Genomic alterations in lung adenocarcinoma displayed in Figure 2A.

Supplementary Table 1. Mutation hotspots interrogated by Sequenom.

Previously characterized MET exon 14 alterations.

Summary of MET exon 14 alterations.

Supplementary Figure 1. Distribution of median exon coverage depths for the samples included in this study, all passing the minimum threshold of 250x with an average median exon coverage depth >700x. Supplementary Figure 2. Median exon coverage depth and percent of exons >100x for 34 samples included in this study.

Co- and anti-occurrence of genomic alterations in lung adenocarcinoma.

Supplementary Table 1: Results for samples obtained (less than or equal to) 12 months prior to therapy (optimal) vs. those obtained >12 months from treatment or shortly after starting therapy (non-optimal). Supplementary Table 2: Cox proportional hazards model of OS adjusting for baseline variables. Supplementary Table 3: Cox proportional hazards model of PFS adjusting for baseline variables. Figure S1: (A) Performance of mutational load across a range of potential thresholds. (B) Receiver Operating Curve (ROC) for mutational loads cutoffs of 3.3 mutations/MB (low mutational load group) and 23.1 mutations/MB (high mutational load group). Figure S2: (A) Mutational load in tumors with cutaneous/unknown primaries in responders vs. non-responders. (B) Mutational load in tumors with non-cutaneous primaries (acral, mucosal, uveal) in responders vs. non-responders. (C) Gene amplifications and deletions in responders vs. non-responders. Figure S3: (A) LRP1B mutations/variants of unknown significance in responders vs. non-responders. (B) Total number of mutations among melanomas with and without LRP1B mutations. (C) Association between number of LRP1B mutations and total mutations in the melanoma TCGA. Figure S4: (A) T cell receptor (TCR) clonality in responders vs. non-responders. (B) T cell fraction in responders vs. non-responders. (C) TCR clonality in responders vs. non-responders in ideal samples, defined as those obtained within 4 months of anti-PD-1/PD-L1 treatment without other prior therapies. (D) T cell fraction in these ideal samples. Figure S5: Mutation load correlated with (A) T cell receptor (TCR) clonality and (B) T cell fraction.

BACKGROUND & AIMS:It has been a challenge to select treatment for patients with pancreatic ductal adenocarcinomas (PDACs) based on genome alterations. We performed targeted genomic profile analyses of a large number of PDACs to assess the full spectrum of actionable genomic alterations. METHODS:We performed targeted genomic profile analyses of 3594 PDAC samples from an international cohort, including capture-based targeted genomic profiling of as many as 315 cancer-associated genes and intron regions of 28 genes that are rearranged in cancer cells. Tumor mutation burden (TMB) and microsatellite instability (MSI) status were also assessed. TMB was calculated across a 1.14-megabase region; TMB-high was defined as ≥20 mutations/megabase. MSI-high status was assigned based on analysis of 114 intron homopolymer loci. RESULTS:KRAS, TP53, CDKN2A, and SMAD4 were the most frequently altered genes in PDAC. We found KRAS mutations in 88% of samples. Among PDACs without mutations in KRAS, we found alterations in genes whose products are in the mitogen-activated protein kinase signaling pathway and are candidate drug targets (actionable targets, n = 132; 4%), as well as gene fusions (n = 51), gene amplifications (n = 35), genes with missense mutations (n = 30), and genes that contain deletions (n = 16). Many of these encode proteins in receptor tyrosine kinase, RAS, or mitogen-activated protein kinase signaling pathways. Aside from TP53, alterations in genes encoding DNA damage repair proteins (BRCA and FANC) were detected in 14% of PDACs. Among PDACs evaluated for MSI (n = 2563) and TMB (n = 1021), MSI-high and/or TMB-high phenotypes were detected in 0.5% of samples. Alterations in FGF23, CCND2, PIK3CA, and FGF6 were more commonly detected in intraductal papillary mucinous neoplasm-associated PDACs. CONCLUSIONS:In targeted genomic profile analyses of 3594 PDACs, we found 17% to contain genomic alterations that might make the tumor cells susceptible to currently used anticancer agents. We identified mutations in genes that could contribute to progression of intraductal papillary mucinous neoplasms into malignancies. These alterations might be used as biomarkers for early detection.

271 Background: Mutations in oncogenic KRAS have been widely accepted as the signature genomic alteration (GA) in sporadic PDAC, but therapeutic efforts aimed at targeting constitutively activated KRAS have been disappointing. We examined somatic GAs in KRAS WT PDAC utilizing a CGP platform to identify actionable targets. Methods: DNA was extracted from formalin fixed paraffin embedded (FFPE) PDAC clinical specimens and CGP was performed on hybrid-capture, adaptor ligation based libraries to a mean coverage depth of > 600 unique reads. Alterations in the RAS/ RAF/ MEK pathway genes ( KRAS, NRAS, HRAS, ARF, BRAF, EGFR, MAP2K2, MAP2K1, MAPK1) and DNA Damage Repair (DDR) pathway genes ( BRCA1/2, ATM, ATR, BRIP1, RAD50, RAD51, RAD52, PALB2, CHEK1, CHEK2) were examined. Tumor mutational burden (TMB) was determined on 1.1 Mbp of sequenced DNA and microsatellite instability (MSI) was determined on 114 loci. TMB was categorized based on mutations(m)/Mbp of DNA - high (H; > 20), Intermediate (I; 8-20) and low (L; < 8). Results: CGP was performed on 3426 PDAC specimens; 1815 (53%) were male, 390 (11.3%) were KRAS WT. GAs in the RAS/ RAF/ MEK pathway were identified in 90.6% of all cases, while 68 (17.4%) KRAS WT cases had one or more GAs in RAS/ RAF/ MEK pathway genes. DDR pathway GAs were identified in 1405 (41%) cases for a total of 2050 GAs, and 180 (46%) KRAS WT cases for a total of 285 GAs. DDR pathway alterations were common in KRAS WT PDAC compared to KRAS mutated PDAC (p = 0.028). Among the 842 (24.6%) cases with available TMB data, 5 (0.6%), 104 (12.3%) and 733 (87.1%) pts had H, I and L, TMB respectively. Among 88 (22.6%) KRAS WT cases with available TMB data, 2 (2.3%), 12 (13.6%) and 74 (84.1%) pts had H, I and L, TMB, respectively. MSI status was available in 2314 (67.5%) cases, 13 (0.6%) were MSI-high (MSI-H); among the KRAS WT cases, 222 (57%) had MSI status available, 3 (1.3%) were MSI-H. Conclusions: MSI-H status and high TMB are rare in PDAC, regardless of KRAS mutation status. GAs in the DDR pathway are relatively common in PDAC and may serve as predictive biomarkers for platinum chemotherapeutic agents and/or PARP inhibitors. Prospective validation of such predictive gene signatures will improve therapeutic efficacy and minimize toxicities.

Abstract Breast invasive lobular carcinoma (ILC) is the second most common type of breast cancer, making up approximately 10% of all invasive breast cancers. Breast ILC is characterized by loss of e-cadherin protein expression (CDH1), which is usually caused by loss of function mutations in the CDH1 gene. A subset of breast ILC cases have no identifiable CDH1 mutation, but presumably inactivate CDH1 through currently unknown mechanisms. Tumors from more than 150,000 unique patients including more than 15,000 breast cancers were examined by comprehensive genomic profiling to detect base substitutions, indels, copy number alterations, and genomic rearrangements in the full coding regions of up to 465 genes. Statistical significance of the disease distributions was calculated using Fisher's exact test with false discovery rate multiple hypothesis testing correction. Of 760 breast ILC cases, 583 (76.7%) had a known loss-of-function mutation in the CDH1 gene. To identify previously unknown mutations that result in or substitute for CDH1 loss-of-function we performed statistical analysis on all mutations, in any gene, occurring more than once in the 177 ILC cases without an identifiable CDH1 mutation. The most statistically significant mutation was very near to a splice site of CDH1 (chr16:68863554 c.2296-3A>G), occurring in 7 CDH1 negative ILC cases as well as 9 other cases in cancer types that frequently harbor CDH1 mutations (7 breast, 2 gastroesophageal junction) and 1 colorectal cancer. Based on tumor-only test based zygosity modeling, this mutation is only observed as somatic and usually homozygous in the tumor. Four other positions near to splice sites of CDH1 (chr16:68844247 c.832+3A>G, chr16:68849667 c.1565+5G>A, chr16:68856133 c.1936+5G>A, chr16:68857532 c.2164+3A>C), mutated in 13 CDH1 negative ILC cases, were also statically significant, somatic, and usually homozygous. No statistically significant mutations were found in genes other than CDH1. Previously unappreciated somatic, non-canonical splice site mutations in CDH1 occur in ~10% (20/177) of breast ILC cases with no known CDH1 mutation. These mutations occur almost exclusively in tumor types known to harbor frequent CDH1 mutations and are usually homozygous in the tumor, strongly indicating that they are CDH1 loss-of-function mutations. Citation Format: Garrett M. Frampton, Ryan Hartmaier, Ethan Sokol, Anika Gupta, Joel Greenbowe, Steven Roels, Laurie Gay, Philip J. Stephens. Novel CDH1 mutations in breast invasive lobular carcinoma [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr 2363.

After sequential treatment with first- and third-generation EGFR tyrosine kinase inhibitors (TKIs), EGFR-mutant non-small cell lung cancers frequently harbor multiple resistance mutations in exon 20 of EGFR including T790M, mediating resistance to first-generation TKIs, and at codons 792, 796, or 797 mediating resistance to third-generation TKIs. However, whether these resistance mutations are in cis or trans has therapeutic implications for patients. We analyzed a cohort of 29 patients with NSCLC harboring EGFR mutations at codons 792, 796, or 797 to establish the configuration of these mutations. We performed hybrid capture-based, next-generation sequencing on formalin-fixed paraffin-embedded biopsy tissue or liquid biopsy. 27 samples had both a T790M mutation and a mutation at codons 792, 796, or 797. In all of these cases, the mutations were found in the cis configuration; the trans configuration was not observed. Two patients’ samples harbored a mutation at codon 797 but no T790M mutation. In these two cases, longitudinal analysis showed earlier biopsies harbored EGFR T790M, which was undetectable following osimertinib treatment. Treatment of one these patients with both first- and third-generation EGFR TKIs resulted in a mixed response. Here we describe multiple configurations of EGFR T790M and third-generation TKI resistance mutations at codons 792, 796, and 797. These mutations are most commonly found in cis, which confers resistance to all current EGFR TKIs. We also describe two patients that exhibited T790M loss with acquisition of a mutation at codon 797. In addition, one of these patients, with an EGFR C797S in a lung biopsy was subsequently found to have EGFR C797N in a later biopsy of pleural fluid, highlighting the dynamic multiclonal nature of advanced NSCLC.

Despite a therapeutic paradigm shift into targeted-driven medicinal approaches, resistance to therapy remains a hallmark of lung cancer, driven by biological and molecular diversity. Using genomic and expression data from advanced non-small cell lung cancer (NSCLC) patients enrolled in the BATTLE-2 clinical trial, we identified RICTOR alterations in a subset of lung adenocarcinomas and found RICTOR expression to carry worse overall survival. RICTOR-altered cohort was significantly enriched in KRAS/MAPK axis mutations, suggesting a co-oncogenic driver role in these molecular settings. Using NSCLC cell lines, we showed that, distinctly in KRAS mutant backgrounds, RICTOR blockade impairs malignant properties and generates a compensatory enhanced activation of the MAPK pathway, exposing a unique therapeutic vulnerability. In vitro and in vivo concomitant pharmacologic inhibition of mTORC1/2 and MEK1/2 resulted in synergistic responses of anti-tumor effects. Our study provides evidence of a distinctive therapeutic opportunity in a subset of NSCLC carrying concomitant RICTOR/KRAS alterations.

BACKGROUND:The clinical application of PD1/PD-L1 targeting checkpoint inhibitors in colorectal cancer (CRC) has largely focused on a subset of microsatellite instable (MSI-high) patients. However, the proposed genotype that sensitizes these patients to immunotherapy is not captured by MSI status alone. Estimation of tumor mutational burden (TMB) from comprehensive genomic profiling is validated against whole exome sequencing and linked to checkpoint response in metastatic melanoma, urothelial bladder cancer and non-small cell lung carcinoma. We sought to explore the subset of microsatellite stable (MSS) CRC patients with high TMB, and identify the specific genomic signatures associated with this phenotype. Furthermore, we explore the ability to quantify TMB as a potential predictive biomarker of PD1/PD-L1 therapy in CRC.METHODS:Formalin-fixed, paraffin embedded tissue sections from 6,004 cases of CRC were sequenced with a CLIA-approved CGP assay. MSI and TMB statuses were computationally determined using validated methods. The cutoff for TMB-high was defined according to the lower bound value that satisfied the 90% probability interval based on the TMB distribution across all MSI-High patients.RESULTS:MSS tumors were observed in 5,702 of 6,004 (95.0%) cases and MSI-H tumors were observed in 302 (5.0%) cases. All but one (99.7%) MSI-H cases were TMB-high (range, 6.3-746.9 mut/Mb) and 5,538 of 5,702 (97.0%) MSS cases were TMB-low (range, 0.0-10.8 mut/Mb). Consequently, 164 of 5,702 (2.9%) MSS cases were confirmed as TMB-high (range, 11.7-707.2 mut/Mb), representing an increase in the target population that may respond to checkpoint inhibitor therapy by 54% (466 vs. 302, respectively). Response to PD-1 inhibitor is demonstrated in MSS/TMB-high cases.CONCLUSIONS:Concurrent TMB assessment accurately classifies MSI tumors as TMB-high and simultaneously identifies nearly 3% or CRC as MSS/TMB-high. This subgroup may expand the population of CRC who may benefit from immune checkpoint inhibitor based therapeutic approaches.

292 Background: Kinases activated by gene fusions represent an important class of oncogenes that have been detected in hematologic and solid malignancies. These fusion genes have been well-described in non-small cell lung cancer, and such patients often benefit from matched kinase inhibitors. Recent work in pancreatic ductal adenocarcinoma (PDAC) has also identified recurrent ALK and BRAF rearrangements. Further, patients harboring the former have benefited from ALK inhibitors, as previously reported. We present a survey of kinase fusion genes identified in a large, consecutive series of PDACs. Methods: In the context of clinical care, hybrid-capture based CGP was performed prospectively on 3,426 PDACs for up to 315 cancer-related genes and intronic regions of up to 28 genes that are commonly rearranged in cancer. Results: Overall, kinase fusion genes were identified in 32 (1%) patients with PDAC. The patients consisted of 8 females and 23 males, and ranged in age between 31 and 83 years (median, 58 years). PDACs harboring kinase fusions were predominantly KRAS-wildtype (97%). In fact, the prevalence of kinase fusion genes among KRAS-wildtype PDACs was 8%. Gene fusions involved BRAF (n = 22), ALK (n = 5), NTRK (n = 3) and MET (n = 2). Concurrent genomic alterations included CDKN2A/B (41%), TP53 (28%), GNAS (19%), SMAD4 (16%), MCL1 (16%), ARID1A (13%) and MYC (13%). Updated clinical history will be presented on patients, who previously benefitted from ALK inhibitors. Conclusions: Recurrent gene fusions involving BRAF, ALK, NTRK and MET occur in a subset of PDACs. Although rare, these kinase fusions potentially represent actionable targets and screening should be considered, especially in KRAS-wildtype PDACs.

Pancreatic ductal adenocarcinoma (PDAC) is one of the most lethal cancers, with a 5-year survival of 8%. Current therapeutic regimens are largely ineffective and underscore the need for novel treatment strategies. Chromosomal rearrangements involving the anaplastic lymphoma kinase (ALK) gene have been identified in several neoplasms. In addition, ALK protein inhibitors have proven efficacy in patients with ALK-rearranged tumors. However, ALK translocations in PDAC have not been described. Through comprehensive genomic profiling of 3,170 PDACs, we identified 5 cases (0.16%) that harbored an ALK fusion gene: an exon 6 EML4-exon 20 ALK translocation (n=3), an exon 13 EML4-exon 20 ALK translocation (n=1), and an exon 3 STRN-exon 20 ALK translocation (n=1). Among the most prevalent PDAC-related genes, activating KRAS mutations were absent in all 5 cases, who were <50 years of age. Among patients aged <50 years in our study cohort, ALK translocations constituted 1.3% of PDACs. Four of 5 patients were treated with an ALK inhibitor, and 3 of these patients demonstrated stable disease, radiographic response, and/or normalization of serum CA 19-9. Although rare, ALK fusions occur in PDAC, and screening for ALK rearrangements should be considered in young patients with PDAC.

Abstract Background Patients across a range of disease types have demonstrated robust and durable responses using checkpoint inhibitor therapies (CPITs). Given the limitations of immuno-histochemical based testing, identifying a unified, quantitative metric to determine potential response to CPITs remains an urgent need. Tumor mutational burden (TMB) measures the number of somatic protein coding mutations per target sequence in a tumor specimen. This measure has been associated with response and survival for multiple CPITs across an array of indications. In this study we describe Foundation Medicine’s (FMI) work to develop and validate a TMB result as part of our comprehensive genomic profiling assays and summarize clinical feasibility in NSCLC, melanoma and bladder cancer. Methods We developed an analysis method to determine TMB based on data from our comprehensive genomic profiling assays. TMB is calculated by counting all synonymous and non-synonymous somatic variants across 315 or 405 genes, excluding germline alterations and known or likely driver alterations. The mutation count is normalized by the coding target territory to achieve a mutation density of mutations per megabase (mut/Mb). To determine accuracy, we compared TMB values from our comprehensive genomic profiling assay against a CLIA-validated whole-exome sequencing (WES) method on 29 patients. Precision was assessed over 10 clinical samples replicated 4-6 times. Lower limit of sample tumor purity was determined through dilutions of tumor/normal pairs from 80% to 5% tumor. Clinical feasibility was assessed by analyzing TMB versus immunotherapy-based survival in a cohort of 65 metastatic melanoma patients, 150 urothelial carcinoma patients and 463 NSCLC patients. Additionally, we examined the relationship between TMB and microsatellite instability status (MSI), an independent biomarker associated with response to CPITs. Results Foundation Medicine’s TMB measure provides accurate and precise results across a range of tumor mutational burden values on samples with as little as 20% tumor purity. Using cohort specific thresholds, TMB was significantly associated with improved survival to CPITs in NSCLC, melanoma and bladder cancer. Using data from over 40,000 patient samples, we also show significant overlap between high TMB and high MSI samples and show that MSI-High specimens represent a subset of TMB-High specimens. Conclusions We have developed and validated the tumor mutational burden (TMB) biomarker as part of our comprehensive cancer genomic profiling assays. Initial clinical feasibility results demonstrate that TMB can be used to predict the likely response to anti-PD-1/PD-L1 CPITs across a growing number of indications including NSCLC, melanoma and bladder cancer. Citation Format: Daniel S. Lieber, Mark R. Kennedy, Douglas B. Johnson, Jonathan E. Rosenberg, Marcin Kowanetz, Joel R. Greenbowe, Garrett M. Frampton, Caitlin F. Connelly, Alexa B. Schrock, Jeffrey S. Ross, Philip J. Stephens, Siraj M. Ali, Vincent A. Miller, David A. Fabrizio. Validation and clinical feasibility of a comprehensive genomic profiling assay to identify likely immunotherapy responders through tumor mutational burden (TMB) [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2017; 2017 Apr 1-5; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2017;77(13 Suppl):Abstract nr 2987. doi:10.1158/1538-7445.AM2017-2987

Abstract Background: The ability of tumors to evade immune surveillance by overexpressing immune checkpoint proteins has been exploited for therapeutic intervention through antibodies designed to interrupt their signaling. A number of patients across a range of disease types, including melanoma, lung, renal and bladder cancer, have demonstrated robust and durable responses using checkpoint inhibitor therapies (CPITs). Still, identifying the most likely responders remains an urgent need for proper clinical management. Tumor mutational burden (TMB) measures the overall number of somatic protein coding mutations per area of sequence counted occurring in a tumor specimen. This measure has been associated with both response and survival for multiple CPITs across an array of indications. It is hypothesized that immunotherapies are more effective for tumors with high TMB because these cells are more likely to express immune-reactive neoantigens. In this study we describe Foundation Medicine's (FMI) work to develop and validate a TMB result as part of the current FoundationOne (F1) and FoundationOne Heme (F1H) comprehensive genomic profiling assays. Methods: We developed an analysis method to determine TMB based on data from both the F1 and F1H comprehensive genomic profiling assays. TMB is calculated by counting all synonymous and non-synonymous somatic variants across 315 or 405 genes. Germline alterations and known and likely driver alterations are excluded to avoid sample bias, as both F1 and F1H specifically target genes with cancer associations. The resulting mutation count is normalized by expressing the number as a mutation density with units of mutations per megabase (mut/Mb) of coding target territory. Analytic validation of TMB focused on accuracy, precision and sensitivity, while initial clinical feasibility was assessed in a cohort of 65 metastatic melanoma patients receiving immunotherapy. To determine accuracy, we compared the TMB values generated from F1 against a CLIA validated whole-exome sequencing (WES) method on 29 patients with TMB values ranging from <1 mut/Mb up to 600 mut/Mb. Precision was defined as the reliability of the TMB metric when determined from 10 clinical samples replicated 4-6 times. Sensitivity was evaluated by determining the lower limit of sample tumor purity at which a TMB value could be reliably assessed through a dilution series of tumor/normal pairs ranging from 80% to 5% tumor. We also assessed the clinical feasibility of the F1 TMB result by examining its ability to predict clinical response to anti-PD1 or PD-L1 immunotherapy in a cohort of 65 metastatic melanoma patients. The patients were evaluated for best response per RECIST criteria, progression free survival (PFS) and overall survival (OS). Results: Foundation Medicine's TMB measure provides accurate and precise results across a range of tumor mutational burden values on samples with as little as 20% tumor purity. In a cohort of 65 metastatic melanoma patients, the median TMB value was 37.9 mut/Mb in the responder group and 6.6 mut/Mb in the non-responder group (p<0.0001, Mann-Whitney test). Additionally, TMB-high (≥20 mut/Mb) patients demonstrated superior PFS and OS compared to TMB non-high patients (median PFS and OS not reached for TMB-High through 66 months vs. medians of 3 months PFS and 12 months OS for TMB non-high, p-value <0.001). Conclusions: We have developed and validated a TMB result as part of the FoundationOne and FoundationOne Heme platforms. Initial clinical feasibility results demonstrate that the FoundationOne TMB value can be used to predict the likely response of metastatic melanoma patients to anti-PD1/PD-L1 checkpoint inhibitors, while feasibility in NSCLC and bladder cancer have been presented elsewhere. Citation Format: Daniel S. Lieber, Mark R. Kennedy, Douglas B. Johnson, Joel R. Greenbowe, Garrett M. Frampton, Alexa B. Schrock, Jeffrey S. Ross, Phillip J. Stephens, Siraj M. Ali, Vincent A. Miller, David A. Fabrizio. Validation and clinical feasibility of a Foundation Medicine assay to identify immunotherapy response potential through tumor mutational burden (TMB). [abstract]. In: Proceedings of the AACR Special Conference on Tumor Immunology and Immunotherapy; 2016 Oct 20-23; Boston, MA. Philadelphia (PA): AACR; Cancer Immunol Res 2017;5(3 Suppl):Abstract nr B16.

OBJECTIVE:Small cell carcinoma of the ovary, hypercalcemic-type (SCCOHT) is a rare, extremely aggressive neoplasm that usually occurs in young women and is characterized by deleterious germline or somatic SMARCA4 mutations. We performed comprehensive genomic profiling (CGP) to potentially identify additional clinically and pathophysiologically relevant genomic alterations in SCCOHT.METHODS:CGP assessment of all classes of coding alterations in up to 406 genes commonly altered in cancer and intronic regions for up to 31 genes commonly rearranged in cancer was performed on 18 SCCOHT cases (16 exhibiting classic morphology and 2 cases exhibiting exclusive a large cell variant morphology). In addition, a retrospective database search for clinically advanced ovarian tumors with genomic profiles similar to SCCOHT yielded 3 additional cases originally diagnosed as non-SCCOHT.RESULTS:CGP revealed inactivating SMARCA4 alterations and low tumor mutational burden (TMB) (<6mutations/Mb) in 94% (15/16) of SCCOHT with classic morphology. In contrast, both (2/2) cases exhibiting only large cell variant morphology were hypermutated (TMB scores of 90 and 360mut/Mb) and were wildtype for SMARCA4. In our retrospective search, an index ovarian cancer patient harboring inactivating SMARCA4 alterations, initially diagnosed as endometrioid carcinoma, was re-classified as SCCOHT and responded to an SCCOHT chemotherapy regimen.CONCLUSION:The vast majority of SCCOHT demonstrate genomic SMARCA4 loss with only rare co-occurring alterations. Our data support a role for CGP in the diagnosis and management of SCCOHT and of other lesions with overlapping histological and clinical features, since identifying the former by genomic profile suggests benefit from an appropriate regimen and treatment decisions, as illustrated by an index patient.

3587 Background: Biomarkers capable of predicting response to checkpoint inhibitor therapies represent a significant clinical need. Increased tumor neo-antigenic burden has been linked to PD1/PD-L1 therapeutic response in several conditions including metastatic melanoma, non-small cell lung carcinoma and microsatellite instable (MSI-H) colorectal cancer (CRC). However, the challenges and high cost associated with neo-antigen discovery has shifted focus towards more efficient methods of response stratification. As such, mutational burden determination from whole exome sequencing and comprehensive genomic profiling (CGP) has emerged as a potential solution. In CRC, MSI-H may serve as a predictive biomarker for activity of PD1/PD-L1 therapy, reflecting an unquantified immunogenic mutational burden. Herein, we explore the ability to quantify tumor mutational burden (TMB) as a potential predictive biomarker of PD1/PD-L1 therapy in CRC. Methods: Formalin-fixed, paraffin embedded tissue sections from 2013 cases of CRC were sequenced with CGP (FoundationOne assay). MSI status and PD-L1 gene amplification were determined as previously described. TMB was calculated by counting mutations across a 1.25Mb region spanning 315 genes. Patients were classified as TMB high or low using the top quartile threshold and microsatellite instable (MSI-H) or stable (MSS) using a computational algorithm developed by Foundation Medicine. Results: MSS tumors were observed in 1934 of 2013 (95.2%) cases and MSI-H tumors were observed in 79 of 2013 (3.9%) cases. 79 of 79 (100%) MSI-H cases were TMB high (range 16.8-72.7 mut./Mb) and 1510 of 1934 (78.1%) MSS cases were TMB low (range 0.0-8.0 mut./Mb). Consequently, 424 of 1934 (21.9%) MSS cases were confirmed as TMB high (range 8.8-43.1 mut./Mb). Of note, less than 0.1% of cases featured PD-L1 gene amplifications. Conclusions: Tumor mutational burden as estimated by CGP, as opposed to MSI assessment alone, significantly increases the number of patients with metastatic CRC who may benefit from checkpoint inhibitor based therapeutic approaches. Investigations to validate this biomarker are ongoing.