown samples (e.g., "test samples") according to the subgroups.

[0070] Pattern recognition methods have been widely used to characterize various types of problems, such as in linguistics, fingerprint analysis, chemistry, and psychology. In the context of the methods described herein, pattern recognition refers to analyzing data using both parametric and nonparametric multivariate statistics to classify samples and predict the value of some dependent variable based on the range of observed measurements. There are mainly two approaches.

[0071] The first approach to pattern recognition is called "unsupervised" modeling. These simply simplify the complexity of the data in a reasonable way and generate a display plot that can be interpreted by the human eye. However, this type of approach may not be suitable for developing clinical assays that can be used to classify samples from targets independent of the initial sample population used to train the predictive algorithm.

[0072] The second method of pattern recognition is called "supervised," which generates a mathematical model using a training set of samples with known classes or output values, and evaluates it on an independent validation dataset. Here, a statistical model is built using a "training set" of gene expression data to correctly predict a "subgroup" of each sample. This training set is then tested on independent data (called a "test set" or "validation set") to determine the robustness of the computer-based model. These models are sometimes called "expert systems," but may be based on different mathematical techniques such as support vector machines, decision trees, k-nearest neighbors, and naive Bayes. Supervised methods can use dimensionality-reduced datasets (e.g., the first number of principal components), but typically use unreduced data containing all dimensions. In either case, it is possible to quantitatively describe the multivariate boundaries that characterize and separate each subtype based on their unique gene expression profiles. It is also possible to obtain confidence limits in arbitrary predictions, such as the probability levels applied to the goodness of fit. The robustness of the predictive model can also be confirmed using cross-validation by excluding selected samples from the analysis.

[0073] As used herein, “training set” refers to a portion of a dataset used to fit or train a model to predict or classify values ​​that are known within the training set but unknown in (future) test data. In some embodiments, the training set includes gene expression levels of at least eight oncogenic genes from at least 50, at least 100, at least 150, at least 200, at least 250, at least 300, at least 350, at least 400, at least 500, at least 600, at least 700, at least 800, or at least 900 samples from mammalian cancer patients. In some embodiments, the training set may be a dataset from the Cancer Genome Atlas (TCGA). In some embodiments, the training set may be the CPI1000 dataset (Litchfield et al., 2021), which includes a collection of datasets from patients treated with ICB (anti-PD(L)1, anti-CTLA4) across multiple cancer types, along with whole exome sequencing (WES) and RNA sequencing (RNAseq) data.

[0074] After stratifying training samples according to subtype, centroids can be constructed using a centroid-based prediction algorithm based on the gene set expression profiles listed in Table 1.

[0075] "Transformations" of descriptor coordinate axes can be useful in some cases. Examples of such transformations include normalization and mean centering. "Normalization" can be used to remove variability between samples. Common methods for calculating normalization coefficients include: (i) whole-gene normalization using all genes on a microarray or nanostring code set; (ii) housekeeping gene normalization using constitutively expressed housekeeping / invariant genes; and (iii) internal control normalization using known amounts of exogenous control genes added during hybridization.

[0076] In one embodiment, the expression levels of each PTI cancer-promoting gene listed in Table 1 can be normalized, for example, to one or more control housekeeping genes. Representative housekeeping genes include ACTB(60), GAPDH(2597), and TBP(6908), where the number in parentheses after each gene name is the NCBI gene ID number for that gene. The nucleotide sequences of each gene disclosed for the NCBI gene ID number as of January 31, 2023, are expressly incorporated herein by reference. Those skilled in the art will understand that the methods disclosed herein are not constrained by normalization to any particular housekeeping gene, and that any suitable housekeeping gene known in the art can be used.

[0077] Many normalization approaches are possible, and they can often be applied at any of several points in the analysis. In one embodiment, microarray data are normalized using the LOWESS method, which is a scatter plot smoothing normalization function that is locally weighted overall. In another embodiment, qPCR and NanoString nCounter analysis data are normalized to the geometric mean of multiple housekeeping gene sets. Furthermore, qPCR can be analyzed using the double-variance method.

[0078] Mean centering can also be used to simplify data visualization and interpretation of calculations. Typically, for each descriptor, the mean value of that descriptor across all samples is subtracted. This causes the mean of the descriptors to coincide with the origin, and all descriptors are "centered" to zero. Unit variance scaling can scale data to have equal variances. Typically, the value of each descriptor is scaled by 1 / StDev, where "StDev" is the standard deviation of that descriptor across all samples. In another embodiment, Z-score normalization can be used to normalize all values ​​in a dataset, so that the mean of all values ​​is zero and the standard deviation is equal to the Z-score of 1. Pareto scaling is, in a sense, an intermediate method between mean centering and unit variance scaling. In Pareto scaling, the value of each descriptor is scaled by l / sqrt(StDev), where "StDev" is the standard deviation of that descriptor across all samples, and "sqrt" represents the square root. This makes the variance of each descriptor numerically equal to the initial standard deviation. Pareto scaling can be performed, for example, on raw data or mean centered data.

[0079] Logarithmic scaling can be used to aid interpretation when data has a positive bias and / or when data spans a wide range, such as differences of several orders of magnitude. Typically, the value of each descriptor is replaced with the logarithm of its value. In equirange scaling, each descriptor is divided by the range of that descriptor across all samples. This results in all descriptors having the same range, i.e., 1. However, this method is sensitive to the presence of outliers. In autoscaling, each data vector is mean-centered and scaled by the unit variance. This technique is very useful because each descriptor is then weighted equally, with large and small values ​​being treated equally. This can be important for genes that are expressed at very low levels but are still detectable.

[0080] When comparing data from multiple analyses (for example, comparing the expression profiles of one or more test samples with centroids constructed from samples collected and analyzed in independent studies), it is necessary to normalize the data across these datasets. In one embodiment, distance-weighted discriminant analysis (DWD) is used to combine these datasets together (Benito et al. (2004) Bioinformatics 20(1):105-114, the entire work of which is incorporated herein by reference). DWD is a multivariate analysis tool that can identify systematic biases present within individual datasets and then perform global adjustments to correct for these biases. Essentially, each individual dataset is a multidimensional cloud of data points, and DWD takes two point clouds and shifts one so that it more optimally overlaps the other.

[0081] In some embodiments described herein, the prognostic performance of a gene expression score (PTI score) can be evaluated using Cox proportional hazards (CoxPH) model analysis, a regression method on survival data that provides estimates of hazard ratios and their confidence intervals. The Cox model is a well-known statistical method for investigating the relationship between patient survival and specific variables. This statistical method allows for the estimation of an individual's hazard (i.e., risk) based on a prognostic variable (e.g., a gene expression ratio as described herein). The "hazard ratio" is the risk of death at any given time for a patient exhibiting a particular prognostic variable. A CoxPH regression analysis can be performed using the PTI score as input and disease stage and / or cancer type as stratification.

[0082] A logistic regression model can be calculated using PTI scores as input to predict a patient's treatment response outcome. This logistic regression model can be calculated for both tumor type-independent and tumor type-specific data.

[0083] In some embodiments, logistic regression models predicting binary patient treatment response outcomes for tumor type-independent and tumor type-specific approaches are determined by measuring individual gene expression values ​​for at least eight PTI genes, adding a pseudo-count of 1 to each value before log2 normalizing, then normalizing all samples in the dataset by Z-scores ("tumor type-independent"), or by grouping samples by cancer type ("Tum_Type"), normalizing the Z-scores by cancer type ("tumor type-specific"), and then calculating a reference gene expression score by taking the mean of the scaled expression data for at least eight PTI genes within the "tumor type-independent" or "tumor type-specific" group. In some embodiments, patients who achieved disease stability as the RECIST best response (metadata feature: Response_RECIST) are excluded from the PTI score calculation. In some embodiments, "in treatment" samples (metadata feature: Prior_IO) are excluded from the PTI score calculation.

[0084] The predictive power of the PTI score surpasses that of known resistance signatures. The ability of the claimed PTI score to predict poor overall survival in ICB-treated patients is higher than that of known resistance signatures, such as the preceding COX-IS ratio disclosed in International Publication WO2019 / 243567. As illustrated in Figure 6F and examined in Example 11, the overall cancer survival analysis of the PTI score (HR 1.38 [1.15-1.65], P=0.0004) surpasses that of the COX-IS ratio (HR 1.18 [0.96-1.45], P=0.12).

[0085] Prediction of treatment response Individuals classified into the favorable prognosis group may be identified as having cancer that is sensitive to immunotherapy, such as immune checkpoint blockade (ICB) therapy. They may also be referred to as responders, individuals who respond well to immunotherapy, or individuals who show a partial response (PR) or complete response (CR) to immunotherapy, such as immune checkpoint blockade (ICB) therapy (collectively referred to as PR / CR).

[0086] Individuals classified into the poor prognosis group may be identified as having cancer that is resistant to immunotherapy, such as immune checkpoint blocker (ICB) therapy. These may be referred to as non-responders, individuals with a poor response to immunotherapy, individuals who have not experienced benefits from immunotherapy, or individuals who have progressive disease (PD) after immunotherapy, such as immune checkpoint blocker (ICB) therapy.

[0087] If an individual is classified into the favorable prognosis group, they may be selected for treatment with appropriate immunotherapy (e.g., immune checkpoint blockade therapy), as will be explained in more detail below. If an individual is classified into the unfavorable prognosis group, they may be deselected from the aforementioned immunotherapy treatments and may receive, for example, surgical intervention, radiotherapy, and / or other forms of anticancer drugs (e.g., one or more non-immunochemotherapeutic agents or anti-angiogenic agents).

[0088] In some embodiments, the method is a computer implementation for predicting the therapeutic response of mammalian cancer patients to anti-cancer immunotherapy, the method comprising: a) providing gene expression data including gene expression levels of at least eight, at least nine, at least ten, at least eleven, at least twelve or more (e.g., all) of the following oncogenic genes previously measured in a sample obtained from a tumor of a mammalian cancer patient: PTGS2, CXCL1, CXCL2, CXCL5, CXCL6, IL1A, HAS2, IL11, IL1B, EREG, IL6, CXCL8 (IL8), MMP12, NRG1, VCAN, AREG; b) calculating a gene expression score (PTI score) for at least eight or more oncogenic genes provided in a); and c) predicting the therapeutic response of the mammalian cancer patient based on the PTI score calculated in step b). In some embodiments, the PTI score is calculated based on the gene expression levels of all 16 oncogenic genes.

[0089] In some embodiments, the computer implementation method further includes evaluating the tumor burden and / or the prevalence of neoantigens in cancer patients. In some embodiments, predicting the treatment response in mammalian cancer patients involves comparing a calculated PTI score to the median reference gene expression score from a responder sample cohort and the median reference gene expression score from a non-responder sample cohort, where the responder and non-responder sample cohorts are i) cancer patients having the same cancer type as the aforementioned cancer patient, or ii) cancer patients whose cancer is tumor type independent.

[0090] In some embodiments, mammalian cancer patients are selected for anti-cancer immunotherapy if their PTI score indicates that the cancer is predicted to respond to anti-cancer immunotherapy. In some embodiments, a method for predicting the therapeutic response of a mammalian cancer patient to anti-cancer immunotherapy, the method comprising: a) measuring the gene expression of at least eight oncogenic genes selected from PTGS2, CXCL1, CXCL2, CXCL5, CXCL6, IL1A, HAS2, IL11, IL1B, EREG, IL6, CXCL8 (IL8), MMP12, NRG1, VCAN, and AREG in a sample obtained from the patient's tumor; b) determining the PTI score of the gene expression of the at least eight oncogenic genes based on the gene expression scores measured in a); and c) predicting the patient's therapeutic response based on the gene expression levels determined in step b).

[0091] In some embodiments, determining a PTI score to predict a subject's treatment response includes the steps of inputting data representing the gene expression levels of at least eight cancer-promoting genes in the subject into a machine learning classifier, wherein the machine learning classifier is trained using a training dataset containing the gene expression levels of at least the same cancer-promoting genes that form the subject's PTI score, and the training dataset is derived from at least two sample cohorts; one cohort includes multiple samples from multiple mammalian cancer patients (responders) known to have responded to anti-cancer immunotherapy, and the other cohort includes multiple samples from mammalian cancer patients (non-responders) known not to have responded to anti-cancer immunotherapy; and causing the machine learning classifier to classify the subject's PTI score as belonging to either the responding group or the non-responding group. In some embodiments, the training dataset includes the gene expression levels of at least eight cancer-promoting genes from at least 50, at least 100, at least 150, at least 200, at least 250, and at least 300 samples from mammalian cancer patients known to have responded to anti-cancer immunotherapy (known as responders), and the gene expression levels of at least eight cancer-promoting genes from at least 50, at least 100, at least 150, at least 200, at least 250, at least 300, at least 350, at least 400, at least 450, at least 500, and at least 550 samples from mammalian cancer patients known not to have responded to anti-cancer immunotherapy (known as non-responders).

[0092] In some embodiments, the machine learning classifier is trained with multiple PTI scores from at least three sample cohorts, one cohort comprising multiple samples from multiple mammalian cancer patients known to have fully responded to anti-cancer immunotherapy (known as full responders), another cohort comprising multiple samples from multiple mammalian cancer patients known to have partially responded to anti-cancer immunotherapy (known as partial responders), and another cohort comprising multiple samples from mammalian cancer patients known to have not responded to anti-cancer immunotherapy (known as non-responders).

[0093] A method for classifying a sample containing nucleic acids and / or proteins from a mammalian cancer patient comprises: analyzing the sample to obtain the gene expression levels of at least eight oncogenic genes in the mammalian cancer patient sample; analyzing the gene expression levels of at least eight oncogenic genes to determine a PTI score; and analyzing the PTI score of the sample to classify the sample as matching a PTI score for an anti-cancer immunotherapy response score or an anti-cancer immunotherapy non-response score, wherein if the PTI score is classified as matching a PTI score for an anti-cancer immunotherapy response score, the sample is characterized as originating from an anti-cancer immunotherapy response subject; and if the PTI score is classified as matching a PTI score for an anti-cancer immunotherapy non-response score, the sample is characterized as originating from an anti-cancer immunotherapy non-response subject. Optionally, the sample may originate from one or more cancer cells from a mammalian cancer patient.

[0094] A system for predicting the therapeutic response of mammalian cancer patients to anti-cancer immunotherapy is also provided. The system comprises at least one processor and at least one non-transient computer-readable medium, the medium containing instructions that, when executed by at least one processor, cause at least one processor to predict the therapeutic response of mammalian cancer patients to anti-cancer immunotherapy and / or classify samples obtained from mammalian cancer subjects as PTI scores for responders or non-responders to anti-cancer immunotherapy. In some embodiments, the present invention comprises one or more computer-readable mediums containing instructions that, when executed by one or more processors, cause one or more processors to predict the therapeutic response of mammalian cancer patients to anti-cancer immunotherapy and / or classify samples obtained from mammalian cancer subjects as PTI scores for responders or non-responders to anti-cancer immunotherapy.

[0095] Whether a prognosis is considered good or bad may vary between cancers and stages. Generally speaking, a good prognosis is when overall survival (OS) and / or progression-free survival (PFS) are longer than the average for that stage and cancer type. A poor prognosis may be considered when PFS and / or OS are lower than the average for that stage and cancer type. The average may be the median OS or PFS of the survival rate. As used herein, the term “overall survival” (OS) may be interpreted as the length of time a patient diagnosed with a disease, such as cancer, is still alive from the date of diagnosis or the start of treatment for the disease. As used herein, the term “progression-free survival” (PFS) may be understood as the period a patient is alive without progression in the disease, such as cancer, during and after treatment for the disease.

[0096] In some embodiments, the machine learning classifier is a logistic regression model. For example, a PFS of more than 6 months and / or an OS of more than 18 months may be considered a good prognosis. Similarly, a PFS of less than 6 months or an OS of less than 18 months may be considered a poor prognosis. In particular, in advanced cancer, a PFS of more than 6 months and / or an OS of more than 18 months may be considered a good prognosis.

[0097] Generally speaking, a “good prognosis” means that an individual patient’s survival rate (OS and / or PFS) is favorable compared to the survival rate expected in a comparable disease population, such as one with similar disease stage and cancer type. This may be defined as being better than the median survival rate (i.e., a survival rate greater than 50% of patients in the population).

[0098] Generally speaking, a “poor prognosis” is when an individual patient’s survival rate (OS and / or PFS) is unfavorable compared to the survival rate expected in a comparable disease population, such as one with similar disease stage and cancer type. This may be defined as being worse than the median survival rate (i.e., a survival rate not exceeding that of 50% of patients in the population).

[0099] "Predicting a cancer patient's response to a selected treatment" is intended to mean assessing the likelihood that a patient will experience a positive or negative outcome with a particular treatment. As used herein, “positive treatment outcome” means that the patient is more likely to experience beneficial results from the treatment of choice, such as a reduction in tumor size, a “good” prognosis, or an improvement in disease-related symptoms and / or quality of life.

[0100] "Negative treatment outcome" is intended to mean that the patient is more likely to not enjoy the benefits of the aforementioned positive treatment outcomes (e.g., no reduction in tumor size, a "poor" prognosis, no improvement in disease-related symptoms and / or quality of life).

[0101] In some embodiments, a cancer patient's PTI score can predict the treatment response and / or prognosis of the cancer patient by referring to or comparing it to the median reference gene expression score obtained from a sample cohort of cancer patients having the same cancer type as the cancer patient (and optionally age-matched, time since diagnosis, and / or stage matched) ("tumor type-specific" prediction).

[0102] In other embodiments, the PTI score of a cancer patient may be compared to the median reference gene expression score obtained from a sample cohort of cancer patients with the same and different cancer types as the patient ("tumor type independent" prediction) (and optionally matched for age, time since diagnosis, and / or stage).

[0103] treatment A method for treating cancer in mammalian cancer patients may include the steps of: performing a method to predict the therapeutic response of a mammalian cancer patient to anti-cancer immunotherapy according to the method of the present invention; determining that the mammalian cancer patient is predicted to respond to the anti-cancer immunotherapy; and administering a therapeutically effective dose of the anti-cancer immunotherapy to a subject in need. A method for treating cancer in mammalian cancer patients may include the steps of: performing a method to predict the therapeutic response of a mammalian cancer patient to anti-cancer immunotherapy according to the method of the present invention; determining that the PTI score of the mammalian cancer patient indicates that the mammalian cancer patient is predicted to respond to the anti-cancer immunotherapy; and administering the immunotherapy to the mammalian cancer patient in need. Some methods for treating mammalian cancer patients may further include analyzing one or more additional markers of anti-cancer immunotherapy response derived from mammalian cancer patients to supplement and / or confirm predictions of anti-treatment response or non-response.

[0104] A pharmaceutical composition comprising an anti-cancer immunotherapy for use in the treatment of mammalian cancer patients is provided, wherein the patients are expected to respond to the anti-cancer immunotherapy. Mammalian cancer patients may have one of the following cancer types, or the cancer itself may be one of the following: renal cancer, e.g., renal cell carcinoma; gastroesophageal tumor; melanoma; bladder cancer; stomach cancer; lung cancer, e.g., lung adenocarcinoma (LUAD); pancreatic cancer (PAAD); liver cancer (LIHC); cervical cancer (CESC); head and neck cancer (HNSC); thyroid cancer (THCA); uterine cancer (UCEC); tenosynovial giant cell tumor (TGCT). Renal cell carcinoma may be clear cell renal cell carcinoma (ccRCC). Mammalian cancer may have one of the following immune subtypes, as defined by Thorsson et al. (2018), or may be one of the following: wound healing type, IFNγ dominant type, TGFβ dominant type, inflammatory type, and / or lymphocyte depletion type.

[0105] Anti-cancer immunotherapy may include IL-2 immunotherapy or immune checkpoint blockade therapy alone or in combination with anti-VEGF therapy. In some cases, anti-VEGF therapy may be bevacizumab. Immune checkpoint blockade therapy may include one of the following: programmed cell death receptor-1 (PD-1) blockade, programmed cell death ligand 1 (PD-L1) blockade, and / or cytotoxic T lymphocyte-associated protein 4 (CTLA-4) blockade. Immune checkpoint blockade therapy may include treatment with nivolumab (anti-PD-1 blocker), pembrolizumab (anti-PD-1 blocker), atezolizumab (anti-PD-L1 blocker), and / or ipilimumab (anti-CTLA-4 blocker).

[0106] The features disclosed in the foregoing description, or in the following claims, or in the accompanying drawings, can be appropriately expressed in their specific forms, or in relation to means for performing the disclosed functions, or as methods or processes for obtaining the disclosed results, and can be used separately or in any combination of those features to realize the present invention in various forms.

[0107] While the present invention has been described in conjunction with the exemplary embodiments described above, many equivalent modifications and variations will become apparent to those skilled in the art given this disclosure. Therefore, the exemplary embodiments of the present invention described above are considered examples and not limiting. Various modifications can be made to the described embodiments without departing from the spirit and scope of the invention.

[0108] To avoid any doubt, the theoretical explanations provided herein are provided for the purpose of enhancing the reader's understanding. The inventors do not intend to be bound by any of these theoretical explanations.

[0109] The section headings used herein are for structural purposes only and should not be construed as limiting the subject matter described herein. Throughout this Specification (including the subsequent claims), unless otherwise required by context, the words “comprise” and “include,” as well as variations such as “comprises,” “comprising,” and “including,” shall be understood to mean that a specified integer or step, or group of integers or steps, is included, but not that other integers or steps, or groups of integers or steps, are excluded.

[0110] It should be noted that the singular forms “a,” “an,” and “the” used in the specification and the appended claims include multiple references unless the context clearly requires otherwise. Herein, ranges may be expressed as “about” a particular value and / or “about” another particular value. Where such ranges are expressed, another embodiment includes starting from one particular value and / or up to the other particular value. Similarly, where a value is expressed as an approximation by the use of the antecedent “about,” it is understood that a particular value forms another embodiment. The term “about” in relation to a number is arbitrary and means, for example, ±10%.

[0111] The following are presented as examples only and do not limit the scope of the claims. [Examples]

[0112] material and method Data and code availability Transcriptome data from in vitro PGE2-treated mouse bone marrow macrophages will be made public through GEO (https: / / www.ncbi.nlm.nih.gov / geo / ). Appropriate code for performing the analysis is available at https: / / gitlab.com / christian.bromley / protumorigenic-inflammation.

[0113] statistical methods All in silico analyses were performed using R version 4.0.3. All statistical tests were performed using R, or in some cases within GraphPad Prism (7.0d). Unless otherwise noted, p-values ​​less than 0.05 (*), less than 0.01 (**), and less than 0.001 (***) were labeled. The Benjamini-Hochberg method was used to adjust p-values ​​in multiple hypothesis tests. These are referred to as FDR values ​​throughout this specification. When it was not appropriate to assume a normal distribution, two-sided nonparametric tests were used. χ² was used for comparing categorical variables. 2 We used statistical tests. For groups with more than two variables, we used the Kruskal-Wallis test with Dunn's multiple comparison test, including p-value adjustment for multiple hypothesis testing using FDR. For univariate survival analysis, we used Wald test p-values ​​and log-Lenck test p-values ​​to assess statistical significance. For multivariate survival analysis, we used the coxph function from the survival R package (3.1-12) to collect Wald test p-values ​​for individual covariates.

[0114] Differential gene expression and network analysis of TGCA Gene expression data (raw data count format) from the TCGA (Cancer Genome Atlas) was downloaded from the SAGE Synapse Database (syn4311114) along with clinical data (syn12026747). Raw data counts were TMM normalized using calcNormFactors in edgeR (3.24.3). Genes that did not reach 0.25 count per million (CPM) in 10% of the samples were excluded from the analysis. Patients belonging to individual TCGA studies were stratified into high and low PTGS2 subgroups using the upper and lower quartiles of log2 normalized expression levels. Non-solid tumor types, such as LAML (Acute Myeloid Leukemia) and DLBC (Diffuse Large B-Cell Lymphoma), were excluded from subsequent analyses. A model matrix was designed to compare patients with high and low PTGS2 while adjusting for cancer type. This was done using the top 10 cancer types with the highest median PTGS2 values. DGE analysis was performed using limma-trend (Ritchie et al., 2015). Upregulated genes (P<0.05, FC>1.2) were filtered based on a ligand list (Ramilowski et al., 2015), and the resulting 16 genes were used as the final PTI signatures (Table 1).

[0115] In gene set enrichment analysis (GSEA), directional log 10 Genes were ranked based on their p-values. The ranks were used as input to the fgsea(1.8.0) package (500 repeats). Hallmark gene signatures were downloaded from the molecular signature database (https: / / www.gsea-msigdb.org / gsea / msigdb / ). Genes upregulated by DGE analysis were imported into the STRING database (https: / / string-db.org) to construct protein-protein interaction networks. Nodes without any edges were removed. A table containing significantly enriched gene ontology biological processes was exported from STRING. The networks were also exported and analyzed using the igraph package in R (1.2.7).

[0116] TCGA survival analysis The signature score of the target gene signature was calculated from the mean expression level of the signature gene using Z-score normalized expression data from all cases. Pairwise Spearman correlation coefficients were calculated using the rcorr function of the Hmisc package (4.3.0). Survival analysis was performed in R using the survival(3.1-12) package, and KM graphs were plotted using survminer(0.4.6). For survival analysis, OS days and OS status information from the complete TCGA clinical data table downloaded from SAGE synapse (syn12026747) was utilized. In the overall cancer survival analysis, stratification by interquartile or median within each overall cancer immune subtype was used in addition to continuous PTI scores (Thorsson et al., 2018). HR was adjusted for cancer type and stage as stratification factors in the CoxPH model. To obtain detailed immunological and genomic annotations for TCGA patients, we used the supplemental file (Supp1a.txt) by Thorsson et al. (2018). For cancer type-specific survival analysis, we used continuous signature scores from a multivariate CoxPH model adjusted for available clinical parameters. Patient disease stages were converted to continuous variables (e.g., stages Ia and Ib = 1). For TMB estimation in the TCGA data, we used the "non-silent mutation rate" annotated by Thorsson et al. (2018). For HNSCs, we extracted patients' final HPV status from supplemental data of the TCGA study (Lawrence et al., 2015).

[0117] Analysis of PGE2-treated cells Mouse BMDM was generated as previously reported (Zelenay et al., 2015). Macrophages were treated with synthetic PGE2 for 24 hours, after which RNA was extracted and sequenced.

[0118] Reads obtained from RNA sequencing experiments were quality-checked using FastQC (http: / / www.bioinformatics.bbsrc.ac.uk / projects / fastqc / ), and base trimming was performed. Next, using STAR (Dobin et al., 2013), the reads were aligned to the mouse genome (mm10) with default parameters. Then, using the featureCounts function in the R package Rsubread (Liao, Smyth and Shi, 2014), gene-level count data was generated using the Ensembl GTF annotation (Mus_musculus.GRCm38.75.gtf). The DESeq2 (1.24.0) package was used for DGE analysis, and genes satisfying FDR<0.05 and 1<log2FC<-1 were defined as DEGs. Public microarray datasets from the Gene Expression Omnibus (GEO) were downloaded for PGE2-treated human fibroblasts (GSE63659) and CD4 + T cells (GSE17200). Using the limma package, DGE analysis was performed by fitting an array-weighted linear model. Significant DEGs were defined as FDR<0.05 and 1<log2FC<-1.

[0119] TCGA Mutation Analysis The whole cancer mutation annotation file (mc3.v0.2.8.PUBLIC.maf.gz) was downloaded from SAGE Synapse (syn7214402). The R package maftools (2.2.10) was used for plotting and data exploration. Mutation data was filtered to exclude non-solid tumor types and variants classified as RNA, silent, 3' flank, or 5' flank. The total number of mutations was calculated for each patient sample and log2 normalized. Multivariate linear regression analysis using the following formula: lm(PTI ~ mutations + log2 - total number of mutations + cancer type) was used to identify all cancer predictors of PTI expression. For the selection of known cancer-driving genes (Lawrence et al., 2014), the same formula was applied to individual cancer types. P values ​​from the linear model were adjusted using the Benjamini-Hochberg method. For the KRAS gene, TCGA mutations, structural variants, and copy number estimates for all cancer types were downloaded from cBioPortal (https: / / www.cbioportal.org).

[0120] CCLE PTI expression analysis Gene expression data (CCLE_RNAseq_rsem_genes_tpm_20180929.txt) and cell line annotations (Cell_lines_annotations_20181226.txt) were downloaded from the Cancer Cell Line Encyclopedia (CCLE) (https: / / portals.broadinstitute.org / ccle / data). Transcript IDs were converted to gene symbols using the EnsDb.Hsapiens.v79 and ensembldb R packages. Cell lines named "FIBROBLAST" or "HAEM" were excluded. TPM values ​​were converted to a log2 scale using a pseudo-count of 1. Pairwise Spearman correlation coefficients between PTI genes were calculated, and the mean Spearman coefficient was calculated for the mean pairwise correlation coefficient of PTI genes, and separately for 1000 random gene signatures (n=16) of the same length as the PTI signatures. MMP12 was not present in the CCLE dataset. To compare cell lines with high and low cancer-specific PTGS2 expression, DGE analysis using edgeR(3.28.1) was performed to compare the upper and lower quartile groups. Only genes with more than 1 CPM in 10% of the samples were included in the analysis. Cancer types, defined by the TCGA code variable in the cell line annotation file, were adjusted in the modeling. Cells satisfying adjusted p-values ​​<0.05 and log2FC>2 were defined as upregulated DEGs. Using a pipeline similar to that for PTGS2, PTI was performed. high Cell lines and PTI low Cell lines were compared. Long non-coding RNAs and pseudogenes were excluded from genes that showed high upregulation in PTI comparison (FC>5, FDR<0.05). Motif enrichment analysis was performed using the Rcistarget(1.6.0) package with the final list of highly expressed genes (n=500). To search for conserved motifs in the 500 base pairs immediately upstream of the transcription start site (TSS) of the target gene, the motif ranking file hg19-500bp-upstream-10species.mc9nr.feather downloaded from https: / / resources.aertslab.org / cistarget / was used.

[0121] CCLE mutation and somatic cell copy number analysis Mutation calls (CCLE_DepMap_18q3_maf_20180718.txt) and gene-specific copy number estimates (CCLE_copynumber_byGene_2013-12-03.txt) were downloaded from the DepMap portal (https: / / depmap.org / portal / download), and information on cell line aneuploidy and chromosomal arm-level copy number was extracted from Supplementary Table 1 of a recent study (Cohen-Sharir et al., 2021). For the cell line mutation analysis, the following variant classes were excluded: RNA, silent, 3' flank, and 5' flank, and the analysis was filtered to include 568 publicly available cancer-driving genes (Martinez-Jimenez et al., 2020). Signature scores for PTI signatures were modeled by multivariate linear regression using driving gene mutation status, mutation rate, and cancer type as covariates. For cancer type, each level of the categorical variable in the TCGA code was converted to a series of binary variables (e.g., HNSC=1, Other=0). maftools(2.2.10) was used to plot individual mutations in the KRAS gene (RefSeq ID:NM_004985). For gene-level SCNA estimates, a multivariate logistic regression model was fitted to predict high and low cancer cell-specific PTI scores using gene copy number, aneuploidy, and cancer type as covariates. A significant predictor was defined as FDR<0.25.

[0122] CCLE RPPA Data Analysis RPPA data (CCLE_RPPA_20181003.csv) was downloaded from the CCLE database. Filtering was performed to include only cell lines with KRAS mutations. Next, the limma package was used to perform PTI analysis. high (n=44) Cell lines and PTI low(n=18) Cell lines were compared. Upregulated proteins and phosphorylated proteins were defined as FC>1.5 and FDR<0.05, while downregulated proteins were defined as FC<-1.5 and FDR<0.05.

[0123] Drug screening data analysis We downloaded an R data file (E-MTAB-783.eSet.r) containing gene expression data for cell lines screened by Garnett et al. (2012). We normalized the microarray data using the rma function from the affy(1.62.0) package and converted the IDs (affy_hg_u133a) to gene symbols using biomaRt(2.40.5). For observations mapped to single genes, we selected the row with the highest variance. Cancer cell-specific PTI signature scores for individual cell lines were calculated similarly to CCLE. The top 25% of cell lines with the highest expression were selected for PTI. high This was defined as follows. Drug screening data for cell lines were downloaded from the supplemental files of Garnett et al. (2012). Similar to Jerby-Arnon et al. (2018), susceptible cell lines were defined as the top 25% with the lowest IC50 for each drug. Cell line susceptibility to drugs was modeled as a function of the cell line's PTI status (high / low), adjusting for cancer type as a covariate, within a linear mixed-effects model using the lme4 package (1.1-23). ​​The following equation was used: glmer(sensitivity~PTIstatus+(1+cancertype),family=binomial(link=”logit”),control=glmerControl(optimizer=”Nelder_Mead”,optCtrl=list(maxfun=20000))) PTI was used. highDrugs that preferentially kill cell lines were predicted by filtering models in which PTI status is a key predictor of cell line susceptibility (coefficient > 0 and P < 0.05). The IC25, IC50, IC75, and IC90 concentrations of the drug refametinib against the target cell line were extracted from supplemental data of Garnett et al. (2012).

[0124] TCGA Immuno-Deconvolution The MCPcounter (Becht et al., 2016) immune deconvolution algorithm was applied to a TCGA expression dataset encompassing all cancer types. The resulting immune cell scores were correlated with PTI signatures within individual cancer type datasets, and the Spearman correlation coefficients for the PTI signatures were plotted using corrplot(0.9).

[0125] Single-cell RNA sequencing data analysis The entire analysis of scRNAseq data was performed using the Seurat package (3.1.5). Unnecessary variability factors were removed using the sctransform package (0.2.1) and the SCTransform function. Signature scores for individual cells were calculated using the AddModuleScore function. The original cluster annotations were used whenever possible. The FindMarkers function was used for inter-cluster or inter-group DGE analysis, setting a 25% expression threshold across all cells in the target cluster. To plot the expression of individual PTI genes in specific clusters, the mean log2-normalized TPM or CPM expression levels across each cluster were calculated. Hierarchical clustering was performed on the mean expression matrix to cluster PTI genes by cell type-specific expression.

[0126] Lambrechts et al. (2018) Non-small cell lung cancer (NSCLC): R data files containing log2CPM data and t-SNE coordinates were downloaded from ArrayExpress using access code E-MTAB-6149. Total count per cell and mitochondrial content were regression-removed using sctransform(0.2.1). Clusters from the original publication were re-annotated into a broad range of cell types. Dendritic cells were a combination of cells from annotated clusters, including "cross-presenting dendritic cells (14)", "Langerhans cells (3)", "plasmacytoid dendritic cells (24)", and "monocyte-derived dendritic cells (19)". Clusters containing "fibroblasts" were grouped as fibroblasts, and similar principles were applied to endothelial cells and cancer cells. Cell types specific to the lung were excluded from the analysis. Patient 4 (PTI) high ) Tumor-derived CD8 + The cells in the T cell cluster were compared to all other tumor-derived cells.

[0127] HNSC by Puram et al. (2017): Expression data and annotations were obtained using data collected in recent studies and downloaded as R data files from Zenodo (https: / / zenodo.org / record / 3260758 / files / hnscc_expression.rds). Lymph node cell samples (HN10, HN12, HN13, HN24, HN7, HN8, HN23) were excluded. The same cell filtering procedure as the original study (Puram et al., 2017) was employed. Variables representing cell counts and the enzyme used for reverse transcription were regression-removed using sctransform. Six PCs were used for graph-based clustering, and the resulting clusters were annotated using marker genes and annotations from the original study. CD8 + T cells were compared between tumors with high and low cancer cell-specific PTI expression, and the FindMarkers function was used for DGE analysis.

[0128] Jerby-Arnon et al. (2018) Melanoma: Gene expression data (GSE115978_tpm.csv) and cell annotations (GSE115978_cell.annotations.csv) were downloaded from GEO (GSE115978). Cells with at least 500 genes were retained, and genes expressed in two or more cells passed the filtering. Twenty PCs were selected for graph-based clustering. Twenty-five novel clusters were defined and annotated using known marker genes and annotations from the original study. Clusters were grouped into broad cell categories to reflect classifications from other scRNAseq datasets. All T cells were subclustered and CD4 + T cells, naive-like T cells and CD8 + T cells were defined. This allowed for the distinction between tumors with high and low cancer cell-specific PTI expression and CD8 + It has become possible to compare T cells.

[0129] TCGA CIBERSORT and CD8 + T cell signature analysis Using a TCGA expression dataset of all solid tumors, the inventors first selected so-called "hot" tumors by filtering to include only patients with "IFNγ-dominant" immune subtype tumors (Thorsson et al., 2018). Next, they selected PTI signatures and dysfunctional signatures (Singer et al., 2016), and tolerant autoantigen-specific CD8 + CD8, which includes the signatures for anergy (Azizi et al., 2018) and depletion (Guo et al., 2018), is a gene set derived from early T cell research (Schietinger et al., 2012). + Several signatures related to the T cell phenotype were calculated. CD8 +Spearman correlation coefficients were calculated between T cell phenotypic signatures and intratumor PTI scores. Using CIBERSORT data from Thorsson et al. (2018), the proportions of various immune populations among total immune cells in tumors with high, moderate, and low PTI were compared. Nonparametric t-tests with p-values ​​adjusted for multiple hypothesis testing were used to compare the mean proportions of each individual population between tumors with high and low PTI.

[0130] Analysis of immune checkpoint blocker datasets Transcriptome data from patients treated with ICB were matched from 12 publications. For datasets requiring normalization from counts to TPM values, genes showing expression above 0.25 CPM in less than 10% of samples across the entire cohort were first excluded. For three datasets (Kim et al., 2018; Mcdermott et al., 2018; Gide et al., 2019), FASTQ files were downloaded. RNAseq reads were quality-checked and base-trimmed using FastQC software (http: / / www.bioinformatics.bbsrc.ac.uk / projects / fastqc / ). Next, the reads were aligned to the human genome assembly (Homo_sapiens.GRCh38) using the STAR aligner (Dobin et al., 2013) with default parameter settings. The mapped data was converted to transcript-level integer read counts using featureCounts (Liao, Smyth, and Shi, 2014) and Ensemble GTF annotation (Homo_sapiens.GRCh38.85.gtf). The transcript counts were then converted to TPM.

[0131] For each cohort, patients were stratified into high and low signature score groups based on the median cutoff value. Next, the χ² was calculated between the stratified patient groups. 2The proportions of responders (PR / CR) and non-responders (PD) were compared using statistical tests. Patients with SD were excluded from this analysis. To model patient response to ICB in each dataset, a multivariate logistic regression model was fitted using only the TIS signature as the baseline predictive model. Each resistance signature was combined with the TIS signature in a multivariate logistic regression model that included a two-variable linear multiplicative interaction term within the model (e.g., glm(response~TIS+PTI+TIS:PTI)). The AUC value for each model was calculated using the pROC(1.17.0.1)R package, and the AUCs across the 16 datasets were compared using nonparametric t-tests. In addition, OS was compared for patients in each quadrant using baseline pre-ICB datasets for which OS data was available. A multivariate Cox regression model was fitted to this entire cancer immunotherapy metadata set, including the dataset and TIS status as covariates. The same analysis was performed on 1000 random gene signatures and publicly available ICB resistance signatures. The Wald test p-values ​​for each resistance signature / random signature obtained from each multivariate CoxPH model were adjusted using the Benjamin-Hochberg method.

[0132] Analysis of the developmental stages of squamous cell carcinoma of the lung Microarray data from different developmental stages of LUSCs were downloaded from the GEO database (GSE33479) (Mascaux et al., 2019). When multiple probes mapped to the same gene, the probe with the greatest variance across the entire sample cohort was selected. PTI signature scores were calculated in the same manner as in TCGA analysis.

[0133] In vitro assay using lung cancer cell lines Human lung cancer cell lines NCIH1339 and NCIH3255 were cultured in RPMI 1640 medium supplemented with L-glutamine (Lonza:LZBE12-702F), 10% fetal bovine serum (Merck:F7524), and 5% penicillin / streptomycin (Thermo:11548876) at 37°C in Nunc T75 cell culture flasks (Life Cells were cultured in (Technologies: 156499). Cells were washed with Dulbecco's phosphate-buffered saline (Fisher: 12559069), suspended in trypsin-EDTA, and seeded at 1 × 10⁵ cells per well in 12-well plates and 2 × 10⁴ cells per well in 96-well flat-bottom plates (Starlab: CC7682-7596). After 24 hours, the medium was removed from the plates and fresh medium containing 0.1% dimethyl sulfoxide was used as a control, or 1 μM or 10 μM trametinib or cereal. The culture medium was replaced with lamtinib-containing medium. Supernatants were then collected at 1, 2, 4, 6, 8, and 24 hours and stored at -20°C. Cells were suspended in trypsin-EDTA and centrifuged at 1500 rpm for 5 minutes. The supernatant was discarded, and the cell pellet was stored at -80°C. To monitor and estimate cell proliferation, drug-treated 96-well plates were imaged at 2-hour intervals for 3 days using an Incucyte instrument. Confluence rate and change in confluence relative to baseline were used as surrogate markers of cell proliferation.

[0134] The Luminex assay protocol was used to measure soluble cytokines and chemokines in cell culture supernatant. A 96-well flat-bottom Luminex plate was washed with purified H2O. Lyophilized protein standards were diluted to 50 μl in cell culture medium, centrifuged at 15000 rpm for 10 minutes, and then left on ice for 10 minutes. Protein standards were diluted 4-fold in cell culture medium. Beads were diluted 1:100 with assay wash buffer and 50 μl was added to each well. After holding the plate on a magnet for 2 minutes, the liquid was removed by flicking, and 50 μl of sample / standard diluted 1:2 with medium was plated. After shaking the plate for 2 hours, it was washed three times with assay wash buffer. Antibodies were diluted 1:200 with detection antibody diluent and 25 μl per well was plated. After shaking the plate at room temperature for 30 minutes, it was washed three more times with wash buffer. 30 μl of streptavidin-PE was added to each well, and the plate was left on a shaker for 30 minutes. Next, the plates were washed four more times with washing buffer, and 75 μl of measurement buffer was added to each well. The plates were read using a Luminex MAGPIX instrument. Protein concentrations were extrapolated from standard curves fitted by plotting expected concentrations against net MFI levels in GraphPad Prism v8.

[0135] Analysis of the CPI1000 dataset The inventors were granted access to individual PTI gene expression, clinical, and genomic biomarker data from the CPI1000 dataset (Litchfield et al., 2021). Expression data was available for a total of 523 patients. For each gene, the Log2-normalized TPM value (log2(TPM+1)) was Z-score normalized, and PTI signatures were calculated using conventional methods. Due to missing MMP12 data, 15 of the 16 PTI genes were available. Continuous variables were cocorrelated, and Spearman Rho values ​​were used for hierarchical clustering using the default parameters in the corrplot function of the corrplot package (0.84). SD patients and samples undergoing treatment were excluded again. For each biomarker, univariate logistic regression models were fitted to predict PD versus PR / CR patients. Effect sizes, standard errors, and Wilcoxon test p-values ​​were extracted from six studies, and meta-analysis was performed using the metagen function of the R meta package (4.18-0). To compare the combined effects of PTI measurement with other biomarkers, a multivariate logistic regression model was fitted with four variables (tumor type, clonal TMB, CXCL9, and PTI), including a linear multiplicative interaction term for the numerical variables. Nested models were compared using likelihood ratio tests. The variance of patient responses explained by each model was estimated by calculating the deviation using the anova function. In the survival analysis on the CPI1000 dataset, patients were stratified based on the 75% cutoff value of PTI and the median values ​​of clonal TMB, CXCL9, and CD274. To determine the optimal combination of individual PTI genes for ICB response prediction on the CPI1000 dataset, multiple elastic net regression models were trained using the glmnet R package. Two scaled expression matrices were created using log2-normalized TPM values ​​from the CPI1000 dataset. The first scaled the data for all samples (tumor type independent), and the second scaled after grouping by cancer type (tumor type specific). Patients with SD and patients with samples undergoing treatment were excluded. Each matrix was split into 70% training data and 30% test data.Using 10-fold cross-validation, we trained an elastic network regression model that optimizes the λ hyperparameter while increasing the α hyperparameter value between 0 and 1 in increments of 0.1. For each trained model, we selected the optimal model using AUC as the metric. Next, we used each trained model to predict patients' ICB responses in the test dataset.

[0136] Manchester ccRCC cohort patients Ethical approval was obtained from the Manchester Cancer Research Centre Biobank Research Tissue Bank Ethics Committee (reference number: 18 / NW / 0092) for retrospective analysis of tumor FFPE and plasma from RCC patients. Forty-eight surgical blocks were identified from patients who received immunotherapy (including high-dose IL-2 treatment) at some stage of the treatment process. Pathological evaluation of H&E stained sections from three tumors revealed no tumor tissue. Two blocks came from the same patient who underwent two brain metastasis resection surgeries. In another patient, separate blocks were obtained from adrenal metastasis and lung metastasis resection surgeries. Pathologists marked tumor and normal tissue areas on H&E sections and defined histological grade and morphology.

[0137] Formalin-fixed paraffin-embedded (FFPE) tissue treatment, RNA extraction, and gene expression profiling. FFPE surgical excision tissue is the most clinically available specimen and can be used, for example, as input material for molecular inflammation profiling using the NanoString nCounter platform. H&E (hematoxylin and eosin) stained FFPE ccRCC blocks were sectioned and placed on glass slides. After macro-dissection to remove most necrotic or normal tissue, thrombi, cartilage, and other unwanted tissues, the samples were scraped into 1.5 mL Eppendorf tubes. On average, 15 × 5 μm tissue sections were used. Smaller samples that did not require macro-dissection were cut directly from the block into 1.5 mL Eppendorf tubes. RNA was extracted from up to 10 slides using a commercially available Qiagen RNeasy FFPE kit (RNeasy FFPE kit for RNA extraction; RSP-54) (Figure 17). First, the total RNA concentration was determined using a NanoDrop instrument (Thermo Fisher). RBA quantitative measurements were also performed using a Qubit 2.0 fluorometer (Invitrogen), and quality assessments were calculated using the Bioanalyser (Agilent) platform. The RNA RIN values ​​ranged from 1 to 6.4, with an average value of 2.2. Next, the extracted RNA was used as input material for a molecular inflammation profiling system (e.g., the Nanostring nCounter platform) compatible with samples exhibiting low RNA quality and high fragmentation (e.g., FFPE tissue).

[0138] Nanostring gene expression analysis Gene expression profiling was performed using the Nanostring nCounter protocol and a commercially available Pan-Immune coding set including 30 additional custom genes. In the nCounter workflow, appropriate housekeeper genes were selected and added to the PTI gene coding set. To quantify gene expression, 150 ng of total RNA was hybridized overnight at 65°C to capture probes and reporter probes specific to the PTI and housekeeper genes. The reaction system, containing the hybridized samples and master mix reagents, was then supplied to a Nanostring nCounter preparation station where the samples were positioned for imaging. The samples were then scanned using an nCounter digital analyzer, and the fluorescent barcodes were counted. This method quantifies target RNA transcripts by directly measuring the fluorescence signal from the reporter-capture probe complex held on the cartridge surface. Nanostring output data (RCC files) were analyzed in R using the NanoStringQC function within the nanostringr package (0.1.4.9000). This analysis included exporting the obtained gene expression counts to NanoString nSolver 4.0 software for quality control analysis and normalization (for internally possible controls and / or housekeeper genes). NanoString nSolver 4.0 software allows for rapid and intuitive analysis and correction of batch effects using a reference signal without requiring advanced computational skills. This analysis identified samples that did not meet the required data quality thresholds (signal-to-noise ratio > 50, detected gene percentage > 50%). A total of 3 samples failed the quality control step due to low signal-to-noise ratios. We followed the published pipeline for normalizing NanoString count data (Bhattacharya et al., 2020). First, we used the MASS package (7.3-53) to fit a negative binomial model to each housekeeper gene expression with a binary variable (primary renal cancer vs. other) as the predictor variable. Nine housekeeper genes were excluded based on the criterion P < 0.05.Next, the RUVseq package (1.24.0) was used to identify and remove unwanted sources of variation with parameter k=1. The Limma (3.46.0) and DESeq2 (1.30.0) packages were used for DGE analysis. Batch-normalized expression data were used for downstream correlation analysis. Gene signature scores were calculated as the average expression levels of Z-score normalized and batch-corrected expression data.

[0139] Example 1: PTI: Pro-tumorogenic inflammatory signature centered on human-derived COX-2 Given the vast amount of literature suggesting that the COX-2 / PGE2 pathway promotes hallmarks in multiple cancers (including our own previous research establishing this inflammatory axis as central to immune evasion (Zelenay et al., 2015; Bonavita et al., 2020; Pelly et al., 2021)), we hypothesized that transcripts associated with high COX-2 mRNA levels capture conserved features across cancers in the pro-tumorogenic inflammatory signaling network.

[0140] Mining TCGA data revealed heterogeneity in the transcription levels of PTGS2, which encodes COX-2, both across and within cancer types (Figure 1A). As a result, the inventors devised a computational pipeline to compare high and low PTGS2 levels in the top 10 cancer types expressing PTGS2 (Figures 1A and 1B). Among the 54 genes overexpressed in PTGS2-highly expressing cancers were classical secretory factors involved in local and systemic inflammation, such as IL6, CXCL8, IL1B, and CXCL2; cell surface receptors (CSF3R, IL7R, CD69); neutrophil markers (G0S2); growth factors such as EGFR ligands, AREG, and EREG; and ECM modifiers HAS2 and MMP12 (Figure 1B). Gene set enrichment analysis (GSEA) using Hallmark pathway signatures revealed that COX-2 highIn cancer, the TNF-NFκB pathway and other inflammatory pathways (IL6-STAT3) were found to be upregulated (Figure 8A). Interestingly, the pathways of epithelial-mesenchymal transition (EMT), complement, hypoxia, KRAS signaling, and apoptosis were also overexpressed (FDR<0.05, NES>1) (Figure 8A). Therefore, COX-2 expression is a relevant cancer node for inflammatory tumors exhibiting several other cancer hallmarks. Next, genes encoding extracellular molecules, including secretory factors, were considered to be strong candidate regulators for cellular and molecular immune TME profiles. Ptgs2 high Genes upregulated in tumors were filtered by a list of “predictive ligands” (Ramilowski et al., 2015). The resulting genes were cytokines (IL1A, IL1B, IL6), chemokines (CXCL1, CXCL2, CXCL5, CXCL6, CXCL8), and growth factors (NRG1). These, along with PTGS2, were named the 15 gene signatures of pro-tumorogenic inflammation (PTI). In human cancer cell lines with high PTGS2 expression, the expression of 13 of the 15 evaluable PTI genes was significantly elevated (Figure 8B). Two of the most abundant non-cancerous cell types that make up the tumor mesenter are fibroblasts and macrophages. To investigate whether PGE2 signaling to fibroblasts and macrophages can explain the association of these ligands with COX-2 in tumors, transcriptome data of cells treated with synthetic PGE2 in vitro were analyzed. Notably, each ligand was regulated by PGE2 in either fibroblasts or macrophages (Figure 1D). Human CD4 + Treatment of T cells with PGE2 also induced numerous ligands (Figure 1D). These findings validate our bioinformatics pipeline and suggest that ligand PTI signatures represent hierarchically downstream nodes on the COX-2 / PGE2 axis, functioning as "hubs" within the network.

[0141] COX-2 highTo investigate whether genes upregulated in tumors are members of biologically meaningful aggregated signaling cascades, a protein-protein interaction network was constructed using a curated STRING database (Szklarczyk et al., 2019) (Figure 1C). Consistent with a common involvement in the signaling pathway, COX-2 high Proteins encoded by genes upregulated in tumors showed significantly more interactions than would have been expected by chance (P<1E-16), with an average of 9.7 edges per node. These proteins were overexpressed in biological processes of the gene ontology, including inflammatory responses, myeloid chemotaxis, cell death, and stress responses (Figure 1D). This data highlights how the genes identified in the pipeline are involved in co-processing and signaling cascades with COX-2, which is central to the network.

[0142] Example 2: PTI signature as a strong indicator of poor prognosis Next, we investigated whether the PTI signature was associated with poor prognosis. Using cancer genome atlas (TCGA) data, we performed all-cancer Cox proportional hazards (CoxPH) regression analysis, using disease stage and cancer type as stratification. The PTI signature was considered both as a continuous variable (HR=1.23, P=1E-05) and as stratified by quartiles (PTI). high vs PTI low Even with an HR of 1.25 and a p-value of 0.02, it was strongly correlated with a shortened survival time for cancer patients (Figure 1E).

[0143] Example 3: PTI signature is a strong indicator of poor prognosis across cancer immune subtypes. The interaction between PTI and the signature of cancer-inhibitory inflammation in predicting overall survival (OS) in cancer patients was studied using a multivariate CoxPH model including an interaction term (Ayers et al., 2017; Danaher et al., 2017). By including a linear multiplicative interaction term (e.g., product of PTI and TIS signatures) in the model, we estimated whether the effect of PTI on survival depends on the degree of IFNγ signaling and T-cell inflammation. As expected, PTI was generally associated with increased mortality (HR=1.25, P<0.0001) (Figure 2A). Therefore, for a particular patient with mean levels of IFNγ signaling, a one-unit increase in PTI corresponds to a 25% increase in risk. In contrast, for patients with mean levels of PTI, a scalar increase in IFNγ signaling corresponds to a 16% reduction in mortality risk (HR=0.84, P<0.0001) (Figure 2A). A significant interaction term was found between PTI and IFNγ signature (HR=1.13, P=0.004). To clarify this interaction, hazard ratios were simulated for high and low levels of IFNγ signaling (Figure 2B). When IFNγ signaling levels were low, the hazard ratio consistently increased regardless of PTI level, but in patients with high IFNγ signaling levels, the mortality risk was low and increased significantly with PTI expression (Figure 2B). In fact, when PTI levels were high (e.g., PTI score = 1.5, Figure 2B), patients with high IFNγ signaling in TME (which is usually a marker indicating a good tumor prognosis) had a mortality risk similar to patients with non-T cell inflammatory tumors. Thus, high intratumoral IFNγ signaling correlates with a favorable outcome in patients with simultaneously low PTI levels. PTI and CD8 + Similar interaction effects were observed between T cells and TIS signatures (Table 2). Taken together, these findings reveal a close interaction between pro-tumorogenic and anti-tumorogenic inflammatory phenotypes, highlighting the potent antagonistic effect of PTI signatures against T-cell inflammatory tumors.

[0144] [Table 2]

[0145] While immune infiltration and TME vary significantly depending on tumor type, previous whole-cancer analyses have defined six broad immune subtypes: "wound healing (C1)," "IFNγ dominant (C2)," "inflammatory (C3)," "lymphocyte depletion (C4)," "immunologically quiescent (C5)," and "TGFβ dominant (C6)" (Thorsson et al., 2018). These subtypes were found to possess unique genomic, transcriptome, and clinical characteristics (Thorsson et al., 2018). It was hypothesized that PTI maintains prognostic predictive ability and antagonistic effects against antitumor immunity under diverse immunological conditions. PTI was highest in the "wound healing," "IFNγ dominant," and "TGFβ dominant" subtypes (Figure 9A), and significantly predicted prognosis in all three subtypes, as well as the "inflammatory" and "lymphocyte depletion" subtypes (Figure 2C). Meanwhile, TIS and CD8... + T cell and IFNγ signaling signatures were more heterogeneous, being highest in the "IFNγ-dominant" subtype (Figure 8B). Furthermore, these biomarkers of so-called "hot" T-cell inflammatory tumors, as well as antigenic biomarkers such as the non-silent mutation rate, here referred to as tumor mutational burden (TMB), and single-nucleotide variant neoantigen burden (SNV NAB), were inconsistent across subtypes and less frequently indicated prognosis (Figure 2C) (Thorsson et al., 2018). Thus, while PTI may be a prognostic indicator across tumor immune subtypes, which as a whole have very contrasting characteristics, typical markers of so-called "hot" tumors, such as IFNγ signaling, correlated with extended OS in a single immune subtype.

[0146] Example 4: The PTI signature is a strong indicator of poor prognosis across all IFNγ-dominant subtypes. To extend this analysis, the inventors investigated the case where T cells have a highly inflamed tumor, as is common in patients treated with immunotherapy, and CD8 +We focused on the "IFNγ-dominant" subtype, which included patients in whom T cells showed a clear association with favorable outcomes. In these highly invasive tumors, PTI was strongly associated with low survival rates, whether high or low IFNγ levels (Figure 2D). In fact, PTI high IFNγ high While the median OS for patients was approximately 5 years, PTI low IFNγ high The group uses IFNγ and CD8 + Despite similarly high levels of T cell infiltration, the median OS exceeded 15 years (Figure 2D). These results suggest that the benefits of high levels of IFNγ signaling are strongly antagonized by coexisting PTI expression. Similar findings were observed when the IFNγ signaling signature was replaced with various established immunosignatures indicating T cell inflammation (TIS, IFNγ, CD8, PDL1) or the percentage of tumor-infiltrating lymphocytes (TILs) and PD-L1 (CD274) mRNA expression determined by immunohistochemistry (Thorsson et al., 2018). All of these have been reported as indicators of highly immunogenic and ICB-responsive tumors (Figure 9B). Therefore, COX-2-centered PTI is antagonistic to a variety of indicators of more immunogenic tumors.

[0147] To account for the potential confounding effects of organ-specific immune characteristics (Salmon et al., 2019), we used a multivariate CoxPH model to investigate the interaction of PTI and IFNγ signaling in individual cancer types. In nine cancer types, after adjusting for clinical factors such as age, sex, and stage, the PTI signature was found to be a statistically significant prognostic factor, either as a standalone variable or through its interaction with IFNγ signaling (Figure 2E (ccRCC (n=531), LIHC (n=374), PAAD (n=176), glioblastoma (GBM, n=168), CESC (n=306), adrenocortical carcinoma (ACC, n=79), UCEC (n=277), THCA (n=513), HNSC (n=240)). Increased PTI was associated with pancreatic cancer (PAAD), In ccRCC, lung adenocarcinoma (LUAD), liver cancer (LIHC), and cervical cancer (CESC), PTI was associated with increased mortality regardless of the level of IFNγ signaling (Figure 2E). In thyroid cancer (THCA), head and neck cancer (HNSC), and uterine cancer (UCEC), a significant interaction between PTI and IFNγ signaling was observed, mimicking observations in the overall cancer dataset (Figure 2E). In conclusion, the prognostic ability of PTI extends to individual malignancies, and, most importantly, is frequently associated with its inhibitory effect on IFNγ signaling.

[0148] Example 5: Prognostic capacity of individual PTI genes within a PTI signature By performing survival analyses of individual PTI genes, we evaluated whether there were any dominant factors explaining the prognostic ability of PTI signatures. The COX-2 gene (PTGS2) itself showed a low survival rate in ccRCC patients (Figure 9C). The EGFR ligands EREG and AREG were associated with low survival rates in cancers prevalent with KRAS mutations, such as PAAD, or in cancers prevalent with abnormalities in the receptor tyrosine kinase-RAS pathway, such as GBM and THCA. The cytokines IL11 and IL1A, as well as the chemokines CXCL5 and CXCL1, were prognostic factors in CESC and ccRCC. These data highlight how individual inflammatory ligands contribute heterogeneously to the association between PTI and poor survival in multiple malignancies, and that no single gene is as consistently a prognostic factor as the signature itself.

[0149] Example 6: Relationship between PTI level and cancer stage The association between PTI levels and cancer stage was questioned. Therefore, a dataset of squamous cell carcinoma of the lung (LUSC), including gene expression data from nine stages of advanced cancer development, was mined to compare PTI levels in healthy, pre-tumor, and malignant states (Mascaux et al., 2019). PTI levels were lowest in normal lung tissue and gradually increased (Figure 9D). Elevated expression of the PTI signature was first prominent in moderately dysplastic lesions and peaked in cancer samples (Figure 9D). These findings suggest that highly proliferative precancerous lesions begin to incorporate inflammatory transcriptional networks before complete malignant transformation.

[0150] Finally, the correlation between disease progression stage and PTI levels was investigated using the TCGA's TNM cancer staging criteria, which consider primary tumor size, lymph node metastasis, and the presence of metastasis to other organs. A weak but statistically significant correlation was found between tumor stage and PTI expression across all cancers, independent of cancer type (P=0.005). In certain cancer types, PTI levels increased significantly with disease progression (Figure 9E). In most malignancies, PTI increased in the pre-metastatic tumor state (I-III), but in LUSC, PTI levels were similar between stages I-III and elevated in stage IV tumors with distant metastasis (Figure 9E). In conclusion, PTI correlates with tumorigenesis, neoplasm formation, and progression to more advanced metastatic disease stages. These data collectively support the idea that inflammation is essential for tumor progression and establish the PTI signature as a broad and consistent indicator of poor prognosis.

[0151] Example 7: MEK inhibition reduces the secretion of tumorigenic inflammatory mediators by cancer cells. Researchers had previously reported observations that numerous cancer cell lines express COX-2 and secrete PGE2 (Zelenay et al., 2015; Bell et al., 2022), and they evaluated whether COX-2 is co-expressed with other PTI genes in cancer cells. By mining transcriptomics data from over 800 human cancer cell lines in the Cancer Cell Line Encyclopedia (CCLE), they found that PTI genes are expressed heterogeneously across cell lines but are positively correlated with each other (Figure 3A).

[0152] To explore methods for targeting cancer cells expressing the PTI factor, data from a whole-cancer drug screening was analyzed in which more than 600 human cancer cell lines with matched transcription data were treated with a series of chemotherapy and targeted drugs (Garnett et al., 2012). Cell lines were grouped by PTI expression, and the relationship between PTI status and drug sensitivity was screened according to cancer type. highThe cell line was found to be selectively sensitive to four different MEK inhibitors (refametinib, CI1040, PD0325901, AZD6244) (Figure 3B, Figure 10A). In contrast, PTI low The cell line was sensitive to PARP1 / 2 inhibitors, as well as receptor tyrosine kinase (RTK) inhibitors targeting VEGFR, PDGFR, and FGFR (Figure 3B). Therefore, PTI high Cell lines are largely unaffected by the overall inhibition of RTKs, which are thought to be mostly signaled via ERK. In fact, more targeted inhibition of the MAPK cascade by MEK1 / 2 inhibitors is unaffected by PTI. high Although it resulted in strong growth inhibition of the cell line, PTI low It did not result in inhibition of the cell line. For example, in cell lines treated with the MEK1 / 2 inhibitor refametinib, a significant inverse correlation was observed between PTI and the semi-maximal inhibitory concentration (IC50) of refametinib (Figure 3C, upper panel). Refametinib-sensitive PTI in the upper left quadrant. high The cell lines (Figure 3C, upper panel) included multiple cancer types, including 27% NSCLC, 18% esophageal cancer, 8% pancreatic cancer, and 8% bladder cancer. Refametinib-sensitive PTI high The group contained significantly more NSCLCs compared to all other groups (21% vs 7%, FDR=0.006). Focusing solely on NSCLC cell lines, an even more pronounced inverse correlation was found between refametinib IC50 and PTI (Figure 3C, bottom panel).

[0153] To determine whether pharmacological inhibition of MEK suppresses the production of PTI factors by cancer cells, we studied two human lung cancer cell lines (NCIH1339, NCIH3255) with high PTI levels according to CCLE transcription data. Trametinib, an approved selective MEK1 / 2 inhibitor, suppressed the proliferation of these cells within 24 hours post-treatment (Figure 10C, Figure 10D). Analysis of the supernatant by a multiple cytokine / chemokine assay along the transcriptome showed that both cell lines secreted PTI factors such as IL-1α, CXCL1, IL-6, and IL-8 (Figure 3C). Notably, between 6 and 24 hours post-treatment, before any apparent effect on cell proliferation, trametinib inhibited the production of all these inflammatory mediators in both KRASG12R (NCIH1339) and KRASWTEGFR-amplified cell lines (NCIH3255) (Figure 3C). Similar results were obtained with selumetinib, a second MEK1 / 2 inhibitor (Figure 10E). Notably, selumetinib did not alter the production of osteopontin, an inflammatory mediator that is not part of PTI. This indicates that MEK inhibition does not indiscriminately inhibit cytokine production (Figure 10E). Taken together, these data suggest that inhibition of MEK kinase activity can rapidly slow down the production of PTI factors by cancer cells.

[0154] Example 8: NFκB links MAPK pathway mutations to cancer cell-specific PTI expression. PTI highThe differing sensitivities of cell lines to MEK inhibitors suggested that oncogenic changes leading to increased MAPK activity may correlate with cancer cell lines and tumor PTI levels. Consistent with this hypothesis, KRAS mutant cell lines showed high PTI expression (Figure 4A). Next, we screened for copy number estimates of cancer-driving genes that predict high PTI expression specific to cancer cells, independent of cancer type and aneuploidy (Martinez-Jimenez et al., 2020). Genes positively correlated with PTI levels included HRAS, RASA1, RAS GTPase-activated protein (GAP), RTK PDGFRB and FLT4 which transmit signals via MAPK signaling, and RELA, which encodes the NFκB subunit p65 that can be transactivated by MEK (Figure 4B, Figure 11A) (Chang et al., 2003). Taken together, these results suggest that KRAS mutations and / or amplifications in MAPK-NFκB signaling frequently drive PTI expression by cancer cells.

[0155] Given the intersection of the MAPK and NFκB pathways, convergence around NFκB activation is likely the primary driver of PTI levels. Indeed, analysis of reverse-phase protein array (RPPA) data showed that high-PTI KRAS mutant cell lines exhibited significantly increased levels of activated p65(RELA) serine 536 phosphorylation, in addition to higher levels of EGFR, the ECM component fibronectin, transglutaminase, and annexin-1, which regulates NFκB activity (Moraes, Ampomah, and Lim, 2018) (Figure 11C). The p65 serine 536 phosphorylation level gradually increased in correlation with PTI expression (Figure 4C). Furthermore, motif enrichment analysis showed that NFκB-binding motifs were significantly enriched in the promoter region of the PTI gene extension set (see Methods). Among the top 10 motifs prominently featured in the promoter region of the PTI gene, there were three NFκB transcription factors containing highly conserved motifs to which both the standard (NFKB1) and non-standard (NFKB2) forms of NFκB bind (Figure 4D, Figure 4E). Taken together, these findings at the gene and protein levels suggest that NFκB activity promotes the expression of the PTI transcription program in cancer cells with KRAS mutations.

[0156] To test whether intratumoral PTI levels correlate with genomic alterations leading to hyperactivation of the MAPK and NFκB pathways, mutational calls from the whole cancer TCGA dataset were analyzed. Consistent with cancer cell line data, mutations in MAPK pathway components (MAP2K1 encoding MEK1, NF1, and BRAF) were important predictors of PTI expression in all cancers, regardless of mutational load and cancer type (FDR < 0.25). Notably, TP53 mutations also showed a strong positive correlation with PTI levels (Figure 4F), while mutations in the oncogene NFE2L2, or mutations in tumor suppressor genes in the NRF2 (AXIN2) and WNT (KEAP1) pathways, were associated with PTI. lowTumor-rich PTI was found (Figure 4F). Similar analyses in individual cancer types identified more than 30 significant mutations linked to high PTI expression (FDR < 0.25). The majority of these mutations were in MAPK pathway genes, including KRAS in LUAD, ccRCC, and PAAD, NF1 in LUSC and GBM, and BRAF in THCA (Figure 4G). Therefore, MAPK activation is linked to PTI in multiple tissue types. high This is a hallmark of the tumor. Furthermore, EGFR and RELA amplification in solid tumors was found to be positively correlated with intratumoral PTI signature scores (Figure 4H). Thus, analysis of genomic and transcriptome data from cancer cell lines and bulk tumors showed that both MAPK / NFκB pathway mutations and SCNA drive PTI factor expression.

[0157] Example 9: Diverse malignant and non-malignant cell types emphasize high PTI signature expression. PTI high vs PTI low To investigate the invasive composition, researchers calculated pairwise Spearman correlations between PTI levels and multiple immune and non-immune cell types estimated by the MCPcounter algorithm for the bulk tumor transcriptome (Becht et al., 2016). PTI signature levels showed significant positive correlations with multiple cell types, including endothelial cells (26 / 33), fibroblasts (30 / 33), neutrophils (31 / 33), and monocytic cells (28 / 33), in most cancer types. In contrast, CD8 + Cytotoxic cells such as T cells (X / 33) or NK cells (X / 33) showed a far less consistent and weaker correlation with PTI expression (Figure 12A). This data suggests that PTI high This study demonstrates that the main tumor-infiltrating cellular components of tumors consist of various myeloid cell subsets and mesenchymal cells such as fibroblasts, suggesting that cancer cell-specific PTI expression leads to the expansion and / or recruitment of diverse cell types.

[0158] To identify the cellular origin of PTI factors, three independent scRNAseq datasets obtained from three different cancer types (HNSC, MEL, NSCLC) were analyzed (Puram et al., 2017; Jerby-Arnon et al., 2018; Lambrechts et al., 2018). Single cells from tumor samples were reclustered using unsupervised graph-based clustering and assigned states guided by metadata from the original studies. T-distributed stochastic neighbor embedding (t-SNE) identified multiple diverse cell types as PTI sources (Figure 5A, Figure 12B). Measuring the mean expression of each PTI gene within each original cluster revealed high levels of expression in neutrophils, macrophages, fibroblasts, mast cells, and endothelial cells (Figure 5B, Figure 12C). Some PTI genes were highly expressed in specific cell types. For example, IL6 and HAS2 were predominantly expressed in fibroblasts, while CXCL5 was specific to macrophages (Figure 5B, Figure 12C). Other PTI genes, including VCAN, AREG, PTGS2, and CXCL2, were expressed across various immune and non-immune cell types (Figure 5B, Figure 12C). Taken together, these data suggest that PTI factors may not only be directly expressed by cancer cells but may also be heterogeneously generated by multiple immune and non-immune cell types.

[0159] Next, PTI levels were compared within single cancer cells isolated from 10 independent HNSC patients. Variation in PTI expression was observed among patients (Figure 5D). The researchers compared PTI levels within cancer cells derived from individual patients and based PTI expression on cancer cell-specific characteristics. high Tumors and PTI low The tumor was defined (Figure 5D). Taken together, these findings support a model in which inflammatory mediators (many of which are coordinately expressed by cancer cells and induced by PGE2) are also expressed by multiple cell types within the TME.

[0160] Example 10: PTI signatures affect the percentage and transcriptional phenotype of tumor-infiltrating cytotoxic lymphocytes. To investigate whether high PTI levels are associated with a unique leukocyte infiltration composition in “hot” tumors, we analyzed CIBERSORT estimates of immune cell proportions in “IFNγ-dominant” subtype tumors (Newman et al., 2015; Thorsson et al., 2018). Comparing mean cell type ratios among patients with different PTI levels, we found that leukocyte ratios were nearly consistent across groups (Figure 5E). high In tumors, the proportion of lymphocytes to total immune cells was significantly lower (FDR = 0.00003, Figure 5E). PTI high In tumors, CD8 + Although the proportion of T cells (FDR=0.002) and activated NK cells (FDR=8.6E-08) was statistically significantly lower, activated memory CD4 + The proportions of T cells (FDR=1.2E-19) and resting NK cells (FDR=1.2E-30, Figure 5E) were high. These results suggest that the overall immune infiltration of "hot" tumors with simultaneously high PTI levels is subtly altered by a reduction in the effectors of antitumor immunity.

[0161] CD8 infiltrating the tumor + T cells exhibit a diverse range of transcriptional states, from naive to dysfunctional and cytotoxic (van der Leun, Thommen, and Schumacher, 2020). Based on previous analyses, PTI expression is associated with CD8 + Promotes T cell dysfunction, and therefore CD8 + It was hypothesized that PTI would show a positive correlation with the suboptimal transcriptional signature of T cells. Consistent with this hypothesis, PTI and dysfunctional autoantigen-specific CD8 + A strong positive correlation was observed between T cell signatures (Schietinger et al., 2012), which is called the T cell anergy signature (Azizi et al., 2018) (Figure 5F). Therefore, PTIs are transcriptionally similar to CD8 cells that cannot induce an effective antitumor immune response. + It is strongly associated with the presence of T cells. Furthermore, PTI is CD8 +CD8 showed a positive correlation with the signature of T cell dysfunction and depletion (Singer et al., 2016; Guo et al., 2018) (Figure 12D). + Previous studies analyzing T cells have shown that naive-like CD8 single-cell ICB response-related T cells, composed of genes involved in memory, survival, and activation, are found in T cells. + T cell signatures have been identified (Sade-Feldman et al., 2018). PTI low CD8 infiltrates MEL and NSCLC + In T cells, PTI high Compared to other cells, the expression of this CD8 ICB response signature was found to be significantly higher. Therefore, PTI high CD8 in tumor + T cells in MEL were not similar to naive / memory-like cells that promote the ICB response. Tumor-derived CD8 + Analysis of the single-cell transcriptome of T cells supports our findings on the tumor as a whole and establishes a strong relationship between the expression of pro-tumor inflammatory mediators and the suboptimal T cell transcription state.

[0162] Example 11: The PTI signature predicts the lack of efficacy with checkpoint inhibitors and outperforms other resistance signatures. To date, no strong predictive factor for ICB resistance has been identified. Given that the PTI signature has strong prognostic predictive power relative to current immune biomarkers of T cell inflammation, many of which are associated with ICB response, the utility of PTI as a biomarker for ICB resistance was investigated. Sixteen transcriptomic datasets comprising over 1000 pre-treatment samples (n=1032) from patients receiving ICB treatment were collated (Figure 6A, Figure 6B). This pan-cancer ICB metadata set included patients with 5 cancer types (ccRCC, mUC, MEL, GCA and NSCLC) treated with monotherapy and combination therapy with anti-CTLA-4, anti-PD-1 or anti-PD-L1 antibodies. 25% of the samples were from patients who responded to treatment (partial or complete response (PR / CR)), and 43% were from patients who showed no benefit (progressive disease (PD)). By stratifying patients according to the median pre-treatment PTI level of each dataset, PTI low was found to be significantly enriched for PR / CR among patients (χ 2 , P=0.004). Conversely, consistent with the reported values of T cell inflammation markers, patients with low TIS (χ 2 , p=4×10 -7 ), CD8 + T cells (χ 2 , p=4×10 -6 ), IFNγ signaling (χ 2 , p=3×10 -6 ) signature, or PD-L1 gene expression (CD274, χ 2 , p=0.003) were overrepresented among PD patients (Figure 6C) (Tumeh et al., 2014; Ayers et al., 2017). Importantly, PTI showed co-predictive power when combined with all current biomarkers of "hot" T cell inflamed tumors (TIS, CD8, IFNγ and / or CD274), such that the proportion of CR / PR was lowest in PTI high TIS low patients, and PTI low TIS highIt gradually increased in patients and reached the maximum (Figure 6D). Similarly, the combination of PTI and TIS also improved the predictive value of OS in these ICB-treated patients (Figure 6E). PTI high TIS high Patients showed significantly worse OS than PTI low TIS high patients, with a difference of about 1 year in median survival time. Notably, PTI high TIS high patients and PTI low TIS low patients had equivalent survival rates, which is consistent with our previous TCGA data analysis showing the antagonistic interaction between PTI and anti-tumor immune mediators (Figure 6E). Taken together, this analysis shows that the PTI signature helps to distinguish ICB responders and non-responders, including patients with equally high levels of intratumoral T cell inflammation.

[0163] Next, the researchers attempted to compare the PTI signature with other signatures reported to be associated with ICB resistance. These included the WNT / β-catenin signaling signature (Gajewski et al., 2019), which has previously been shown to correlate with tumors with weak "cold" T-cell inflammation; gene signatures associated with fibroblast-mediated immunosuppressive TGFβ signaling (F-TBRS, Chakravarthy et al., 2018) and ECM remodeling (TGFβ-ECM, Mariathasan et al., 2018); the EMT signature linked to reduced response rates in T-cell inflammation mUC (Stroma-EMT, Wang et al., 2019); and the signature of angiogenesis-related factors highly expressed during the inflammation-to-cancerous (I2C) transition associated with ICB resistance in melanoma (Chen et al., 2019). To evaluate the additional predictive value when each of these signatures is combined with T-cell inflammation signatures, the TIS signature, one of the most robust and widely validated gene expression biomarkers for ICB, was used as a baseline predictive model (Ayers et al., 2017; Cristescu et al., 2018). Each consecutive signature score was combined with TIS in a collaborative predictive model, and the area under the curve (AUC) was calculated as an indicator of model performance for each of the 16 patient cohorts treated with ICB. In particular, the PTI / TIS model had the highest mean AUC compared to other resistance signature models and TIS alone (Figure 13A). Importantly, the mean AUC of the PTI+TIS combined model (0.79) was significantly higher than that of TIS alone (mean AUC=0.69, P=0.02), highlighting the predictive utility of combining PTI and TIS values ​​across the entire cancer ICB dataset. The researchers also compared the performance of all resistance signatures to predict overall survival (OS) in patients after ICB therapy, which does not necessarily correlate with response rates. Patients were stratified by the median cutoff value for each signature and fitted to a multivariate CoxPH model adjusted for TIS status and ICB datasets. 1000 randomly generated signatures of the same length as the PTI were also included as a control.Consistent with objective response data, PTI was the strongest resistance signature in predicting poor overall survival (OS) in ICB-treated patients (Figure 6F). These combined data further demonstrate the unique and superior utility of combining PTI with "hot" tumor immunobiomarkers to predict patient response and survival after ICCB.

[0164] The currently claimed PTI signature shows improved predictive power compared to the COX-IS ratio disclosed in the inventors' previous application, International Publication WO2019 / 243567. Figure 6F shows the predictive power of the resistance signature in predicting the decline in overall survival in patients treated with ICB. The PTI signature (HR 1.38 [1.15-1.65], P=0.0004) outperforms the inventors' previous COX-IS ratio (HR 1.18 [0.96-1.45], P=0.12), thus demonstrating improved predictive power.

[0165] Next, we examined the CPI1000 dataset (Litchfield et al., 2021), which consists of a collection of datasets (substantially overlapping with the metadata set above) from patients with multiple cancer types treated with ICB (anti-PD(L)1, anti-CTLA4) using whole exome sequencing (WES) and RNA sequencing (RNAseq) data. This unique dataset underwent extensive harmonization of WES, RNAseq, and clinical data across several cohorts. First, this analysis focused on 523 patients from seven studies targeting three cancer types (MEL, BLCA, ccRCC) annotated with multiple genomic and transcriptome tumor characteristics previously associated with ICB responses (Snyder et al., 2014, 2017; Van Allen et al., 2015; Hugo et al., 2017; Riaz et al., 2017; Mariathasan, 2018; Mcdermott et al., 2018). APOBEC signature (Spearman = 0.26, P = 9 × 10) -7 ), TP53 mutant state (Spearman = P = 0.0004) and CD274 expression (Spearman = 0.22, P = 3 × 10) -7In addition to a moderate positive correlation with ), PTI levels were not strongly associated with any single parameter (Figures 13B, 13C). This supports the idea that PTI captures a distinct and independent feature of tumor inflammation and its potential use as an independent immunobiomarker.

[0166] To evaluate the predictive power of PTI in this harmonized dataset, benchmarks were performed against multiple ICB resistance signatures and parameters such as F-TBRS and Stroma-EMT (Mariathasan et al., 2018; Wang et al., 2019), or against responses such as the IMPRES algorithm, TIS score, CXCL13 expression, TLS signature, CYT score, and various TMB measurements (Rizvi et al., 2015; Rooney et al., 2015; McGranahan et al., 2016; Ayers et al., 2017; Auslander et al., 2018; Mariathasan et al., 2018; Cabrita et al., 2020; Litchfield, Reading et al., 2020). Notably, PTI levels were higher in pre-treatment samples from PD patients than in PR / CR patients (n=552, Figure 6G). Next, the inventors fitted univariate logistic regression models for PTI and other biomarkers in seven individual datasets comprising a broader overall cancer cohort (Van Allen et al., 2015; Hugo et al., 2017; Riaz et al., 2017; Snyder et al., 2017; Mariathasan et al., 2018; Mcdermott et al., 2018; Liu et al., 2019). A meta-analysis was performed to calculate the overall effect size to assess the consistency of each estimated biomarker. PTI was the best predictor of ICB resistance (odds ratio = 0.67, P = 0.03, Figure 6H). PTI across the entire CPI1000 cohort, including CR / PR, PD, and stable disease (SD) patients. highOverall survival (OS) was significantly worse in patients with moderate and low PTI compared to the moderate and low PTI groups (HR=1.41[1.1-1.8], P=0.007) (Figure 13D). Furthermore, even after adjusting for cancer type and prior exposure to anti-CTLA-4 therapy, PTI remained a significant prognostic factor (HR=1.29[1.006-1.65], P=0.04), further demonstrating the independent predictive value of the PTI signature as an immunobiobitemer.

[0167] A model combining clonal TMB and CXCL9 expression was shown to be optimal for predicting patient responses to ICB in the CPI1000 dataset (Litchfield et al., 2021). We tested whether PTI levels provided significant predictive utility for the combination of clonal TMB and CXCL9 by fitting predictive models encompassing cancer type, clonal TMB, CXCL9, PTI, and their combinations. Notably, PTI added significant predictive value in addition to cancer type, clonal TMB, and CXCL9 (Figure 13E, likelihood ratio test, P=0.01). Clonal TMB had the highest proportion of ICB responders. hi , or CXCL9 hi Or TMB hi CXCL9 hi Patients (Litchifield et al.) were further stratified by PTI signature (Figure 6I, Figure 13F, Figure 13G). These findings further demonstrate the value of measuring PTI concurrently with other immunobiomarkers in pre-treatment samples. Furthermore, this data positions PTI as a useful tool for identifying patients who should receive more aggressive combination therapy, such as COX-2 inhibitors or other drugs that suppress PTI, in order to maximize the benefits of ICB.

[0168] Example 12: A model for predicting binary patient response outcomes for tumor type-independent and tumor type-specific approaches. The logistic regression model was trained on the CPI1000 dataset. First, TPM values of PTI genes were obtained from the CPI10000 dataset. A pseudo-count of 1 was added before log2 normalization of the values. Next, a) the data were scaled (Z-score normalized) across all samples in the dataset (tumor type-independent), or b) the data were grouped by cancer type (metadata feature: "Tum_Type"), and the expression values were scaled by tumor type (tumor type-specific). Next, the mean of the PTI genes in the scaled expression data was determined, and the PTI score was calculated. Patients who achieved stable disease as the best RECIST response (metadata feature: Response_RECIST) were excluded, along with "on-treatment" samples (metadata feature: Prior_IO).

[0169] In both the tumor type-independent approach and the tumor type-specific approach, three different logistic regression models (discussed in detail below) were trained to predict the binary patient response outcome (metadata feature: Response_Responder).

[0170] The three logistic regression models are as shown in Table 3. The results of logistic regression models A to C are shown in Table 4. For tumor-specific Model A, the coefficients of PTGS2, CXCL2, CXCL6, IL1A, AREG, EREG, CXCL1, CXCL5, HAS2, IL11, IL1B, IL6, CXCL8, NRG1, VCAN, Clonal_TM and CXCL9 were extracted according to Table 5. The area under the curve (AUC) was calculated as an indicator of the model performance, and a larger AUC indicates better predictive utility. The receiver operating characteristic curve of this logistic regression Model A (tumor type-specific) is shown in Figure 15.

[0171] Model C, which combines the expression of PTGS2, CXCL1, CXCL2, CXCL5, CXCL6, CXCL8, IL1A, IL1B, IL6, IL11, NRG1, HAS2, AREG, EREG and VCAN, was shown to predict the binary response outcome of patients in both tumor type-independent approaches and tumor type-specific approaches even in the absence of TMB or CXCL9, as shown in Table 4 below.

[0172] [Table 3]

[0173] [Table 4]

[0174] [Table 5]

[0175] Example 13: Clinical value of the PTI signature as an immune biomarker for ICB outcomes Finally, the predictive utility and potential clinical value of PTI signatures as immunobiomarkers for analyzing formalin-fixed paraffin-embedded (FFPE) tissue from immunotherapy-treated patients were evaluated. The focus was on ccRCC, a cancer species where ICB is the mainstream treatment, but current immunobiomarkers cannot predict outcomes. As a proof of concept, the researchers obtained archived FFPE samples with matched overall survival data from 42 patients (91% ccRCC) treated for progressive disease with high-dose IL-2 immunotherapy (n=23), ICB as monotherapy (n=11), or in combination with anti-CTLA-4 (n=4) or anti-VEGF (n=1) therapy (Figure 14A). This diverse real-world clinical cohort included nephrectomy samples as well as samples from surgical resections of brain and lung metastases (Figure 14B). RNA was extracted, and multiple gene expression analysis was performed on the Nanostring nCounter platform using a whole cancer immune profile panel containing 800 immune-related genes, including most of the PTI signature genes (12 / 16). 83% of the present PTI genes were expressed beyond the detection limit in over 50% of the samples and were generally correlated (mean Spearman = 0.3, Figure 7A). Therefore, regardless of the quantification method, PTI represents a conserved TME transcriptional program. PTI signatures correlated with the presence of various myeloid cell subsets (neutrophils, dendritic cells, macrophages), as well as stromal and complement signatures (Figure 7B). Thus, consistent with previous bulk RNA-seq findings, PTI signatures were associated with the presence of multiple immune populations and the molecular characteristics of the disease. Importantly, PTI levels were associated with CD8 + It was inversely correlated with T cell signatures (CD8A, CD8B, FLT3LG, GZMM, PRF1) and IFNγ signaling signatures (CXCL9, CXCL10, IDO1, STAT1, IFNG, HLA-DRA) (Figure 7C).

[0176] Of these 42 RCC tumor biopsies, 25 were pre-treatment samples from patients who had received IL-2 immunotherapy, anti-PD(L)1 antibodies, or anti-PD1 / anti-CTLA-4 blockers. When these samples were stratified according to PTI levels, patients with significantly different survival outcomes were observed. high Overall survival (OS) was significantly shortened in these patients (Figures 7D and 7E). Importantly, none of the immunosignatures for “hot” T-cell inflammatory tumors showed a significant association with patient survival in this patient cohort (Figure 7F). The “bone marrow cell signature” (Mcdermott et al., 2018), which had previously been suggested to predict ICB outcomes in ccRCC patients, also showed no significant association. These remarkable findings, obtained in a small, heterogeneous cohort of ccRCC patients actually treated with immunotherapy using clinically fitted assays, further highlight the potential power of monitoring PTI to predict ICB outcomes and guide patient treatment.

[0177] Example 14: PTI as a predictor of outcomes in ccRCC patients after immunotherapy Finally, to further investigate the specific utility of PTI in predicting patient outcomes after immunotherapy in ccRCC, transcriptome data from Checkmate trials 009, 010, and 025 were analyzed. In all of these trials, the anti-PD-1 blocker nivolumab demonstrated clinical benefit in previously treated patients with advanced ccRCC, including when compared to the mTOR inhibitor everolimus in a Phase III trial (Checkmate 025) (Braun et al., 2020). Notably, patient stratification based on pre-treatment PTI expression levels showed a negative correlation with survival in nivolumab-treated patients (HR=1.46 [1.02-2.08], P=0.04), but not in everolimus-treated patients (HR=1.03 [0.71-1.50], P=0.87) (Figure 7G). To test the predictive ability of PTI in patients treated with ICB as first-line therapy, previously untreated patients with advanced ccRCC were randomized into one of three trial arms and administered atezolizumab (anti-PD-L1) and bevacizumab (anti-VEGF), atezolizumab monotherapy, or sunitinib (Mcdermott et al., 2018). Data from the Phase II IMMotion150 study were analyzed. Similar to the inventor's analysis in patients treated with second-line ICB, non-responders who received atezolizumab or atezolizumab plus bevacizumab as first-line therapy had higher baseline PTI levels compared to responders (Figure 7H). In contrast, there was no difference in baseline PTI expression between responders and non-responders to the multitarget tyrosine kinase inhibitor sunitinib (Figure 7H). Taken together, these findings demonstrate that PTI selectively identifies ccRCC patients resistant to current immunotherapy regimens, and that its predictive utility from ICB exceeds prognostic predictions.

[0178] Example 15: Minimum number of PTI genes required to calculate a score that predicts treatment response and / or prognosis Elastic network regression is a statistical model that helps narrow the feature space and select the most predictive features in classification tasks such as predicting ICB responses. To demonstrate this, we first trained an elastic network regression model using PTI genes and clonal TMB and CXCL9 expression levels as input features. We did this using a normalized expression matrix. In the first model (Model D), expression levels were normalized individually for each tumor type (tumor type specific), and in the second model (Model E), expression levels were normalized in a tumor-independent manner. Furthermore, we trained an elastic network regression model on members of the PTI gene set in a tumor-independent manner without using clonal TMB and CXCL9 as covariates (Model F). The elastic network regression model was trained with different α hyperparameter values ​​from 0 (ridge regression) to 1 (lasso regression). The main difference is that increasing α causes the coefficients of less important features to converge to 0, thereby enabling a more rigorous selection of important features. method First, tumor-type-independent (Model E) and tumor-type-specific (Model D) datasets were split into training (70%) and test (30%) sets. Elastic network classifier models were trained to predict responses (CR / PR) using 10-fold cross-validation, with AUC as the performance metric. Models were trained with various α hyperparameter values, incrementing by 0.1 from 0 to 1. For each training run, the λ value corresponding to the best training performance was selected as the best model. Next, patient responses to ICBs in the test dataset were predicted using each optimized model, and the AUC was calculated. The AUCs from the training and test sets were then plotted (Figures 16A-B for Models D and E, respectively). An alpha value of 0 indicates that all available features (all 16 PTI genes, clonal_TMB, and CXCL9) were used.

[0179] The elastic net regression model was also trained for members of the PTI gene set without clonal_TMB and CXCL9 as covariates (Model F) (Figure 16C). result The plots in Figures 16A and 16B show the trial and training AUCs for a tumor type-specific model (Model D) and a tumor type-independent model (Model E), respectively. AUC was calculated using the gene expression levels of clonal_TMB and all 16 PTI genes and CXCL9. The points represent models trained with different α hyperparameter values ​​ranging from 0 to 1. The corresponding α values ​​are shown next to each point. Generally, higher α values ​​indicate a closer resemblance to Lasso regression and a smaller number of features utilized by the model. As is evident from Figure 16, models trained with α values ​​greater than 0 tended to show improved training and trial performance. Of particular note is that in many models, the number of critical features decreased compared to the number of features the model was trained with. Critical features are defined as any feature whose regression coefficient is non-zero after applying regularization.

[0180] Table 6 illustrates the key features and coefficients of the best tumor-type-independent dataset model (Model E) with an alpha value of 0.3. This best model E has a reduced feature set. Four genes derived from the PTI signature (EREG, HAS2, CXCL8, and VCAN), along with Clonal_TMB and CXCL9, were identified as key features.

[0181] [Table 6]

[0182] Figure 16C exemplifies the training AUC values and test AUC values of the tumor type-independent model (Model F) characterized only by the PTI genes (CXCL9 and Clonal_TMB are not included as covariates). The alpha value is shown next to the corresponding point (the trained model). The best model had an alpha hyperparameter of 0.2, and had 7 important features (AREG, CXCL1, CXCL2, EREG, HAS2, IL1A and PTGS2) corresponding to 7 out of the 16 PTI genes. The coefficient of each of these 7 features in the best Model F (α=0.2) is exemplified in Table 7.

[0183] [Table 7]

[0184] The elastic regression model trained only on members of the PTI gene set (in the absence of clonal_TMB and CXCL9) does not perform as well as the model trained on the PTI genes in the presence of clonal_TMB and CXCL9, but still explains a small proportion of the variance in patients' ICB response.

[0185] Discussion Determining the potential patient benefit of ICB is a key challenge in this field (Hegde and Chen, 2020). We hypothesized that differences in patient outcomes after ICB are supported by biological mechanisms beyond current biomarkers. Subsequently, we showed that pro-tumor inflammatory ligands, particularly COX-2, frequently control such divergences, including cases where patients with clonal TMB and high CXCL9 expression are more likely to benefit from ICB (Litchfield et al., 2021). These findings indicate the need for multiple combinations of biomarkers, requiring both WES and gene expression assays that must include PTI genes as important in addition to genes measuring T-cell inflammation. A key consideration in gene signature-based biomarker development is intratumor heterogeneity (ITH). The extent to which sampling bias affects PTI-based patient classification remains unclear. However, one way to avoid these problems is to select PTI genes with generally low ITH levels, which are less susceptible to sampling bias, as has been done previously for prognostic signatures of NSCLC (Biswas et al., 2019).

[0186] Using a published method (Jerby-Arnon et al., 2018), we analyzed cell line drug screening data (Garnett et al., 2012) and found that PTI highCell lines have been shown to be sensitive to MEK inhibitors. However, the dataset does not include next-generation inhibitors of historically difficult targets such as KRAS (Canon et al., 2019). Furthermore, in vivo experiments targeting all interacting cell types are needed to identify core nodes that can be targeted to disrupt the PTI signaling network. Nevertheless, this work indicates that the COX-2 / PGE2 axis is a promising candidate (Pelly et al., 2021). Numerous studies have reported favorable immunomodulatory properties of MEK inhibitors (Hu-Lieskovan et al., 2015; Ebert et al., 2016; Bommareddy et al., 2018; Baumann et al., 2020). Further work is needed to test whether MEK inhibitors specifically repress the PTI transcription program in vivo, but current data further support the idea that MEK inhibitors can be used in a broader range of applications, not just as cell proliferation inhibitors (Templeton and Musib, 2015; Hamarsheh et al., 2020).

[0187] By utilizing bulk tumor RNA sequencing, immune deconvolution algorithms, and scRNA sequencing data (Becht et al., 2016; Puram et al., 2017; Jerby-Arnon et al., 2018; Lambrechts et al., 2018), it is possible to identify that not only cancer cells themselves, but also diverse cell types within the tumor express the PTI gene, and PTI high These cells are shown to infiltrate tumors. Neutrophils, in particular, are highly correlated with PTI levels and have been shown to be elevated in the KRAS mutant setting, consistent with the working model (Hamarsheh et al., 2020; Polidoro et al., 2020). However, it remains unclear whether these cells play an active role in maintaining cancer-promoting inflammation. Furthermore, it remains unclear whether the generation of cancer cell-specific inflammatory mediators propagates the PTI signature to adjacent non-cancer cells, or vice versa. In the future, mechanistic studies using spatially degraded transcriptomics data obtained from human tumors should help answer these questions.

[0188] We investigated the underlying rationale for the association between PTI ligands and ICB resistance. PTI factors play diverse roles within TMEs and contribute to the hallmark of all cancers. It has been revealed that PTI is deeply involved in T cell infiltration and IFNγ signaling. In fact, direct regulation of the activation state of intratumor lymphocytes is likely to contribute to the ICB resistance phenotype. CD8 + Patients with tumors in which T cells were inflamed, or those exhibiting high levels of IFN-γ signaling, showed different outcomes depending on the PTI level. This trend was also observed in patients with high plasma levels of the pro-tumorogenic chemokine IL-8 (Yuen et al., 2020). The COX-2 / PGE2 axis, a central node in the PTI phenotype, has previously been associated with T cell depletion (Harris et al., 2002; Chen et al., 2015). CD8 ICB response signature (Sade-Feldman et al., 2018) and CD8 + Using the signature of T cell dysfunction (Azizi et al., 2018; Guo et al., 2018), PTI high CD8 in tumor + T cells have been shown to be in a suboptimal activation state.

[0189] In conclusion, 16 COX-2-centered inflammatory gene signatures were defined that are associated with molecularly distinct T cell infiltration and disastrous outcomes following treatment with immune checkpoint inhibitors.

[0190] References To more fully explain and disclose the present invention and the state of the art to which it relates, numerous publications have been cited above. A complete list of these references is provided below. The entirety of each of these references is incorporated herein. Alfaro C. et al. (2016) "Tumor-Produced Interleukin-8 Attracts Human Myeloid-Derived Suppressor Cells and Elicits Extrusion of Neutrophil Extracellular Traps (NETs)," Clinical Cancer Research, 22(15), pp.3924 LP~3936. doi:10.1158 / 1078-0432.CCR-15-2463. Van Allen EM et al. (2015) "Genomic correlates of response to CTLA-4 blockade in metastatic melanoma," Science, 350(6257), pp.207~211. doi:10.1126 / science.aad0095. Andrea S. et al. (2012) "Rescued Tolerant CD8 T Cells Are Preprogrammed to Reestablish the Tolerant State," Science, 335(6069), pp.723-727. doi:10.1126 / science.1214277. Auslander N. et al. (2018) "Robust prediction of response to immune checkpoint blockade therapy in metastatic melanoma," Nature Medicine. doi:10.1038 / s41591-018-0157-9. Ayers M. et al. (2017) "IFN-γ-related mRNA profile predicts clinical response to PD-1 blockade," The Journal of Clinical Investigation, 127(15), pp.1~11. doi:10.1172 / JCI91190. Azizi E. et al. (2018) "Single-Cell Map of Diverse Immune Phenotypes in the Breast Tumor Microenvironment," Cell, 174(5), pp.1293~1308.e36.doi:https: / / doi.org / 10.1016 / j.cell.2018.05.060. Baumann D. et al. (2020) "Proimmunogenic impact of MEK inhibition synergizes with agonist anti-CD40 immunostimulatory antibodies in tumor therapy," Nature Communications, 11(1).doi:10.1038 / s41467-020-15979-2. Becht E. et al. (2016) "Erratum to Estimating the population abundance of tissue-infiltrating immune and stromal cell populations using gene expression [Genome Biol., 17(2016), (218)]" (correction to "Estimating the population abundance of tissue-infiltrating immune and stromal cell populations using gene expression" [Genome Biol., 17(2016), (218)]), Genome Biology, 17(1), pp.1~20. doi:10.1186 / s13059-016-1113-y. Bhattacharya A. et al. (2020) "An approach for normalization and quality control for NanoString RNA expression data," Briefings in Bioinformatics, 00 (June), pp.1~13. doi:10.1093 / bib / bbaa163. Biswas D. et al. (2019) "A clonal expression biomarker associates with lung cancer mortality," Nature Medicine, 25 (October). doi:10.1038 / s41591-019-0595-z. Bommareddy PK et al. (2018) "MEK inhibition enhances oncolytic virus immunotherapy through increased tumor cell killing and T cell activation," Science Translational Medicine, 10(471).doi:10.1126 / scitranslmed.aau0417. Bonavita E. et al. (2020) "Antagonistic Inflammatory Phenotypes Dictate Tumor Fate and Response to Immune Checkpoint Blockade," Immunity, 53(6), pp.1215~1229. e8.doi:10.1016 / j.immuni.2020.10.020. Braun DA et al. (2020) "Interplay of somatic alterations and immune infiltration modulates response to PD-1 blockade in advanced clear cell renal cell carcinoma," Nature Medicine.doi:10.1038 / s41591-020-0839-y. Browaeys R., Saelens W., and Saeys Y. (2020) "NicheNet: modeling intercellular communication by linking ligands to target genes," Nature Methods. doi:10.1038 / s41592-019-0667-5. Cabrera RM et al. (2018) "A novel neuregulin-jagged1 paracrine loop in breast cancer transendothelial migration," Breast Cancer Research, 20(1), pp.1~10. doi:10.1186 / s13058-018-0960-8. Cabrita R. et al. (2020) "Tertiary lymphoid structures improve immunotherapy and survival in melanoma," Nature, 577(7791), pp.561~565. doi:10.1038 / s41586-019-1914-8. Calon A. et al. (2012) "Dependency of Colorectal Cancer on a TGF-β-Driven Program in Stromal Cells for Metastasis Initiation," Cancer Cell, 22(5), pp.571~584. doi:10.1016 / j.ccr.2012.08.013. Canon J. et al. (2019) "The clinical KRAS(G12C) inhibitor AMG 510 drives anti-tumor immunity," Nature, 575(7781), pp.217~223. doi:10.1038 / s41586-019-1694-1. Chang F. et al. (2003) "Signal transduction mediated by the Ras / Raf / MEK / ERK pathway from cytokine receptors to transcription factors: Potential targeting for therapeutic intervention," Leukemia, 17(7), pp.1263~1293. doi:10.1038 / sj.leu.2402945. Chen JH et al. (2015) "Prostaglandin E2 and programmed cell death 1 signaling coordinately impair CTL function and survival during chronic viral infection," Nature Medicine, 21(4), pp.327-334. doi:10.1038 / nm.3831. Chen X. et al. (2019) "Immune Cell Types and Secreted Factors Contributing to Inflammation-to-Cancer Transition and Immune Therapy Response," Cell Reports, pp. 1965-1977. doi:10.1016 / j.celrep.2019.01.080. Chittezhath M. et al. (2014) "Molecular Profiling Reveals a Tumor-Promoting Phenotype of Monocytes and Macrophages in Human Cancer Progression," Immunity, 41(5), pp.815~829. doi:10.1016 / j.immuni.2014.09.014. Cohen-Sharir Y. et al. (2021) "Aneuploidy renders cancer cells vulnerable to mitotic checkpoint inhibition," Nature, 590(7846), pp.486~491. doi:10.1038 / s41586-020-03114-6. Coussens LM, Zitvogel L., and Palucka AK (2013). "Neutralizing tumor-promoting chronic inflammation: A magic bullet?", Science, 339(6117), pp.286-291. doi:10.1126 / science.1232227. Cristescu R. et al. (2018) "Pan-tumor genomic biomarkers for PD-1 checkpoint blockade-based immunotherapy," Science, 362(6411), p.eaar3593.doi:10.1126 / science.aar3593. Danaher P. et al. (2017) "Gene expression markers of Tumor Infiltrating Leukocytes", pp. 1-15. doi:10.1186 / s40425-017-0215-8. Danaher P. et al. (2018) "Pan-cancer adaptive immune resistance as defined by the Tumor Inflammation Signature (TIS): results from The Cancer Genome Atlas (TCGA)", pp. 1-17. Dannenberg AJ and Subbaramaiah K. (2003) "Targeting cyclooxygenase-2 in human neoplasia: Rationale and promise," Cancer Cell, 4(6), pp.431-436. doi:https: / / doi.org / 10.1016 / S1535-6108(03)00310-6. Davis AA and Patel VG (2019) "The role of PD-L1 expression as a predictive biomarker: An analysis of all US food and drug administration (FDA) approvals of immune checkpoint inhibitors," Journal for ImmunoTherapy of Cancer, 7(1), pp.1~8. doi: 10.1186 / s40425-019-0768-9. Dobin A. et al. (2013) "STAR: Ultrafast universal RNA-seq aligner", Bioinformatics. doi:10.1093 / bioinformatics / bts635. Ebert PJR et al. (2016) "MAP Kinase Inhibition Promotes T Cell and Anti-tumor Activity in Combination with PD-L1 Checkpoint Blockade," Immunity, 44(3), pp.609~621. doi:10.1016 / j.immuni.2016.01.024. Feng B. et al. (2020) "Integrative analysis of multi-omics data identified EGFR and PTGS2 as key nodes in a gene regulatory network related to immune phenotypes in head and neck cancer," Clinical Cancer Research. doi:10.1158 / 1078-0432.ccr-19-3997. Fridman WH et al. (2012) "The immune contexture in human tumors: impact on clinical outcome," Nature Reviews Cancer, 12(4), pp.298-306. doi:10.1038 / nrc3245. Gajewski TF et al. (2019) "WNT / β-catenin pathway activation correlates with immune exclusion across human cancers," Clinical Cancer Research, p.clincanres.1942.2018.doi:10.1158 / 1078-0432.ccr-18-1942. Garnett MJ et al. (2012) "Systematic identification of genomic markers of drug sensitivity in cancer cells," Nature, 483(7391), pp.570-575. doi:10.1038 / nature11005. Georgouli M. et al. (2019) "Regional Activation of Myosin II in Cancer Cells Drives Tumor Progression via a Secretory Cross-Talk with the Immune Microenvironment," Cell, 176(4), pp.757~774. e23.doi:10.1016 / j.cell.2018.12.038. Gerlinger M. et al. (2014) "Genomic architecture and evolution of clear cell renal cell carcinomas defined by multiregion sequencing," Nature Genetics.doi:10.1038 / ng.2891. Gide TN et al. (2019) "Distinct Immune Cell Populations Define Response to Anti-PD-1 Monotherapy and Anti-PD-1 / Anti-CTLA-4 Combined Therapy," Cancer Cell, 35(2), pp.238~255. e6.doi:10.1016 / j.ccell.2019.01.003. Green JA et al. (2017) "A nonimmune function of T cells in promoting lung tumor progression," Journal of Experimental Medicine, 214(12), pp.3565~3575. doi:10.1084 / jem.20170356. Guo X. et al. (2018) "Global characterization of T cells in non-small cell lung cancer by single-cell sequencing," Nature Medicine, 24 (July), pp. 1-18. doi:10.1038 / s41591-018-0045-3. Hamarsheh S. et al. (2020) "Immune modulatory effects of oncogenic KRAS in cancer," Nature Communications, 11(1), p.5439. doi:10.1038 / s41467-020-19288-6. Harris SG et al. (2002) "Prostaglandins as modulators of immunity," Trends in Immunology.doi:10.1016 / S1471-4906(01)02154-8. Havel JJ (2019) "The evolving landscape of biomarkers for checkpoint inhibitor immunotherapy," Nature Reviews Cancer, 19 (March). doi:10.1038 / s41568-019-0116-x. Hegde PS and Chen DS (2020). "Perspective: Top 10 Challenges in Cancer Immunotherapy," Immunity, 52(1), pp. 17-35. doi: 10.1016 / j.immuni.2019.12.011. Hennequart M. et al. (2017) "Constitutive IDO1 Expression in Human Tumors Is Driven by Cyclooxygenase-2 and Mediates Intrinsic Immune Resistance," Cancer Immunology Research, 5(8), pp.695 LP~709. doi:10.1158 / 2326-6066.CIR-16-0400. Hu-Lieskovan S. et al. (2015) "Improved antitumor activity of immunotherapy with BRAF and MEK inhibitors in BRAFV600E melanoma," Science Translational Medicine, 7(279).doi:10.1126 / scitranslmed.aaa4691. Hugo W. et al. (2017) "Genomic and transriptomic features of anti-PD1 response", 165(1), pp.35~44. doi:10.1016 / j.cell.2016.02.065.Genomic. Jerby-Arnon L. et al. (2018) "A Cancer Cell Program Promotes T Cell Exclusion and Resistance to Checkpoint Blockade," Cell, 175(4), pp.984~997. e24.doi:10.1016 / j.cell.2018.09.006. Keenan TE, Burke KP, and Van Allen EM (2019) "Genomic correlates of response to immune checkpoint blockade," Nature Medicine. doi:10.1038 / s41591-019-0382-x. Kim ST et al. (2018) "Comprehensive molecular characterization of clinical responses to PD-1 inhibition in metastatic gastric cancer," Nature Medicine.doi:10.1038 / s41591-018-0101-z. Lambrechts D. et al. (2018) "Phenotype molding of stromal cells in the lung tumor microenvironment," Nature Medicine, 24(8), pp.1277~1289. doi:10.1038 / s41591-018-0096-5. Lawrence MS et al. (2014) "Discovery and saturation analysis of cancer genes across 21 tumor types," Nature. 2014 / 01 / 05, 505(7484), pp.495~501. doi:10.1038 / nature12912. Lawrence MS et al. (2015) "Comprehensive genomic characterization of head and neck squamous cell carcinomas," Nature, 517(7536), pp.576~582. doi:10.1038 / nature14129. van der Leun AM, Thommen DS, and Schumacher TN (2020). "CD8+ T cell states in human cancer: insights from single-cell analysis," Nature Reviews Cancer, 20(4), pp.218-232. doi:10.1038 / s41568-019-0235-4. Liao Y., Smyth GK, and Shi W. (2014) "FeatureCounts: An efficient general-purpose program for assigning sequence reads to genomic features," Bioinformatics.doi:10.1093 / bioinformatics / btt656. Liberzon A. et al. (2015) "The Molecular Signatures Database Hallmark Gene Set Collection," Cell Systems, 1(6), pp.417~425. doi:10.1016 / j.cels.2015.12.004. Litchfield K., Reading JL et al. (2020) "Escape from nonsense-mediated decay associates with anti-tumor immunogenicity," Nature Communications, 11(1), pp.1~11. doi:10.1038 / s41467-020-17526-5. Litchfield K., Stanislaw S., et al. (2020) "Representative Sequencing: Unbiased Sampling of Solid Tumor Tissue," Cell Reports, 31(5).doi:10.1016 / j.celrep.2020.107550. Litchfield K. et al. (2021) "Meta-analysis of tumor- and T cell-intrinsic mechanisms of sensitization to checkpoint inhibition," Cell, pp.1~19. doi:10.1016 / j.cell.2021.01.002. Liu David et al. (2019) "Integrative molecular and clinical modeling of clinical outcomes to PD1 blockade in patients with metastatic melanoma," Nature Medicine, 25 (December). doi:10.1038 / s41591-019-0654-5. Mantovani A. et al. (2008) "Cancer-related inflammation," Nature, 454(7203), pp.436~444. doi:10.1038 / nature07205. Mariathasan, S. et al. (2018) "TGFβ attenuates tumor response to PD-L1 blockade by contributing to exclusion of T cells," Nature, 554(7693), pp.544~548. doi:10.1038 / nature25501. Markosyan N. et al. (2019) "Tumor cell-intrinsic EPHA2 suppresses antitumor immunity by regulating PTGS2(COX-2)", Journal of Clinical Investigation. doi:10.1172 / JCI127755. Martinez-Jimenez F. et al. (2020) "A compendium of mutational cancer driver genes," Nature Reviews Cancer, 20(10), pp.555~572. doi:10.1038 / s41568-020-0290-x. Mascaux C. et al. (2019) "Immune evasion before tumor invasion in early lung squamous carcinogenesis," Nature, 571(7766)pp.570~575. doi:10.1038 / s41586-019-1330-0. McCarthy DJ, Chen Y., and Smyth GK (2012). "Differential expression analysis of multifactor RNA-Seq experiments with respect to biological variation," Nucleic Acids Research, 40(10), pp.4288~4297. doi:10.1093 / nar / gks042. Mcdermott DF et al. (2018) "Clinical activity and molecular correlates of response to atezolizumab alone or in combination with bevacizumab versus sunitinib in renal cell carcinoma," Nature Medicine, 24 (June). doi:10.1038 / s41591-018-0053-3. McGranahan N. et al. (2016) "Clonal neoantigens elicit T cell immunoreactivity and sensitivity to immune checkpoint blockade," Science, 351(6280), pp.1463 LP~1469, available at: http: / / science.sciencemag.org / content / 351 / 6280 / 1463.abstract. McLane LM, Abdel-Hakeem MS, and Wherry EJ (2019) "CD8 T Cell Exhaustion During Chronic Viral Infection and Cancer," Annual Review of Immunology. doi:10.1146 / annurev-immunol-041015-055318. Moraes LA, Ampomah PB, and Lim LHK (2018) "Annexin A1 in inflammation and breast cancer: a new axis in the tumor microenvironment," Cell Adhesion & Migration, 12(5), pp.417~423. doi:10.1080 / 19336918.2018.1486143. Motzer RJ et al. (2020) "Avelumab plus axitinib versus sunitinib in advanced renal cell carcinoma: biomarker analysis of the phase 3 JAVELIN Renal 101 trial," Nature Medicine, 26(11), pp.1733~1741. doi:10.1038 / s41591-020-1044-8. Newman AM et al. (2015) "Robust enumeration of cell subsets from tissue expression profiles," Nature Methods, 12(5), pp.453-457. doi:10.1038 / nmeth.3337. Okuda H. et al. (2012) "Hyaluronan Synthase HAS2 Promotes Tumor Progression in Bone by Stimulating the Interaction of Breast Cancer Stem-Like Cells with Macrophages and Stromal Cells," Cancer Research, 72(2), pp.537 LP~547.doi:10.1158 / 0008-5472.CAN-11-1678. Pelly VS et al. (2021) "Anti-inflammatory drugs remodel the tumor immune environment to enhance immune checkpoint blockade efficacy," Cancer Discovery, p.candisc.1815.2020.doi:10.1158 / 2159-8290.CD-20-1815. Polidoro MA et al. (2020) "Impact of RAS mutations on the immune infiltrate of colorectal liver metastases: A preliminary study," Journal of Leukocyte Biology, 108(2), pp.715~721. doi:10.1002 / JLB.5AB0220-608R. Puram SV et al. (2017) "Single-Cell Transcriptomic Analysis of Primary and Metastatic Tumor Ecosystems in Head and Neck Cancer," Cell, pp. 1-14. doi:10.1016 / j.cell.2017.10.044. Ramilowski JA et al. (2015) "A draft network of ligand-receptor-mediated multicellular signaling in human," Nature Communications, 6. doi:10.1038 / ncomms8866. Riaz N. et al. (2017) "Tumor and Microenvironment Evolution during Immunotherapy with Nivolumab," Cell, 171(4), pp.934~949. e15.doi:10.1016 / j.cell.2017.09.028. Ritchie ME et al. (2015) "limma powers differential expression analyses for RNA-sequencing and microarray studies," Nucleic Acids Research, 43(7), pp.e47~e47. doi:10.1093 / nar / gkv007. Rizvi, NA et al. (2015) "Mutational landscape determines sensitivity to PD-1 blockade in non-small cell lung cancer," Science, 348(6230), pp. 124 LP~128. Available at: http: / / science.sciencemag.org / content / 348 / 6230 / 124.abstract. Rooney MS et al. (2015) "Molecular and genetic properties of tumors associated with local immune cytolytic activity," Cell, 160(1-2), pp.48-61. doi:10.1016 / j.cell.2014.12.033. Rosenthal R. et al. (2019) "Neoantigen-directed immune escape in lung cancer evolution," Nature.doi:10.1038 / s41586-019-1032-7. Sade-Feldman M. et al. (2018) "Defining T Cell States Associated with Response to Checkpoint Immunotherapy in Melanoma," Cell.doi:10.1016 / J.CELL.2018.10.038. Salmon H. et al. (2019) "Host tissue determinants of tumor immunity," Nature Reviews Cancer.doi:10.1038 / s41568-019-0125-9. Shalapour S. and Karin M. (2019) "Pas de Deux: Control of Anti-tumor Immunity by Cancer-Associated Inflammation," Immunity.doi:10.1016 / j.immuni.2019.06.021. Singer M. et al. (2016) "A Distinct Gene Module for Dysfunction Uncoupled from Activation in Tumor-Infiltrating T Cells," Cell, 166(6), pp.1500~1511. e9.doi:10.1016 / j.cell.2016.08.052. Snyder A. et al. (2014) "Genetic Basis for Clinical Response to CTLA-4 Blockade in Melanoma," New England Journal of Medicine, 371(23), pp.2189~2199. doi:10.1056 / NEJMoa1406498. Snyder A. et al. (2017) "Contribution of systemic and somatic factors to clinical response and resistance to PD-L1 blockade in urothelial cancer: An exploratory multi-omic analysis," PLoS Medicine, 14(5), pp.1~24. doi:10.1371 / journal.pmed.1002309. Szklarczyk D. et al. (2019) "STRING v11: protein-protein association networks with increased coverage, supporting functional discovery in genome-wide experimental datasets," Nucleic Acids Research, 47(D1), pp.D607~D613. doi:10.1093 / nar / gky1131. Templeton IE and Musib L. (2015) "MEK inhibitors beyond monotherapy: Current and future development," Current Opinion in Pharmacology.doi:10.1016 / j.coph.2015.05.012. Thorsson V. et al. (2018) "The Immune Landscape of Cancer," Immunity, pp. 1-19. doi:10.1016 / j.immuni.2018.03.023. Todoric J. and Karin M. (2019) "The Fire within: Cell-Autonomous Mechanisms in Inflammation-Driven Cancer," Cancer Cell, 35(5), pp.714~720. doi:10.1016 / j.ccell.2019.04.001. Tumeh PC et al. (2014) "PD-1 blockade induces responses by inhibiting adaptive immune resistance," Nature, 515(7528), pp.568~571. doi:10.1038 / nature13954. Wang D. and Dubois RN (2010). "Eicosanoids and cancer." Nature Reviews Cancer, 10(3), pp. 181-193. doi:10.1038 / nrc2809. Wang L. et al. (2019) "EMT- and stroma-related gene expression and resistance to pd-1 blockade in urothelial cancer," Journal of Urology, 202(3), p.458. doi:10.1038 / s41467-018-05992-x. Wang Y. et al. (2019) "Epiregulin reprograms cancer-associated fibroblasts and facilitates oral squamous cell carcinoma invasion via JAK2-STAT3 pathway," Journal of Experimental & Clinical Cancer Research, 38(1), p.274, doi:10.1186 / s13046-019-1277-x. Yuen KC et al. (2020) "High systemic and tumor-associated IL8 correlates with reduced clinical benefit of PD-L1 blockade," Nature Medicine, 26 (May). doi:10.1038 / s41591-020-0860-1. Zelenay S. et al. (2015) "Cyclooxygenase-Dependent Tumor Growth through Evasion of Immunity," Cell, 162(6), pp.1257~1270. doi:10.1016 / j.cell.2015.08.015. For standard molecular biology techniques, please refer to Sambrook J., Russell DW, Molecular Cloning, A Laboratory Manual, 3rd edition, 2001, Cold Spring Harbor, New York: Cold Spring Harbor Laboratory Press.

Claims

1. A computer implementation method for predicting the therapeutic response of mammalian cancer patients to anti-cancer immunotherapy, a) Providing a set of features including the expression levels of at least eight, at least nine, at least ten, at least eleven, at least twelve or more (e.g., all) of the following cancer-promoting genes previously measured in samples obtained from tumors of mammalian cancer patients: PTGS2, CXCL1, CXCL2, CXCL5, CXCL6, IL1A, HAS2, IL11, IL1B, EREG, IL6, CXCL8 (IL8), MMP12, NRG1, VCAN, AREG; b) A step of calculating a score using the features provided in a), c) A step in which the treatment response of a mammalian cancer patient is predicted based on the score calculated in step b) and A method that includes this.

2. The computer implementation method according to claim 1, wherein the features further include Clonal_TMB and CXCL9.

3. The computer implementation method according to claim 1 or claim 2, wherein the features include the expression levels of all genes: PTGS2, CXCL1, CXCL2, CXCL5, CXCL6, IL1A, HAS2, IL11, IL1B, EREG, IL6, CXCL8 (IL8), MMP12, NRG1, VCAN, and AREG.

4. The computer implementation method according to any one of claims 1 to 3, wherein the gene expression level is a normalized gene expression level.

5. The computer implementation method according to any one of claims 1 to 4, wherein the score is calculated by normalizing the individual gene expression values ​​of the features using Z-scores and calculating the average of the normalized measurements.

6. The score is calculated using the formula: [Math 1] It is calculated according to the formula, where n ｐ This is the number of the aforementioned gene expression features, [Math 2] (i) is a positively correlated gene within a unit value interval, and (e) represents the gene expression value expressed in log2 count per million (CPM); at any choice [Math 3] The computer implementation method according to any one of claims 1 to 5, wherein is the average expression value of the log2 converted count per million (read per kilobase million (FPKM)) of a positive gene.

7. The score is calculated using the formula: [Math 4] This is a probability score calculated using, where β ０ The method according to any one of claims 1 to 5, wherein x is the intercept weight, β is a vector of weights for each of the k variables, x is a vector of variables related to the sample, and the variables include the plurality of features.

8. The method according to any one of claims 1 to 7, wherein the computer implementation method further comprises evaluating tumor burden and / or neoantigen frequency in cancer patients.

9. Predicting the treatment response in mammalian cancer patients includes comparing the score calculated from step b) with the reference median score derived from the responder sample cohort and the reference median score derived from the non-responder sample cohort. The method according to any one of claims 1 to 8, wherein the sample cohort of responders and non-responders is i) cancer patients having the same type of cancer as the cancer patient, or ii) cancer patients independent of tumor type.

10. The method according to any one of claims 1 to 9, wherein the score calculated in step b) indicates that the cancer is predicted to respond to anti-cancer immunotherapy, and the method further comprises selecting mammalian cancer patients for anti-cancer immunotherapy.

11. The method according to any one of claims 1 to 10, wherein the gene expression levels of at least eight cancer-promoting genes are obtained by measuring the gene expression levels of at least eight, at least nine, at least ten, at least eleven, at least twelve or more (e.g., all) of the following cancer-promoting genes in a sample obtained from a tumor of a mammalian cancer patient: PTGS2, CXCL1, CXCL2, CXCL5, CXCL6, IL1A, HAS2, IL11, IL1B, EREG, IL6, CXCL8 (IL8), MMP12, NRG1, VCAN, AREG.

12. A method for predicting the therapeutic response to anti-cancer immunotherapy in mammalian cancer patients, a) A step of measuring multiple features, including measuring the gene expression of at least eight, at least nine, at least ten, at least eleven, at least twelve or more (e.g., all) of the following cancer-promoting genes in a sample obtained from a patient's tumor: PTGS2, CXCL1, CXCL2, CXCL5, CXCL6, IL1A, HAS2, IL11, IL1B, EREG, IL6, CXCL8 (IL8), MMP12, NRG1, VCAN, AREG, b) A step of determining the score using the features measured in a), c) A step in which the patient's treatment response is predicted based on the score determined in step b) A method that includes this.

13. The method according to claim 12, wherein the features further include Clonal_TMB and CXCL9.

14. The step of determining a score to predict the treatment response of the subject is The step involves inputting data representing features including the gene expression levels of at least eight cancer-promoting genes into a machine learning classifier, wherein the machine learning classifier is trained using a training dataset containing the gene expression levels of at least the same cancer-promoting genes that form the score of the target, and the training dataset is derived from at least two sample cohorts. One cohort includes multiple samples from mammalian cancer patients known to have responded to anti-cancer immunotherapy (responders), and the other cohort includes multiple samples from mammalian cancer patients known not to have responded to anti-cancer immunotherapy (non-responders), in the step of... The steps include: having a machine learning classifier classify the target score as belonging to either the response group or the non-response group; The method according to any one of claims 1 to 13, including the method described in any one of claims 1 to 13.

15. The method according to claim 14, wherein the training dataset includes gene expression levels of at least eight cancer-promoting genes from at least 50, at least 100, at least 150, at least 200, at least 250, and at least 300 samples from mammalian cancer patients known to have responded to anti-cancer immunotherapy (responders), and gene expression levels of at least eight cancer-promoting genes from at least 50, at least 100, at least 150, at least 200, at least 250, at least 300, at least 350, at least 400, at least 450, at least 500, and at least 550 samples from mammalian cancer patients known not to have responded to anti-cancer immunotherapy (non-responders).

16. The machine learning classifier is trained using multiple scores from at least three sample cohorts. The method according to claim 14 or 15, wherein one cohort comprises multiple samples from multiple mammalian cancer patients known to have fully responded to anti-cancer immunotherapy (full responders), one cohort comprises multiple samples from multiple mammalian cancer patients known to have partially responded to anti-cancer immunotherapy (partial responders), and the other cohort comprises multiple samples from mammalian cancer patients known not to respond to anti-cancer immunotherapy (non-responders).

17. The method according to any one of claims 14 to 16, wherein the machine learning classifier is a logistic regression model, an elastic network regression model, or a tree-based model, such as a random forest classifier.

18. A method for classifying samples obtained from mammalian cancer subjects, wherein the samples include nucleic acids and / or proteins. The method is, a) A step of analyzing the sample to obtain the gene expression levels of at least eight cancer-promoting genes selected from PTGS2, CXCL1, CXCL2, CXCL5, CXCL6, IL1A, HAS2, IL11, IL1B, EREG, IL6, CXCL8 (IL8), MMP12, NRG1, VCAN, and AREG, b) A step of performing the method according to any one of claims 1 to 17 using the gene expression level obtained in step a); A method that includes this.

19. The method according to claim 18, wherein the sample is derived from one or more target cancer cells.

20. A method for treating cancer in mammalian cancer patients, a) A step of predicting the therapeutic response of a mammalian cancer patient to anti-cancer immunotherapy by performing the method according to any one of claims 1 to 19, b) step a) a step of determining that the mammalian cancer patient is expected to respond to the anti-cancer immunotherapy, c) The step of administering a therapeutically effective dose of the anti-cancer immunotherapy to a subject in need thereof. A method that includes this.

21. A method for treating cancer in mammalian cancer patients, a) the step of performing the method according to any one of claims 1 to 20; b) A step of determining whether the score of the mammalian cancer patient indicates that the mammalian cancer patient is predicted to respond to anti-cancer immunotherapy; c) The step of administering immunotherapy to mammalian cancer patients who require immunotherapy. A method that includes this.

22. The method according to any one of claims 1 to 21, further comprising analyzing one or more additional markers of anti-cancer immunotherapy response derived from mammalian cancer patients to supplement and / or confirm predictions of treatment resistance or non-response.

23. The method according to any one of claims 1 to 22, wherein the sample obtained from the patient's tumor is a formalin-fixed paraffin-embedded (FFPE) sample.

24. A system for predicting the therapeutic response to anti-cancer immunotherapy in mammalian cancer patients, At least one processor, When executed by at least one processor, at least one non-temporary computer-readable medium includes an instruction causing at least one processor to perform the method according to any one of claims 1 to 17, and A system that includes this.

25. One or more computer-readable media, which, when executed by one or more processors, include instructions causing one or more processors to perform steps according to any one of claims 1 to 17.

26. A pharmaceutical composition for use in the treatment of a mammalian cancer patient, including anti-cancer immunotherapy, wherein the patient is expected to respond to the anti-cancer immunotherapy by the method according to any one of claims 1 to 19.

27. A method, system, and / or pharmaceutical composition according to any one of claims 1 to 26, wherein a mammalian cancer patient has, and / or is any of the following cancers: kidney cancer, e.g., renal cell carcinoma or renal cell carcinoma (KIRC); gastroesophageal tumor; esophageal cancer (ECA); melanoma; bladder cancer (BLCA); stomach cancer; lung cancer, e.g., adenocarcinoma of the lung (LUAD), squamous cell carcinoma of the lung (LUSC), or non-small cell lung cancer (NSCLC); pancreatic cancer (PAAD); liver cancer (LIHC); glioblastoma (GBM); adrenal cortical cancer (ACC); cervical cancer (CESC); head and neck cancer, e.g., squamous cell carcinoma of the head and neck (HNSC); thyroid cancer (THCA); uterine cancer (UCEC); giant cell tumor of the tendon sheath (TGCT); melanoma; BRCA; sarcoma and pancancerous cancer.

28. The method, system, and / or pharmaceutical composition according to claim 27, wherein the renal cell carcinoma is clear cell renal cell carcinoma (ccRCC).

29. The method, system, and / or pharmaceutical composition according to any one of claims 1 to 28, wherein the immune subtype of cancer is one of the following subtypes: wound healing, IFNγ dominant, TGFβ dominant, inflammatory, and / or lymphocyte depletion.

30. The method, system, and / or pharmaceutical composition according to any one of claims 1 to 29, wherein the anti-cancer immunotherapy includes IL-2 immunotherapy or immune checkpoint blockade therapy alone or in combination with anti-VEGF therapy.

31. The method, system, and / or pharmaceutical composition according to claim 30, wherein the immune checkpoint blockade therapy comprises blockade of programmed cell death-1 (PD-1), blockade of programmed cell death ligand 1 (PD-L1), and / or blockade of cytotoxic T lymphocyte-associated protein 4 (CTLA-4).

32. The method, system, and / or pharmaceutical composition according to claim 30 or 31, wherein the immune checkpoint blockade therapy comprises treatment with nivolumab, pembrolizumab, atezolizumab, and / or ipilimumab.

33. The method, system, and / or pharmaceutical composition according to claim 30, wherein the anti-VEGF therapy is bevacizumab.

Images