Search

Tumor mutation burden is an imperfect proxy of tumor foreignness and has therefore failed to consistently demonstrate clinical utility in predicting responses in the context of immunotherapy. We evaluated mutations in regions of the genome that are unlikely to undergo loss in a pan-cancer analysis across 31 tumor types (n = 9,242) and eight immunotherapy-treated cohorts of patients with non-small-cell lung cancer, melanoma, mesothelioma, and head and neck cancer (n = 524). We discovered that mutations in single-copy regions and those present in multiple copies per cell constitute a persistent tumor mutation burden (pTMB) which is linked with therapeutic response to immune checkpoint blockade. Persistent mutations were retained in the context of tumor evolution under selective pressure of immunotherapy and tumors with a high pTMB content were characterized by a more inflamed tumor microenvironment. pTMB imposes an evolutionary bottleneck that cancer cells cannot overcome and may thus drive sustained immunologic tumor control in the context of immunotherapy.

Aberrant alternative splicing can generate neoantigens, which can themselves stimulate immune responses and surveillance. Previous methods for quantifying splicing-derived neoantigens are limited by independent references and potential batch effects. Here, we introduce SpliceMutr, a bioinformatics approach and pipeline for identifying splicing derived neoantigens from tumor and normal data. SpliceMutr facilitates the identification of tumor-specific antigenic splice variants, predicts MHC-binding affinity, and estimates splicing antigenicity scores per gene. By applying this tool to genomic data from The Cancer Genome Atlas (TCGA), we generate splicing-derived neoantigens and neoantigenicity scores per sample and across all cancer types and find numerous correlations between splicing antigenicity and well-established biomarkers of anti-tumor immunity. Notably, carriers of mutations within splicing machinery genes have higher splicing antigenicity, which provides support for our approach. Further analysis of splicing antigenicity in cohorts of melanoma patients treated with mono-or combined immune checkpoint inhibition suggest that the abundance of splicing antigens is reduced post-treatment from baseline in patients who progress, likely because of an immunoediting process. We also observe increased splicing antigenicity in responders to immunotherapy, which may relate to an increased capacity to mount an immune response to splicing-derived antigens. We find the splicing antigenicity to be higher in tumor samples when compared to normal, that mutations in the splicing machinery result in increased splicing antigenicity in some cancers, and higher splicing antigenicity is associated with positive response to immune checkpoint inhibitor therapies. Further, this new computational pipeline provides novel analytical capabilities for splicing antigenicity and is openly available for further immuno-oncologic analysis.

Computational prediction of binding between neoantigen peptides and major histocompatibility complex (MHC) proteins can be used to predict patient response to cancer immunotherapy. Current neoantigen predictors focus on in silico estimation of MHC binding affinity and are limited by low predictive value for actual peptide presentation, inadequate support for rare MHC alleles, and poor scalability to high-throughput data sets. To address these limitations, we developed MHCnuggets, a deep neural network method that predicts peptide–MHC binding. MHCnuggets can predict binding for common or rare alleles of MHC class I or II with a single neural network architecture. Using a long short-term memory network (LSTM), MHCnuggets accepts peptides of variable length and is faster than other methods. When compared with methods that integrate binding affinity and MHC-bound peptide (HLAp) data from mass spectrometry, MHCnuggets yields a 4-fold increase in positive predictive value on independent HLAp data. We applied MHCnuggets to 26 cancer types in The Cancer Genome Atlas, processing 26.3 million allele–peptide comparisons in under 2.3 hours, yielding 101,326 unique predicted immunogenic missense mutations (IMM). Predicted IMM hotspots occurred in 38 genes, including 24 driver genes. Predicted IMM load was significantly associated with increased immune cell infiltration (P < 2 × 10−16), including CD8+ T cells. Only 0.16% of predicted IMMs were observed in more than 2 patients, with 61.7% of these derived from driver mutations. Thus, we describe a method for neoantigen prediction and its performance characteristics and demonstrate its utility in data sets representing multiple human cancers.

Supplemental Legend for Figure S1

Supplementary Table 1-9

Figure S1

Mesothelioma is a rare and fatal cancer with limited therapeutic options until the recent approval of combination immune checkpoint blockade. Here we report the results of the phase 2 PrE0505 trial (NCT02899195) of the anti-PD-L1 antibody durvalumab plus platinum-pemetrexed chemotherapy for 55 patients with previously untreated, unresectable pleural mesothelioma. The primary endpoint was overall survival compared to historical control with cisplatin and pemetrexed chemotherapy; secondary and exploratory endpoints included safety, progression-free survival and biomarkers of response. The combination of durvalumab with chemotherapy met the pre-specified primary endpoint, reaching a median survival of 20.4 months versus 12.1 months with historical control. Treatment-emergent adverse events were consistent with known side effects of chemotherapy, and all adverse events due to immunotherapy were grade 2 or lower. Integrated genomic and immune cell repertoire analyses revealed that a higher immunogenic mutation burden coupled with a more diverse T cell repertoire was linked to favorable clinical outcome. Structural genome-wide analyses showed a higher degree of genomic instability in responding tumors of epithelioid histology. Patients with germline alterations in cancer predisposing genes, especially those involved in DNA repair, were more likely to achieve long-term survival. Our findings indicate that concurrent durvalumab with platinum-based chemotherapy has promising clinical activity and that responses are driven by the complex genomic background of malignant pleural mesothelioma., In a phase 2 trial, the combination of chemotherapy with durvalumab, an anti-PD-L1 antibody, exhibited promising clinical activity in patients with previously untreated, unresectable mesothelioma, with additional analyses providing insights into genomic and immunologic features potentially associated with response.

Identifying neoepitopes that elicit an adaptive immune response is a major bottleneck to developing personalized cancer vaccines. Experimental validation of candidate neoepitopes is extremely resource intensive, and the vast majority of candidates are non-immunogenic, making their identification a needle-in-a-haystack problem. To address this challenge, we present computational methods for predicting MHC-I epitopes and identifying immunogenic neoepitopes with improved precision. The BigMHC method comprises an ensemble of seven pan-allelic deep neural networks trained on peptide-MHC eluted ligand data from mass spectrometry assays and transfer learned on data from assays of antigen-specific immune response. Compared with four state-of-the-art classifiers, BigMHC significantly improves the prediction of epitope presentation on a test set of 45,409 MHC ligands among 900,592 random negatives (AUROC=0.9733, AUPRC=0.8779). After transfer learning on immunogenicity data, BigMHC yields significantly higher precision than seven state-of-the-art models in identifying immunogenic neoepitopes, making BigMHC effective in clinical settings. All data and code are freely available athttps://github.com/KarchinLab/bigmhc.

INTRODUCTION: Tumor mutation burden (TMB) is a commonly used biomarker for cancer immunotherapy however TMB only partially captures tumor foreignness. We hypothesized that mutations in single-copy regions of the genome or mutations present in multiple copies (hereafter referred to as persistent mutations) are retained during cancer evolution and immunoediting, may render the tumor continuously visible to the immune system and promote sustained tumor control during immune checkpoint blockade (ICB). METHODS: We performed pan-cancer analyses of whole exome sequencing data across 31 tumor types in TCGA to quantify the landscape of persistent mutations (n=9,242). We then evaluated the association between persistent tumor mutation burden (pTMB) and ICB response compared to TMB in eight ICB-treated cohorts of patients with NSCLC, melanoma, mesothelioma, and head and neck cancer (n=524). To investigate the clonal evolution of persistent mutations we serially analyzed whole exome sequence data from NSCLCs prior to and at emergence of acquired resistance to ICB. Finally, we evaluated the composition of the tumor microenvironment (TME) in baseline and on-ICB melanomas by RNA sequencing differential enrichment analyses and deconvolution. RESULTS: Integration of sequence alterations in only-copy and multi-copy states for 9,242 tumors across 31 tumor types revealed a cancer lineage-dependent distribution of persistent mutations that was largely independent of the overall TMB. In evaluating differential classification based on pTMB- vs TMB-high, we found re-classification rates as high as 53% in individual tumor types, with a median reclassification rate of 33% across all tumor types (range 15% - 53%). We then evaluated the clonal composition of persistent mutations and found a wide range of correlations between pTMB and fraction of clonal mutations (Spearman ρ range: -0.11 - 0.59). In ICB-treated cohorts, pTMB better distinguished responding tumors compared to TMB, and a number of mutation and copy-number related features including tumor aneuploidy (melanoma: Mann-Whitney p=2.3e-06, NSLC: p CONCLUSIONS: Persistent mutations represent a biologically distinct subset within the overall TMB that is unlikely to be lost under selective pressure of ICB and may function as an intrinsic driver of sustained immunologic tumor control. Citation Format: Noushin Niknafs, Archana Balan, Christopher Cherry, Karlijn Hummelink, Kim Monkhorst, Xiaoshan M. Shao, Zineb Belcaid, Kristen A. Marrone, Joseph Murray, Kellie N. Smith, Benjamin Levy, Josephine Feliciano, Christine L. Hann, Vincent Lam, Drew M. Pardoll, Rachel Karchin, Tanguy Y. Seiwert, Julie R. Brahmer, Patrick M. Forde, Victor E. Velculescu, Valsamo K. Anagnostou. Persistent mutation burden drives sustained anti-tumor immune responses in human cancers [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr 1393.

Introduction: The magnitude of benefit from immune checkpoint inhibitors (ICI) may be sex-dependent, however the mechanisms underlying this sex dimorphism in ICI response are unknown. Methods: To explore sex-specific genomic features linked with ICI response, we analyzed whole-exome sequence data of 89 patients with non-small cell lung cancer (NSCLC) treated with ICI. Sequence alteration, mutation signatures and clonality analyses were combined with HLA class I and II genotyping, and computationally derived mutation-associated neoantigens were evaluated based on the peptide-HLA binding affinity (2]. Missense sequence alterations were characterized as putatively immunogenic mutations (IMM), and HLA class I and II matched IMM loads were calculated. Durable clinical benefit ≥6 months was used to define response. Our findings were validated in an independent cohort of ICI treated NSCLC (n=34). Results: There were no differences in the genomic features assessed or in clinical outcomes between male (n=46) and female (n=43) patients. Female responders had higher tumor mutational burden (TMB; p=0.004), in contrast to male patients (p=0.15). Similarly, a mutation spectra-derived smoking signature was associated with response in females (p=0.001) but not in males (p=0.2). Responding female patients had a more clonal TMB compared to non-responders (p=0.008). Importantly, female responders harbored a significantly higher class I and II IMM load (p=0.008 and p=0.004 respectively), which was not appreciated in males. These findings were supported by analyses of a second independent cohort, where HLA class I and II IMM load differentiated responders from non-responders only in the female patient group (p=0.008 and p=0.004 for class I and II IMM, respectively). In the primary cohort, germline HLA class I zygosity was not associated with outcome for either sex, however a trend towards HLA DQ-B1 homozygosity and worse outcome was evident only in females. Importantly, when HLA class II diversity was combined with class II IMM load, a survival benefit was noted for tumors harboring a high IMM load and maximal HLA II germline diversity in males (log rank p=0.015) but not in females. Similarly, HLA class I diversity was important in males when combined with class I IMM load, such that low IMM load combined with reduced antigen presentation capacity was predictive of non-durable clinical benefit (p=0.003 vs p=0.68 for males and females respectively) and was linked with shorter overall survival (log rank p= 0.062 vs p=0.58 for males and females respectively). Conclusions: Our findings support a sex dimorphism in the genomic determinants of response to immune checkpoint blockade, which may point to differences in immune surveillance between males and females in the context of therapy and may be used to tailor therapeutic delivery. Citation Format: Susan C. Scott, Xiaoshan M. Shao, Noushin Niknafs, Archana Balan, Mara Lanis, James White, Patrick Forde, Kristen Marrone, Vincent Lam, Josephine Feliciano, Benjamin Levy, Julie Brahmer, David Ettinger, Victor Velculescu, Rachel Karchin, Christine Hann, Jarushka Naidoo, Valsalmo Anagnostou. Sex-specific genomic determinants of response to immunotherapy [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2021; 2021 Apr 10-15 and May 17-21. Philadelphia (PA): AACR; Cancer Res 2021;81(13_Suppl):Abstract nr 1617.

Computational prediction of binding between neoantigen peptides and major histocompatibility complex (MHC) proteins is an emerging biomarker for predicting patient response to cancer immunotherapy. Current neoantigen predictors focus onin silicoestimation of MHC binding affinity and are limited by low positive predictive value for actual peptide presentation, inadequate support for rare MHC alleles and poor scalability to high-throughput data sets. To address these limitations, we developed MHCnuggets, a deep neural network method to predict peptide-MHC binding. MHCnuggets is the only method to handle binding prediction for common or rare alleles of MHC Class I or II, with a single neural network architecture. Using a long short-term memory network (LSTM), MHCnuggets accepts peptides of variable length and is capable of faster performance than other methods. When compared to methods that integrate binding affinity and HLAp data from mass spectrometry, MHCnuggets yields a fourfold increase in positive predictive value on independent MHC-bound peptide (HLAp) data. We applied MHCnuggets to 26 cancer types in TCGA, processing 26.3 million allele-peptide comparisons in under 2.3 hours, yielding 101,326 unique candidate immunogenic missense mutations (IMMs). Predicted-IMM hotspots occurred in 38 genes, including 24 driver genes. Predicted-IMM load was significantly associated with increased immune cell infiltration (p2 patients, with 61.7% of these derived from driver mutations. Our results provide a new method for neoantigen prediction with high performance characteristics and demonstrate its utility in large data sets across human cancers.SynopsisWe developed a newin silicopredictor of Major Histocompatibility Complex (MHC) ligand binding and demonstrated its utility to assess potential neoantigens and immunogenic missense mutations (IMMs) in 6613 TCGA patients.

Computational prediction of binding between neoantigen peptides and major histocompatibility complex (MHC) proteins is an emerging biomarker for predicting patient response to cancer immunotherapy. Current neoantigen predictors focus on in silico estimation of MHC binding affinity and are limited by low positive predictive value for actual peptide presentation, inadequate support for rare MHC alleles, and poor scalability to high-throughput data sets. To address these limitations, we developed MHCnuggets, a deep neural network method to predict peptide-MHC binding. MHCnuggets is the only method to handle binding prediction for common or rare alleles of MHC Class I or II, with a single neural network architecture. Using a long short-term memory network (LSTM), MHCnuggets accepts peptides of variable length and is capable of faster performance than other methods. When compared to methods that integrate binding affinity and HLAp data from mass spectrometry, MHCnuggets yields a fourfold increase in positive predictive value on independent MHC-bound peptide (HLAp) data. We applied MHCnuggets to 26 cancer types in TCGA, processing 52.6 million allele-peptide comparisons in under 2.3 hours, yielding 103,587 candidate immunogenic missense mutations (IMMs). IMM hotspots occurred in 36 genes, including 22 driver genes. Predicted IMM load was significantly associated with increased immune cell infiltration (p2 patients, with 65% of these derived from driver mutations. Our results provide a new method for neoantigen prediction with high performance characteristics and demonstrate its utility in large data sets across human cancers. Citation Format: Xiaoshan M. Shao, Rohit Bhattacharya, Justin Huang, Ashok Sivakumar, Collin Tokheim, Lily Zheng, Dylan Hirsch, Ben Kaminow, Ashton Omdahl, Maria Bonsack, Angelika B. Riemer, Victor E. Velculescu, Valsamo Anagnostou, Kymberleigh Pagel, Rachel Karchin. High-throughput prediction of MHC Class I and Class II neoantigens with MHCnuggets [abstract]. In: Proceedings of the AACR Special Conference on Tumor Immunology and Immunotherapy; 2019 Nov 17-20; Boston, MA. Philadelphia (PA): AACR; Cancer Immunol Res 2020;8(3 Suppl):Abstract nr A33.