Somatic Variant Calling Research Articles

Whole Exome Sequencing of Intracranial Epidermoid Cysts Reveals Immune-Associated Mechanistic and Potential Targets.

Background/Objectives: Intracranial Epidermoid Cysts (IECs) are rare intracranial tumors primarily treated through surgery. Cyst adherence complicates complete removal, leading to high rates of tumor progression after subtotal resection. The molecular drivers of IEC remain unknown. Consequently, advances in treatment have fallen short. Tumor genetic profiling has revealed potential targets for drug development, including FDA-approved options and reshaping treatment. The genetic landscape of IECs has not been explored. We applied Whole Exome Sequencing (WES) to IECs to gain insights into the mechanisms of oncogenesis and identify potential therapeutic targets. Methods: We performed WES on tumor tissue and matched blood samples, when available. Following GATK best practices, we conducted read processing, quality control, somatic variant calling, and copy-number inference. Data analyses and visualization were conducted in R. Results: Top altered genes are associated with the immune system and tumor microenvironment, suggesting a mechanism of immune evasion. Gene and pathway enrichment revealed a high mutation burden in genes associated with Extracellular Matrix (ECM) and PI3K-AKT-mTOR cascades. Recurrent and deleterious alterations in NOTCH2 and USP8 were identified in 50% and 30% of the cohort, respectively. Frequent amplifications in deubiquitinases and beta-defensins strengthened the involvement of immune mechanisms for oncogenic transformation. Conclusions: Top altered genes and recurrent mutations may play a role in shaping the microenvironment and modulating immune evasion in IECs. USP8 and NOTCH2 may serve as clinically relevant target for IECs. Finally, we present evidence that the crosstalk between the PI3K-Akt-mTOR and ECM signaling pathways may play a role in modulating the immune escape mechanism in IECs.

Performance of somatic structural variant calling in lung cancer using Oxford Nanopore sequencing technology

BackgroundLung cancer is a heterogeneous disease and the primary cause of cancer-related mortality worldwide. Somatic mutations, including large structural variants, are important biomarkers in lung cancer for selecting targeted therapy. Genomic studies in lung cancer have been conducted using short-read sequencing. Emerging long-read sequencing technologies are a promising alternative to study somatic structural variants, however there is no current consensus on how to process data and call somatic events. In this study, we preformed whole genome sequencing of lung cancer and matched non-tumour samples using long and short read sequencing to comprehensively benchmark three sequence aligners and seven structural variant callers comprised of generic callers (SVIM, Sniffles2, DELLY in generic mode and cuteSV) and somatic callers (Severus, SAVANA, nanomonsv and DELLY in somatic modes).ResultsDifferent combinations of aligners and variant callers influenced somatic structural variant detection. The choice of caller had a significant influence on somatic structural variant detection in terms of variant type, size, sensitivity, and accuracy. The performance of each variant caller was assessed by comparing to somatic structural variants identified by short-read sequencing. When compared to somatic structural variants detected with short-read sequencing, more events were detected with long-read sequencing. The mean recall of somatic variant events identified by long-read sequencing was higher for the somatic callers (72%) than generic callers (53%). Among the somatic callers when using the minimap2 aligner, SAVANA and Severus achieved the highest recall at 79.5% and 79.25% respectively, followed by nanomonsv with a recall of 72.5%.ConclusionLong-read sequencing can identify somatic structural variants in clincal samples. The longer reads have the potential to improve our understanding of cancer development and inform personalized cancer treatment.

Abstract C102: Polyethnic-1000: Advancing cancer genomics by studying ethnically diverse, underserved patient populations in New York

Abstract Prompted by the considerable biases toward European ancestry cases in large cancer genomics repositories, the New York Genome Center initiated an effort to diversify the public datasets that are forming the foundations of current cancer research, in order to optimize the unbiased applications of modern genomically-driven discoveries to a larger population. Our effort was focused on 8 cancer types (breast, prostate, bladder, lung, pancreatic, colon and endometrial cancer, as well as multiple myeloma) and generated Illumina tumor-normal whole-genome sequencing and tumor transcriptome sequencing. The majority of the samples were retrospectively obtained in clinics and hospitals of the New York region and the analyses are conducted with collaborators in these institutions. Our dataset is currently composed of 851 tumor-normal WGS and 471 RNASeq, and is progressively being distributed to the scientific community in partnership with ICGC-ARGO. Overall, 75% of the cohort is predominantly of non-European ancestry, as estimated by genetic ancestry analysis using the 1000Genomes continental references.We analyzed the samples with NYGC’s somatic variant calling pipeline and will present results from the 8 cancer types related to germline and somatic variants (single-nucleotide variants, indels, copy-number alterations, complex structural variants) as well as mutational signatures, neoantigen prediction, fusion genes and expression profiles, in relation to genetic ancestry estimation and clinical features of the patients. We anticipate that the dataset will significantly diversify the genetic ancestry of tumor profiles in public databases, and serve to form hypotheses regarding the contribution of ancestry on cancer disparities. The experience of the Polyethnic-1000 project informs our contribution to the Cancer Grand Challenge SAMBAI, which focuses on breast, prostate and pancreatic cancers in the African Diaspora. SAMBAI will further investigate the causes of cancer disparity by combining genomic data with social determinants of health and exposomics information, deeper immune profiling and new genomics technologies such as long reads sequencing and the use of a pangenome reference. Citation Format: Nicolas Robine, Lara Winterkorn, Tim Chu, Zoe Goldstein, Will Hooper, Ali Oku, Heather M. Geiger, Melissa B. Davis, Onyinye Balogun. Polyethnic-1000: Advancing cancer genomics by studying ethnically diverse, underserved patient populations in New York [abstract]. In: Proceedings of the 17th AACR Conference on the Science of Cancer Health Disparities in Racial/Ethnic Minorities and the Medically Underserved; 2024 Sep 21-24; Los Angeles, CA. Philadelphia (PA): AACR; Cancer Epidemiol Biomarkers Prev 2024;33(9 Suppl):Abstract nr C102.

Benchmarking whole exome sequencing in the German network for personalized medicine

IntroductionWhole Exome Sequencing (WES) has emerged as an efficient tool in clinical cancer diagnostics to broaden the scope from panel-based diagnostics to screening of all genes and enabling robust determination of complex biomarkers in a single analysis. MethodsTo assess concordance, six formalin-fixed paraffin-embedded (FFPE) tissue specimens and four commercial reference standards were analyzed by WES as matched tumor-normal DNA at 21 NGS centers in Germany, each employing local wet-lab and bioinformatics. Somatic and germline variants, copy-number alterations (CNAs), and complex biomarkers were investigated. Somatic variant calling was performed in 494 diagnostically relevant cancer genes. The raw data were collected and re-analyzed with a central bioinformatic pipeline to separate wet- and dry-lab variability. ResultsThe mean positive percentage agreement (PPA) of somatic variant calling was 76 % while the positive predictive value (PPV) was 89 % in relation to a consensus list of variants found by at least five centers. Variant filtering was identified as the main cause for divergent variant calls. Adjusting filter criteria and re-analysis increased the PPA to 88 % for all and 97 % for the clinically relevant variants. CNA calls were concordant for 82 % of genomic regions. Homologous recombination deficiency (HRD), tumor mutational burden (TMB), and microsatellite instability (MSI) status were concordant for 94 %, 93 %, and 93 % of calls, respectively. Variability of CNAs and complex biomarkers did not decrease considerably after harmonization of the bioinformatic processing and was hence attributed mainly to wet-lab differences. ConclusionContinuous optimization of bioinformatic workflows and participating in round robin tests are recommended.

433. UNRAVELLING THE MOLECULAR MECHANISMS BEHIND TUMOUR DIFFERENTIATION IN ESOPHAGEAL ADENOCARCINOMA

Abstract Background Esophageal adenocarcinoma tumors are divided into three grades based on the tumour’s histological differentiation: well, moderate and poor. Poorly differentiated tumours have a worse survival rate than moderate and well tumours. Understanding the molecular programs of this differentiation may lead to the identification of novel therapeutic interventions specific to tumour differentiation. We have utilized laser-capture microdissection to enrich tumour cells followed by gene expression profiling (RNA-seq) to identify gene expression programs and whole genome sequencing for differentiating specific mutations and copy number changes. Collectively, these results will enable us to unravel the molecular drivers of tumour differentiation. Methods Laser capture microdissection was applied to N=127 RNA-seq samples from N=74 patients and N=103 from N=81 patients from a mix of primary tumour biopsies, resections, and metastatic biopsies. Most samples have a matching RNA-seq and WGS sample. We used a standard pipeline to analyze the WGS data and produce somatic mutation, structural variant, and copy number calls. The gene expression data was segregated into two sets a test set consisting of N=74 samples and a test set of N=53 samples. Non-negative matrix factorization was used to identify eleven gene expression programs. Results Our testing RNA-seq cohort consisted of N=74 samples from N=74 patients with N=4 G1, N=26 G2, N=35 G3, and N=9 missing differentiation data. Our initial goal was to unravel the gene expression programs that correlate with tumour differentiation. Our non-negative matrix factorization analysis yielded 11 gene signatures, N=3 programs enriched in glandular gene expression, N=3 enriched in EMT pathways, N=2 with fibroblasts, and N=3 associated with immune / inflammation genes (not shown) (Figure 1). Moreover, the glandular signatures were associated with G1/G2 and the EMT and fibroblast signatures with G3. Moreover, the glandular 2 signature was associated with HER2 amplifications. Conclusion In this work we have begun to unravel the gene expression and genomic changes associated with tumour differentiation. We have found signatures enriched for both G1/G2 and G3 tumours and from these signatures we have observed gene expression heterogeneity within the different tumour differentiation categories. Moreover, the G3 tumours are enriched in fibroblasts despite our laser-capture microdissection. We are currently working on a classification model to predict tumor differentiation from these gene expression programs and are looking to further integrate our whole genome data to find additional genomic drivers.

ONCOLINER: A new solution for monitoring, improving, and harmonizing somatic variant calling across genomic oncology centers

The characterization of somatic genomic variation associated with the biology of tumors is fundamental for cancer research and personalized medicine, as it guides the reliability and impact of cancer studies and genomic-based decisions in clinical oncology. However, the quality and scope of tumor genome analysis across cancer research centers and hospitals are currently highly heterogeneous, limiting the consistency of tumor diagnoses across hospitals and the possibilities of data sharing and data integration across studies. With the aim of providing users with actionable and personalized recommendations for the overall enhancement and harmonization of somatic variant identification across research and clinical environments, we have developed ONCOLINER. Using specifically designed mosaic and tumorized genomes for the analysis of recall and precision across somatic SNVs, insertions or deletions (indels), and structural variants (SVs), we demonstrate that ONCOLINER is capable of improving and harmonizing genome analysis across three state-of-the-art variant discovery pipelines in genomic oncology.

Genomic landscape of gallbladder cancer: insights from whole exome sequencing.

Gallbladder carcinoma (GBC) is a common gastrointestinal malignancy noted for its aggressive characteristics and poor prognosis, which is mostly caused by delayed detection. However, the scarcity of information regarding somatic mutations in Indian patients with GBC has hampered the development of efficient therapeutic options. In the present study, we attempted to bridge this gap by revealing the mutational profile of GBC. To evaluate the somatic mutation profile, whole exome sequencing (WES) was performed on 66 matched tumor and blood samples from individuals with GBC. Somatic variant calling was performed using GATK pipeline. Variants were annotated at pathogenic and oncogenic levels, using ANNOVAR, VEP tools and the OncoKB database. Mutational signature analysis, oncogenic pathway analysis and cancer driver genes identification were performed at the functional level by using the maftools package. Our findings focused on the eight most altered genes with pathogenic and oncogenic mutations: TP53, SMAD4, ERBB3, KRAS, ARID1A, PIK3CA, RB1, and AXIN1. Genes with pathogenic single nucleotide variations (SNVs) were enriched in oncogenic signaling pathways, particularly RTK-RAS, WNT, and TP53 pathways. Furthermore, our research related certain mutational signatures, such as cosmic 1, cosmic 6, and cosmic 18, 29, to known characteristics including patient age and tobacco smoking, providing important insights into disease etiology. Given the scarcity of exome-based sequencing studies focusing on the Indian population, this study represents a significant step forward in providing a framework for additional in-depth mutational analysis. Genes with substantial oncogenic and pathogenic mutations are promising candidates for developing targeted mutation panels, particularly for GBC detection.

Novel Methodology for the Design of Personalized Cancer Vaccine Targeting Neoantigens: Application to Pancreatic Ductal Adenocarcinoma.

Personalized cancer vaccines have emerged as a promising avenue for cancer treatment or prevention strategies. This approach targets the specific genetic alterations in individual patient's tumors, offering a more personalized and effective treatment option. Previous studies have shown that generalized peptide vaccines targeting a limited scope of gene mutations were ineffective, emphasizing the need for personalized approaches. While studies have explored personalized mRNA vaccines, personalized peptide vaccines have not yet been studied in this context. Pancreatic ductal adenocarcinoma (PDAC) remains challenging in oncology, necessitating innovative therapeutic strategies. In this study, we developed a personalized peptide vaccine design methodology, employing RNA sequencing (RNAseq) to identify prevalent gene mutations underlying PDAC development in a patient solid tumor tissue. We performed RNAseq analysis for trimming adapters, read alignment, and somatic variant calling. We also developed a Python program called SCGeneID, which validates the alignment of the RNAseq analysis. The Python program is freely available to download. Using chromosome number and locus data, SCGeneID identifies the target gene along the UCSC hg38 reference set. Based on the gene mutation data, we developed a personalized PDAC cancer vaccine that targeted 100 highly prevalent gene mutations in two patients. We predicted peptide-MHC binding affinity, immunogenicity, antigenicity, allergenicity, and toxicity for each epitope. Then, we selected the top 50 and 100 epitopes based on our previously published vaccine design methodology. Finally, we generated pMHC-TCR 3D molecular model complex structures, which are freely available to download. The designed personalized cancer vaccine contains epitopes commonly found in PDAC solid tumor tissue. Our personalized vaccine was composed of neoantigens, allowing for a more precise and targeted immune response against cancer cells. Additionally, we identified mutated genes, which were also found in the reference study, where we obtained the sequencing data, thus validating our vaccine design methodology. This is the first study designing a personalized peptide cancer vaccine targeting neoantigens using human patient data to identify gene mutations associated with the specific tumor of interest.

Whole‐exome sequencing‐based mutational profiling of hepatocellular adenoma malignant transformation to hepatocellular carcinoma

AbstractHepatocellular adenoma (HCA) represents a rare benign hepatic neoplasm with potential for malignant transformation into hepatocellular carcinoma (HCC), yet the underlying mechanism remains elusive. In this study, we investigated the genomic landscape of this process to identify therapeutic strategies for blocking malignant transformation. Using micro‐detection techniques, we obtained specimens of adenoma, cancerous neoplasm and adjacent normal liver from three patients undergoing hepatic resection surgery. Whole‐exome sequencing (WES) was performed, and genomic interactions between HCA and HCC components within the same tumour were evaluated using somatic variant calling, copy number variation (CNV) analysis, clonality evaluation and mutational signature analysis. Our results revealed genomic heterogeneity among patient cases, yet within each sample, HCA and HCC tissues exhibited a similar mutational landscape, suggesting a high degree of homology. Using nonnegative matrix factorization and phylogenetic trees, we identified shared and distinct mutational characteristics and uncovering necessary pathways associated with HCA–HCC malignant transformation. Remarkably, we found that HCA and HCC shared a common monoclonal origin while displaying significant genetic diversity within HCA–HCC tumours, indicating fundamental genetic connections or evolutionary pathways between the two. Moreover, elevated immune therapy‐related markers in these patients suggested heightened sensitivity to immune therapy, providing novel avenues for the treatment of hepatic malignancies. This study sheds light on the genetic mechanisms underlying HCA–HCC progression, offering potential targets for therapeutic intervention and highlighting the promise of immune‐based therapies in managing hepatic malignancies.

Integrated approach to generate artificial samples with low tumor fraction for somatic variant calling benchmarking

BackgroundHigh-throughput sequencing (HTS) has become the gold standard approach for variant analysis in cancer research. However, somatic variants may occur at low fractions due to contamination from normal cells or tumor heterogeneity; this poses a significant challenge for standard HTS analysis pipelines. The problem is exacerbated in scenarios with minimal tumor DNA, such as circulating tumor DNA in plasma. Assessing sensitivity and detection of HTS approaches in such cases is paramount, but time-consuming and expensive: specialized experimental protocols and a sufficient quantity of samples are required for processing and analysis. To overcome these limitations, we propose a new computational approach specifically designed for the generation of artificial datasets suitable for this task, simulating ultra-deep targeted sequencing data with low-fraction variants and demonstrating their effectiveness in benchmarking low-fraction variant calling.ResultsOur approach enables the generation of artificial raw reads that mimic real data without relying on pre-existing data by using NEAT, a fine-grained read simulator that generates artificial datasets using models learned from multiple different datasets. Then, it incorporates low-fraction variants to simulate somatic mutations in samples with minimal tumor DNA content. To prove the suitability of the created artificial datasets for low-fraction variant calling benchmarking, we used them as ground truth to evaluate the performance of widely-used variant calling algorithms: they allowed us to define tuned parameter values of major variant callers, considerably improving their detection of very low-fraction variants.ConclusionsOur findings highlight both the pivotal role of our approach in creating adequate artificial datasets with low tumor fraction, facilitating rapid prototyping and benchmarking of algorithms for such dataset type, as well as the important need of advancing low-fraction variant calling techniques.

Abstract 4926: Advancements in somatic variant calling from UG100 whole genome and whole exome sequencing data

Abstract Somatic variant calling involves the identification of genomic alterations that occur in somatic cells, requiring deep coverage to enable high sensitivity for low-frequency variants. Characterizing somatic variants across the entire genome therefore benefits from novel cost-efficient sequencing platforms, such as UG100. Here, we present optimization of variant calling tools for short and structural variants on WGS and WES data from UG100. For calling short variants, we optimized DeepVariant (DV) for somatic calling using data from matched tumor-normal sample pairs, improving both variant calling accuracy and pipeline running time (up to 10-fold). We defined the task of somatic variant calling as deciding if the pileup image containing reads from the tumor and normal samples represents a true somatic variant (vs a germline variant or artifact). The challenge of robust variant calling using deep learning models is exacerbated in somatic calling, where sequencing depth and coverage variability are typically high. Our optimized DV overcomes these challenges by several data sampling strategies. First, allele-frequency preserving down-sampling reduces randomness of read sub-sampling in high coverage regions. Second, alternative allele prioritization samples alt-allele supporting reads first allowing to call variants at very high coverage loci without sacrificing sensitivity and computational efficiency. Finally, a Panel-of-Normals based on targeted WES data provides an additional improvement of precision for this assay type. We used these strategies to train two models, one for tumor characterization using WGS (T/N coverage: 40x-150x/40x-100x), and one for deep WES (T/N coverage: >500x/>120x). We called variants on simulated tumors using the WGS model. For VAF>10% the model showed SNV recall >98% and indel recall >95% with false-positive rate of 0.2/Mb. For VAF range of 5-10%, indel recall was 67% and SNV recall was 86%. To demonstrate the utility of our somatic variant calling, we applied the models to call somatic variants from well characterized cancer cell lines: COLO829, HCC1395 and HCC1143. Results showed F1>90% for variants with VAF>10%. The WES model was used to reliably call variants at VAF>5% on simulated tumors with average SNV recall of 99% with precision >99% and indel recall >86% with precision >94%. To analyze structural and copy-number variations, we optimized the assembly engine of GRIDSS to enable fast calling of structural variations and demonstrate that Control-FREEC can be used to call copy number variants. SV calling on COLO829/COLO829BL achieved sensitivity >95%. In conclusion, our research highlights the utility of UG100 within the field of oncology, demonstrating its capacity for comprehensive and precise somatic variant detection, both on WGS and WES data. Citation Format: Doron Shem-Tov, Maya Levy, Gil Hornung, Ilya Soifer, Hila Benjamin, Ariel Jaimovich, Adam Blattler, William Brandler, Robert Sugar, Isaac Kinde, Omer Barad, Doron Lipson. Advancements in somatic variant calling from UG100 whole genome and whole exome sequencing data [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 4926.

Abstract 326: Matched fresh frozen and FFPE patient tissues reveal the enhanced sensitivity and data quality of a novel DNA library prep method

Abstract In cancer genomics, a common source of DNA is formalin-fixed, paraffin-embedded (FFPE) tissue from patient surgical samples, where in most cases high quality fresh or frozen tissue samples are not available. FFPE DNA poses many notable challenges for preparing NGS libraries, including low input amounts and highly variable damage from fixation, storage, and extraction methods. It is difficult to obtain libraries with sufficient coverage and the sequencing artifacts arising from damaged DNA bases confound somatic variant detection. Additionally, many laboratories process FFPE tumor samples alongside matched, high quality, normal DNA and many library prep workflows are not readily compatible with both sample types. We developed a novel NGS library prep method compatible with both high quality and very low quality FFPE DNA samples, employing a novel enzymatic DNA repair mix, enzymatic fragmentation mix, and PCR master mix. To validate this workflow on real patient sample sets, we obtained DNA extracted from matched tumor and normal tissue of various tissue types preserved by both fresh frozen and FFPE methods. The FFPE DNA samples ranged in quality from DNA integrity number (DIN) 1.5 to 6.8, including samples stored for nearly 3 decades. We prepared libraries using this method compared against other library prep workflows and sequenced them by WGS and target capture. We assessed the performance of these workflows by library yield, library quality metrics, depth and evenness of coverage, and somatic variant calling with fresh frozen-extracted DNA providing the gold standard for library quality and mutation content. This new enzymatic fragmentation-based library prep workflow not only reduced the false positive rate in somatic variant detection by repairing damage-derived mutations in FFPE DNA samples but also improved the library yield, library quality metrics (including mapping, chimeras, and properly-paired reads), library complexity, coverage depth and uniformity, and hybrid capture library quality metrics. Comparing the variant calls from matched FFPE and frozen tissues revealed an improved sensitivity and accuracy of variant calling using this library prep method compared to mechanical shearing and other enzymatic fragmentation library prep approaches. This new suite of enzyme mixes allows even highly damaged FFPE samples to achieve high quality libraries with a greater sensitivity for somatic variant identification. The workflow is robust and flexible, compatible with both FFPE DNA and matched high quality DNA samples as well as automation-friendly for convenience in sample processing. Citation Format: Margaret R. Heider, Jian Sun, Brittany Sexton, Bradley W. Langhorst, Laura Blum, Pingfang Liu. Matched fresh frozen and FFPE patient tissues reveal the enhanced sensitivity and data quality of a novel DNA library prep method [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 326.

Abstract 1769: Pan-cancer genomic characterization of human papillomavirus associated tumors reveals patterns of somatic alteration that associate with virus status and anatomic site

Abstract Human papilloma virus (HPV) infection causes over 600,000 human cancers yearly and accounts for nearly all cervical cancers, increasing rates of head and neck squamous cell carcinomas (HNSCC), and many anogenital cancers - all with varying clinical outcomes due to a lack of personalized care. While recent integrative genomic studies have described molecular features of individual cancer types, few studies have compared genomic changes between HPV(+) and HPV(-) cancers across anatomic sites. Here, we conducted the first pan-cancer genomic analysis of HPV-associated tumors across multiple anatomic tumor types using whole exome and transcriptome data from The Cancer Genome Atlas (TCGA) cohorts of cervical (n=254) and HNSCCs (n=514), and targeted exome sequencing of 800 cancer genes plus full length HPV16/18 genomes in a clinical cohort of squamous tumors from the head and neck (n=458), cervix (n=78), vulva (n=23), anal canal (n=5), and vagina (n=2). Somatic variant calling and filtering, followed by an integrative pathway analysis of commonly altered targets, defined the catalog of somatic mutations and copy number alterations (CNAs) that drive HPV(+) and HPV(-) tumorigenesis. Sequencing reads from viral RNA or DNA determined HPV status, and HPV type, genome structure, integration events, and viral load were characterized in a subset of samples via de novo assembly of viral aligned reads followed by copy number analysis, breakpoint identification, and the calling of structural variants. Overall HPV positivity was 50% (668 out of 1334 total), with HPV16 accounting for 96%, 89%, and 60% of all HPV(+) anogenital, head and neck, and cervical tumors respectively. Significant differences in somatic mutation frequency between HPV(+) and HPV(-) tumors were observed in the full cohort, as well as in analyses stratified by anatomic site. Interestingly, we noticed recurrently mutated “hotspots” attributable to increased APOBEC-mutagenesis in HPV(+) samples across anatomic sites (PIK3CA:E545K, FGFR3:S249C, EP300:D1399N), while hotspot mutations likely caused by tobacco smoking predominate HPV(-) HNSCC (PIK3CA:H1047R/L, CDKN2A:R80*, TP53:R282W). Focal and arm level CNAs were distinctive, including gains of 11q22 in HPV(+) cervical and HPV(-) HNSCC, and losses of 11q22 in HPV(+) HNSCC. Biological pathways commonly altered include epithelial differentiation, cell death, innate immunity, growth factor/kinase signaling, and cell cycle control. In summary, pan-cancer genomic analysis revealed distinct patterns of somatic alteration of conserved biological pathways that associate with HPV status and anatomic site. These findings improve our understanding of tumor biology unique to HPV-associated cancers and may lead to novel treatment and classification strategies to improve patient outcomes in the clinical setting. Citation Format: Jeremiah Ray Holt, Paul Little, Heejoon Jo, Xiaobei Zhao, Hyo Young Choi, Vonn Walter, Benjamin Wahle, Jose P. Zevallos, Angela Mazul, Katherine A. Hoadley, Michele Hayward, David N. Hayes. Pan-cancer genomic characterization of human papillomavirus associated tumors reveals patterns of somatic alteration that associate with virus status and anatomic site [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 1769.

Abstract 1760: Mutational landscape of the canine acute myeloid leukemia exome

Abstract Over 21,000 newly diagnosed cases of human acute myeloid leukemia (hAML) and 10,000 hAML-related deaths are reported in the United States every year. hAML is considered a highly diverse aggressive hematopoietic malignancy, with over 16 recurrent gene rearrangements, 60 recurrent point mutations, and hundreds of rarely mutated genes. There are large ongoing multi-institutional studies devoted to advancing medical management for hAML by providing targeted therapeutics to patients based on their molecular characteristics. A preclinical spontaneous large animal model for testing novel therapies could accelerate the development of better treatments in people. Dogs spontaneously develop AML with similar clinical features and outcomes. As such, the dog may be a useful pre-clinical model, but little is known about the molecular landscape and/or drivers of canine AML (cAML). The goal of this work was to identify somatic mutation candidates in cAML and determine the overlapping features with hAML. We utilized whole exome sequencing at 300x coverage to identify candidate mutations in 51 cAML samples. Somatic variant calling was performed per the GATK best practice recommendations using Mutect2 and a reference panel of normals (n=77) and a variant call file of canine germline variants. We focused our analysis on novel variants predicted to have a moderate to high impact and that occurred in genes implicated in cancer. At least one variant was identified in genes involved in the RTK/RAS pathway (e.g. NRAS, KRAS, PTPN11, FLT3, KIT) in over 75% of cAML samples, a trend that is also found in people. Additionally, recurrent point mutations were predicted in NRAS and KRAS, and some of which, such as the codon changes G13R and Q61R/H in NRAS, were noted to be shared between hAML and cAML. Mutations in DNMT3A were uncommon (n=3/44) and NPM1 mutations were not identified in our cohort, both of which are common hAML-associated genes. However, mutations in other epigenetic modifiers, such as KDM5C and KDM6A, occurred at a high frequency. This data provides insight into the candidate mutations potentially driving cAML pathogenesis and highlights the overlapping molecular features with hAML. Citation Format: Robert Harris, Evan Conaway, Anne Avery. Mutational landscape of the canine acute myeloid leukemia exome [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 1760.

Abstract 5014: Analytical validation of a complete system for detection and characterization of genomic aberrations in circulating tumor DNA

Abstract Cell-free DNA (cfDNA) fragments are readily detected at a low concentration in the human blood stream. In healthy patients cfDNA is primarily derived from apoptosis of hematopoietic cells. In cancer patients circulating DNA derived from tumor cells (ctDNA) undergoing apoptosis, shedding, and necrosis contribute to total cfDNA. Next generation sequencing (NGS) of cfDNA isolated from patient plasma can be used to fully characterize cancer driven changes in the tumor DNA including the relative contribution of cfDNA, single nucleotide variants (SNVs), small insertions and deletions (Indels), larger deletions, regions of amplified DNA, and structural rearrangements. Frontage has developed a complete lab system for analysis of circulating cfDNA including automated sample extraction, preparation and sequencing of libraries targeting exon sequences of 293 oncology-related genes, and a bioinformatic reporting system. The Frontage sample extraction process uses a post-extraction size selection step to remove high molecular weight DNA derived from lysed blood cells in the plasma sample to increase the sensitivity of detection of ctDNA. The NGS platform couples hybridization-based selection of genomic DNA fragments derived from protein-coding exons with deep sequencing to ~4,000X gross coverage on the NextSeq 2000. The bioinformatic pipeline integrates unique molecular barcode-based generation of consensus reads for each template molecule, somatic variant calling with GATK Mutect2 and FreeBayes, and removal of false positive variants by comparison to a 60 samples healthy patient data set. Analytical validation was performed for each component of the system. For verification of the sample extraction process, cfDNA was extracted in duplicate from four lots of pooled K2 EDTA plasma of healthy patients as well as individual K2 EDTA plasma of colorectal cancer (CRC) patients in four independent extraction runs. The mean yields of cfDNA from healthy samples ranged from 2.3 - 4.2 nanograms per ml plasma and all 32 of the isolated cfDNA aliquots from healthy patient plasma showed a prominent peak around 167 nt with minimal contamination from high molecular weight DNA. For validation of the NGS and bioinformatics platform a set of 8 different reference cfDNAs were processed through four independent runs of library preparation, sequencing, and analysis. For each of the validation runs, greater than 97% of 208 SNVs and Indels with expected allele frequencies (AF) between 2-3% were detected by the platform with higher levels of accuracy for variants at above 3% AF. Two or less false positive variants per run were detected out of 278,000 nucleotides of verified non-mutated sequence that overlapped with the 293 gene panel. Four known DNA amplification events were detected in each of the validation runs. Data from characterization of CRC and other clinical samples with the system will be presented. Citation Format: Ericca Stamper, Weiming Shen, Aman Pruthi, Casey Nagel, Marcin Piechota, Katarzyna Tomala, Qi Wei, David Willoughby. Analytical validation of a complete system for detection and characterization of genomic aberrations in circulating tumor DNA [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 5014.

Abstract 6654: Oncolytic virotherapy targeting high-grade glioma tumor antigens

Abstract Glioblastoma, the most common primary brain malignancy in adults, is an aggressive cancer with five-year median survival of ~6.8%. New treatment options are urgently needed for glioblastoma and other high-grade gliomas (HGGs). Oncolytic viruses (OVs) are an emerging class of cancer therapy currently being evaluated as treatments for a variety of malignancies, including glioblastoma. Our labs developed OV C134, now under evaluation in a phase I clinical trial of recurrent glioblastoma (NCT03657576). In addition to directly infecting and lysing tumor cells, OVs are capable of stimulating anti-tumor immune responses. We previously showed in the CT2A syngeneic mouse model of HGG that treatment with a version of C134 engineered to express the shared tumor antigen Ephrin-A2 (EPHA2; commonly upregulated in glioblastoma) significantly increased survival and produced EPHA2-specific T cell responses. This finding led us to hypothesize that C134 could be further engineered to stimulate antigen-specific immune responses against tumor-specific neoantigens. We first assessed the impact of C134 treatment on immune cell dynamics in HGGs, by performing single cell RNA sequencing of CD45+ immune cell infiltrate from CT2A HGGs treated with C134 or saline (control). Clustering of these data identified 17 distinct immune cell types, and differential expression analyses revealed several C134-induced changes. These included increased expression of genes involved in antigen presentation and STAT signaling in microglia and macrophage subsets, the proliferation marker Mki67 in germinal center B cells, the chemokine Ccl12 from microglia, and tissue-resident memory marker genes Fos and Jun in T cells. These data suggest C134 treatment poises the tumor immune microenvironment for antigen presentation and antigen-specific immune responses. Next, we evaluated CT2A to identify candidate tumor neoantigens by isolating DNA and RNA from orthotopic tumors and DNA from tail snips of mice treated with saline or C134. We carried out RNA sequencing (tumors) and whole exome sequencing (tumor and normal), and employed the Personalized Variant Antigens by Cancer Sequencing (pVAC-Seq) computational pipeline. This pipeline takes somatic variant calls from the exome comparison of tumor to normal DNA, validates RNA expression of called variants, and feeds corresponding peptide sequences into algorithms that predict the binding of variant-containing peptides to major histocompatibility complex (MHC), thereby assessing for potential neoantigens. Our pVAC-Seq analysis revealed a conserved set of 1,264 tumor-exclusive variants expressed at the RNA level with high (ic50<500nm) predicted binding to mouse MHC. I will present details of C134-induced immune remodeling in syngeneic HGG models, and describe a plan to engineer C134 to express a prioritized set of neoantigens to evaluate how efficacy from OV therapy may be further enhanced in a patient-specific manner. Citation Format: Jack Hedberg, Alexia Martin, Doyeon Kim, Jason Navarro, Adithe Rivaldi, Yeaseul Kim, Ilse Hernandez-Aguirre, Elizabeth Garfinkle, Katherine Miller, Kevin Cassady, Elaine Mardis. Oncolytic virotherapy targeting high-grade glioma tumor antigens [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 6654.

Abstract 1763: Long-read sequencing of pediatric leukemia identifies clinically relevant genomic rearrangements

Abstract Clinical detection of genomic rearrangements currently depends on a range of complementary and overlapping methods including fluorescence in situ hybridization (FISH), karyotyping, and DNA and/or RNA sequencing. This requirement for multiple tests results in higher total costs, extended analysis time, and increased usage of sometimes limited tumor material. In our laboratory, we apply long high-fidelity reads to human samples using PacBio circular consensus sequencing (HiFi-GS) of 12-18kb DNA fragments, which offers better structural variant detection relative to short-read sequencing. Here, we assess the ability of HiFi-GS to detect clinically relevant genomic rearrangements (CRGRs) in pediatric leukemia. We selected 11 pediatric leukemia cases with corresponding normal (remission) samples, including 7 cases of B-cell acute lymphoblastic leukemia (B-ALL) and 4 cases of acute myeloid leukemia (AML); 5 of these 11 leukemia cases had known CRGRs as below. The remaining 6 leukemia cases had also undergone clinical genetic profiling via karyotyping, FISH, microarray, and sequencing yet a clinical grade genomic driver was not identified. DNA from each sample was sequenced on a Revio instrument (PacBio, Menlo Park, CA) to a target depth of 30x. The PacBio Human Whole Genome Sequencing (WGS) workflow was used to process HiFi reads for haplotagged alignment and phasing, followed by somatic structural variant calling using Severus. Subsequent analysis focused on the presence of break ends in known cancer-relevant genes. We found that HiFi-GS detected the known CRGRs in all 5 cases with prior findings, providing precise break ends and clarifying sometimes unclear cytogenetic observations (such as exact partner genes), in these cases. These included detection of an ETV6-RUNX1 fusion, NUP98-NSD1 fusion, KMT2A-AFF1 fusion, KMT2A-AFDN fusion, and RUNX1-RUNX1T1 fusion. HiFi sequencing also detected CRGRs in 2 of the 6 “undiagnosed cases”; namely, one case of ZNF384-TCF3 fusion and one case of KMT2A-MLLT10 fusion. The former can be cytogenetically cryptic, while the latter can often be missed due to its potentially complex nature. Our results demonstrate that HiFi-GS can reliably detect CRGRs, bringing us closer to the promise of a single comprehensive genetic test for cancer characterization. Additional work examining other somatic structural and copy number variants (i.e., inversions, deletions, and duplications) in these cases, as well as oncogenic sequence variants, is ongoing. Citation Format: Byunggil Yoo, Ayse Keskus, Chengpeng Bi, Lisa Lansdon, Tanveer Ahmad, Irina Pushel, Adam Walter, Margaret Gibson, Erin Guest, Tomi Pastinen, Mikhail Kolmogorov, Midhat S. Farooqi. Long-read sequencing of pediatric leukemia identifies clinically relevant genomic rearrangements [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 1763.

Abstract 874: Expanded detection of BAP1 alterations in cancer and comparison of tumor type-specific gene expression scores

Abstract BAP1 is a tumor suppressor gene that was originally studied in uveal melanoma (UVM), kidney renal cell clear cell carcinoma (KIRC), and malignant mesothelioma (MESO). Early analyses focused on single-nucleotide variants, but other alteration types such as larger indels and gene-level copy number (CN) loss can also lead to loss of BAP1 expression. By incorporating these, we can improve our assessment of cancer type-specific consequences of BAP1 loss and broaden our findings beyond the initial small subset of cancer types which were studied in the context of mutations alone. We performed integrated multi-omic analyses using data from The Cancer Genome Atlas (TCGA), for 33 cancer types and more than 10,000 individuals. By combining and manually reviewing existing variant calls and new calls derived from a de novo local realignment pipeline across multiple independent variant callers including indel callers, we increased detection of high-quality somatic variant calls by 30% from 85 to 111, including 14 indels ≥5bp. Including CN loss alterations, we found a total of 1561 samples from 32 cancer types to be BAP1-altered, with alterations being predominantly CN-driven. We confirmed that expression of BAP1 was lower in altered versus unaltered tumors (p<0.001). Alteration frequencies across cancer types were highly variable, ranging from 0% in thyroid cancer to >30% in KIRC, cholangiocarcinoma (CHOL), UVM, MESO, lung squamous carcinoma (LUSC), and head-neck squamous cell carcinoma (HNSC). Tumor types such as CHOL, UVM, MESO, and liver hepatocellular carcinoma (LIHC) were relatively more mutation-driven whereas other tumor types like KIRC, LUSC, and HNSC were almost entirely CN-driven. To better understand the tissue and context-specific consequences of BAP1 alteration, we performed differential expression analyses across BAP1-altered and -unaltered samples. For each cancer type, we separately computed per-sample BAP1 alteration signature z-scores summarizing expression of the set of differentially expressed genes. In UVM, we observed distinct patterns of RNA expression, consistent with established literature on BAP1 loss and serving as a baseline for comparison for the biological effects of BAP1 loss in other less-studied cancer types. The BAP1 altered scores for 66% of cancer types were positively correlated with the UVM BAP1 altered signature, suggesting shared biological responses and affected pathways including immune responses, epithelial-mesenchymal transition, oxidative phosphorylation, and proliferation – all known roles of BAP1. In LIHC, we noted enrichment of bile duct gene expression in BAP1-altered tumors, suggesting a possible role for BAP1 in maintaining cell identity distinguishing hepatocytes and cholangiocytes. Our findings broadly emphasize the improvements that are gained by using new computational approaches in large cancer-genome studies such as TCGA. Citation Format: Ian R. Sturgill, Jesse R. Raab, Katherine A. Hoadley. Expanded detection of BAP1 alterations in cancer and comparison of tumor type-specific gene expression scores [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 874.

Abstract 2275: Reducing artifactual somatic variant calls from formalin-fixed paraffin-embedded specimens by using DEEPOMICS FFPE, a bioinformatic approach based on deep neural networks

Abstract Formalin-fixed, paraffin-embedded (FFPE) tissue specimens are routinely used in pathological diagnosis. However, formalin induced artificial mutations make it challenging to accurately analyze next-generation sequencing data. Existing FFPE filtering tools tend to categorize all low-frequency variants as artifacts, making it difficult to identify true variants with low allele frequencies, compromising analysis of tumors with low tumor cell content. To address this issue, we developed DEEPOMICS FFPE, an AI model designed to distinguish true variants from artifacts. We gathered paired exome genome sequencing data from both fresh frozen and FFPE samples of 24 tumors from public sources. These data were used to train and validate the model, with a split of 70% for training and 30% for validation. The deep neural network model, comprising three hidden layers, was trained using input features derived from the outputs of the MuTect2 caller. We employed the SHapley Additive exPlanations algorithm to identify relevant features, and then fine-tuned them based on the training outcomes. We assessed the performance of the final DEEPOMICS FFPE model in comparison to existing models (MuTect filter, FFPolish, and SOBDetector) using well-defined test datasets. Our findings revealed 43 distinguishing characteristics for FFPE artifacts. Refining the quantification of these characteristics resulted in improved model performance. DEEPOMICS FFPE successfully filtered out 99.6% of artifacts while retaining 85.2% of true variants, achieving an F1-score of 87.8 in the validation dataset, surpassing the performance of existing tools. Furthermore, the model demonstrated consistent performance even for variants with low allele frequencies, exhibiting a specificity of 0.998, suggesting its capability to identify subclonal variants. Future enhancement of the model will include the ability to use other variant callers and analysis of whole genome sequencing data. This newly developed tool has potential applications in designing capture panels for personalized circulating tumor DNA assays and identifying candidate neoepitopes for personalized vaccine development. DEEPOMICS FFPE is freely accessible for research purposes on the web at http://deepomics.co.kr/ffpe. Citation Format: Dong-hyuk Heo, Inyoung Kim, Heejae Seo, Seong-Gwang Kim, Minji Kim, Jiin Park, Hongsil Park, Seungmo Kang, Juhee Kim, Soonmyung Paik, Seong-Eui Hong. Reducing artifactual somatic variant calls from formalin-fixed paraffin-embedded specimens by using DEEPOMICS FFPE, a bioinformatic approach based on deep neural networks [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 2275.

Abstract 3675: Design of high-performance tumor-informed minimal residual disease (MRD) panels from low FFPE tumor input

Abstract Background: Minimal residual disease (MRD) testing can detect cancer recurrence months to years earlier than the current standard of care, enabling earlier treatment of recurrence and improved patient outcomes. Tumor-informed MRD assays typically utilize formalin-fixed paraffin-embedded (FFPE) tumor tissue, which is available in limited quantities for some patients, for example, following core needle biopsy (CNB), after neoadjuvant treatment or when patients need multiple tests from the same tumor sample. To determine the lower limit of tissue input, we evaluated our MRD assay performance across a range of extracted tumor volumes. Methods: Resected tumors and CNBs were sectioned, H&E stained, and macro-dissected. Tumor gDNA was extracted, quantified, prepared into libraries and sequenced. Sequenced libraries were aligned and evaluated for depth of coverage, variation of coverage, and duplication rate. Somatic calling was performed on matched tumor and normal samples. Up to 1000 target sites were selected for high-depth targeted sequencing of the tumor and normal gDNA for confirmation of somatic variant calls. The positive predictive value (PPV) was computed as the percent of putative somatic variant sites that were present in the tumor capture library and absent in the normal capture library. Results: Extracted tumor volumes varied by almost two orders of magnitude, from 0.06mm3 (equivalent to needle core or fine needle aspirate biopsies) to 5mm3 (achievable with resected tumor). gDNA amount varied linearly (3.6ng to 1549ng) with tumor tissue input, indicating the low tissue to paraffin ratio did not have an adverse effect on yield. Tumor gDNA inputs into library prep ranged from 2.5ng to 100ng. DNA amounts above 100ng into library prep had no discernable benefit. Below 10 ng, depth of coverage and the coefficient of variation in coverage indicated poor-quality libraries. Samples with ~10 ng of gDNA input into library prep showed depth of coverage comparable to higher inputs and saturated the achievable library complexity. Additionally, PPV of somatic calling was consistent across the range of gDNA inputs from 10-100 ng, demonstrating equivalent assay performance. Conclusion: Tumor-informed MRD assays have immense potential for increasingly sensitive treatment response and recurrence monitoring that can inform better treatment decisions. FFPE tumor tissue is a critical input into MRD assays but is a limited resource. This study supports a minimum DNA input of 10 nanograms for a single attempt at extraction, corresponding to a tissue volume of 0.2mm3 or a single 10µm slide with a 20mm2 area, representing one of the lowest tissue input requirements for an MRD assay. Low FFPE tissue requirements expand the patient population that benefit from MRD testing by utilizing samples that have low tumor content, are post-neoadjuvant therapy or do not meet the tumor volume requirements of competing MRD offerings. Citation Format: Matt LaBella, Ashley Acevedo, Ravi Patel, Kiefer Haug, Sangita Ganesh, Elise Buser, Nafei Xu, Shalee Carlson, Kyle Trettin, Sarah Ratzel, Kieko Hara, Pavlos Msaouel, Kanishka Sircar, Chad Tang, Dale Muzzey, Genevieve Gould. Design of high-performance tumor-informed minimal residual disease (MRD) panels from low FFPE tumor input [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 3675.