Epigenomic regulation of human T-cell leukemia virus by chromatin-insulator CTCF

Human T-cell leukemia virus type 1 (HTLV-1) is a retrovirus that causes an aggressive T-cell malignancy and a variety of inflammatory conditions. The integrated provirus includes a single binding site for the epigenomic insulator, CCCTC-binding protein (CTCF), but its function remains unclear. In the current study, a mutant virus was examined that eliminates the CTCF-binding site. The mutation did not disrupt the kinetics and levels of virus gene expression, or establishment of or reactivation from latency. However, the mutation disrupted the epigenetic barrier function, resulting in enhanced DNA CpG methylation downstream of the CTCF binding site on both strands of the integrated provirus and H3K4Me3, H3K36Me3, and H3K27Me3 chromatin modifications both up- and downstream of the site. A majority of clonal cell lines infected with wild type HTLV-1 exhibited increased plus strand gene expression with CTCF knockdown, while expression in mutant HTLV-1 clonal lines was unaffected. These findings indicate that CTCF binding regulates HTLV-1 gene expression, DNA and histone methylation in an integration site dependent fashion.

Human T-cell leukemia virus type 1 (HTLV-1) is a cause of leukemia and lymphoma as well as several inflammatory medical disorders. The virus integrates in the host cell DNA, and it has a single binding site for a protein designated CTCF. This protein is important in the regulation of many DNA viruses, as well as many properties of normal and malignant cells. In order to define the role of CTCF binding to HTLV, we analyzed a mutant virus lacking the binding site. We found that this mutation variably affected gene expression, DNA and histone modification, suggesting a key role in regulation of virus replication in infected cells.

Autologous humanized PDX modeling for immuno-oncology recapitulates features of the human tumor microenvironment

BACKGROUND

Interactions between immune and tumor cells are critical to determining cancer progression and response. In addition, preclinical prediction of immune-related drug efficacy is limited by interspecies differences between human and mouse, as well as inter-person germline and somatic variation. To address these gaps, we developed an autologous system that models the tumor microenvironment (TME) from individual patients with solid tumors.

METHOD

With patient-derived bone marrow hematopoietic stem and progenitor cells (HSPCs), we engrafted a patient's hematopoietic system in MISTRG6 mice, followed by transfer of patient-derived xenograft (PDX) tissue, providing a fully genetically matched model to recapitulate the individual's TME. We used this system to prospectively study tumor-immune interactions in patients with solid tumor.

RESULTS

Autologous PDX mice generated innate and adaptive immune populations; these cells populated the TME; and tumors from autologously engrafted mice grew larger than tumors from non-engrafted littermate controls. Single-cell transcriptomics revealed a prominent vascular endothelial growth factor A (VEGFA) signature in TME myeloid cells, and inhibition of human VEGF-A abrogated enhanced growth.

CONCLUSIONS

Humanization of the interleukin 6 locus in MISTRG6 mice enhances HSPC engraftment, making it feasible to model tumor-immune interactions in an autologous manner from a bedside bone marrow aspirate. The TME from these autologous tumors display hallmarks of the human TME including innate and adaptive immune activation and provide a platform for preclinical drug testing.

Abstract NG11: Autologous humanized PDX modeling for immuno-oncology recapitulates the human tumor microenvironment

The immune milieu within tumors, consisting of diverse cell types including adaptive immune cells as well as macrophages, dendritic cells, natural killer and other innate immune cells, is critical to determining cancer outcome. However, the immune tumor microenvironment (TME) has been challenging to model, owing to inherent inter-species differences. While humanized mice can support human immune cells, the hematopoietic stem and progenitor cells (HSPCs) used for transplantation have been largely limited to fetal or neonatal stem cell sources, necessitating allogeneic experiments with limited applicability. We sought to develop a method to pre-clinically model an individual adult cancer patient, capturing the unique features of an individual such as germline genetic determinants of immune function and somatic tumor heterogeneity, and creating an autologous system.

MISTRG6 may be engrafted with low numbers of HSPCs. When engrafted with equivalent numbers of CD34+ cells from human fetal liver (FL), neonatal cord blood (CB), adult mobilized peripheral blood (MPB), or adult bone marrow (BM), MISTRG6 mice harbored greatly increased human hematopoietic cells as a proportion of total hematopoietic cells in peripheral blood compared with NOD-scid-gamma (NSG) and MISTRG mice (p<0.0001). We found that MISTRG6 mice could be engrafted with as few as 1,000 human HSPCs, arguably 100x more efficient than other models, and achieve robust hematopoietic transplantation after 10-12 weeks, indicating the efficiency of this strain in supporting the growth of hematopoietic cells. To better elucidate the mechanism responsible for this enhanced human engraftment, we enumerated human and mouse hematopoietic progenitors in BM of NSG, MISTRG, and MISTRG6 mice. Human progenitors, including CD34+ and CD34+CD38+ cells, were significantly increased in both frequency and absolute numbers in MISTRG and MISTRG6 mice compared with NSG mice (p<0.001), and mouse hematopoietic lin(-)cKit+ (LK) and lin(-)Sca1+cKit+ (LSK) progenitor populations were significantly diminished (p<0.0001), suggesting that the enhanced hematopoietic engraftment observed in MISTRG6 is, in part, a consequence of increased human progenitor frequency and reduced mouse competition.

MISTRG6 allows efficient engraftment of patient derived HSPCs. We sought to apply this improved engraftment prospectively to model individual patients’ TME through collection of BM-derived CD34+ cells from patients under active treatment along with tumor tissue from the same patient. At two cancer centers, we enrolled patients with melanoma, NSCLC, PDAC, and HNSCC to provide BM aspirate, peripheral blood, and tumor tissue. CD34+ cells were isolated from BM aspirates and tumor tissue was utilized to generate PDXs. Overall, 71 patients were enrolled, 46 melanoma, 19 NSCLC, 4 PDAC, 2 HNSCC, ages 22-85, 39% females. These yielded autologous, immune-reconstituted MISTRG6 hosts from 14 melanoma, 5 NSCLC, 2 PDAC, and 1 HNSCC patients. Autologously engrafted MISTRG6 mice displayed the gamut of human immune cells of adaptive and innate types in PB at 7 weeks of age. Notably, this included CD33+ myeloid cells such as CD14+CD16− classical, CD14+CD16+ intermediate, and CD14−CD16+ non-classical monocytes. Moreover, human dendritic cells (DCs), key innate immune cells for initiation of anti-tumor responses were readily detected by flow cytometry in spleens of autologously-engrafted mice, including cDC1, cDC2, and pDC cells.

MISTRG6 mice bearing a patient’s hematopoietic cells support autologous PDX growth. Having achieved successful engraftment of patient hematopoietic systems in MISTRG6 hosts, we next subcutaneously introduced the patient’s matched PDX tumor tissue to generate autologously engrafted PDX mice. For most patients, tumors grown in autologous HSPC-engrafted hosts were significantly larger than in non-engrafted hosts. Multicolor immunofluorescence staining of PDX tumors demonstrated that human immune cells, including CD3+ T cells, CD14+ and HLA-DR+ myeloid cells, penetrated deeply into the tumor and co-localized with tumor cells as well as with other engrafted immune cells. Indeed, HLA-DR+CD14+macrophages and HLA-DR+CD14(-) dendritic cells were present, and direct physical interaction between T cells and macrophages was evident. Using whole-exome sequencing, we found that 225 somatic changes were shared between patient Mel738’s surgical resection sample, two PDX tumors from non-engrafted mice lacking human immune cells, and two PDX tumors from mice with autologous engraftment. 5 additional changes were shared among the tumor samples and absent from the cell line, with 36 additional mutations being specific to the cell line. These data underscore the capacity of the autologous PDX method to recapitulate the somatic heterogeneity that the patient tumor possesses.

Autologous MISTRG6 mice display diverse human immune cell populations and recapitulate an immunosuppressive TME. To fully characterize the autologous MISTRG6 model and investigate mechanisms by which autologous human immune cells enhance tumor growth, we performed single cell transcriptomics on hCD45+-enriched cells from blood and tumor isolated from autologous mice. This revealed 16 distinct cell subtypes, including 3 myeloid, 2 NK cell, 2 CD8 T cell, 3 CD4 T cell, 2 cycling lymphocyte, 1 B cell, and 3 melanoma cell clusters. Subclustering of myeloid cells revealed 9 distinct clusters including 4 monocyte, 4 macrophage, and 1 DC cluster. Comparing CD8 T cells present in blood versus tumor revealed that the most differentially expressed genes (DEGs) found in blood were characteristic of naïve T cells, while genes present in the TME were consistent with activated T cell phenotypes. In addition, sub-clustering revealed 3 distinct CD8 T cell types that included two activated-like populations, with one of these populations also expressing an activated/exhausted program typified by expression of PDCD1, LAG3, and GZMA. Naïve-like T cells were most highly represented in the blood, while activated and activated/exhausted-like genes were more present in the TME.

Inhibiting the actions of human VEGF-A blocks the enhanced tumor growth in autologously engrafted mice. Notably, IPA Upstream Regulator Analysis identified VEGFA, a central player in tumor growth and vascularization, as a key upstream inducer of genes in the TME (FDR p= 5.65 × 10−13). Indeed, expression VEGFA itself was nearly absent in blood but induced in the TME, especially in macrophages and VEGFA targets were highly represented among the DEGs between tumor and blood.To test the relevance of VEGF-A in the TME, we selectively blocked human VEGF-A by treating autologous mice humanized from Mel2 with the anti-hVEGF-A antibody bevacizumab that has high affinity for human VEGF-A yet low affinity for mouse VEGF-A. PDXs grown in untreated autologously engrafted MISTRG6 mice grew significantly larger than those in non-engrafted littermate control hosts (p<0.05). When treated with bevacizumab, the enhanced tumor growth was significantly abrogated, with bevacizumab-treated mice bearing significantly smaller tumors compared with controls (p<0.001).

Future Directions: Thus, these in silico and in vivo results suggest that human VEGF-A production in the autologous TME enhances tumor growth in MISTRG6 PDX models and underscores the utility of the MISTRG6 system for pre-clinical testing of drugs that act on human immune components of the TME. By engrafting mice with bone marrow derived stem cells followed by implantation of tumor derived from the same donor, we have demonstrated that autologous MISTRG6 models recapitulate important features of the human TME, including sufficient immunosuppression to prevent tumor clearance, presence of activated/exhausted T cells, and innate immune cells including DCs, monocytes, NK cells, and macrophages, the latter especially relevant to the production of VEGF-A.

Citation Format: Michael Chiorazzi, Jan Martinek, Bradley Krasnick, Yunjiang Zheng, Keenan Robbins, Rihao Qu, Gabriel Kaufmann, Zachary Skidmore, Laura Henze, Frederic Brösecke, Adam Adonyi, Jun Zhao, Liang Shan, Esen Sefik, Jacqueline Mudd, Ye Bi, S Peter Goedegebuure, Malachi Griffith, Obi Griffith, Abimbola Oyedeji, Sofia Fertuzinhos, Roland Garcia-Milian, Daniel Boffa, Frank Detterbeck, Andrew Dhanasopon, Justin Blasberg, Benjamin Judson, Scott Gettinger, Katerina Politi, Yuval Kluger, A Karolina Palucka, Ryan Fields, Richard A. Flavell. Autologous humanized PDX modeling for immuno-oncology recapitulates the human tumor microenvironment. [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr NG11.

Abstract 4739: The immunogenomic landscape of HPV-associated oropharyngeal squamous cell carcinoma by smoking and treatment response

Background: Human papilloma virus (HPV)-associated oropharyngeal squamous cell carcinoma (OPSCC) is an increasingly common malignancy and public health epidemic. Overall, outcomes in HPV-associated OPSCC are very good but treatment failure is associated with a devastating prognosis. Although tobacco use is one of the few clinical biomarkers associated with worse prognosis, the mechanisms underlying this disparity are not well understood. The goal of this study was to investigate the genomic and immune related alterations associated with tobacco use in HPV-associated OPSCC.

Methods: We prospectively identified a cohort of 53 patients with HPV-associated OPSCC with corresponding clinical and demographic data. Tumor DNA and RNA were extracted from FFPE samples along with whole blood as a control for variant calling. We performed whole exome sequencing, mutational signature analysis and a 770 gene RNA-based NanoString immuno-oncology assay to further interrogate these tumors.

Results: We identified that smokers and non-smokers had remarkably well conserved copy number alterations and mutational alterations. Mutational alterations in tumors across the cohort including in heavy smokers had frequent mutations in PIK3CA, FGFR3, ZNF750 that largely reflect the known alterations common in HPV-associated OPSCC. Regardless of smoking status, we rarely observed mutations typical of non-HPV associated head and neck cancers such as TP53, CDKN2A, and FAT1. COSMIC mutational signature analysis revealed that a majority of the tumors expressed an APOBEC signature (#2 and #13) consistent with viral etiology of tumorigenesis. Strikingly, none of the tumors expressed mutational signatures associated with tobacco use (#4 and #29). We then performed an RNA-based NanoString immuno-oncology assay to investigate differences within the immune microenvironment. We identified significantly upregulated expression of genes associated with myeloid derived suppressor cell (MDSC) recruitment and activity including S100A8/A9, S100A12, CXCR2, CXCL1, and ARG2, suggesting a potential association between tobacco use, MDSC-driven immunosuppression and poorer prognosis.

Conclusion: In this genomic analysis of HPV-associated OPSCC, we demonstrate that tobacco use does not clearly induce structural or mutational differences within the tumor cells. Even in heavy tobacco users, the mutational signatures reflect a tumor driven by its viral etiology and more closely aligns to non-tobacco using HPV-associated OPSCC than it does HPV-negative OPSCC. The poorer prognosis in smokers may be related to alterations in the immune microenvironment including the recruitment and activity of MDSCs. This provides further evidence that strategies to target MDSC activity in the immune microenvironment may be effective in high-risk HPV-OPSCC smokers.

Citation Format: Paul Zolkind, Ben Wahle, Zachary Skidmore, Angela Mazul, Obi Griffith, Malachi Griffith, Jose P. Zevallos. The immunogenomic landscape of HPV-associated oropharyngeal squamous cell carcinoma by smoking and treatment response [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 4739.

Abstract A1-44: Clinical cancer sequencing and integrated analysis of whole genomes, exomes and transcriptomes

Deep and comprehensive tumor sequencing in a clinical context presents unique challenges compared to discovery-based cancer genomics. To explore these challenges, we have developed a comprehensive approach for identification of clinically actionable events in patient tumors by integrated analysis of whole genome, exome, and transcriptome sequencing. To demonstrate the utility of this approach we sequenced the DNA and RNA for both tumor and matched normal tissue of a diverse set of 21 cancer cases (1 ALL, 2 AML, 6 breast, 1 gastrointestinal adenocarcinoma, 1 gastrointestinal stromal tumor, 1 lung, 1 low grade glioma, 1 high grade glioma, 1 leiomyosarcoma, 1 signet ring gastric and 5 pancreatic). Each case represented a patient with advanced disease. These tumors varied substantially in their purity, heterogeneity, extraction method, sample quality, and sample amount. Each tumor/normal pair was sequenced to ~30-90X whole genome coverage, ~150-300X exome coverage for tumor and normal, and varying transcriptome coverage depending on sample quality (at least one lane of Illumina HiSeq2000 data each). Integrating the analysis of all three data types allowed for more sensitive and interpretable identification of clinically relevant tumor associated mutations than any single approach. For example, combining exome and whole genome data increased detection of variants in sub-clones and low purity tumors. Combining WGS and RNA-seq data allowed confirmation of the expression effect of focal amplifications, identification of variant biased allele-specific expression and confirmation of gene fusion products predicted by structural variants. To maximize the potential for at least one clinically actionable finding in each case, our analysis goal was to identify, annotate, visualize and prioritize single nucleotide variants (SNVs), small indels, translocations, copy number variants, gene fusions, and expression of aberrant mRNA isoforms. We accomplished these tasks by creating a clinical sequencing pipeline that incorporates existing and novel bioinformatics methods into the analysis infrastructure of the Genome Institute's Genome Modeling System (GMS). A maximum turnaround time of 30 days was targeted for every case from sample receipt to complete report generation. Events were prioritized according to potential clinical relevance with particular attention paid to focal amplifications, SNVs and indels with ‘driver’ characteristics, gene fusions, and aberrantly expressed genes. These candidates were further prioritized by a suite of tools we are developing to help researchers and clinicians assess clinical actionability including: DGIdb (www.dgidb.org) a drug-gene interaction resource created to facilitate mining the druggable genome, DoCM (www.docm.info) a database of canonical mutations, and CIViC (www.civicdb.org) an open interface for clinical interpretation of variants in cancer.

Citation Format: Malachi Griffith, Obi L. Griffith, Avinash Ramu, J Benjamin Ainscough, Kilannin Krysiak, Mayank Choudhary, Zachary Skidmore, Benjamin Tan, Govindan Ramaswamy, Brian Van Tine, Matthew J. Ellis, Timothy J. Ley, Richard K. Wilson, Elaine R. Mardis. Clinical cancer sequencing and integrated analysis of whole genomes, exomes and transcriptomes. [abstract]. In: Proceedings of the AACR Special Conference on Translation of the Cancer Genome; Feb 7-9, 2015; San Francisco, CA. Philadelphia (PA): AACR; Cancer Res 2015;75(22 Suppl 1):Abstract nr A1-44.

Abstract IA20: Genomics of a STAT1 knockout mouse model of human ER+ breast cancer

Estrogen receptor alpha positive luminal breast cancers are the most frequent subtype of breast cancer. Previous work has established that Stat1-/- mouse mammary tumor model recapitulates signaling, expression and phenotypic alterations observed in this subtype of human breast cancers. To identify transforming events that contribute to tumorigenesis, we performed whole genome sequencing of 22 Stat1-/- primary mammary tumors and cell lines. We discovered a novel hotspot of somatic mutations in 100% of tumors that resulted in a truncated form of the prolactin receptor (Prlr). Targeted sequence analysis identified similar mutations in 77.8% of ductal carcinoma in situ. Co-expression of truncated and full-length Prlr in normal cells led to activation of oncogenic Stat3 and Stat5 as well as cellular transformation. In conclusion, truncating mutations of Prlr drive tumor development in a model of human ERa+ breast cancer and should be considered as novel antitumor targets.

Citation Format: Elaine Mardis, Obi L. Griffith, Ruby Chan Szeman, Malachi Griffith, Kilannin Krysiak, Zachary Skidmore, Jasreet Hundal, Julie A. Allen, Arthur Cora, Alexander P. Miceli, Heather Schmidt, Lee Trani, Krishna-Latha Kanchi, Christopher A. Miller, David E. Larson, Robert S. Fulton, Richard K. Wilson, Robert D. Schreiber. Genomics of a STAT1 knockout mouse model of human ER+ breast cancer. [abstract]. In: Proceedings of the AACR Special Conference on Advances in Breast Cancer Research; Oct 17-20, 2015; Bellevue, WA. Philadelphia (PA): AACR; Mol Cancer Res 2016;14(2_Suppl):Abstract nr IA20.

Abstract P2-02-14: Circulating tumor DNA predicts clinical outcome in early stage triple negative breast cancer

Background- Triple negative breast cancer (TNBC) is the most aggressive subtype of breast cancer as these patients have the highest risk of recurrence and death. Only 35% of TNBC patients achieve a pathologic complete response (pCR) following neoadjuvant chemotherapy. Patients who do not achieve pCR have a 27% risk of distant recurrence and ultimate death at 3 years compared to 9% for pCR. Unidentified micrometastases are responsible for ultimate overt progression and death. Developing strategies to identify patients with minimal residual disease following curative treatment is an unmet need. Circulating tumor DNA (ctDNA) can characterize and monitor advanced cancers. In this study, we sought to assess if ctDNA can predict clinical outcome in TNBC.

Methods-Biospecimens were obtained from patients with stages II and III TNBC enrolled on a neoadjuvant trial (NCT02124902). Patients have a research biopsy and plasma for ctDNA collected at baseline, cycle 1 day 3, definitive surgery for those with residual disease, and at recurrence for those who relapse. Plasma for ctDNA is also collected every 6 months for 5 years after treatment. Patients receive docetaxel and carboplatin every 3 weeks X 6 cycles. Surgery is 3-5 weeks after chemotherapy. Six patients' serial tumor samples and germline DNA were studied by whole exome sequencing. The median sequencing depth was 90.13x. Sequencing was performed on samples with high cellularity (≥50%). All 6 patients also had serial ctDNA analyzed using Swift Biosciences Accel-Amplicon™ 56G Oncology Panel v2. After identifying somatic mutations in each breast tumor series, we determined the subset of mutations that intersected with the regions targeted by the Swift 56 gene panel. We then evaluated whether corresponding mutations could be detected in ctDNA, and if ctDNA predicted clinical outcome.

Results-Four of the 6 patients were non-pCR with residual disease following chemotherapy. We identified 627 somatic variants by exome analysis that were called by at least two somatic variant callers and passed additional quality filtering steps. Of these, 10 variants overlapped with the Swift panel. TP53 variants were identified in all 6 patients' tumor tissue samples. At least one TP53 variant was identified in 4 patients' baseline pre-chemotherapy ctDNA samples. Both pCR patients had either no detectable ctDNA TP53 mutations (NTN007-ref. in baseline tumor tissue was 19.58% variant allele frequency [VAF]); or clearance of ctDNA following chemotherapy from 4.45% VAF at baseline to 0.06% following chemotherapy (NTN004-ref. in baseline tumor tissue 37.34% VAF). Three non-PCR patients had persistent TP53 mutations in ctDNA during the treatment course. One non-pCR patient did not have detectable mutations in ctDNA. The only patient with recurrent disease whose ctDNA TP53 mutation persisted during the treatment course (baseline VAF-1.65%, cycle 1 day 3-0.78%, definitive surgery-0.09%), was found to have a higher ctDNA VAF at recurrence (29.55%).

Conclusion-In this pilot study, mutation tracking by ctDNA is sensitive and distinguishes pCR from non-pCR in TNBC patients receiving neoadjuvant chemotherapy. ctDNA also identifies recurrence following curative therapy. Evaluating ctDNA as a biomarker of outcome in TNBC is warranted.

Citation Format: Ademuyiwa F, Feng Y-Y, Skidmore Z, Kunisaki J, Walker J, Fulton R, Krysiak K, Skinner T, Weilbaecher K, Ma C, Griffith O, Griffith M. Circulating tumor DNA predicts clinical outcome in early stage triple negative breast cancer [abstract]. In: Proceedings of the 2017 San Antonio Breast Cancer Symposium; 2017 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2018;78(4 Suppl):Abstract nr P2-02-14.

Abstract PO-06: Ultradeep sequencing of classical Hodgkin lymphoma (cHL) identifies recurrent somatic mutations and demonstrates the production of reproducible data from rare malignant cells

Purpose/Background: cHL patients who receive standard therapy have a high rate of event-free and overall survival. However, some patients (~10%) will be refractory to initial therapy and up to 1/3 will relapse. Thus, improved methods of prognostication and new treatment targets are needed. High-throughput sequencing can identify recurrent somatic mutations that drive lymphomagenesis and impact treatment response. However, Hodgkin-Reed-Sternberg (HRS) cells have a low (~1%) abundance in cHL biopsies, creating a challenge for comprehensive and accurate detection of somatic mutations in bulk lymphoma biopsies. Genomic studies of cHL have characterized HRS somatic mutations through the analysis of malignant cells obtained using purification techniques, cell-free DNA, or DNA amplified through whole-genome amplification. We hypothesized that ultradeep sequencing of bulk lymphoma biopsies provides a more accessible approach to HRS characterization while also creating robust and reproducible data.

Methods: We performed exome sequencing on 32 fresh frozen samples from 31 cHL patients obtained prior to treatment (27) or after relapse (4) with paired normal skin samples (31). The Illumina HiSeq platform (2 x 150bp reads) was used with multiple independent library constructions and a 1,000X median coverage goal. Sequence data were aligned to GRCh38. SNVs and INDELs were called using multiple algorithms. We employed several variant filtering strategies, including manual review, to remove common polymorphisms and false positives. Because we discovered mutations with VAFs close to the platform error rate (~1%), we used an orthogonal sequencing strategy (Haloplex) to validate all somatic variants.

Results: We observed 4,020 somatic variants. On average, we observed 32 protein-coding mutations/case, excluding one hypermutated case in which 3,084 variants were observed. We identified a potential loss-of-function insertion in MSH6 that could explain the hypermutated phenotype. We achieved a 99% validation rate across the cohort for somatic variants discovered in exomes. We confirmed known recurrently mutated cHL genes (e.g., SOCS1 [43%], STAT6 [20%], TNFAIP3 [40%]). We identified several significantly recurrent mutated genes not well characterized in cHL, including IGLL5 [26%] and IL4R [13%]. All IL4R mutations are potential loss-of-function mutations that could result in greater activation of STAT6 through ablation of ITIM negative modulation. We identified an enrichment of SOCS1 and IGLL5 mutations that is likely the result of aberrant somatic hypermutation. Pathway analysis also identified an enrichment of mutations in MAPK pathways.

Conclusion: These data suggest that cHL somatic mutations can be confidently identified via ultradeep exome sequencing without cell purification. We show that cHL genomes harbor somatic variation that inform new targets for treatment and prognostication.

Citation Format: Felicia Gomez, Matthew Mosior, Zachary Skidmore, Alina Schmidt, Fernanda Rodrigues-Martins, Kilannin Krysiak, Cody Ramirez, Eric Duncavage, Grace Triska, Lee Trani, Nancy Bartlett, Amanda Cashen, Neha Mehta-Shah, Friederike Kreisel, Malachi Griffith, Todd Fehniger, Obi Griffith. Ultradeep sequencing of classical Hodgkin lymphoma (cHL) identifies recurrent somatic mutations and demonstrates the production of reproducible data from rare malignant cells [abstract]. In: Proceedings of the AACR Virtual Meeting: Advances in Malignant Lymphoma; 2020 Aug 17-19. Philadelphia (PA): AACR; Blood Cancer Discov 2020;1(3_Suppl):Abstract nr PO-06.