Supplementary Table from Prediction of Immunotherapy Response in Melanoma through Combined Modeling of Neoantigen Burden and Immune-Related Resistance Mechanisms

Tumors harbor a complex ecosystem of malignant, immune, and stromal cells. While malignant cells dictate much of the tumor biology, there is evidence that the tumor microenvironment (TME) also plays a major role in disease etiology. Given the complexity and abundance of the TME cellular composition, investigating the role of immune cell types will yield novel biomarkers for tumor progression and response to therapies. The role of B cells as a prognostic biomarker remains elusive. For instance, infiltrating B cells in CRC have both positive and negative prognostic value. Thus, a scalable approach to quantify B cells and the B-cell receptor (BCR) repertoire could yield novel insights into the role of B cells in tumor biology. To address this, we have developed immune cell quantification (InfiltrateID࣪) and immune receptor repertoire profiling (RepertoireID࣪) methods as part of the ImmunoID NeXT Platform®, an augmented, immuno-oncology-optimized exome/transcriptome platform. We estimate B cell abundance and BCR repertoire by profiling FFPE and PBMC samples using ImmunoID NeXT࣪. In expanding upon InfiltrateID to further estimate B cell abundance, here we regress the bulk RNA-seq readout from a reference signature from purified immune cell types. We also generate orthogonal quantifications of B cell abundance by profiling samples with cytometry by time of flight, single-cell RNA-seq, flow cytometry, and immunohistochemistry (IHC). We compare BCR results from ImmunoID NeXT to a standalone sequencing approach to evaluate the concordance of top clones. We then utilize BCR profiling from ImmunoID NeXT to analyze clonality and isotype composition in tumor samples. We first use InfiltrateID to estimate absolute B cell fractions in over 50 samples. Overall, we observe a high correlation between InfiltrateID results and orthogonal data sets in both PBMC and tumor FFPE samples (R2=0.90). When comparing BCR results from RepertoireID to a standalone BCR sequencing method that profiles IgM and IgG, we identify 475 and 387 of the top 500 clones in IgG and IgM, respectively, with highly concordant abundances across all clones (R2>0.72 and R2>0.82 in IgM and IgG, respectively). Next, we use InfiltrateID to estimate absolute B cell fractions in over 650 samples from 14 tumor types. On average, samples display B cell fractions in agreement with the literature and IHC quantifications, with higher B cell fractions in lung, breast, and cervical tumors. We also observe a range of BCR clonality values across tumor types. Finally, we observe differences in B cell composition and repertoire diversity in tumor samples from patients who underwent checkpoint blockade therapy. We show that InfiltrateID and RepertoireID accurately capture the composition and clone diversity of infiltrating B cells in tumor samples. Citation Format: Fabio Navarro, Eric Levy, Pamela Milani, Qiang Li, Shruti Bhide, Upasana Dutta, Charles W. Abbott, Jose Jacob, Rena McClory, John West, John Lyle, Sean Boyle, Richard O. Chen. Accurate quantification of infiltrating B cell composition and clone diversity in tumor samples [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr 5021.

Abstract Gastrointestinal stromal tumors (GIST) are lethal tumors characterized by constitutively activating mutations to KIT or PDGFRA. Transient disease control in the first-line setting is achieved via inhibition of tyrosine kinase signaling using the KIT inhibitor imatinib. As patients progress through subsequent lines of therapy a molecularly heterogeneous disease evolves, characterized by distinct subtypes and shifting repertoires of exon-specific KIT variants which directly impact treatment outcomes. Here, we use tumor-informed exome-scale liquid biopsy to identify and track the evolution of multiple resistance mechanisms in patients receiving tyrosine kinase inhibitors (TKIs) to address the unmet need of comprehensive understanding of GIST evolution in response to therapy. Matched tumor, normal and serial plasma samples were obtained from 15 heavily pretreated metastatic GIST patients. Following baseline sample collection, all patients received systemic TKI therapy, and were monitored until disease progression. Exome-scale detection of somatic variants in cfDNA from longitudinal matched plasma samples was achieved using the NeXT Liquid BiopsyTM platform. The ImmunoID NeXT PlatformⓇ, an augmented exome/transcriptome platform and analysis pipeline which generates comprehensive tumor and immune data was used to profile paired tumor and normal samples. Longitudinal whole exome sequencing of plasma identified dynamic shifts in existing clones harboring exon-specific KIT mutations, and evolution of new KIT mutations arising prior to identification of tumor progression using standard imaging techniques. We detected a correlation between the number of damaging mutations detected in baseline ctDNA and tumor exon 11 KIT mutation status, suggesting that plasma mutation profiles may be KIT-variant dependent. ctDNA from patients with shorter overall survival (OS) was enriched for variants in the PI3K-AKT and MAPK pathway, potentially contributing to immune evasion observed in those patients. Additional associations were observed between gene copy-number changes and OS (P = .0097). Previous studies have demonstrated that immune infiltration and activity may be KIT variant specific, here we broaden those findings, identifying a significant correlation between TCRɑ clonality and variants detected only in plasma (P = .04), as well as a significant association between TCRβ diversity and OS (HR = 2.55, log rank P = .04). Comprehensive profiling of paired tumor tissue (WES and RNA-Seq) and WES of serially collected ctDNA sensitively and repeatedly identified evolving KIT mutations and other molecular alterations prior to radiologically confirmed disease progression. These findings suggest plasma-based monitoring of late-stage GIST malignancies may be useful for non-invasive disease tracking, providing treatment guidance prior to traditional approaches. Citation Format: Charles W. Abbott, Niamh Coleman, Jing Wang, Josette Northcott, Jason Pugh, Dan Norton, Fábio C. Navarro, Lee D. McDaniel, Eric Levy, Rachel Marty Pyke, John Lyle, Jason Harris, Gabor Bartha, Filip Janku, John West, Richard O. Chen, Sean Boyle. Exome-scale longitudinal tracking of emerging therapeutic resistance in GIST via analysis of circulating tumor DNA [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr 5161.

Human leukocyte antigen loss of heterozygosity (HLA LOH) allows cancer cells to escape immune recognition by deleting HLA alleles, causing the suppressed presentation of tumor neoantigens. Despite its importance in immunotherapy response, few methods exist to detect HLA LOH, and their accuracy is not well understood. Here, we develop DASH (Deletion of Allele-Specific HLAs), a machine learning-based algorithm to detect HLA LOH from paired tumor-normal sequencing data. With cell line mixtures, we demonstrate increased sensitivity compared to previously published tools. Moreover, our patient-specific digital PCR validation approach provides a sensitive, robust orthogonal approach that could be used for clinical validation. Using DASH on 610 patients across 15 tumor types, we find that 18% of patients have HLA LOH. Moreover, we show inflated HLA LOH rates compared to genome-wide LOH and correlations between CD274 (encodes PD-L1) expression and microsatellite instability status, suggesting the HLA LOH is a key immune resistance strategy.

HLA loss of heterozygosity (LOH) is increasingly being recognized as an important immune escape mechanism in response to checkpoint inhibitor therapy. HLA LOH reduces the repertoire of neoantigens displayed on the cell surface of cancer cells, limiting the efficacy of the immune system to detect and eliminate them. Though highly accurate HLA LOH detection algorithms are needed to allow clinical utility, the field lacks robust, allele-specific validation approaches. Moreover, algorithms of unknown sensitivity and specificity have led to significant discrepancies in the estimated occurrence of HLA LOH as an immune escape mechanism across tumor types. To address these challenges, we have developed a machine learning algorithm to detect HLA LOH (DASH - Deletion of Allele-Specific HLAs), established the accuracy of the algorithm with an allele-specific PCR validation strategy, investigated the frequencies of HLA LOH across 14 tumor types in a cohort of over 800 patients and observed allele-specific neoantigen expansion in response to immunotherapy. To build DASH, we profiled 279 patients on the ImmunoID NeXT Platform to create a training dataset. Our novel features, which account for allele-specific differences in exome probe capture and capitalize on our whole exome platform by including information about copy number alterations in the regions flanking the HLA genes, were used to train an XGBoost model. Orthogonal, allele-specific validation was required to accurately assess sensitivity and specificity for clinical utility. Thus, we profiled over 30 paired tumor-normal cell lines on the ImmunoID NeXT Platform® and identified cell lines with HLA LOH. Using in silico mixtures, we found 100% sensitivity and specificity for tumors with at least 36% tumor purity. Next, we designed a digital PCR (dPCR) assay using patient-specific, allele-specific primers that target a single HLA allele while avoiding all other HLA alleles and tested the limit of detection of the assay in the same cell lines. Then, we performed dPCR with patient-specific primers on 20 tumor and normal sample pairs and found 94% sensitivity. After establishing the high sensitivity and specificity of DASH, we profiled over 800 patients spanning 14 tumor types on the ImmunoID NeXT Platform. We found that over 25% of patients in the majority of tumor types had at least one HLA LOH event. Further, we observed that novel neoantigens that arose during checkpoint treatment were significantly more likely to bind to deleted HLA alleles as compared to the remaining HLA alleles in a head and neck carcinoma cohort treated with anti-PD-1 therapy, shedding light on the mechanism of immune escape in response to checkpoint inhibitors. In summary, we introduced an HLA LOH detection method, performed allele-specific validation, exposed widespread HLA across tumor types and observed the mechanism of immune escape in response to immunotherapy. Citation Format: Rachel Marty Pyke, Datta Mellacheruvu, Charles Abbott, Steven Dea, Eric Levy, Simo V. Zhang, Nikita Bedi, A. Dimitrios Colevas, Devayani Bhave, Manju Chinnappa, Gabor Bartha, John Lyle, John West, Michael Snyder, John Sunwoo, Richard Chen, Sean Michael Boyle. Pan-cancer survey of HLA loss of heterozygosity using a robustly validated NGS-based machine learning algorithm [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 399.

BackgroundComprehensive profiling of both the tumor and tumor microenvironment (TME) can help further our understanding of tumor progression and response to treatment. Many immune features can be extracted from transcriptomic data, including characterization of the immune infiltrate and profiling the diversity of immune receptors. To address this, we have developed multiple TME profiling features as part of the ImmunoID NeXT Platform®, an augmented, immuno-oncology-optimized exome/transcriptome platform designed to provide comprehensive information regarding the tumor and TME from a single FFPE tumor sample. These features including quantification of immune cell infiltration and profiling of the T-cell receptor (TCR) and B-cell receptor (BCR).MethodsTo develop our immune infiltrate quantification method, we profiled the transcriptomes of eight purified immune cell types using ImmunoID NeXT™ to develop platform-specific gene sets, and compared our transcriptome quantification to immune cell quantification with IHC. For TCR and BCR methods, we analyzed the reproducibility of clone results, and compared top clones to standalone TCR and BCR sequencing approaches. In addition, we characterized the immune content of over 800 tumor samples across 14 cancer types. Finally, we analyzed the immune features in a cohort of melanoma patients who underwent PD-1 blockade.ResultsWe observe significant concordances between cell fractions by IHC and ImmunoID NeXT’s transcriptome-based scores in tumor FFPE samples for B cells, CD8+ T cells, and macrophages (R2>0.82, R2>0.75, and R2>0.52, respectively). For TCR and BCR methods, abundances of clones shared between subsequent curls of a tumor FFPE sample have very high concordances (R2>0.89, R2>0.92, and R2>0.76 for TRB, IgG, and IgA, respectively). Compared to the standalone approaches, we identify 100% of the top 500 TRB clones and 95% of the top 500 IgG clones, with highly concordant abundances (R2>0.94 and R2>0.82 for TRB and IgG, respectively) in a PBMC sample. We identify biologically-relevant immune signatures across tumor types by characterizing the immune features across over 800 tumor samples. Finally, in a melanoma cohort, TRB clonality and CD8+ T cell scores are significantly different in responders to checkpoint inhibition.ConclusionsRNA sequencing can be used as a scalable approach to profile the immune composition in tumors. Such analysis can add to our understanding of the tumor-immune interaction, including studies of response to immunotherapy. We show that immune infiltrate quantification and TCR and BCR profiling – all part of the ImmunoID NeXT Platform – are able to accurately and effectively evaluate the composition and diversity of tumor-infiltrating immune cells.

Background Typical liquid biopsy panels offer a limited understanding of tumor biology, potentially under-representing the heterogeneity of resistance in late-stage cancers. Here, diminished scope can result in undetected, therapeutically-relevant biomarkers which respond dynamically to treatment, as well as potentially missed resistance mechanisms and pathway-level events. To address the challenges associated with identifying multiple concurrent heterogeneous resistance mechanisms in individual patients, we evaluated longitudinal exome-scale tumor-informed cell-free DNA (cfDNA) data from head and neck squamous cell carcinoma (HNSCC) patients receiving anti-PD1 therapy. Methods Pre- and post-intervention matched tumor, normal and plasma samples were retrospectively obtained from 15 stage II-IV HNSCC patients. Following baseline sample collection, all patients received a single dose of nivolumab or pembrolizumab. The primary tumor was then resected approximately one month later when possible, or a second biopsy collected where resection was impractical. Paired tumor and normal samples were then profiled using ImmunoID NeXT Platform®, an augmented exome/transcriptome platform and analysis pipeline. Exome-scale cfDNA profiling of matched plasma samples was performed using the NeXT Liquid BiopsyTM platform to detect somatic variants. Results Patient neoantigen presentation score (NEOPSTM) rapidly and significantly contracted following therapy (p=.00098). Novel neoantigens arising post-treatment which were predicted to be presented on lost HLA alleles were significantly higher in patients with longer overall survival (p=.019). Variant detection across same-patient serial cfDNA samples revealed significantly correlated VAFs (R=.62, p<.0001) despite significant contraction of mutational burden in solid tumor (p=.0039), suggesting complex clonal/subclonal dynamics. Investigation of the evolving tumor and cfDNA subclonal architecture revealed significant association between decreasing cellular prevalence and NOTCH signaling (q=.001) and the innate immune system (q=.002), while increasing cellular prevalence was associated with p53 signalling (q=.02) and hypoxia (q=.02). These findings were complimented by transcriptomic data which showed significant enrichment of multiple immune pathways across treatment. Conclusions We found that immune checkpoint blockade precipitates rapid evolution of the HNSCC tumor microenvironment. By leveraging comprehensive, tumor-informed liquid biopsy data we were able to identify contracting cellular populations enriched for NOTCH pathway mutations. Longer OS following either intervention was associated with an expansion of novel neoantigens predicted to be presented by lost HLA alleles. Our results suggest that tumor-informed liquid biopsy provides a more robust understanding of therapeutic response and resistance mechanisms than that attainable with typical liquid biopsy panels alone. Ethics Approval This study obtained ethics approval from Human Subjects Research at Stanford University. ID number is 40425. All participants gave informed consent prior to enrollment.

e21548 Background: While immune checkpoint blockade (ICB) has become a pillar of cancer treatment, biomarkers that consistently predict patient response remain elusive due to the complex mechanisms driving immune response to tumors. We hypothesized that a multi-dimensional approach modeling both tumor and immune-related molecular mechanisms would better predict ICB response than simpler mutation-focused biomarkers. Methods: Pre-treatment samples from 51 unresectable, stage III/IV melanoma patients who underwent anti-PD-1 therapy were characterized to assess factors influencing response. RECIST criteria were used to evaluate tumor response to therapy, with a median follow-up of 24 months. For each patient, a single paired FFPE tumor and normal blood sample was collected and profiled using Personalis’ ImmunoID NeXT Platform; an augmented exome/transcriptome platform and analysis pipeline which produces comprehensive tumor mutation information, gene expression profiling, neoantigen characterization, HLA typing and LOH, TCR repertoire profiling, and tumor microenvironment profiling. These data were then integrated to form a composite neoantigen presentation score (NEOPS) for each patient. Results: We demonstrate increasing predictive power with deeper modeling of neoantigens and immune-related resistance mechanisms to ICB. Neoantigen burden score (NBS) was derived from SHERPA, our exome and transcriptome integrative neoantigen prediction model. We found that NBS more significantly stratified responders and non-responders (P = 0.016) than TMB alone (P = 0.049). A broader score, NEOPS, incorporated DASH, to capture HLA allele-specific loss of heterozygosity and additional antigen presenting machinery resistance mechanisms. NEOPS consequently demonstrated a stronger association with response (p = 0.002). We used area under the curve (AUC) to benchmark NEOPS performance against established mutational and transcriptomic predictive models, demonstrating the model was more predictive of response in both the treatment naive setting, as well as in a mixed cohort of treatment-naive and experienced patients. These findings were confirmed in an independent cohort of patients (n = 110), suggesting that NEOPS is a robust, novel biomarker of ICB response in melanoma. Conclusions: Given the complex nature of resistance to immunotherapy, as well as potential toxicities associated with treatment, there is a need for biomarkers that can more accurately predict response. Here we demonstrate that NEOPS can significantly improve stratification of patient response. We also demonstrate that data intensive biomarkers like NEOPS can be clinically practical, with comprehensive tumor profiling in our clinical cohort achieved using very limited tumor tissue.

Abstract Due to the complexity of tumor-immune interactions, comprehensive profiling of both the tumor and tumor microenvironment (TME) can help further our understanding of tumor progression and response to treatment. One significant application is quantification of the immune infiltrate, which has the potential to characterize response to checkpoint blockade therapy. While common experimental approaches exist to profile tumor-infiltrating immune cells, they can have significant practical limitations. Instead, RNA sequencing can be used to comprehensively profile the immune composition of the TME in a scalable way. To address this, we have developed an approach to quantify eight immune cell types in tumor samples, and compare it to quantification by orthogonal methods. This module adds to the ImmunoID NeXT Platform®, an augmented, immuno-oncology-optimized exome/transcriptome platform designed to provide comprehensive information regarding the tumor and TME from a single FFPE tumor sample. To generate our reference data, we profiled the transcriptomes of eight purified immune cell types using ImmunoID NeXT™. Then, we utilized multiple methodologies and sample types to generate orthogonal quantification of immune cells for testing. These include creating in vitro cell mixtures and profiling by flow cytometry, profiling healthy donor PBMCs with cytometry by time of flight (CyTOF), profiling of dissociated tumor cells (DTCs) with FACS, and immunofluorescence (IF) characterization of FFPE tumor samples. We also used ImmunoID NeXT to profile the immune infiltrate of over 800 tumor samples across 13 cancer types. We utilized the transcriptomes of the purified immune cells to develop reference expression signatures specific for each cell type. Then, we compared ImmunoID NeXT's transcriptome-based approach to the orthogonal quantification from the variety of samples. We observed a strong concordance of the quantification from our transcriptome-based approach to the flow cytometry results of the cell mixtures, demonstrating accuracy in purified immune cell populations. We also observed strong concordance to CyTOF results from healthy donor PBMC samples, showing accuracy in real samples with diverse immune populations. Further, we demonstrated strong concordance of our methodology to FACS results in DTCs and IF in FFPE tumor samples, ensuring that our approach is able to profile the immune composition in various tumor samples. Finally, we highlight the diversity of immune populations across cancer types by applying ImmunoID NeXT to over 800 tumor samples. RNA sequencing can be used as a scalable approach to profile the immune infiltrate in tumors. Such analysis can add to our understanding of the tumor-immune interaction, including studies of response to immunotherapy. We show that the ImmunoID NeXT Platform is able to accurately evaluate the composition of tumor-infiltrating immune cells. Citation Format: Eric Levy, Pamela Milani, Fabio Navarro, Charles W. Abbott, Robert Power, John West, John Lyle, Sean M. Boyle, Richard Chen. Profiling tumor-infiltrating immune cells using an augmented transcriptome [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 2241.

BackgroundHuman Leukocyte Antigen (HLA) genes are critical for the presentation of neoantigens to the immune system by cancer cells. Deletion of HLA alleles, known as HLA loss of heterozygosity (LOH), has been highlighted as a key immune escape mechanism. Validated algorithms to detect HLA LOH from sequencing data are critical for exploring the biological impact of HLA LOH and assessing its utility as a clinical biomarker.MethodsWe developed DASH (Deletion of Allele-Specific HLAs), a machine learning algorithm trained on data from 279 patients on the ImmunoID NeXT Platform using features that account for probe capture variability between alleles and incorporate information from the regions flanking each HLA gene. To understand the contribution of boosted sequencing in the HLA region of the ImmunoID NeXT Platform, we performed an in silico downsampling analysis. To assess DASH’s performance at variable tumor purities and HLA LOH subclonalities we identified three tumor-normal cell lines with HLA LOH and created in silico mixtures. Furthermore, after designing patient-specific primers for 21 patients that target specific alleles, we applied digital PCR (dPCR) to validate the HLA allele copy number status of the patients. Finally, we applied DASH to 611 patients spanning 15 tumor types.ResultsIn cross validation analyses across patient samples, DASH achieved 98.7% specificity and 92.9% sensitivity while LOHHLA, a widely used algorithm, only reached 94.3% and 78.8%, respectively (figure 1). Downsampling analyses demonstrated that DASH benefits significantly from the boosted HLA sequencing on the ImmunoID NeXT Platform, dropping 0.06 in F-score after downsampling to the sequencing depth of other exome platforms. In cell line mixture analyses, DASH demonstrates greater than 99% specificity across all tumor purity and sub-clonality levels and greater than 98% sensitivity for above 27% tumor purity. Moreover, DASH demonstrated 100% sensitivity and specificity in dPCR experiments across 21 tumor samples with stable controls. We applied DASH to a large pan-cancer cohort and found that 18% of patients had HLA LOH (figure 2). We identified strong associations between HLA LOH and genomic instability. Moreover, we demonstrated relationships between HLA LOH and markers of immune pressure, such as a correlation with CD274 (PD-1) expression and allele-specific neoantigen enrichment for deleted HLA alleles.ConclusionsDASH, a highly sensitive HLA LOH algorithm that has been extensively validated using cross validation, in silico downsampling, cell line mixtures and dPCR, has demonstrated the widespread impact of HLA LOH in a large pan-cancer cohort.Abstract 79 Figure 1Bar plots showing the sensitivity and specificities scores across ImmunoID NeXT cross validation samples between LOHHLA (blue) and DASH (green).Abstract 79 Figure 2Bar plots denoting the number of patients and the frequency of HLA LOH in each tumor type cohort. 95% confidence intervals are shown with the thin dark grey bars. Only cohorts with at least 10 patients are shown

BackgroundMetastatic gastrointestinal stromal tumors (GIST) are lethal tumors of the GI tract characterized by gain of function mutations in KIT or PDGFRα. Transient first-line control is achieved through the inhibition of tyrosine kinase signaling using the KIT inhibitor imatinib, though most patients progress after 2–3 years. Progression through successive lines of therapy results in a molecularly heterogeneous disease with diverse subtypes, driven by distinct collections of exon-specific KIT mutations which directly inform therapy decisions. To address the unmet need of comprehensive understanding of GIST, we used tumor-informed whole exome liquid biopsy to identify and track the evolution of multiple concurrent heterogeneous resistance mechanisms in individual patients receiving tyrosine kinase inhibitors (TKIs).MethodsBaseline matched tumor, normal and longitudinal plasma samples were obtained from 15 metastatic, heavily pretreated GIST patients. Following baseline sample collection, all patients received systemic TKI therapy, and were monitored until disease progression. Paired tumor and normal samples were profiled using the ImmunoID NeXT Platform®, an augmented exome/transcriptome platform and analysis pipeline which generates comprehensive tumor and immune information. Exome-scale cfDNA profiling of matched plasma samples was performed using the NeXT Liquid BiopsyTM platform to detect somatic variants.ResultsBaseline solid tumor WES confirmed primary sensitizing KIT mutations in all 15 (100%) patients, and secondary KIT mutations in 7/15 patients (47%). Serial plasma whole exome sequencing identified evolution and expansion of clones harboring newly formed, druggable, exon-specific KIT mutations which evolved prior to identification of tumor progression using standard imaging techniques. In addition to these variants, we detected node-specific enrichment of PI3K-AKT and MAPK pathway mutations in plasma of patients with shorter overall survival (OS), which may contribute to the observed immune evasion. Accompanying these changes, we also detected significant association between gene copy-number profiles and duration of OS (P = 0.0097). Investigation of immune signatures using univariate cox modeling revealed a significant association between TCRβ diversity and reduced OS (HR = 2.55, log rank P = 0.04).ConclusionsComprehensive genomic profiling (WES and RNA-Seq) of paired tumor tissue and WES of serially collected ctDNA identified evolving druggable KIT mutations and other molecular alterations which preceded clinical disease progression. These findings suggest liquid biopsy-based monitoring of late-stage GIST malignancies may be useful for early identification of treatment resistance, providing treatment guidance prior to traditional approaches.Ethics ApprovalEthics approval was granted by the MD Anderson Human Research Protection Program, and all participants gave informed consent prior to participation.

AbstractPurpose: While immune checkpoint blockade (ICB) has become a pillar of cancer treatment, biomarkers that consistently predict patient response remain elusive due to the complex mechanisms driving immune response to tumors. We hypothesized that a multi-dimensional approach modeling both tumor and immune-related molecular mechanisms would better predict ICB response than simpler mutation-focused biomarkers, such as tumor mutational burden (TMB). Experimental Design: Tumors from a cohort of patients with late-stage melanoma (n = 51) were profiled using an immune-enhanced exome and transcriptome platform. We demonstrate increasing predictive power with deeper modeling of neoantigens and immune-related resistance mechanisms to ICB. Results: Our neoantigen burden score, which integrates both exome and transcriptome features, more significantly stratified responders and nonresponders (P = 0.016) than TMB alone (P = 0.049). Extension of this model to include immune-related resistance mechanisms affecting the antigen presentation machinery, such as HLA allele-specific LOH, resulted in a composite neoantigen presentation score (NEOPS) that demonstrated further increased association with therapy response (P = 0.002). Conclusions: NEOPS proved the statistically strongest biomarker compared with all single-gene biomarkers, expression signatures, and TMB biomarkers evaluated in this cohort. Subsequent confirmation of these findings in an independent cohort of patients (n = 110) suggests that NEOPS is a robust, novel biomarker of ICB response in melanoma.

e14526 Background: The ubiquity of tumor genotyping has identified hundreds of clinically actionable somatic mutations such as KRAS mutations predicting anti-EGFR response in colorectal cancer. Here, we describe an integrative pan-cancer study leveraging the Personalis NeXT Platform across 13 tumor types to reveal a broad portrait of clinically relevant mutations across 36 genes. Given the ascendance of tumor immunotherapy, there is an urgent need to better elucidate associations between clinically relevant gene mutations, patient demographics, and tumor immunobiology. Methods: Paired tumor-normals (n > 450) were sequenced on the Personalis NeXT Platform, a diagnostic platform which simultaneously profiles tumor and immune microenvironments via exome and transcriptome sequencing. Samples were evaluated for the presence of KRAS variants at the G12, G13, and Q61 positions, requiring a minor allele frequency of more than 2%. Additionally, more than 200 clinically relevant amino acid changes were considered across 36 genes. DNA sequencing data was harnessed to profile MSI status, oncovirus status, HLA allele-specific LOH, SNVs, and neoantigens. RNA sequencing data furnished profiles for gene expression, TILs, TCR, BCR, and immune signatures. Statistical tests compared both the presence of clinically relevant mutations, and specific amino acid changes, to each of these DNA and RNA derived features. Results: Immunogenomic analysis with the Personalis NeXT Platform identified KRAS mutations at the G12, G13, and Q61 positions in 6 of the 13 tumor types, including bladder, CRC, liver, NSCLC, pancreatic, and uterine cancers, consistent with known KRAS allelic incidence. 11 different KRAS amino acid substitutions were identified in this cohort, including G12(A/C/D/R/S/V), G13(C/D), and Q61(H/K/L), with 14% of samples positive for KRAS variants across our pan-cancer cohort, including malignancies not associated with KRAS. Immune signatures, including TCRB clones, mutational burden, and TILs signatures demonstrated variability in a KRAS allele-dependent manner. Pancreatic cancers demonstrated a statistically significant increase in macrophage enrichment score (p < 0.001) in KRAS-mutant tumors. Conclusions: Utilizing the Personalis NeXT Platform, our analysis characterized the tumor microenvironment of over 460 samples across 13 tumor types, 36 genes, and over 200 clinically relevant amino acid substitutions. This comprehensive characterization demonstrated significant differences in immune cell signatures in pancreatic cancer, suggestive of an association between the presence of KRAS mutations and immune cell infiltrate composition.

Background Typical liquid biopsy panels capture a relatively small number of variants, and likely under-represent the heterogeneity of resistance in late-stage cancers. This reduced scope can result in overlooked therapeutic biomarkers which respond dynamically to treatment, as well as potentially missed resistance mechanisms and pathway-level events. To address the challenges associated with identifying multiple concurrent heterogeneous resistance mechanisms in individual patients, we evaluated longitudinal whole exome sequencing of cell free DNA (cfDNA) and solid tumor biopsies from head and neck squamous cell carcinoma (HNSCC) patients that received anti-PD1 therapy. Using this approach, we identified evolving variant and pathway-level resistance mechanisms in cfDNA, as a complement to tumor biopsy derived information, and identified differences in putative neoantigens found in tissue and cfDNA. Methods Pre- and post-intervention matched tumor, normal and plasma samples were obtained from a pilot cohort of 13 patients with HNSCC. Following baseline sample collection, all patients received a single dose of nivolumab. The primary tumor was then resected, approximately one month later when possible, or a second biopsy was collected where resection was impractical. Paired tumor and normal samples were profiled using ImmunoID NeXTTM, an augmented exome/transcriptome platform and analysis pipeline. Exome-scale cfDNA profiling of matched plasma samples was performed using the NeXT Liquid BiopsyTM platform to detect somatic variants. Data from these two platforms were then compared with corresponding clinical findings. Results We observed a rapid evolution of the tumor microenvironment and disease mutation profile following therapy, with strong concordance detected between plasma and tumor variants at each timepoint. Post-therapy interrogation of cfDNA revealed dynamic changes in numerous oncogenes and clinically relevant pathways, such as ERK1/2 and MAPK, that were not observed in solid tumor. These findings suggest that single-lesion biopsy of the primary tumor misses co-occurring, clinically relevant resistance alterations. Median post-treatment neoantigen count was reduced in solid tumor, but increased in cfDNA. HLA-specific loss of heterozygosity (LOH) was identified in a number of subjects, likely resulting in reduced neoepitope presentation in those cases. Conclusions Exome-wide somatic events were reliably detected in cfDNA, providing additional potential biomarkers to complement those identified in solid tumor. As we increase our cohort size, we expect that identification of biomarkers from both exome scale tissue biopsy and cfDNA will provide a more comprehensive view into therapeutic response and resistance mechanisms in HNSCC patients missed with typical liquid biopsy panels. Citation Format: Charles W. Abbott, Nikita Bedi, Simo V. Zhang, Josette Northcott, Robin LI, Rachel Marty Pyke, Eric Levy, Rebecca Chernock, Mena Mansour, A. Dimitrios Colevas, John Lyle, John B. Sunwoo, Sean Boyle, Richard Chen. Longitudinal exome-scale liquid biopsy monitoring of evolving therapeutic resistance mechanisms in head and neck squamous cell carcinoma patients receiving anti-PD-1 therapy [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 555.

e18030 Background: Human leukocyte antigen loss of heterozygosity (HLA LOH) restricts immune recognition of tumors by limiting the major histocompatibility complex (MHC) presentation of neoantigens to T cells and correlates with reduced response to immune checkpoint blockade therapy (ICB) in non-small cell lung cancer. To explore the mechanism behind the impairment of HLA LOH on ICB, we analyzed the relationship between the antigen presentation pathway, neoantigen presentation and response to ICB in a head and neck squamous cell carcinoma (HNSCC) cohort. Methods: Following baseline sample collection, a cohort of 14 HNSCC patients recieved a single dose of PD-1 inhibitor. The primary tumor mass was definitively resected approximately one month later. If resection was impractical, a second biopsy was taken. Response to therapy was evaluated using RECIST criteria. Each pre- and post-intervention tumor sample and normal PBMC sample were profiled using Personalis’ ImmunoID NeXT Platform, an HLA-enhanced exome/transcriptome platform. HLA LOH was detected using a digital PCR validated machine learning algorithm (DASH). Neoantigen presentation was computationally predicted using a machine learning algorithm (SHERPATM) trained on mono-allelic immunopeptidomics data. Results: We found that 50% of the HNSCC cohort had HLA LOH, a larger percentage than in a large pan-cancer cohort (23%, n=611) and a distinct HNSCC cohort (40%, n=20). Further, two patients had B2M LOH and one patient had a deleterious mutation in an HLA allele. Despite the high frequency of somatic alteration in the antigen presentation pathway, we did not find an association between HLA LOH and ICB response. However, if HLA LOH was still shaping tumor evolution in response to ICB, we would expect to see immune pressure against subclonal tumor populations with neoantigens presentable by the retained HLA alleles but not the deleted HLA alleles. Indeed, we found that significantly more novel post-treatment neoantigens were predicted to bind to deleted HLA alleles compared to their homologous alleles (p=0.045). Conclusions: Given the high prevalence of HLA LOH across tumor types, a greater understanding is needed regarding the impact of HLA LOH on tumor evolution during ICB treatment. Though HLA LOH does not correlate with response to ICB, the consistent shift in neoantigen composition suggests that it acts as an evolutionary force in resistance to response during immunotherapy.

Background Human leukocyte antigen (HLA) genes facilitate communication between tumor cells and the immune system through the cell surface presentation of a diverse set of peptides. HLA loss of heterozygosity (LOH) has been associated with reduced immune pressure on neoantigens and impaired response to checkpoint blockade immunotherapy. Although HLA LOH is emerging as a key biomarker for response to immunotherapy, few tools exist to detect HLA LOH. Moreover, the accuracy of these tools is not well understood due to lack of orthogonal validation approaches. Here, we briefly describe DASH (Deletion of Allele-Specific HLAs), an algorithm to detect HLA LOH from exome sequencing data, and present a three-pronged validation approach to assess its performance. Methods In-silico evaluation of the limit of detection (LOD) of DASH was performed by deeply sequencing a tumor-normal paired cell line with HLA LOH and mixing reads at different proportions to simulate variable tumor purity and clonality. Direct genomic validation was performed using digital PCR (dPCR) with allele-specific primers targeting both predicted kept and lost alleles in ten patient samples and one cell line. Quantitative immunopeptidomics was performed to compare peptides presented by HLA alleles in tumor cells and adjacent normal cells. The relative increase or decrease of peptide presentation per allele was estimated by predicting the binding of each peptide to the patient-specific alleles. Results DASH is a machine learning model built upon the HLA-enhanced ImmunoID NeXT Platform®. We validated the performance of DASH using three orthogonal approaches to better understand the factors driving sensitivity and specificity of the algorithm. Evaluation using cell line mixtures that simulate LOH at various dilutions helped establish the LOD of DASH. For fully clonal tumors, DASH had 100% sensitivity at all tumor purity levels above 8% and 100% specificity at tumor purity levels higher than 24%. Patient-specific and allele-specific dPCR assays provided sensitive, direct evidence of HLA LOH. All samples predicted to have HLA LOH by DASH with high confidence were confirmed by dPCR. Finally, a quantitative immunopeptidomics experiment in one patient with HLA LOH revealed a large decrease in the peptides presented by deleted alleles, revealing the functional implications of HLA LOH. Conclusions HLA LOH detection methods need to be rigorously validated in order to be used as a clinical biomarker. Here, we introduced three methods to assess performance, demonstrated the strong predictive power of DASH, and highlighted the need to consider tumor purity in such assessments.

6557 Background: The reduced scope, and number of genes profiled by typical liquid biopsy panels can result in missed biomarkers including neoantigens, which may change with treatment, as well as potentially undetected resistance mechanisms and pathways beyond the scope of targets typically captured by panels. To address these limitations, we used a whole-exome scale liquid biopsy monitoring platform, NeXT Liquid Biopsy, to analyze head and neck squamous cell carcinoma (HNSCC) patients that have received anti-PD1 therapy. Presently, we sought to (1) monitor neoantigen changes in cfDNA as a complement to tumor biopsy-derived neoantigens, (2) compare the impact of tumor escape mechanisms, including HLA-LOH, on neoantigens identified in tissue and cfDNA and (3) to identify novel biological signatures that combine information from both solid tumor and liquid biopsies. Methods: Pre- and post-intervention matched normal, tumor and plasma samples were collected from a cohort of 12 patients with HNSCC. Following baseline sample collection all patients received a single dose of nivolumab, followed by resection approximately one month later when feasible, or a second biopsy where resection was impractical. Solid tumor and matched normal samples were profiled using ImmunoID NeXT, an augmented exome/transcriptome platform and analysis pipeline. Exome-scale somatic variants were identified in cfDNA from plasma samples using the NeXT Liquid Biopsy platform. Data from these two platforms were compared with corresponding clinical findings. Results: Concordant somatic events were detected between plasma and tumor at pre- and post-treatment timepoints. Neoantigens predicted to arise from these somatic events were reduced in solid tumor post-treatment, but increased in cfDNA, when compared to pre-treatment timepoints. HLA LOH was identified in a number of subjects, likely resulting in reduced neoepitope presentation in those cases. Immune cell infiltration increased in the tumor following treatment, with no changes to the CD8+/Treg cell ratio, suggesting consistent immunoregulation. Conclusions: Exome-wide neoantigen burden was reliably predicted from cfDNA, providing additional insight complementing data from solid tumor. Analyzing HLA LOH, and neoantigen burden from both solid and liquid biopsies together over the course of treatment creates a more comprehensive profile of therapeutic response and resistance mechanisms in HNSCC patients missed with typical liquid biopsy panels.

Abstract Checkpoint inhibitor therapy has demonstrated meaningful, if varied antitumor activity, with patient response influenced by a variety of biological factors, including complex interactions between the tumor and immune system. Thus, it is of increasing interest to identify composite biomarkers integrating multiple biological features to better predict immunotherapy response. In this study we use a comprehensive tumor immungenomics profiling platform to examine the effectiveness of our composite neoantigen score for stratifying patient response to checkpoint blockade therapy compared to tumor mutational burden and other biomarkers.Pre-treatment tumor/normal samples from 55 unresectable, stage III/IV melanoma patients who underwent anti-PD-1 therapy were characterized to assess factors influencing response. RECIST criteria were used to evaluate tumor response to therapy, with a median follow-up of 18 months. For each patient, a single paired FFPE tumor and normal blood sample was collected and profiled using Personalis' ImmunoID NeXT platform; an augmented exome/transcriptome platform and analysis pipeline, which produces comprehensive tumor mutation information, gene expression quantification, neoantigen characterization, HLA typing and LOH, TCR repertoire profiling and tumor microenvironment profiling. These data were then analyzed together with clinical outcome, and a composite neoantigen score computed for each patient along with other biomarkers such as tumor mutational burden (TMB).In this cohort, an elevated pretreatment composite neoantigen score combining neoantigen predictions adjusted based on resistance mechanisms that affect neoantigen presentation on the MHC complex was more strongly predictive of response to PD-1 blockade than TMB alone. This was true for both response and non-response via RECIST criteria and progression free survival. We also found that the composite neoantigen score was a stronger predictor of patient response when compared to neoantigen burden alone. Additionally, we observed increased response to anti-PD-1 therapy in patients with elevated pretreatment TCR clonality. Combining the composite neoantigen score and TCR clonality data revealed a significant association with non-response to therapy. Patients with high composite neoantigen score and TCR clonality that failed to achieve complete response revealed potential resistance mechanisms to anti-PD-1 therapy. Specifically, we identified patients with high expression of IDO1 or CTLA4, which may facilitate PD-1-independent immune escape. Additionally, we found patients with mutations within the antigen presentation machinery (APM), likely leading to loss of surface expression of the proteins, and in the case of B2M mutations, improper HLA class I folding and antigen presentation. These APM mutations likely result in reduced neoantigen presentation in these patients, facilitating tumor escape.In summary, our composite neoantigen score which integrates multiple components of MHC class I presentation into a single score, is more significantly associated with response to therapy than individual biomarkers such as tumor mutational burden. These findings highlight the promise of composite biomarkers for the optimization of anti-PD-1 therapy patient selection. Citation Format: Charles W. Abbott, Eric Levy, Rachel Marty Pyke, Rena McClory, Sekwon Jang, Richard Chen, Sean Boyle. A composite neoantigen score is more strongly associated with therapeutic response than tumor mutational burden in a cohort of late-stage anti-PD-1-treated melanoma patients [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 4278.

Abstract Comprehensive characterization of the tumor and tumor microenvironment (TME) can improve our understanding of tumor progression and treatment outcomes. For example, quantification of the immune infiltrate can inform mechanisms of immune escape and predict response to checkpoint blockade. Standard experimental approaches exist to enumerate tumor-infiltrating immune cells, but they can have practical limitations of throughput, number of markers, or sample requirements. RNA sequencing can be used as a scalable solution to comprehensively profile the immune cell composition of the TME. However, care must be taken to ensure that the computational analysis accurately reflects the underlying immune cell composition. To address these challenges, we developed ImmunoID NeXT, an augmented, immuno-oncology-optimized exome/transcriptome platform designed to provide comprehensive information regarding the tumor and TME from a single FFPE tumor sample. This includes the quantification of tumor-infiltrating immune cells using RNA-seq analysis, which we compare to quantification by orthogonal methods. To generate our reference data, we profiled the transcriptomes of eight purified immune cell types using ImmunoID NeXT. Then, we analyzed multiple sample types and orthogonally quantified immune cells in each. These include profiling of healthy donor PBMCs with cytometry by time of flight (CyTOF), and immunofluorescence (IF) characterization of FFPE tumor samples. We also used ImmunoID NeXT to profile the immune infiltrate of over 500 tumor samples across 13 cancer types. Finally, we created a set of in vitro cell mixtures and profiled them by flow cytometry. We utilized the transcriptome profiles of eight purified immune cell types to develop reference expression signatures specific for each cell type. Then, we compared ImmunoID NeXT's transcriptome-based approach to CyTOF results of healthy donor PBMCs, showing accuracy in real samples with diverse immune populations. We also compared to FFPE tumor samples with IF, ensuring that our approach is able to profile the immune composition in tumor samples. Next, we highlight the diversity of immune populations across cancer types by applying ImmunoID NeXT to over 500 tumor samples. Finally, to demonstrate concordance across the eight cell types, we compared to flow cytometry results of in vitro cell mixtures. Analysis of the immune infiltrate of tumor samples can add to our understanding of the tumor-immune interaction, with potential applications including studies of response to immunotherapy. RNA sequencing can be utilized as a scalable approach for such analysis. Here, we test the accuracy of our approach using multiple sample sets with orthogonal profiling. We demonstrate that ImmunoID NeXT can accurately evaluate the composition of infiltrating immune cells in tumor samples. Citation Format: Eric Levy, Pamela Milani, Charles W. Abbott, Manxia Lee, Robert Power, John West, Richard Chen, Sean M. Boyle. Quantification of tumor-infiltrating immune cell populations with an augmented transcriptome [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 4430.

Background Comprehensive profiling of the tumor and tumor microenvironment (TME) is a critical tool for furthering our understanding of tumor progression and response to treatment, including immunotherapies. To address this challenge, we developed an augmented, immuno-oncology-optimized exome/transcriptome platform, ImmunoID NeXTTM, which provides a more comprehensive view of the tumor and TME from limited FFPE tumor biopsies. We have recently added the ability to profile the B-cell receptor (BCR) heavy chain. Here, we show that ImmunoID NeXT is now able to accurately and reproducibly profile abundant B-cell clones and provide information on the diversity of B-cells in tumor samples. Methods We analyzed multiple replicates of PBMCs to examine the reproducibility of BCR sequence identification using ImmunoID NeXT. Utilizing a standalone BCR sequencing approach, we further evaluated the concordance of top clones to those identified by ImmunoID NeXT. In addition, we analyzed the reproducibility of BCR sequences in patient-derived FFPE samples. Finally, we used ImmunoID NeXT to profile the B-cell clonal diversity across over 500 solid tumor samples. Results Reproducibility in PBMC samples was very high, with abundances of clones shared between replicates being very concordant (R2>0.92, R2>0.86, and R2>0.97 for IgG, IgM, and IgA, respectively). When comparing to a standalone BCR sequencing method that profiles IgM and IgG, we observed highly concordant abundances (R2>0.72 and R2>0.82 in IgM and IgG, respectively), as well as strong overlaps of top clones. When comparing subsequent curls of a tumor FFPE sample, we also achieved a high concordance of clonal abundances (R2>0.92, R2>0.93, and R2>0.76 for IgG, IgM, and IgA, respectively). Finally, we observed differences in clonal diversity of B-cell repertoires across over 500 solid tumor samples. Conclusions We demonstrate that ImmunoID NeXT can be used to reproducibly, sensitively, and accurately profile high-abundance BCR heavy chain clones, including coverage of all major isotypes. In addition, we show how ImmunoID NeXT can profile the diversity of the BCR repertoire across a variety of tumor samples. Combined with the platform's TCR profiling capabilities, ImmunoID NeXT can provide insight into the diversity of the immune repertoire, contributing to its ability to provide comprehensive analysis of both the tumor and TME from a single FFPE sample.