Method of predicting benefit from immune checkpoint inhibition therapy

Abstract

To provide alternative methods based on alternative promoter utilization to predict patient response to immunotherapy or immune checkpoint inhibition (ICI) therapy.SOLUTION: A method of identifying a cancer patient as benefiting or not benefiting from ICI therapy comprises the steps of: measuring an expression level of one or more preselected markers in a cancerous biological sample obtained from the patient; identifying a differentially expressed alternative promoter based on the expression level of the one or more preselected markers; calculating an alternative promoter usage score (APB score); and identifying the patient as benefiting or not benefiting from immune checkpoint inhibition therapy using the APB score.SELECTED DRAWING: None

Description

[Technical field]

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS This application claims the benefit of priority to Singapore Application No. 10201811546W, filed December 21, 2018, the contents of which are incorporated herein by reference in their entirety for all purposes.

[0002] FIELD OF THEINVENTION The present invention relates generally to the field of cancer. In particular, the present invention relates to the use of the method for selecting patients for immune checkpoint inhibitor therapy. [Background technology]

[0003] Among various cancer treatments, immune checkpoint inhibition (ICI) has provided important breakthroughs in several tumor types. In ICI therapy, immune checkpoint inhibitors such as pembrolizumab and nivolumab block the interaction between the immune checkpoint receptor PD-1 and its ligand, reducing negative costimulatory signals and increasing T effector cell function to induce antitumor responses.

[0004] Although benefiting from certain tumor types, some recent ICI Phase III trials have also demonstrated limited success, particularly in solid epithelial tumors. Responses to immunotherapy are observed in only a minority of patients. Therefore, there is a need for robust biomarkers that can predict patient response to immunotherapy. Currently, the most developed ICI biomarkers are PD-L1 expression, microsatellite instability, and tumor mutation burden. These biomarkers are positive predictive biomarkers aimed at identifying patients who will respond to ICI therapy. However, controversy has arisen surrounding these biomarkers, and ICI responses have been observed in biomarker-negative populations. These observations highlight the complementary requirement of negative predictive biomarkers for ICI therapy that can identify tumors that are likely to be resistant to immunotherapy. However, negative predictive biomarkers for ICI remain poorly described.

[0005] A promoter is a genomic cis-regulatory element upstream of a transcription start site (TSS) that functions to initiate transcription. Promoter activity is epigenetically regulated, and more than half of all human genes have multiple promoters that can be selectively activated as a result of normal biological function or disease states. The use of alternative promoters allows the use of distinct 5' untranslated regions (UTRs) and The first exon and the second exon can be produced, enhancing the diversity of mRNA and protein isoforms. In cancer, alternative promoters may generate cancer-specific isoforms with oncogenic potential. Recent studies have shown that tumors may utilize alternative promoters as a mechanism for immune editing and evasion. Thus, epigenetically driven utilization of alternative promoters is a potential mechanism of resistance to ICI therapy. Summary of the Invention [Problem to be solved by the invention]

[0006] Therefore, there is a need to develop alternative methods based on the use of alternative promoters to predict patient response to immunotherapy or ICI therapy. [Means for solving the problem]

[0007] overview In one aspect, a method of identifying a patient suffering from cancer as benefiting or not benefiting from immune checkpoint inhibition (ICI) therapy, comprising: a) measuring the expression level of one or more preselected markers in a cancer biological sample obtained from a patient; b) identifying alternative differentially expressed promoters based on the expression levels of the one or more pre-selected markers measured in step (a); c) calculating an alternative promoter usage score; d) Using the alternative promoter usage score to identify patients as benefiting or not benefiting from ICI therapy. A method is provided, comprising:

[0008] In another aspect, there is provided a method for determining a prognosis of a patient suffering from cancer, comprising: a) measuring the expression level of one or more preselected markers in a cancerous biological sample obtained from a patient; b) identifying alternative differentially expressed promoters based on the expression levels of the one or more pre-selected markers measured in step (a); c) calculating an alternative promoter usage score; d) determining a patient prognosis using the alternative promoter usage score A method is provided, comprising:

[0009] In another aspect, there is provided a biomarker for excluding a patient suffering from cancer from ICI therapy, wherein the biomarker is an alternative promoter usage score that exceeds the alternative promoter usage score of one or more reference samples, and the alternative promoter usage score is the sum of the alternative promoters that are differentially expressed at each alternative promoter site.

[0010] definition As used herein, the term "promoter" refers to a region of DNA that initiates transcription of a gene. A promoter may be a major promoter, a minor promoter, or an alternative promoter. A major promoter is the promoter most frequently used to transcribe a gene.

[0011] As used herein, the terms "alternate promoter" and "alternative The term "alternative promoter" refers to a region of DNA that initiates transcription of a gene at an alternative transcription start site rather than the major or minor promoter.

[0012] As used herein, the term "gained promoter" refers to a non- It refers to a promoter that has gained or increased activity in a cancerous biological sample compared to a cancerous biological sample. A gained promoter can be a promoter in a cancerous biological sample that is not present in a non-cancerous biological sample. A gained promoter can be a promoter in a cancerous biological sample that has increased promoter activity compared to a promoter in a non-cancerous biological sample.

[0013] As used herein, the term "lost promoter" refers to a non-cancer A lost promoter refers to a promoter that has lost or reduced activity in a cancerous biological sample compared to a cancerous biological sample. A lost promoter can be a promoter that is present in a non-cancerous biological sample but not present in a cancerous biological sample. A lost promoter refers to a promoter that has reduced activity compared to a promoter in a non-cancerous biological sample. The promoter may be a promoter in a cancerous biological sample that has inhibitor activity.

[0014] The term "immunotherapy" in the context of cancer refers to a form of therapy that involves modulation of the immune system to treat cancer. Modulation of the immune system can involve activation or inactivation of the immune system. This can involve the use of components of the immune system such as antibodies, cytokines, and vaccines.

[0015] As used herein, the term "immune checkpoint inhibitor therapy" refers to a form of cancer immunotherapy that targets key regulators of the immune system. Certain proteins made by certain types of cancer cells and immune cells can help suppress immune responses and inhibit T cells from killing cancer cells. Certain cancer cells use these proteins to evade immune surveillance. The term "immune checkpoint inhibitor therapy" refers to therapies that involve the use of agents that block these proteins, restoring immune system function and allowing T cells to mount an effective anti-tumor response.

[0016] The term "prognosis" for purposes of this application refers to a prediction of the expected course and outcome of a clinical condition or disease. The term "prognosis" does not refer to the ability to predict the course or outcome of a condition with 100% accuracy. Instead, the term "prognosis" refers to the probability that a particular course or outcome will occur in a subject exhibiting a given condition as compared to an individual not exhibiting the given condition. A "prognosis" may be made with respect to one or more clinical outcomes, such as the rate of disease progression in a subject, the severity of the disease, survival rate, survival time, likelihood of metastasis, likelihood of disease recurrence, or response to a therapeutic intervention.

[0017] For the purposes of this application, the term "epigenetic" refers to changes that regulate gene expression without altering an individual's DNA sequence. Examples of epigenetic changes are DNA methylation, histone modification, and RNA-associated silencing. Epigenetic changes play an important role in disease development through the control of gene expression.

[0018] The term "biomarker" in the context of this specification refers to a measurable indicator of a biological state or disease. Biomarkers include, but are not limited to, a substance, structure, or process that can be measured in body tissues or fluids.

[0019] As used herein, the terms "alternate promoter usage score" and "alternative promoter usage score" refer to a measure of alternative promoter utilization. The score may be expressed as an "AP" score or an "APB" score. AP 高 and AP 低 The term "AP" in this context refers to alternative promoter utilization. 高 The score indicates high alternative promoter utilization, and AP 低 The score indicates low alternative promoter utilization. APB 高 and A.P.B. 低 The term "APB" in this context refers to Alternative Promoter Benefit Burden. 高 The score indicates high alternative promoter utilization, and APB 低 The score indicates low alternative promoter utilization. [Brief description of the drawings]

[0020] The invention will be better understood by reference to the detailed description when considered in conjunction with the non-limiting examples and the accompanying drawings, in which: [Figure 1]Figure 1 shows alternative promoter utilization burden (APB score) across 33 tumor types. APB scores were calculated for 10,393 samples across 33 tumor types. Each point represents the APB score of a sample (y-axis). Tumor types are ordered on the x-axis from lowest to highest median APB score for each tumor type. The red bars represent the median APB scores of the tumors and the green bars represent the median APB scores of the corresponding normal tissues (only if at least 10 normal samples were available for analysis). The 80th centile cutoff is shown as it was found to be the most ideal cut point for dichotomizing samples into APB-high and APB-low groups. [Figure 2A] ~ [Figure 2C] Figure 1 shows alternative promoter utilization in gastric cancer (discovery cohort, n=24). In particular, (A) shows a heat map of alternative promoter utilization in the discovery cohort of patients treated with nivolumab or pembrolizumab. Transcripts with expression levels 4-fold higher than the median level across all tumors and mapping to previously identified gained alternative promoter sites were considered as gained alternative promoters (marked in red in the heat map). Transcripts with expression levels 4-fold lower than the median level across all tumors and mapping to previously identified lost alternative promoter sites were considered as lost alternative promoters (marked in blue in the heat map). (B) shows the association with T cell immune correlates between AP high (disc) group vs. AP low (disc) group. AP high (disc) group is shown in red and AP low (disc) group in blue. Expression of T cell marker CD8A (P=0.059) and T cell cytolytic markers GZMA (P=0.025) and PRF1 (P=0.011) are shown. The AP high (disc) group shows lower expression of immune markers. (C) Kaplan-Meier curves of progression-free survival comparing AP high vs. AP low groups in the discovery cohort. [Figure 3A] ~ [Figure 3D]Figure 1 shows alternative promoter utilization in gastric cancer (pembrolizumab study cohort, n=37). In particular, (A) shows a heat map of alternative promoter utilization. Transcripts with expression levels higher than 4-fold compared to the median level across all tumors and mapping to previously identified gained alternative promoter sites were considered as gained alternative promoters (marked in red in the heat map). Transcripts with expression levels lower than 0.25-fold compared to the median level across all tumors and mapping to previously identified lost alternative promoter sites were considered as lost alternative promoters (marked in blue in the heat map). (B) shows a graph of alternative promoter utilization score. The alternative promoter utilization score is calculated as the sum of gained and lost alternative promoters in each sample. High alternative promoter utilization was defined as above the 66th centile. (C) shows the association with T cell immune correlates between AP high vs. AP low groups. AP high group is shown in red and AP low group in blue. Expression of T cell marker CD8A (P=0.0037) and T cell cytolytic markers GZMA (P=0.0055) and PRF1 (P=0.016) are illustrated. The AP high group shows lower expression of immune markers. (D) Waterfall plot of response to pembrolizumab according to AP high (red) and AP low (blue) subgroups. The Y-axis represents the percentage of maximum tumor reduction assessed according to RECIST 1.1 criteria. [Figure 4A] ~ [Figure 4C] Figure 1 shows survival curves based on alternative promoter utilization. In particular, (A) shows Kaplan-Meier curves of progression-free survival comparing AP-high group vs. AP-low group. (B) shows a swimmer plot where the x-axis represents the duration of pembrolizumab therapy for each patient. AP-high (red) and AP-low (blue) subgroups are illustrated. (C) shows Kaplan-Meier curves of progression-free survival comparing TCGA subtypes divided by AP-high group vs. AP-low group. [Figure 5-1] ~ [Figure 5-2]Figure 1 shows the bioinformatics workflow of the APB score algorithm. Somatic promoter regions (gains and losses) identified in gastric cancer (2732 regions) are mapped against all promoters identified in the pan-cancer analysis from TCGA (113,076 promoters). In total, 4672 promoters were selected and the relative promoter activity was used to calculate the APB score. Tumors were dichotomized into APB-high and APB-low at the 80th centile of the APB score. [Figure 6A] ~ [Figure 6B]Figure 1 shows correlation between APB score groups and markers of cytolytic T cell activity. In particular, (A) shows the association of T cell immune correlates between APB high vs. APB low groups (all tumor samples, pan-cancer). Tumors were dichotomized into APB high and APB low at the 80th centile of APB score. The APB high group is shown in red and the APB low group in blue. Expression of T cell marker CD8A and T cell cytolytic markers GZMA and PRF1 are illustrated. The APB high group shows lower expression of immune markers (Wilcoxon test p<0.0001 (***) for CD8A, GZMA, and PRF1). (B) shows the associations between APB-high vs. APB-low groups and with CD8A, GZMA, and PRF1 for selected tumor types: bladder cancer (BLCA), breast cancer (BRCA), cervical squamous cell carcinoma (CESC), esophageal cancer (ESCA), head and neck squamous cell carcinoma (HNSC), liver hepatocellular carcinoma (LIHC), lung squamous cell carcinoma (LUSC), ovarian cancer (OV), and stomach adenocarincoma (STAD). APB-high and APB-low groups are marked in red and blue, respectively. (***=p<0.0001, **=p<0.001, *=p<0.01, NS=p>0.05, Wilcoxon test). [Figure 7]Figure 1 shows differential gene expression in APB score groups highlighting immune genes. In this volcano plot of immune genes correlated with APB high and APB low, the x-axis is the log2 fold change (log2FC) of gene expression (RSEM) between APB high and APB low. The y-axis is the resulting -log10 adjusted p-value (Bonferroni corrected). Significant genes defined as log2FC>1 and adjusted p-value<0.05 are represented by red dots. Top genes (log2FC>1.3 and -log10 adjusted p-value>200) are labeled. [Figure 8] Figure 1. Association of APB score groups with other immune correlates. In particular, the figure shows a volcano plot of approximately 20,000 genes in PanCanAtlas that correlate with APB high and APB low. The x-axis is the log2 fold change (log2FC) of gene expression (RSEM) between APB high and APB low. The y-axis is the result of -log10 adjusted p-value (Bonferroni correction). Immune genes are represented by red dots. An equal division of overall gene expression is seen between both groups, with the APB high group showing a bias towards under-representation of immune genes. [Figure 9] Correlation between lymphocyte infiltration signature score and APB score. Lymphocyte infiltration signature is dichotomized by APB score group. APB low group has a statistically higher score compared to APB high (Wilcoxon test p<0.0001). [Figure 10] Correlation between IFN-γ response signature and APB score. IFN-γ response signature defined by dichotomizing by APB score group. APB low group has statistically higher score compared to APB high (Wilcoxon test p<0.0001). [Figure 11]Figure 1 shows the correlation between APB score and pembrolizumab objective response rate (ORR) in 19 tumor types selected by correlation with CD8A, GZMA, and PRF1. The x-axis is the median APB score per tumor type, and the y-axis is the ORR of the tumor type with single-agent pembrolizumab, as described in published studies. The size of the circle corresponds to the number of patients treated in the clinical trials selected for ORR evaluation. The color of the circle corresponds to the number of samples in the TCGA cohort used to evaluate the APB score. COAD includes both COAD and colon and rectal cancer (READ). DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0021] Detailed Description of the Invention In a first aspect, the present invention provides a method for identifying a patient suffering from cancer as benefiting or not benefiting from immune checkpoint inhibition (ICI) therapy, comprising: a) measuring the expression level of one or more preselected markers in a cancerous biological sample obtained from a patient; b) identifying alternative differentially expressed promoters based on the expression levels of the one or more pre-selected markers measured in step (a); c) calculating an alternative promoter usage score; d) Using the alternative promoter usage score to identify patients as benefiting or not benefiting from ICI therapy. The present invention refers to a method, including:

[0022] It will be generally understood by those skilled in the art that the cancer biological sample may be a fresh, frozen, fixed, or preserved sample.

[0023] In one embodiment, the patient suffering from cancer has received or is receiving ICI therapy. In a preferred embodiment, the patient suffering from cancer is not receiving ICI therapy.

[0024] Cancer biological samples may be obtained from patients diagnosed with one or more cancers, including, but not limited to, carcinoma, sarcoma, and melanoma. The biological sample may be a sample of cells, tissues, or bodily fluids.

[0025] In some embodiments, the cancerous biological sample may be obtained from a patient suffering from one or more cancers, including, but not limited to, adrenocortical carcinoma, bladder urothelial carcinoma, breast invasive carcinoma, cervical squamous cell carcinoma and adenocarcinoma, bile duct carcinoma, colon adenocarcinoma, diffuse large B-cell lymphoma, esophageal carcinoma, glioblastoma multiforme, head and neck squamous cell carcinoma, kidney chromophobe cell, kidney clear cell carcinoma, kidney papillary cell carcinoma, acute myeloid leukemia, brain low grade glioma, liver hepatocellular carcinoma, lung adenocarcinoma, lung squamous cell carcinoma, mesothelioma, ovarian serous cystadenocarcinoma, pancreatic adenocarcinoma, pheochromocytoma and paraganglioma, prostate adenocarcinoma, rectal adenocarcinoma, sarcoma, cutaneous melanoma, gastric adenocarcinoma, testicular germinoma, thyroid carcinoma, thymoma, uterine endometrial carcinoma, uterine carcinosarcoma, uveal melanoma, anal carcinoma, and combinations thereof.

[0026] In one embodiment, the cancer is a solid cancer. In another embodiment, the solid cancer is a gastrointestinal cancer. In yet another embodiment, the gastrointestinal cancer is a gastric cancer. In yet another embodiment, the gastric cancer is of the chromosomally unstable (CIN) and / or genomically stable (GS) subtype. In yet another embodiment, the gastric cancer is a metastatic gastric cancer.

[0027] In some embodiments, the preselection marker is one or more nucleic acid molecules. The nucleic acid molecule can be DNA or RNA. In some embodiments, the nucleic acid molecule is selected from the group consisting of mRNA, cDNA, microRNA, and genomic DNA. The preselection marker can be one or more genes or one or more transcripts. In some embodiments, the preselection marker is one or more peptides. In another embodiment, the peptide is a post-translationally modified peptide.

[0028] In one embodiment, the preselected marker is a marker associated with a somatic promoter region in the cancerous biological sample. In a preferred embodiment, the cancerous biological sample is a gastric cancer sample.

[0029] The expression levels of one or more preselected markers in a cancerous biological sample described herein may be measured using a method selected from the group consisting of whole transcriptome sequencing (WTS), NanoString analysis, RNA sequencing, and combinations thereof.

[0030] It will be appreciated by those skilled in the art that the use of alternative promoters allows for the initiation of transcription at different transcription start sites, and therefore may affect gene expression in a variety of ways, including altering gene expression levels and / or resulting in the production of altered mRNA transcripts and protein isoforms.

[0031] In one embodiment, the differentially expressed alternative promoters described herein are identified by comparing the expression level of one or more of said preselected markers in a cancerous biological sample with the expression level of said one or more preselected markers in one or more reference samples to determine an increase or decrease in the expression level of the preselected markers between the cancerous biological sample and the one or more reference samples. It will be understood by those skilled in the art that the expression level of the preselected markers, which may be one or more genes, transcripts or peptides, may be regulated by the promoter and thus is indicative of the activity of the promoter.

[0032] Differentially expressed alternative promoters can be identified by comparing the expression level of one or more preselected markers in a cancerous biological sample with the expression level of one or more preselected markers in one or more reference samples.

[0033] In one embodiment, for the obtained promoter, an increase in the expression level of the one or more pre-selected markers in the cancerous biological sample compared to the expression level of the one or more pre-selected markers in the one or more reference samples is indicative of a differentially expressed alternative promoter.

[0034] In another embodiment, for the missing promoter, a decrease in the expression level of the one or more pre-selected markers in the cancerous biological sample compared to the expression level of the one or more pre-selected markers in the one or more reference samples is indicative of a differentially expressed alternative promoter.

[0035] Differentially expressed alternative promoters can be identified in various ways. In one embodiment, the differentially expressed alternative promoters described herein are identified by gain promoters with at least a 4-fold increase in transcript expression level when compared to the median expression level of the promoter's transcript in a panel of tumor samples. In another embodiment, the differentially expressed alternative promoters described herein are identified by loss promoters with less than 0.25-fold or at least a 4-fold reduction in transcript expression level when compared to the median expression level of the promoter's transcript in a panel of tumor samples. In one embodiment, the median expression level of each promoter's transcript is the median expression level of the promoter across all samples in the panel of tumor samples.

[0036] In another embodiment, the differentially expressed alternative promoters described herein are identified by gain promoters with relative promoter activity greater than the median promoter activity of the panel of tumor samples. In another embodiment, the differentially expressed alternative promoters described herein are identified by loss promoters with relative promoter activity less than the median promoter activity of the panel of tumor samples. In one embodiment, the median expression level of the transcripts of each promoter is greater than the median promoter activity of all the promoters in the panel of tumor samples. Figure 1 shows the median expression level of the promoter across samples.

[0037] In one embodiment, an gained promoter is a promoter that is gained or has increased activity in a cancerous biological sample compared to a non-cancerous biological sample. An gained promoter can be a promoter in the cancerous biological sample that is not present in the non-cancerous biological sample, or a promoter in the cancerous biological sample that has increased promoter activity compared to the promoter in the non-cancerous biological sample.

[0038] In another embodiment, the missing promoter is a promoter that is missing or has reduced activity in a cancerous biological sample compared to a non-cancerous biological sample. A missing promoter can be a promoter that is present in a non-cancerous biological sample but not present in a cancerous biological sample, or a promoter in a cancerous biological sample that has reduced promoter activity compared to the promoter in the non-cancerous biological sample.

[0039] After identifying the differentially expressed alternative promoters described herein, an alternative promoter usage score is then calculated.In one embodiment, the alternative promoter usage score can be an AP score.In another embodiment, the alternative promoter usage score can be an APB score.The alternative promoter usage score can be calculated by determining the sum of the differentially expressed alternative promoters described herein.

[0040] The alternative promoter usage score may be used in a method for identifying a patient suffering from cancer as benefiting or not benefiting from ICI therapy. The method for identifying a patient suffering from cancer as benefiting or not benefiting from ICI therapy may further comprise comparing the alternative promoter usage score described herein with a reference score to identify the patient as benefiting or not benefiting from ICI therapy.

[0041] In one embodiment, the reference score is the median, tertile, or quartile score of the alternative promoter usage scores from one or more reference samples. In another embodiment, the reference score is the 10th, 20th, 30th, 40th, 50th, 60th, 70th, 80th, or 90th percentile score of the alternative promoter usage scores from one or more reference samples. In a preferred embodiment, the reference score may be the 66th or 80th percentile score of the alternative promoter usage scores from one or more reference samples.

[0042] In a preferred embodiment, the reference score is an absolute score, which is a fixed and non-variable score that is compared to the patient's alternative promoter score.

[0043] In one embodiment, an increased alternative promoter usage score compared to the reference score identifies the patient as not benefiting from ICI therapy. In one embodiment, the increased alternative promoter usage score is a score above the 66th percentile of the alternative promoter usage scores from one or more reference samples. In another embodiment, the increased alternative promoter usage score is a score above the 80th percentile of the alternative promoter usage scores from one or more reference samples.

[0044] In another embodiment, a decreased alternative promoter usage score compared to the reference score identifies the patient as benefiting from ICI therapy. In one embodiment, a decreased alternative promoter usage score is a score less than the 66th percentile of the alternative promoter usage scores from one or more reference samples. In one embodiment, a decreased alternative promoter usage score is a score less than the 66th percentile of the alternative promoter usage scores from one or more reference samples. - The usage score is a score less than the 80th percentile of the alternative promoter usage scores from one or more reference samples.

[0045] In one aspect, the present invention provides a method for determining the prognosis of a patient undergoing or who has undergone ICI therapy, comprising the steps of: a) measuring the expression level of one or more preselected markers in a cancerous biological sample obtained from a patient; b) identifying alternative differentially expressed promoters based on the expression levels of the one or more pre-selected markers measured in step (a); c) calculating an alternative promoter usage score; d) determining a patient prognosis using the alternative promoter usage score The present invention provides a method comprising:

[0046] In one embodiment, the method of determining the prognosis of a patient receiving or who has received ICI therapy described herein further comprises comparing the alternative promoter usage score described herein to a reference score to determine the patient's prognosis, wherein an increased alternative promoter usage score compared to the reference score indicates a worse prognosis.

[0047] In one embodiment, the increased alternative promoter usage score is a score above the 66th percentile of the alternative promoter usage scores from one or more reference samples. In another embodiment, the increased alternative promoter usage score is a score above the 80th percentile of the alternative promoter usage scores from one or more reference samples.

[0048] In another embodiment, the method of determining the prognosis of a patient undergoing or having undergone ICI therapy as described herein further comprises comparing the alternative promoter usage score as described herein to a reference score to determine the prognosis of the patient, wherein a decreased alternative promoter usage score compared to the reference score indicates a more favorable prognosis.

[0049] In one embodiment, a reduced alternative promoter usage score is a score that is less than the 66th percentile of the alternative promoter usage scores from one or more reference samples. In another embodiment, a reduced alternative promoter usage score is a score that is less than the 80th percentile of the alternative promoter usage scores from one or more reference samples.

[0050] In another aspect, the present invention provides a biomarker for excluding a patient suffering from cancer from ICI therapy, wherein the biomarker is an alternative promoter usage score that exceeds the alternative promoter usage score of one or more reference samples, wherein the alternative promoter usage score is the sum of the alternative promoters that are differentially expressed at each alternative promoter site.

[0051] In one embodiment, the present invention provides a biomarker for excluding patients suffering from cancer from ICI therapy, wherein the biomarker is an alternative promoter usage score that is greater than the 66th percentile of the alternative promoter usage scores of one or more reference samples, and the alternative promoter usage score is the sum of the alternative promoters that are differentially expressed at each alternative promoter site.

[0052] In another embodiment, the present invention provides a biomarker for selecting patients suffering from cancer for ICI therapy, wherein the biomarker has an alternative promoter usage score that is less than the 66th percentile of the alternative promoter usage scores of one or more reference samples, and the alternative promoter usage score is the sum of the alternative promoters that are differentially expressed at each alternative promoter site.

[0053] In another aspect, the present invention provides a biomarker for excluding a patient suffering from cancer from ICI therapy, wherein the biomarker is an alternative promoter usage score that is greater than the 80th percentile of the alternative promoter usage scores of one or more reference samples, and the alternative promoter usage score is the sum of the alternative promoters that are differentially expressed at each alternative promoter site.

[0054] In another embodiment, the present invention provides a biomarker for selecting patients suffering from cancer for ICI therapy, wherein the biomarker has an alternative promoter usage score that is less than the 80th percentile of the alternative promoter usage scores of one or more reference samples, and the alternative promoter usage score is the sum of the alternative promoters that are differentially expressed at each alternative promoter site.

[0055] The one or more reference samples may be selected from the group consisting of one or more tumor samples obtained from one or more different patients, one or more non-cancerous samples obtained from one or more different patients, one or more non-cancerous samples obtained from the same patient, and combinations thereof. In one embodiment, the one or more non-cancerous samples obtained from the same patient are tissues adjacent to the cancerous tissue. It will be understood by those skilled in the art that multiple combinations of these reference samples can be used in the method of the present invention.

[0056] In another aspect, the present invention provides a method for determining whether a patient suffering from cancer should not be treated with immune checkpoint inhibitor (ICI) therapy, comprising: a) measuring the expression level of one or more preselected markers in a cancerous biological sample obtained from a patient; b) identifying alternative differentially expressed promoters based on the expression levels of the one or more pre-selected markers measured in step (a); c) calculating an alternative promoter usage score; d) identifying patients as not benefiting from ICI therapy using an alternative promoter usage score; e) if a patient is identified as not benefiting from ICI therapy, treating said patient with an alternative therapy. The present invention refers to a method, including:

[0057] In another aspect, the invention provides a method of treating a patient suffering from cancer, comprising: a) measuring the expression level of one or more preselected markers in a cancerous biological sample obtained from a patient; b) identifying alternative differentially expressed promoters based on the expression levels of the one or more pre-selected markers measured in step (a); c) calculating an alternative promoter usage score; d) identifying patients as benefiting or not benefiting from ICI therapy using an alternative promoter usage score; e) treating patients identified according to step (d) as benefiting from ICI therapy with an immune checkpoint inhibitor or treating patients identified according to step (d) as not benefiting from ICI therapy with an alternative therapy. The present invention refers to a method, including:

[0058] In one embodiment, the ICI therapy may include, but is not limited to, ipilimumab, nivolumab, pembrolizumab, atezolizumab, avelumab, durvalumab, cemiplimab, spartalizumab, sintilimab, camrelizumab, and tislelizumab.

[0059] In another embodiment, alternative therapy is any therapy other than ICI therapy suitable for treating cancer.For example, alternative therapy can include chemotherapy, radiotherapy, stem cell transplantation, surgery, hormone therapy, and targeted therapy.Targeted therapy includes the use of drugs that interfere with specific molecules required for tumor growth and progression.Alternative therapy can also include discontinuing ICI therapy in patients who are receiving ICI therapy.

[0060] The invention illustratively described herein can suitably be practiced in the absence of any element or elements, limitation or limitations not specifically disclosed herein. Thus, for example, the terms "comprising," "including," "containing," etc. shall be read expansively and without limitation. Further, as used herein, The terms and expressions employed are used as terms of description rather than of limitation, and there is no intention in the use of such terms and expressions to exclude any equivalents of the features shown and described or portions thereof, but it is recognized that various modifications are possible within the scope of the invention claimed. Thus, while the present invention has been specifically disclosed by preferred embodiments and optional features, it is to be understood that modifications and variations of the invention disclosed and embodied herein may be resorted to by those skilled in the art, and such modifications and variations are considered to be within the scope of the present invention.

[0061] The invention has been described broadly and generically herein. Each narrower species and subgeneric series falling within the scope of the generic disclosure also forms part of the invention. This includes the generic description of the invention with a provisos or negative limitation removing any subject matter from the genus, regardless of whether the excised material is specifically described herein.

[0062] Other embodiments are within the scope of the following claims and non-limiting examples. Furthermore, when features or aspects of the invention are described in terms of a Markush group, one of ordinary skill in the art will recognize that the invention is also thereby described in terms of any individual members or subgroups of members of the Markush group.

[0063] Experimental section The non-limiting examples and comparative examples of the present invention are further illustrated in detail by reference to specific examples, which should not be construed as limiting the scope of the present invention in any way.

[0064] Materials and Methods In this study, the REMARK criteria for validation of tumor biomarkers were followed.

[0065] Clinical cohort Discovery cohort Consecutive patients with metastatic gastric cancer treated with nivolumab or pembrolizumab therapy at Samsung Medical Center, Seoul, Korea, were included in this cohort. ICIs were administered as salvage treatment for patients who had failed at least one cytotoxic regimen. Nivolumab 3 mg / kg was administered as a 1-h infusion every 2 weeks, and pembrolizumab 200 mg was administered as a 30-min intravenous infusion every 3 weeks until disease progression or toxicity became unacceptable. After ethical approval was obtained and all patients provided written informed consent, archival tumor tissue specimens from primary tumors were collected and survival data were followed prospectively.

[0066] Pembrolizumab study cohort Patients with histologically proven metastatic and / or recurrent gastric adenocarcinoma who had failed at least one line of platinum / fluoropyrimidine-containing chemotherapy were enrolled in this study. The study was performed in accordance with the Declaration of Helsinki and Good Clinical Practice guidelines (ClinicalTrials.gov Identifier: NCT#02589496). The study protocol was approved by the Institutional Review Board of Samsung Medical Center (Seoul, Korea), and all patients provided written informed consent before enrollment. Pembrolizumab 200 mg was administered as a 30-minute intravenous infusion every 3 weeks until disease progression, unacceptable toxicity was documented, or for a maximum of 24 months. Tumor response was assessed every 2 cycles according to RECIST 1.1 criteria.

[0067] Nanostring Analysis The NanoString nCounter reporter code set contains promoter-associated genes of 80 recurrent somatic alternatives, as well as immune-related genes corresponding to intra-tumoural cytolytic activity (CYT), cytokines, and immune checkpoints. At least two probes were designed for each gene to measure expression of canonical and alternative promoter-driven transcripts: a canonical probe for unmodified H3K4me3-marked 5' transcripts, and an alternative probe for somatic promoter 5' transcripts. Data analysis was performed using the vendor-provided nCounter software (nSolver). Raw counts were normalized using the geometric mean of the internal positive control probes included in each code set.

[0068] RNA sequencing Tumor tissues were obtained between days -42 and 1 before the start of study treatment. If tumor content was estimated to be greater than 40% after thorough pathological review, tumor DNA and RNA were extracted from freshly obtained tissues using a QIAamp mini kit (Qiagen, Hilden, Germany) according to the manufacturer's instructions. Concentrations and 260 / 280 and 260 / 230 nm ratios were measured using an ND1000 spectrophotometer (Nanodrop Technologies, Thermo-Fisher Scientific, MA, USA), after which DNA / RNA was further quantified using a Qubit fluorometer (Life Technologies, CA, USA).

[0069] RNA transcriptomic analysis RNAseq data were aligned to GENCODE v19 transcript annotations using TopHat, and FPKM abundance measurements were generated using Cufflinks. Transcripts were then fused across all samples and normalized using Cuffnorm. To analyze alternative promoter-associated expression, RNAseq reads were mapped to genomic locations previously identified by epigenome profiling. RNAseq mapping to these epigenome-defined promoter regions was then quantified and normalized by promoter length and library size. Finally, fold changes in expression at each promoter site were calculated between each tumor and median expression levels across all tumor samples.

[0070] PDL1 immunohistochemistry analysis, MSI status, EBV status, TCGA subtyping, and tumor mutation burden were based on the classification used for the phase II study of single-agent pembrolizumab in metastatic gastric cancer.

[0071] statistical analysis Correlation of clinicopathological characteristics to histological subclassification was performed using Fisher's exact test. Progression-free survival (PFS) was significantly longer in the pembrolizumab group than in the control group. Overall survival (OS) was calculated from the time of first administration of pembrolizumab or nivolumab to the time of death. Kaplan-Meier (KM) curves and log-rank tests were used for survival analysis. Hazard ratios (HR) and their 95% confidence intervals (CI) were used. val) was assessed in each analysis using the Cox proportional hazards regression model. All analyses This was done using R (3.4.1). In the validation cohort, samples with alternative promoter usage scores above the 66th percentile were considered to have high alternative promoter usage (AP 高 ), and the rest are defined as low alternative promoter usage (AP 低 ) was defined as

[0072] PDL1 immunochemistry PDL1 immunohistochemistry was performed using the Dako PD-L1 IHC 22C3 pharmDx kit (Agilent Technologies). PD-L1 protein expression was measured using PD-L1 stained cells (tumor The number of viable tumor cells (tumor cells, lymphocytes, and macrophages) was determined using the CPS, which was calculated by dividing the number of viable tumor cells by the total number of viable tumor cells and multiplying the result by 100.

[0073] MSI Status The MSI status of tumor tissues was determined by both IHC for MLH1 and MSH2 and PCR analysis of five markers with mononucleotide repeats.

[0074] EBV status subtype EBV status was determined by EBV-encoded small RNA (EBER) in situ hybridization.

[0075] Tumor mutation burden Mutational load was determined from whole-exome sequencing analysis. Mutation burden was defined as the number of somatic nonsynonymous SNVs that passed all filters. Somatic mutations were annotated with mutation effect predictors. Mutation burden was calculated as the number of nonsynonymous SNVs in the tumor exome data. The ML-H threshold for tissues was set as the upper tertile.

[0076] TGCA Subtype Definitions Gastric cancer subtypes defined by TCGA were based on DNA genomic alterations. These groups included EBV(+), MSI-H, CIN, and genomically stable tumors lacking CIN and highly enriched in diffuse histological subtypes. As a surrogate for CIN, EBV(-), MSS tumors were stratified into CIN and genomically stable based on their TP53 status. Mutational signature analysis was performed using the deconstructSigs package in R (v This was performed using 1.6.0.

[0077] Dataset Promoter activity was inferred from RNA-Seq data available from TCGA's PanCanAtlas, consisting of 10393 samples (9668 tumor and 725 normal samples) across 33 tumor types.

[0078] APB Score Algorithm The first epigenetic promoter alteration study in gastric cancer identified 2732 somatic promoter regions (2053 gains and 679 losses) and used the transcriptomic expression of these isoforms to generate an APB score algorithm.

[0079] For TCGA data, the GENCODE v19 annotation was used to determine a set of promoters. The overlapping first exons of each TSS were combined to obtain a set of promoters. The activity of each promoter was then quantified using junction reads aligning to the first introns of the constituent transcripts. The total junction read counts were then normalized across the entire dataset. The log2-transformed normalized reads were The log counts were used for promoter activity for further downstream analysis. Estimates of gene expression were obtained by summing each promoter activity per gene. Each promoter activity was then normalized by gene expression to obtain relative promoter activity. In total, 113,076 promoters were identified.

[0080] Somatic promoter regions identified in gastric cancer were then selected from this set of 113,076 promoters. 4672 promoters (3263 in gains and 1409 in losses) were located within the 2732 somatic promoter regions identified in gastric cancer. Of these 4672 promoters, the median relative promoter activity of all promoters across all tumor samples (n=9668) was calculated. For all samples, the APB score was calculated as the number of gains in relative promoter activity greater than the median and loss promoters with relative promoter activity less than the median (Figure 1). APs at various centile cutoffs (10, 20, 30…, 90) were calculated as the number of gains in relative promoter activity greater than the median and loss promoters with relative promoter activity less than the median (Figure 1). Using the B score, tumors were classified as APB 高 and A.P.B. 低 The 80th centile was the APB. 高 and A.P.B. 低The ideal cutoff between groups was determined and used for further downstream analysis. Normal samples were only added to the cohort for analyses involving normal tissues and were not used in the calculation of median scores and cutoffs.

[0081] Immune Correlates A selection of over 700 genes, including transcriptional expression levels of CD8A, GZMA and PRF1, as well as genes and antigens related to immune checkpoints, markers of various immune cell types, adaptive and innate immune responses, were selected and analyzed using TCGA data from the Broad GDAC Firehose. The immune subtypes, other immune signatures, and TMB were extracted from the pan-cancer immune landscape analysis. Progression-free survival (PFS) and overall survival (OS) including censored data were extracted from the pan-cancer immune landscape analysis.

[0082] Clinical Data for Immune Checkpoint Inhibitors Data from clinical trials involving immune checkpoint inhibitors are expanding rapidly, with drugs being tested as single agents and in combination with other therapies. Pembrolizumab remains the immune checkpoint inhibitor that has been tested across the broadest spectrum of tumor types for which clinical trial data have been reported. Therefore, an extensive literature review of single-agent pembrolizumab trials was conducted and the largest reported studies for each tumor type were selected. Objective response rates (ORR) from these trials were then correlated with the APB score for each tumor type. TMB was extracted from the pan-cancer immune landscape analysis, and PD-L1 transcript expression was extracted from the Broad GDAC Firehose.

[0083] statistical analysis Using the Wilcoxon rank sum test, APB 高 and A.P.B. 低Expression levels of CD8A, GZMA, and PRF1 were compared between groups. T-SNE plots were generated using the "Rtsne" R package. Pearson's test was used for correlation between APB score and ORR of pembrolizumab. Kaplan-Meier (KM) curves and log-rank tests were used for survival analysis. Hazard ratios (HRs) and their 95% confidence intervals were evaluated for each analysis using the Cox proportional hazards regression model. The Reactome PathwayDatabase was used to analyze the body Network mapping of cellular promoter gene function was performed. All analyses were performed using R (3.5.2).

[0084] Validation of promoter activity estimates using RNA-Seq The accuracy of promoter activity estimation using RNA-Seq was validated by benchmarking against other "gold standard" measurements such as H3K4me3 ChIP-Seq and CAGE tag data available in public datasets. H3K4me3 levels were measured using R The results were strongly correlated with NA-Seq promoter activity, suggesting that epigenetic- and transcript-based promoter activity estimates are concordant (Kruskal-Wallis p<0.001). Findings were confirmed with CAGE-Tag data, with higher CAGE-T Unique promoters were identified in RNA-Seq with ag support.

[0085] The algorithm was also compared to other bioinformatic methods used for RNA-Seq quantification and first exon read counting. The algorithm obtained similar results to these algorithms, with a high level of correlation (Pearson correlation coefficient >0.85). Overall, this analysis demonstrates that quantitative, robust and reproducible estimation of promoter activity from RNA-Seq data is possible with our approach. EXAMPLES

[0086] Alternative promoter utilization in metastatic gastric cancer The first cohort consisted of 24 metastatic gastric cancer patients treated with nivolumab and pembrolizumab (29 subjects were initially included, 24 tumor samples passed sufficient tissue quality control for NanoString analysis). A customized NanoString panel was used to measure transcripts associated with either canonical or alternative promoters. Differentially expressed alternative promoters were defined as promoter sites showing <0.25x fold change (for lost somatic promoters) or >4x fold change (for gained somatic promoters) in expression levels across the median of all samples. Using this algorithm, it was found that one third of tumors (8 / 24) showed high alternative promoter utilization in >10% of sites (>8 / 80). This group used AP 高 and the rest is AP 低 was defined as (Figure 2A).

[0087] Measurement of cytolytic T cell activity was previously described by studying the expression of CD8A (CD8+ tumor-infiltrating lymphocytes), granzyme A (GZMA) and perforin 1 (PRF1). AP 低 The group is 高 demonstrated a significant increase in the expression of GZMA (P = 0.025), PRF1 (P = 0.011), and CD8A (P = 0.059) when compared with the A group. P 低 These findings are consistent with previous results described in early-stage gastric cancer and indicate that alternative promoter utilization in metastatic gastric cancer also correlates inversely with antitumor immunity. Notably, despite the small sample size, heterogeneity of treatment regimens, and nontrial-based nature of this cohort, AP 高 Tumor patients are AP 低There was a trend towards worse progression-free survival (PFS) in patients with BRAF compared with patients with BRAF (129 vs. 389 days, HR 1.96 95% CI: 0.55–6.93, P = 0.29, Fig. 2C). Based on these findings, we further tested this hypothesis in a separate cohort of uniformly treated patients. EXAMPLES

[0088] Alternative promoter utilization as a predictor of response and survival with pembrolizumab treatment For the second cohort, transcriptomic data from the previously described phase II study were used. Transcriptomic data from pretreatment biopsy samples and matched clinical data were available for 37 subjects and were used for analysis. Median age was 57 years, 73% were male (N=27), 4 (11%) were EBV positive, 4 were MSI (11%), and the remainder were defined as CIN or GS TCGA subtype. Complete or partial response to therapy was seen in 11 subjects (30%). Using 2732 somatic alternative promoter sites previously identified in gastric cancer

[10] , differentially expressed alternative promoters were defined similarly to the first cohort (<0.25x fold change for known somatically lost promoters and >4x fold change for known somatically gained promoters). Notably, RNAseq and nanostring platforms were used to assess alternative promoter utilization. Good agreement between the forms has been shown previously. The sum of differentially expressed sites in each sample was calculated to define an alternative promoter usage score (Figure 3A). Scores ranged from 37 to 426 (median 136) (Figure 3B). Data from the first cohort were used to guide cutoff points, and AP 高 The group was defined as samples >66th centile (n=13), AP 低 constituted the remaining samples.

[0089] AP 高 The group is 低There were no statistically significant differences in clinicopathological characteristics with respect to age, sex, or histological subtype compared with the AP group. No differences were detected between the TCGA subtype, mutation burden, and PDL1 CPS score between the two groups (Table 1). 低 The group is 高 demonstrated a significant increase in the expression of CD8A (P = 0.0037), GZMA (P = 0.0055), and PRF1 (P = 0.016) when compared with the AP group. 低 G These results suggest an increase in cytotoxic T cell activity in the AP loop (Figure 3C). The objective response rate (ORR), defined as either a partial or complete response to therapy, was significantly higher in AP 高 Compared with the group, AP 低 The mean age of the AP group was 10 / 24 vs. 1 / 13, P = 0.03 (Fig. 3D). 高 In the group, the only responses were in MSI subtype tumors. Median progression-free survival (PFS) was AP 高 AP 低 The mean time to treatment was 180 days in the AP group (log-rank P = 0.0076) (Figs. 4A and 5B). 低 The group had 17% EBV (n=4) and 12% MSI (n=3) TCGA subtype samples, whereas AP 高 Group 8% had MSI (n=1) and EBV-seropositive As previously shown, PFS among various TCGA subtypes was different (P = 0.0026), with MSI and EBV subtypes showing significantly longer survival (491 days (MSI / EBV) vs. 80 days (CIN / GS)). In particular, within the CIN / GS subtype, PFS was significantly longer than AP. 低 Group and AP 高 There was also a statistically significant difference between the groups (48 days (CIN / GS AP 高 ) vs. 161 days (CIN / GS AP 低 ), P = 0.0019) (Figure 4D). Based on preliminary overall survival data, AP 低There was a trend towards improved survival in the group (340 vs. 292 days, P = 0.16). Multivariate analysis of clinicopathological and alternative promoter utilization revealed high alternative promoter utilization as an independent predictor of PFS with pembrolizumab (HR 0.29, (95% CI 0.099-0.85), P = 0.024) (Table 2).

[0090] [Table 1] * Three samples did not have tissue available for PDL1 CPS scoring by immunohistochemistry, and all were AP 低 It was from the group.

[0091] [Table 2] EXAMPLES

[0092] Progress in the use of alternative promoters following treatment with pembrolizumab Paired biopsy samples were available for eight subjects from the second cohort, providing an opportunity to monitor tumor progression as a result of ICI treatment pressure. Post-treatment biopsies were taken from primary gastric tumors at the time of progression on pembrolizumab. Of these eight subjects, two had a partial response (PR), with response durations of 211 and 491 days (both AP 低 ), one patient had stable disease (SD), and the duration of response was 167 days (AP 低 ), and five patients with progressive disease (AP) as their best response. 高 N=3;AP 低N=2). Interestingly, we observed a very consistent shift in the direction of alternative promoter utilization based on clinical response. Notably, PR and SD tumors showed a ≥1.5-fold increase in alternative promoter utilization score in post-treatment biopsy samples compared to pre-treatment biopsy samples, whereas all five PD tumors showed a decrease in alternative promoter utilization score in post-treatment biopsy samples (Fisher's exact test, P=0.018). These results further highlight the relationship between the alternative promoter landscape and ICI treatment pressure. Further support. EXAMPLES

[0093] Alternative promoter utilization across multiple tumor types To explore the applicability of these findings to other tumor types beyond gastric cancer, we used a recently described algorithm to infer promoter activity in the PanCanAtlas RNA-seq database of 10,393 samples (9668 tumor and 725 normal samples) across 33 tumor types (Table 3). Using the e(release 19) annotation, a set of 113,076 possible promoters was compiled. Promoter activity was inferred by quantifying expression initiated at each promoter using unique junction reads, assuming that isoforms with identical or very close TSSs are regulated by the same promoter.

[0094] [Table 3]

[0095] Of the 113,076 promoters, 4672 promoters were mapped to 2732 tumor-associated promoter regions previously defined in gastric cancer. It was hypothesized that this set of promoter alterations may represent a generalized pan-tumor response to host immune recognition, not limited to gastric cancer. To test this hypothesis, the 4672 promoters were used to calculate the APB level (APB score) for each tumor (Figure 5). As shown in Figure 1, a wide range of APB scores was observed both within and between tumor types. For the entire cohort, the median APB score was 178 (range: 46–241). Tumor types with the lowest median APB scores included thyroid cancer (THCA) (median 165, range: 54–207), diffuse large cell type (LC-ALPHA ... Diffuse large B cell lymphoma (DLBC) (median: 166, range: : 108-196) and lung adenocarcinoma (LUAD, median: 167, In contrast, glioblastoma (GBM) (median: 207, range: 172-241), low grade glioma (LGG) (median: 204, range: 130-230), and testicular germ cell tumours (TGCT) (median: 189, range: 83-218) had the highest median APB scores.

[0096] Tumor APB scores were correlated with markers of T-cell cytolytic activity: CD8A, GZMA, and PRF1. As with other pan-cancer TCGA studies, the primary analysis was performed on the entire cohort of TCGA samples, tumor type agnostic, but tumor type-specific results were also provided. Previously, it was observed that dichotomizing tumors at the median or upper tertiles yielded significant correlations with CD8A, GZMA, and PRF1. However, these studies were performed with smaller sample sizes, and therefore correlations with these three genes were chosen to re-establish the ideal cutoffs for dichotomizing tumors. Therefore, APB score thresholds were tested at various centiles (10, 20, 30…90) to determine the correlation between APB scores and tumor types. 高 Group and APB 低 For the entire cohort (all tumor types), all nine dichotomous cutoffs were associated with APB 高 The tumor is APB 低 Compared to non-tumors, the levels of CD8A, GZMA, and PRF1 were significantly lower (p<0.0001) (Figure 6A). The 80th centile was the strongest correlation with the three genes, with 19 of 33 tumor types found to be significantly correlated with CD8A, GZMA, and PRF1. For subsequent analysis, tumors were classified as APB at the 80th centile (APB score cutoff: 190) of the entire cohort. 高 and A.P.B. 低 Importantly, previous findings in stomach cancer (STAD) were robustly replicated in this analysis, even when using different approaches to interrogate alternative promoters (Figure 6B). To address potential confounding by tumor type, tumor type-specific analyses were also performed, with each tumor type being assigned an APB score of 0.01 at the median APB score for that specific tumor type. 高 and A.P.B. 低The tumors were dichotomized into groups. Twenty of the 33 tumor types had significant correlations with CD8A, GZMA, and PRF1. Eight tumor types had no significant correlation at any cutoff: kidney chromophobe (KICH), DLBC, thymoma and thymic carcinoma (THYM), kidney renal cell cancer (KIRC), acute myeloid leukemia (LAML), cholangiocarcinoma (CHOL), uterine carcinoma sarcoma (UCS), pheochromocytoma, and paraganglioma (PCPG). Of these eight, CHOL, KICH, and UCS had relatively small sample sizes (n<100), and DLBC, THYM, and LAML were of hematological / immune origin. It is of the following.

[0097] Notably, when the APB score was inferred using all 113,076 identified promoters or other similarly sized promoter subsets derived from permutation studies, compared to the set of 4672 promoters originally defined in gastric cancer used to define the APB score, no similar correlation strength was observed with CD8A / GZMA / PRF1 expression (empirical p<0.001). This observation is consistent with the fact that the APB score was originally derived from gastric cancer. Our results support the hypothesis that the APB score algorithm may be applicable across multiple tumor types, which may reflect a conserved pan-cancer response to host immunity. To investigate the specific functions of tumor-associated promoter isoforms, we analyzed 570 promoters (12% of 4672) that were upregulated in at least 100 tumors (~1% of a cohort of 9668 tumors). Different tumor types utilized specific clusters of promoter isoforms, and network analysis of gene functions revealed that alternative promoters affect genes with diverse roles. This suggests that the APB score algorithm may be applicable across multiple tumor types, which may reflect a conserved pan-cancer response to host immunity. On a genome-wide scale in humans, this may suggest that alternative promoter selection may be driven by extrinsic selective pressures (e.g., host antitumor immunity) rather than by endogenous gene function. Tumor APB scores were also compared with non-malignant tissues. Of the 725 matched normal tissue samples analyzed in the PanCanAtlas, only 5% (n=36 normal samples) had APBs. 高 When normal and tumor samples were clustered by alternative promoter usage, the normal samples clustered close to, but separate from, the corresponding tumor type. In 15 of the 16 tumor types with at least 10 normal samples, tumor samples had significantly higher APB expression levels compared to normal samples. 高 were more likely to be classified as ≥ 18 years (Fisher's exact test, p < 0.001). EXAMPLES

[0098] Alternative promoter utilization between tumor and normal samples To further explore the interplay between alternative promoter utilization and tumor immunity, we extended our analysis to study the relationship between APB scores and a broad spectrum of approximately 700 immune-related genes covering multiple immune cell types, immune checkpoints, and antigens. The majority of these genes (78%) were associated with APB scores. 高 Compared with tumors, APB 低 The tumors showed significantly higher expression (Figure 7), which was consistent with the APB 高 Differential expression analysis of all approximately 20,000 genes available in the PanCanAtlas revealed that APB 高It was confirmed that the downregulation of genes in the subgroups was restricted to immune genes only (Figure 8A). In a pan-cancer immune landscape study using TGCA samples, tumors were classified into six immune subtypes characterized by distinct immune signatures - wound-healing, IFN-γ dominant, inflammatory, lymphocyte-depleted, immunologically quiet, and TGF-β dominant. If the APB score is correlated with these immune subtypes, lymphocyte-depleted, or immunologically quiet subtypes are associated with APB scores. 高 The results showed that lymphocyte-depleted APB cells were more likely to be tumor-like (Fisher's exact test, p<0.0001) (Figure 8B). 高 The tumors included a variety of tumor subtypes and were immunologically quiescent APB 高 The tumors were predominantly dominated by LGG. In contrast, IFN-γ and inflammatory subtypes were predominantly expressed in APB. 低 were more likely to be tumors (Fisher's exact test, p < 0.0001) (Figure 8B), and APB 低 Tumors were also more likely to have a lymphocytic infiltration signature score, and an IFN-γ response signature (p<0.0001, Wilcoxon test) (Figure 9, Figure 10). Importantly, APB scores were uncorrelated with TMB levels (r=0.02, p=0.02), suggesting that distinct processes drive alternative promoter utilization and the acquisition of DNA somatic mutations. EXAMPLES

[0099] Alternative promoter utilization and survival outcomes We investigated the association between APB score and progression-free survival (PFS) and overall survival (OS) data obtained from the TCGA pan-cancer analysis. To avoid confounding at the pan-cancer level, where survival outcomes may be influenced by intrinsic tissue- or site-specific characteristics, individual tumor type-specific analyses were performed. When analyzed at the tumor type-specific level, the majority of tumor types did not show any correlation between APB score and PFS or OS. Very few tumor types showed survival differences between APB groups (KIRC, LGG, LUAD, THYM). Multivariate analysis of APB score and disease subtype also suggests that APB score is not an independent prognostic predictor of survival for patients not receiving ICI therapy. EXAMPLES

[0100] Alternative promoter utilization and pembrolizumab resistance As previously shown in metastatic gastric cancer, APB vs ICI 高 To test whether tumors were more resistant, the objective response rates (Os) of pembrolizumab reported in independent studies were compared. The association between ORR and APB score was quantified. Across all tumor types, a significant negative correlation (Pearson's R = -0.46, p = 0.025) was observed between ORR and APB score. A similar correlation was not observed for CD8A (r = 0.18, p = 0.4), GZMA (r = 0.17, p = 0.4), or PRF1 (r = 0.3, p = 0.1). Consistent with previous studies, a positive association was observed with TMB (r = 0.53, p = 0.0078) and PD-L1 expression (r = 0.42, r = 0.042). Thus, These results suggest that APB score and TMB are complementary in identifying distinct subsets of tumors, with TMB selecting types with good ICI response and APB score identifying non-responders, since APB score and TMB are not correlated. Interestingly, when only tumor types showing significant APB score correlation with CD8A, GZMA, and PRF1 were selected for analysis (19 tumor types), the correlation between APB score and ORR increased in strength and significance: Pearson's R = -0.55, p = 0.019 (Figure 11). Taken together, these results extend previous findings shown in gastric cancer and establish a pan-cancer relationship between APB score and pembrolizumab resistance.

[0101] Equivalent The foregoing examples are presented for the purpose of illustrating the present invention and should not be construed as imposing any limitations on the scope of the present invention. It will be readily apparent that numerous modifications and variations can be made to the specific embodiments of the present invention described above and illustrated in the examples without departing from the principles underlying the invention. All such modifications and variations are intended to be included in this application.

Claims

[Claim 1] 1. A method for identifying a patient suffering from cancer as one who would or would not benefit from immune checkpoint inhibition (ICI) therapy, comprising: a) determining the expression level of one or more preselected markers in a cancerous biological sample obtained from said patient; b) identifying alternative differentially expressed promoters based on the expression levels of the one or more pre-selected markers measured in step (a); c) calculating an alternative promoter usage score; d) using said alternative promoter usage score to identify said patient as benefiting or not benefiting from ICI therapy; The method comprising:

Images