Search

PURPOSE Aumolertinib (formerly almonertinib; HS-10296) is a novel third-generation epidermal growth factor receptor tyrosine kinase inhibitor approved in China. This double-blind phase III trial evaluated the efficacy and safety of aumolertinib compared with gefitinib as a first-line treatment for locally advanced or metastatic EGFR-mutated non–small-cell lung cancer (NSCLC; ClinicalTrials.gov identifier: NCT03849768 ). METHODS Patients at 53 sites in China were randomly assigned 1:1 to receive either aumolertinib (110 mg) or gefitinib (250 mg) once daily. The primary end point was progression-free survival (PFS) per investigator assessment. RESULTS A total of 429 patients who were naïve to treatment for locally advanced or metastatic NSCLC were enrolled. PFS was significantly longer with aumolertinib compared with gefitinib (hazard ratio, 0.46; 95% CI, 0.36 to 0.60; P < .0001). The median PFS with aumolertinib was 19.3 months (95% CI, 17.8 to 20.8) versus 9.9 months with gefitinib (95% CI, 8.3 to 12.6). Objective response rate and disease control rate were similar in the aumolertinib and gefitinib groups (objective response rate, 73.8% and 72.1%, respectively; disease control rate, 93.0% and 96.7%, respectively). The median duration of response was 18.1 months (95% CI, 15.2 to not applicable) with aumolertinib versus 8.3 months (95% CI, 6.9 to 11.1) with gefitinib. Adverse events of grade ≥ 3 severity (any cause) were observed in 36.4% and 35.8% of patients in the aumolertinib and gefitinib groups, respectively. Rash and diarrhea (any grade) were observed in 23.4% and 16.4% of patients who received aumolertinib compared with 41.4% and 35.8% of those who received gefitinib, respectively. CONCLUSION Aumolertinib is a well-tolerated third-generation epidermal growth factor receptor tyrosine kinase inhibitor that could serve as a treatment option for EGFR-mutant NSCLC in the first-line setting.

PURPOSEAlthough NRG1 fusions are oncogenic drivers across multiple tumor types including lung cancers, these are difficult to study because of their rarity. The global eNRGy1 registry was thus established to characterize NRG1 fusion–positive lung cancers in the largest and most diverse series to date.METHODSFrom June 2018 to February 2020, a consortium of 22 centers from nine countries in Europe, Asia, and the United States contributed data from patients with pathologically confirmed NRG1 fusion–positive lung cancers. Profiling included DNA-based and/or RNA-based next-generation sequencing and fluorescence in situ hybridization. Anonymized clinical, pathologic, molecular, and response (RECIST v1.1) data were centrally curated and analyzed.RESULTSAlthough the typified never smoking (57%), mucinous adenocarcinoma (57%), and nonmetastatic (71%) phenotype predominated in 110 patients with NRG1 fusion–positive lung cancer, further diversity, including in smoking history (43%) and histology (43% nonmucinous and 6% nonadenocarcinoma), was elucidated. RNA-based testing identified most fusions (74%). Molecularly, six (of 18) novel 5′ partners, 20 unique epidermal growth factor domain–inclusive chimeric events, and heterogeneous 5′/3′ breakpoints were found. Platinum-doublet and taxane-based (post–platinum-doublet) chemotherapy achieved low objective response rates (ORRs 13% and 14%, respectively) and modest progression-free survival medians (PFS 5.8 and 4.0 months, respectively). Consistent with a low programmed death ligand-1 expressing (28%) and low tumor mutational burden (median: 0.9 mutations/megabase) immunophenotype, the activity of chemoimmunotherapy and single-agent immunotherapy was poor (ORR 0%/PFS 3.3 months and ORR 20%/PFS 3.6 months, respectively). Afatinib achieved an ORR of 25%, not contingent on fusion type, and a 2.8-month median PFS.CONCLUSIONNRG1 fusion–positive lung cancers were molecularly, pathologically, and clinically more heterogeneous than previously recognized. The activity of cytotoxic, immune, and targeted therapies was disappointing. Further research examining NRG1-rearranged tumor biology is needed to develop new therapeutic strategies.

PURPOSE Colorectal carcinomas (CRCs) with microsatellite-instability (MSI) are enriched for oncogenic kinase fusions (KFs), including NTRK1, RET, and BRAF, but the mechanism underlying this finding is unclear. METHODS The genomic profiles of 32,218 advanced CRC tumor specimens were analyzed to assess the fusion breakpoints of oncogenic alterations including KFs in microsatellite-stable and microsatellite-unstable CRC. Genomic contexts of such alterations were analyzed to obtain mechanistic insights. RESULTS Genomic analysis demonstrated that oncogenic fusion breakpoints in MSI tumors do not preferentially involve repetitive or low-complexity sequences. Instead, their junction regions showed pronounced guanine and cytosine bias and elevated mutation frequency at G:C contexts. Elevated mutation frequency at G:C bases in relevant introns predicted prevalence of associated oncogenic fusions in MSI CRCs. CRCs harboring mismatch repair signatures had enrichment of butyrate-producing microbial species, reported to be associated with induction of 8-oxoguanine lesions in the intestine. CONCLUSION Detailed analysis of breakpoints in MSI-associated KFs support a model in which inefficient repair and/or processing of microbiome-induced clustered 8-oxoguanine damage in MSI CRC contributes to the increased incidence of specific oncogenic fusions.

PURPOSE To describe the clinical and molecular features of metastatic colorectal cancers (mCRCs) bearing uncommon atypical RAS (At- RAS) mutations at codons other than 12, 13, 59, 61, 117, and 146. MATERIALS AND METHODS By exploiting five next-generation sequencing sources (Italian collaboration, Memorial Sloan Kettering Cancer Center, Samsung Medical Center, the Biomarker Research for Anti-EGFR Monoclonal Antibodies by Comprehensive Cancer Genomics (BREAC) study, and the Foundation Medicine database), we retrieved 175 At- RAS mutated cases. Molecular data were obtained from 163 samples from Memorial Sloan Kettering Cancer Center and the Foundation Medicine database. Clinical data were available for 27 At- RAS–positive and 467 negative cases from the Italian collaboration, Memorial Sloan Kettering Cancer Center, Samsung Medical Center, and the BREAC study. RESULTS At- RAS mutations were identified in 163 (0.9%) of 18,270 mCRCs. Among 133 with evaluable microsatellite instability status, 11 (8%) were microsatellite instability high. POLE exonuclease domain mutations had higher frequency (7%) than expected and were found only in microsatellite-stable tumors with high tumor mutational burden (TMB). Overall, 17% (28 of 163) of At- RAS cases had TMB greater than 20 mutations/Mb. Co-occurring typical RAS/BRAF V600E mutations and NF1 mutations, presumed to cause RAS activation, were found in 30% and 12% of samples, respectively (up to 43% and 50%, respectively, in TMB-high samples). Patients with RAS/BRAF wild-type mCRC achieved a median overall survival (OS) of 42.1 months, whereas those harboring isolated At- RAS, typical RAS, or BRAF V600E mutations showed a median OS of 32.3, 30.0, and 17.9 months, respectively ( P < .001). No significant OS difference ( P = .240) was found between patients with At- RAS versus typical RAS-mutated mCRC. Only one of six patients evaluable for primary resistance to anti–epidermal growth factor receptors achieved tumor response. CONCLUSION At- RAS mutations may be a marker for RAS pathway activation and can be associated with high co-occurrence of POLE exonuclease domain mutations.

Purpose In advanced stages, large-cell neuroendocrine carcinoma of the lung (L-LCNEC) mimics small-cell lung cancer despite its traditional classification as a non–small-cell lung cancer. Here we present a focused analysis of BRAF mutations in this population. Patients and Methods Comprehensive genomic profiling of tumor tissues was performed from a cohort of 300 patients with biopsy-proven L-LCNEC. Specimens were either from a primary lung lesion or metastatic site. Results In 13 patients, 14 unique BRAF alterations (amplifications, mutations) were identified. The importance of biomarker-driven therapy is subsequently highlighted with our case of a 69-year-old man diagnosed with metastatic L-LCNEC who did not respond to cisplatin plus etoposide. A significant durable response was then demonstrated with therapy targeted toward a BRAF non-V600E activating mutation (G469R) associated with biomarker response identified through circulating cell-free tumor DNA analysis. A change in clonal allele frequency from nearly 40% to nondetectable was observed. Conclusion Although uncommon, L-LCNEC does seem to contain activating and therefore actionable alterations. We thus highlight the value of pursuing next-generation sequencing for patients with this disease.

Purpose The promise of precision oncology is that identification of genomic alterations will direct the rational use of molecularly targeted therapy. This approach is particularly applicable to neoplasms that are resistant to standard cytotoxic chemotherapy, like T-cell leukemias and lymphomas. In this study, we tested the feasibility of targeted next-generation sequencing in profiles of diverse T-cell neoplasms and focused on the therapeutic utility of targeting activated JAK1 and JAK3 in an index case. Patients and Methods Using Foundation One and Foundation One Heme assays, we performed genomic profiling on 91 consecutive T-cell neoplasms for alterations in 405 genes. The samples were sequenced to high uniform coverage with an Illumina HiSeq and averaged a coverage depth of greater than 500× for DNA and more than 8M total pairs for RNA. An index case of T-cell prolymphocytic leukemia (T-PLL), which was analyzed by targeted next-generation sequencing, is presented. T-PLL cells were analyzed by RNA-seq, in vitro drug testing, mass cytometry, and phospho-flow. Results One third of the samples had genomic aberrations in the JAK-STAT pathway, most often composed of JAK1 and JAK3 gain-of-function mutations. We present an index case of a patient with T-PLL with a clonal JAK1 V658F mutation that responded to ruxolitinib therapy. After relapse developed, an expanded clone that harbored mutant JAK3 M511I and downregulation of the phosphatase, CD45, was identified. We demonstrate that the JAK missense mutations were activating, caused pathway hyperactivation, and conferred cytokine hypersensitivity. Conclusion These results underscore the utility of profiling occurrences of resistance to standard regimens and support JAK enzymes as rational therapeutic targets for T-cell leukemias and lymphomas.

Purpose Appendiceal neoplasms are heterogeneous and are often treated with chemotherapy similarly to colorectal cancer (CRC). Genomic profiling was performed on 703 appendiceal cancer specimens to compare the mutation profiles of appendiceal subtypes to CRC and other cancers, with the ultimate aim to identify potential biomarkers and novel therapeutic targets. Methods Tumor specimens were submitted to a Clinical Laboratory Improvement Amendments–certified laboratory (Foundation Medicine, Cambridge, MA) for hybrid-capture–based sequencing of 3,769 exons from 315 cancer-related genes and 47 introns of 28 genes commonly rearranged in cancer. Interactions between genotype, histologic subtype, treatment, and overall survival (OS) were analyzed in a clinically annotated subset of 76 cases. Results There were five major histopathologic subtypes: mucinous adenocarcinomas (46%), adenocarcinomas (30%), goblet cell carcinoids (12%), pseudomyxoma peritonei (7.7%), and signet ring cell carcinomas (5.2%). KRAS (35% to 81%) and GNAS (8% to 72%) were the most frequent alterations in epithelial cancers; APC and TP53 mutations were significantly less frequent in appendiceal cancers relative to CRC. Low-grade and high-grade tumors were enriched for GNAS and TP53 mutations, respectively (both χ2 P < .001). GNAS and TP53 were mutually exclusive (Bonferroni corrected P < .001). Tumor grade and TP53 mutation status independently predicted OS. The mutation status of GNAS and TP53 strongly predicted OS (median, 37.1 months for TP53 mutant v 75.8 GNAS- TP53 wild type v 115.5 GNAS mutant; log-rank P = .0031) and performed as well as grade in risk stratifying patients. Conclusion Epithelial appendiceal cancers and goblet cell carcinoids show differences in KRAS and GNAS mutation frequencies and have mutation profiles distinct from CRC. This study highlights the benefit of performing molecular profiling on rare tumors to identify prognostic and predictive biomarkers and new therapeutic targets.

Purpose Inherited germline defects are implicated in up to 10% of human tumors, with particularly well-known roles in breast and ovarian cancers that harbor BRCA1/2-mutated genes. There is also increasing evidence for the role of germline alterations in other malignancies such as colon and pancreatic cancers. Mutations in familial cancer genes can be detected by high-throughput sequencing, when applied to formalin-fixed paraffin-embedded tumor specimens. However, because of the frequent lack of patient-matched control normal DNA and/or low tumor purity, there is limited ability to determine the genomic status of these alterations (germline v somatic) and to assess the presence of loss of heterozygosity (LOH). These analyses, especially when applied to genes such as BRCA1/2, can have significant clinical implications for patient care. Materials and Methods LOHGIC (LOH-germline inference calculator) is a statistical model selection method to determine somatic versus germline status and predict LOH for mutations identified via clinical grade, high-depth, hybrid capture, tumor-only sequencing. LOHGIC incorporates statistical uncertainties inherent to high-throughput sequencing as well as specimen biases in tumor purity estimates, which we used to assess BRCA1/2 mutations in 1,636 specimens sequenced at Rutgers Cancer Institute of New Jersey. Results Evaluation of LOHGIC with available germline sequencing from BRCA1/2 testing demonstrates 93% accuracy, 100% precision, and 96% recall. This analysis highlights a differential tumor spectrum associated with BRCA1/2 mutations. Conclusion LOHGIC can assess LOH status for both germline and somatic mutations. It also can be applied to any gene with candidate, inherited mutations. This approach demonstrates the clinical utility of targeted sequencing in both identifying patients with potential germline alterations in tumor suppressor genes as well as estimating LOH occurrence in cancer cells, which may confer therapeutic relevance.

Purpose Molecular diagnostic testing has become an integral part of the evaluation of patients with metastatic colorectal cancer (CRC). Expanded mutational testing, such as next-generation sequencing (NGS), often identifies mutations with unclear clinical or prognostic implications. One such example is BRAF mutations that occur outside of codon 600 (non-V600 BRAF mutations). Methods We conducted this multicenter, retrospective cohort study to characterize the clinical, pathologic, and survival implications of non-V600 BRAF mutations in metastatic CRC. We pooled patients in whom non-V600 BRAF mutations were identified from NGS databases at three large molecular genetics reference laboratories. Results A total of 9,643 patients with metastatic CRC underwent NGS testing. We identified 208 patients with non-V600 BRAF mutations, which occurred in 2.2% of all patients tested and accounted for 22% of all BRAF mutations identified. Cancers with non-V600 BRAF mutations, compared with cancers with V600E BRAF (V600E BRAF) mutations, were found in patients who were significantly younger (58 v 68 years, respectively), fewer female patients (46% v 65%, respectively), and patients who had fewer high-grade tumors (13% v 64%, respectively) or right-sided primary tumors (36% v 81%, respectively). Median overall survival was significantly longer in patients with non-V600 BRAF-mutant metastatic CRC compared with those with both V600E BRAF-mutant and wild-type BRAF metastatic CRC (60.7 v 11.4 v 43.0 months, respectively; P < .001). In multivariable analysis, non-V600 BRAF mutation was independently associated with improved overall survival (hazard ratio, 0.18; P < .001). Conclusion Non-V600 BRAF mutations occur in approximately 2.2% of patients with metastatic CRC and define a clinically distinct subtype of CRC with an excellent prognosis.

e23530 Background: Malignant myoepithelioma of soft tissue (MM) is a rare tumor affecting patients (pts) of any age without sex predilection. MM has varied histopathologic findings, with epithelioid and spindle components and differentiation along osseous, chondroid or other lineages. Morphologic features suggest MM, however histologic overlap with other lesions exists. Diagnosis requires ancillary immunohistochemical &/or molecular studies; EWSR1 rearrangement with various fusion partners has been reported. Clinical behavior is varied and poorly understood due to disease rarity. Here, we undertook comprehensive genomic profiling (CGP) of MM to evaluate for potential targetable molecular alterations and report clinical outcomes. Methods: We identified pts with MM treated at University of Michigan (UM) from 2000-2017, obtained clinical data and reviewed pathology on 10 available samples and performed integrative sequencing through Michigan Oncology Sequencing Program (MI-ONCOSEQ) on 6. For 31 additional FFPE samples, hybrid capture-based DNA CGP (n = 16) or DNA+RNA CGP (n = 15) was performed (Foundation Medicine, FM). Results: 13 pts were treated at UM; median age at diagnosis was 52 (22-88) with male predominance (9/13). Primary tumor size was 3 cm (2-9.5) and most common location was trunk (4/13) and extremity (3/13). Four pts with metastatic disease (mets) underwent systemic chemotherapy with 0 responses. Median progression free and overall survival in mets was 20 (5-75) and 57 (12-91) months, respectively. CGP analyses of all samples (Table) revealed EWSR1 rearrangement in 8.1% (3/37) and presence of PHF1-TFE3 and CIC:DUX4 rearrangements in 16% (6/37). Independent blinded histopathology review of whole slides and tissue is underway. Conclusions: MM is a rare malignancy with variable clinical course. In our series, pts with mets derived little benefit from systemic therapy. CGP revealed CDKN2A/B alteration as the most prevalent with few harboring EWSR1 rearrangement. PHF1-TFE3 was recently described in ossifying fibromyxoid tumor (OFMT) that may represent a subset of OFMT not yet expressing bone markers. The presence of hypermutation and CIC:DUX4 fusions also suggests potential alternate diagnosis. These results indicate that CGP could complement histopathologic evaluation to aid in diagnoses and treatment of this entity. [Table: see text]

179 Background: The cyclin pathway is comprised of possible targetable alterations as well as resistant alterations that affect treatment, including hormonal agents. We describe the landscape of cyclin genomic alterations in prostate cancer. Methods: Consecutive samples were analyzed in a CLIA-certified laboratory using comprehensive genomic profiling (CGP) performed by next-generation sequencing (315 genes, >500X coverage). We describe alterations in activating genes (Table) and co-alterations in resistant genes ( RB1 and CCNE1) (related to cyclin inhibition) and androgen receptor ( AR). Results: Alterations in any cyclin pathway genes were found in 9.7% of the 5,356 tumors analyzed. Frequent alterations included CCND1 amplification (4.2%) and CDKN2A and B loss (2.4% each). Frequencies were substantially different according to prostate cancer histology (Table). The majority of alterations were copy number changes. Alterations in possible resistance genes, RB1 and CCNE1 were detected in 9.7% (up to 54.6% in neuroendocrine) and 1.2% of cases, respectively, while AR alterations were seen in 20.9% (up to 27.3% in anaplastic). Co-occurrence analysis demonstrated a lower likelihood of concomitant vs. isolated alterations in cyclin activating and resistance pathway (odds ratio (OR), 0.44, p

549 Background: Cyclin pathway genomic alterations can be a possible therapeutic target as well as a resistance mechanism for therapy. We describe the landscape of cyclin alterations (alt) in non-prostate genitourinary (GU) cancers. Methods: Consecutive samples were analyzed in a CLIA-certified laboratory using comprehensive genomic profiling (CGP) performed by next-generation tissuesequencing (315 genes, >500X coverage). We describe alterations in activating genes (Table) and co-alterations in resistant genes ( RB1 and CCNE1) (related to cyclin inhibition). Results: Alterations in any cyclin pathway genes were found in 37.9% of bladder/urothelial, 33.8% of testicular, 25.2% of penile and 24.6% of kidney cancers. Most alterations were copy number changes and frequencies varied substantially for each tumor type (Table). The high number of cases permitted identification of interesting patterns of outliers for each histology (examples: testis rhabdomyosarcoma and leydig tumor, 50% and 54.5% CDK4 alt; ureter urothelial, 17% CCND1 amp; bladder squamous, 41% CDKN2A del; kidney malignant rhabdoid, 90% SMARCB1 alt; urethra urothelial, 14.7% CCNE1 alt). Alterations in possible resistance genes RB1 and CCNE1 were more frequent in bladder cancers (especially with a neuroendocrine component). Co-occurrence analysis demonstrated a lower likelihood of concomitant vs. isolated alt in cyclin activating and resistance genes (odds ratio for bladder 0.17, p

310 Background: Trastuzumab (T) plus chemotherapy is standard of care for pts with HER2+ adv GE cancer. Selection for T relies on HER2+ result by immunohistochemistry (IHC) +/- in-situ hybridization (ISH); ERBB2 amplification by comprehensive genomic profiling (CGP) also predicts benefit. As CGP use increases, it is important to explore associations of IHC/ISH and CGP result agreement with clinical outcomes in pts with adv GE cancer. Methods: Pts with adv GE cancer were selected from the Flatiron Health-Foundation Medicine (FMI) clinico-genomic database (CGDB), a nationwide de-identified EHR-derived clinical database linked to FMI genomic data. Pts treated from 01/2011-12/2018 with CGP data for tissue specimens collected before 1L were included. Agreement between HER2 status by IHC +/- ISH v ERBB2 amplification by CGP was assessed. For 1L containing T, time to tx discontinuation (TTD) and overall survival (OS) from 1L start stratified by HER2:ERBB2 agreement and ERBB2 copy number (CN) [ERBB2+ = CN >4] were estimated with unadjusted Kaplan-Meier analysis and adjusted (practice type, sex, age at adv dx) hazard ratios (aHR) from Cox proportional hazards models. Results: Of 596 HER2-tested pts with adv GE cancer in CGDB, 174 (29%) were HER2+ by IHC/ISH. Median age at adv dx was 63 y; 76.5% were male and 97% had adenocarcinoma. Overall HER2:ERBB2 agreement was 91%. Of 123 HER2+ 1L T-treated pts, median TTD and OS were longer for HER2+:ERBB2+ (concordant; n = 93) v HER2+:ERBB2- pts (discordant; n = 30): TTD, 5.2 v 0.8 months, aHR 0.46 (0.29-0.70); OS 14.8 v 7.5 months, aHR 0.38 (0.21-0.68). Median OS was 22.0 vs 8.4 months for 1L T-treated pts with ERBB2 CN > 30 (median) v CN ≤30 (aHR 0.69 [0.37-1.27]). Conclusions: In this large real-world clinico-genomic dataset, HER2:ERBB2 agreement was high in tested pts with adv GE cancer. 1L T-treated pts with discordant tests (HER2+:ERBB2-) had significantly shorter TTD and OS. Pts with higher ERBB2 CN had longer OS, but this finding was not significant after adjusting for covariates. Further research is needed to explore associations between HER2:ERBB2 agreement and clinical outcomes in this population.

578 Background: Genomic alterations (GA) characteristic of IHCC are well known. We queried whether the GA of IHCC from primary tumor biopsies (p-bx) would differ from IHCC metastasis biopsies (m-bx). Methods: CGP was performed on 1,268 cases of advanced stage IHCC using p-bx in 1,048 cases and m-bx from 220 cases. Tumor mutational burden (TMB) was determined on 0.8-1.1 Mbp of sequenced DNA and microsatellite instability (MSI) was determined on 114 loci. PD-L1 expression in tumor cells (Dako 22C3) was measured by IHC. Results: M-bx sites included: lymph nodes (63), soft tissues (47), peritoneum (34), lung/pleura (27), omentum (15), bone (10), GYN tract (5), brain (2), Upper GI (2), colon (2), bladder (1), abdomen (1) and adrenal (1). The GA per sample were similar as were biomarkers of immuno-oncology (IO) drug response. The KRAS mutation frequency was doubled in the m-bx compared to the p-bx (p < 0.001), and enrichment of the potentially targetable KRAS G12C was also observed. Frequencies of untargetable GA were similar overall. IDH1 (p < 0.001) and FGFR2 GA known to be enriched in IHCC were less frequent in the m-bx cohort. GA in STK11 were more frequently identified in m-bx. Conclusions: GA found in p-bx vs m-bx in IHCC are significantly different; the m-bx cohort featuring greater KRAS and lower IDH1 and FGFR2 GA. This suggests that the m-bx group may contain a significant number of non-IHCC cases whose metastatic lesions were actually derived from other primary sites incorrectly assigned the diagnosis of IHCC. [Table: see text]

757 Background: Pancreas ductal adenocarcinomas (PDA) has a 5-year survival rate of 6% with a need for new therapeutic options. The approval of pembrolizumab for some gastrointestinal cancers shows the potential of immunotherapy (IMT) in PDA. We evaluated the IMT-associated biomarkers of PD-L1 expression, tumor mutational burden (TMB), microsatellite instability (MSI) and PD-L1 amplification in PDAs. Methods: 746 formalin-fixed paraffin embedded samples were evaluated for PD-L1 IHC using the Dako 22C3 pharmDx assay and scored using tumor proportion score (TPS). The cases had comprehensive genomic profiling (CGP) via DNA sequencing, using a hybrid-capture next-generation sequencing assay (FoundationOne and FoundationOneCDx) for genomic alterations (GAs), TMB, and MSI. Results: PD-L1 was positive (TPS ≥ 1%) in 29% (214/746) and negative in 71% (532/746). 43/214 (20%) of positive cases were high positive (TPS ≥ 50%). TMB (590 cases) had a mean of 3.20, 3.46, and 3.61 mutations/Mb for PD-L1 negative, positive, and high positive groups. 3 hypermutated (TMB ≥ 20) were negative for PD-L1 expression. 3/581 cases were MSI-high with a high TMB score (average 23.53 mutations/Mb). 2 MSI-high cases were negative for PD-L1 and 1 was high positive. PD-L1 amplification was not detected (0/746). Only BCOR was significantly different between PD-L1 high positive and PD-L1 negative tumors (Table). Conclusions: Of 729 PDA cases, 29% were positive (TPS ≥ 1%) for PD-L1 expression while only 6% of all cases showed a high level of PD-L1 expression on tumor cells. TMB high (3/729) and MSI-High (3/729) cases were rare. Only 2 of the TMB high cases were also MSI-high. PD-L1 amplification was not detected. Comparing GAs in PD-L1 high positive vs negative cases was only significantly different for BCOR. Further investigation is needed to see if a combined positive score of PD-L1 expression may identify a subset of patients with PDA who are more likely to respond to IMT. [Table: see text]

3099 Background: FGFR2 genomic alterations (GA) have been described in a variety of solid tumors and emerged as biomarkers for investigational agents undergoing clinical trials. Methods: 201,766 primarily relapsed/refractory malignancies were evaluated with a hybrid-capture based sequencing assay Tumor mutational burden (TMB) was determined on 0.8-1.1 Mbp of sequenced DNA and reported as mut/Mb. Microsatellite instability (MSI) was determined on 114 loci. PD-L1 expression was determined by IHC (Dako 22C3 antibody). Results: FGFR2 GA were detected in 2,993 (1.5%) cases featuring short variant (SV) mut (42%), copy number changes (27%), rearrangements/fusions (28%) and multiple GA (3%). The most frequent SV GA were S252W, N549K, C382R, P253R, Y375C, K659E and R664W. A small cohort (2%) of tumors featured the V564I and V564L GA that are associated with resistance to TKI drugs. The FGFR2-altered cases were 69% female/31% male with median age of 61 yrs. Most frequent GA in FGFR2 altered cancers: TP53 (47%), PIK3CA (22%), PTEN (20%), ARID1A (18%), CDKN2A/2B (18/14%) and MYC (12%). FGFR2 SVs most common in endometrial, breast carcinomas (ca) and CUP. FGFR2 amplification most common in breast, gastroesophageal and lung ca. FGFR rearrangement/fusions most common in cholangioca (37%), CUP (15%), pancreatobiliary (12%) and breast ca (6%). The FGFR2- BICC1 was the most frequent fusion followed by fusions with TACC2, AHCYL1, CCDC6, VCL, and KIAA1217. MSI-High status present in 6.8% of evaluable FGFR2 altered cases (63% in endometrial ca). Median TMB was 3.5 mut/Mb with 21.8% featuring ≥ 10 mut/Mb and 12.0% featuring ≥ 20 mut/Mb. Only 63% of MSI-High FGFR2 mut tumors had TMB ≥ 20 mut/Mb. 12.7% FGFR2 mut+ had > 1% PD-L1 staining with 3.4% > 50% staining. 29% of PD-L1 IHC+ cases in NSCLC. FGFR mut ca’s responding to anti-FGFR2 therapies will be presented. Conclusions: FGFR2 GA are most frequent in cholangioca, breast, GI tract, lung ca and CUP, with enrichment of FGFR2 fusions in biliary tract ca. Cases with FGFR2 GA typically do not feature other kinase driver GA and are associated with mut in the MTOR/PIK3CA/AKT pathways. Finally, in contrast with RTK driver GA in EGFR (5.7%) and ERBB2 (7.9%), at 12.0%, across all tumor types, FGFR2 mut cancers may have higher frequency of TMB ≥ 20 mut/Mb suggesting potential immunotherapy responsiveness.

4581 Background: To understand the genomic landscape and inform the therapeutic development of UC, 2463 cases were analyzed by CGP for genomic alterations (GAs) and for genome wide signatures. Methods: 479 UTUC and 1984 BUC FFPE tissues (77%/70% primary [PT] and 23%/30% metastatic tumors [MT] from unmatched patients [pts]) underwent hybrid-capture based CGP to evaluate all classes of GA. Tumor mutational burden (TMB) was determined on 1.1 Mbp of sequenced DNA and microsatellite instability (MSI) was determined on 114 loci. Targetable GA and signatures were assessed according to the ESMO Scale for Clinical Actionability of molecular Targets (ESCAT). Results: 48% of UC pts overall harbored ≥1 tier 1-2 GA suggesting benefit from approved or investigational targeted therapies (TT). Additionally, 17% had a tier 3 GA that provides a strong rationale for clinical trial consideration. Non- FGFR3 kinase fusions were observed in 1% of pts (0.6% UTUC v 1.1% BUC), including BRAF/RAF1 fusions in 0.5%. BRAF mut/fusions were observed in 2% (49/2463) of cases and were mutually exclusive with FGFR3 GA (p = 0.002). In comparing UC from anatomic sites, there were no differences of TMB-H (≥20 mut/mb)/MSI-H for PT and MT but UTUC was enriched for MSI-H (3.4%) relative to BUC (0.77%, p < 0.001, all TMB-H). Excluding MSI-H pts, UTUC has lower median TMB (4.35 mut/mb) than BUC (6.96 mut/mb). FGFR3 GA (26% v 19%, p < 0.05) and specifically short variants (SV) (20% v 13%) were enriched in UTUC vs BUC. HRAS SV were also enriched in UTUC vs BC (7.3% v 3.0%), attributed to an enrichment in renal pelvis UC (10.1%) v ureteral UC (1.8%, p < 0.05). RB1 GA were more frequent in BUC vs UTUC (21% v 7.8% p < 0.001). Conclusions: Against a background of 50% actionability in UC with opportunities for immunotherapy, TT, or combinations thereof, the UTUC cohort is enriched for FGFR3 and HRAS SV relative to BUC, with the observation of HRAS mutations predominantly in UC of the renal pelvis, that warrants further investigation into the distinct modes of oncogenesis for UC as stratified by anatomic origin. These results argue strongly for the routine incorporation of CGP prior to systemic therapy initiation in metastatic UC.

9081 Background: NRG1 fusions are potentially actionable driver events enriched in NSCLCs, particularly invasive mucinous adenocarcinomas (IMAs). These fusions activate HER3/HER2, supporting the therapeutic use of HER3 and/or HER2 inhibitors, but optimal treatment strategies remain unclear. Methods: A global, multicenter network of thoracic oncologists (6 countries, 13 institutions) identified patients with pathologically confirmed NRG1 fusion-positive NSCLCs. Anonymized clinical/pathologic features and clinical outcomes were collected retrospectively. Best response to systemic therapy was determined (RECIST v1.1). PFS was calculated (Kaplan-Meier). Results: 80 NRG1 fusion-positive NSCLCs were identified. RNA-based sequencing identified 66% (n = 53/80), DNA-based sequencing 18% (n = 14/80), and FISH 16% (n = 13/80) of cases. The most common upstream partners were CD74 (45%), SLC3A2 (31%), and SDC (9%). Most patients were female (64%) and never smokers (58%). Histology was adenocarcinoma in 95% (IMA, 91%), squamous 4%, large cell neuroendocrine 1%. At diagnosis, most patients had non-metastatic disease (stage: I 33%, II 27%, III 18%, IV 22%). The lifetime frequency of brain metastases was 15%. 12 patients received the HER2 inhibitor afatinib for stage IV disease. PD was the best response in 55% (n = 6/11) of evaluable patients with 18% PR (n = 2/11) and SD 18% (n = 2/11); median PFS was 3.5 months (range 0.6-16.5 months). 19 patients received platinum-based chemotherapy; most patients had SD as their best response (47%, n = 8); PD 41% (n = 7), PR 12% (n = 2). PD-L1 was negative in the majority of tumors (79%, n = 26/33) and none had high PD-L1 expression (range 0-20%). No responses to single-agent anti-PD-1/L1 therapy were observed (PD n = 5/6, SD n = 1/6: nivolumab/atezolizumab). No responses to chemoimmunotherapy (carboplatin, pemetrexed, pembrolizumab) were observed (SD n = 4/5, PD n = 1/5). Conclusions: RNA-based testing is an important component of NRG1 fusion detection. Novel targeted therapeutic approaches are needed as overall outcomes with afatinib are poor. NRG1 fusion-positive NSCLCs do not highly express PD-L1 and outcomes with immunotherapy ± chemotherapy are poor.

2630 Background: PD-L1 expression and TMB, as a proxy for neoantigen burden, have been correlated with response to IO in advanced NSCLC (aNSCLC) clinical trials, but their combined utility is unclear. We assessed TMB and PD-L1 as predictors of response in aNSCLC patients (pts) after IO monotherapy in a real-world setting. Methods: Pts had a diagnosis of aNSCLC, comprehensive genomic profiling of 186-315 genes/1.1 megabase (Mb), PD-L1 testing of pre-IO specimens, and were treated in the Flatiron Health network (1/2011 - 6/2018). Clinical characteristics and real-world tumor response (rwTR) were obtained via technology-enabled abstraction of clinician notes and radiology/pathology reports, and linked to genomic data in the Flatiron Health-Foundation Medicine Clinico-Genomic Database. A general additive model examined the predictive value of TMB (as continuous measure) and PD-L1 level on rwTR. A reduced PD-L1-only model was compared to the full model using Akaike Information Criterion (AIC). rwTR predictions at representative TMB and PD-L1 levels were calculated. Results: Of 426 pts, PD-L1 expression was high (≥50%) in 140, low (1-49%) in 123, and negative (

9063 Background: EGFR and HER2 ( ERBB2) exon 20 insertion (ex20ins) mutations represent a subset of driver alterations in NSCLC, which historically have largely not responded to available targeted therapies. Recently, inhibitors specifically targeting ex20ins have shown efficacy in the clinic. Previous studies have described the landscape of EGFR ex20ins in NSCLC (PMID: 29981927), but similar descriptions of HER2 ex20ins are lacking. Methods: Hybrid capture-based comprehensive genomic profiling (CGP) was performed on 39,644 tissue and 4,062 blood-based circulating tumor DNA (ctDNA) samples from 43,706 unique patients with advanced NSCLC. Tumor mutational burden (TMB) was determined on 0.8-1.1 Mbp of sequenced DNA for tissue samples and reported as mutations/Mb. Results: HER2 ex20ins were detected in 1.5% (648/43,706) of NSCLC cases (614 tissue and 34 ctDNA). HER2 ex20ins represented 35% (648/1,845) of HER2-altered NSCLCs overall, while 46% (843/1,845) of cases had HER2 amplification (≥5 copies), 17% (320/1,845) had a non-ex20ins HER2 short variant (SV; most commonly S310F in 84 cases and V659E in 29 cases), and 1.8% (34/1,845) had HER2 amplification + SV. There were 28 unique ex20ins including most commonly A775_G776insYVMA (69%, 450/648), G776 > VC (12%, 76/648) and P780_Y781insGSP (8.6%, 56/648). Cases with HER2 ex20ins were significantly enriched for adenocarcinoma histology (89% vs 66%), female gender (64% vs 51%) and low TMB (95% vs 65% TMB < 10 mut/Mb) compared to HER2 wild-type cases (all p < 0.0001). HER2 amplification (7 median copies, range 5-39) co-occurred in 16% (103/648) of HER2 ex20ins cases. Co-occurrence of other known NSCLC drivers in HER2 ex20ins cases was rare (0.8%, 5/648). In contrast, non- HER2ex20ins cases with HER2 amplification or other HER2 SVs each had co-occurring known driver alterations in ~30% of cases. There was no significant difference in histology, gender, age, HER2 co-amplification or TMB in cases with YVMA vs non-YVMA ex20ins. Conclusions: HER2 ex20ins are found in 1.5% of NSCLCs and are generally mutually exclusive of other known drivers. Detection of these alterations may be critical to identify matched targeted therapy options for this subset of patients.

5064 Background: PDC, PAC and PNC are histologic subtypes of prostate cancer (PC). We queried whether these subsets would share similar genomic alterations (GA) reflecting their disease biology and clinical features. Methods: CGP was performed using a hybrid capture-based assay on 61 PDC, 4,132 PAC and 217 PNC. Tumor mutational burden (TMB) was determined on 1.1 Mbp of sequenced DNA and microsatellite instability (MSI) was determined on 114 loci. Results: The age, GA per tumor and TP53 GA of PDC, PAC and PNC were similar (Table). RB1 GA were predominant in PNC. TMPRSS2:ERG fusions were most frequent in PNC, intermediate in PAC and lowest in PDC. AR GA were more often identified in PAC than PDC or PNC whereas PTEN GA were most frequent in PDC than PAC or PNC. Targetable GA were identified in all 3 groups when focused on BRCA2 (PARP inhibitors) and PIK3CA (MTOR inhibitors). ATM GA (PARP inhibitors) were more common in PAC than PDC or PNC. BRAF GA (BRAF/MEK inhibitors) were more frequent in PDC and PAC than PNC. CDK12 GA potentially associated with immunotherapy (IO) benefit were similar in PDC and PAC and low in PNC. Low frequencies of MSI-High and low median TMB levels were similar in all 3 groups. Conclusions: The pathologic features of PDC, PAC and PNC have been classically maintained as representative of 3 different tumor types with potentially contrasting histogenesis. In the current CGP based study, all 3 tumor types did not display significant differences in genomic signatures other than the high RB1 GA. CGP may reveal biomarkers that could direct patients to targeted (PARP, MTOR and BRAF/MEK inhibitors) or immunotherapies ( CDK12 GA, MSI-High or high TMB status) especially in PDC and PAC. [Table: see text]

4087 Background: The management of CCA has evolved as targeted and immune checkpoint inhibitor (ICPI) therapies have emerged. We used comprehensive genomic profiling (CGP) to characterize the genomic alterations (GA) that have potential to personalize therapy for CCA. Methods: 3634 CCA underwent hybrid capture based CGP on 0.8-1.1 Mb of the coding genome to identify GAs in exons and select introns in up to 404 genes, TMB, microsatellite status (MSI) and % monoallelic genome (gLOH). PD-L1 expression was determined by IHC (Dako 22C3). Results: 52% of CCA were female with a median age of 62 years (range 16 - > 89). The most common biopsy sites were liver (74%), lymph node (4%), bile duct (3.3%), and lung (2%). MSI-high was rare (1%), 118 and 47 cases had TMB > 10 and > 20 mut/mb respectively. Of the latter, 51% (24/47) were MSI-H. PD-L1 amplification (AMP) was present in 0.27%. Of 490 CCA tested, 43 (9%) were positive for PD-L1 expression. 11% of cases had gLOH > 16%, only 2 cases had both TMB > 20 and gLOH > 16%. GA were most common in TP53 (31%), CDKN2A (29%), KRAS (20%) and ARID1A (17%). Potentially targetable GAs included FGFR2 (11%, 85% fusions), BRAF (5%, 50% V600E), ERBB2 (5%, 72% AMP), MET (2%, 90% AMP), EGFR (0.52%) and rarer ( < 0.5%) FGFR3, RET, FGFR1, ALK, and ROS1 fusions. The FGFR2 fusions had 128 unique 3’ partner genes including BICC1 (26%), CCDC6 (3.2%), AHCYL1 (2.6%) and KIAA1217 (2.6%). FGFR2 fusions occurred in a mutually exclusive fashion from high gLOH (p < 0.002), but not high TMB. GA in IDH1 (15%) were mutually exclusive of FGFR2 fusions (p < 1e-13), but co-occurred with PBRM1 GA (23%, p < 1e-21), ARID1A (26% p < 1e-10). IDH1 GA had gLOH similar to the overall CCA population but were enriched for low TMB (p < 1e-3). Conclusions: Nearly 20% of CCA cases harbor targetable kinase GA, half of which were FGFR2 fusions. Independently, an additional 10% (gLOH) and 1% (high TMB, MSI and/or PD-L1 AMP) may benefit from PARP inhibitors and ICPI respectively. Independently, co-mutation of IDH1 and PBRM1/ARID1A defines a class of CCA that warrants further investigation for sensitivity to PARP inhibitors and may serve as a paradigm for other tumors (ie. gliomas) with a similar co-occurrence landscape.

4545 Background: Genomic alterations (GA) of EGFR, are well recognized as druggable oncogenic drivers in NSCLC, but the druggable GA EGFR L858R and exon 19 deletion (ex19del), are rarely observed in genitourinary cancer. We reviewed the genomic landscape of advanced upper tract and bladder UC (UTUC and BUC) to assess the frequencies of druggable EGFR GA. Methods: Tissue specimens from patients with UTUC (407) and BUC (2,402) were assayed using hybrid capture-based comprehensive genomic profiling (CGP). Tumor mutational burden (TMB) was determined on up to 1.1 Mbp of sequenced DNA and microsatellite instability (MSI) was determined on 95 or 114 loci. Results: EGFR alterations ( EGFR alt) were present in 17 UTUC and 93 BUC (4.2% and 3.9%). Age distribution between the two subgroups was similar, but UTUC was more prevalent in female patients (47% v 29%). BUC had a higher median TMB (5.2mut/mb v 7.8 mut/mb; p = 0.046) and the prevalence of MSI-H cases was not significantly different. TERT (55% v 71%) and TP53 (59% v 74%) were the most frequently mutated genes in EGFR alt UTUC and BUC. Within EGFRalt, amplifications were the most common alterations in both UTUC and BUC (13/17, 76%; 57/93, 61%). Amplifications were mutually exclusive from cases with EGFR short variants (SV) in BUC (0/34, 0%), and co-occurred with EGFR SV in four UTUC cases (4/8 50%). The majority of EGFR SV were EGFR exon 20 insertions ( EGFRexon20), which made up a larger proportion of EGFR alt in UTUC than BUC (7/17, 41% v 13/93, 14%; p = 0.01). Compared to other EGFR alt , EGFRexon20 trended towards mutual exclusivity of GA in commonly altered UC genes: TP53 (UTUC EGFRexon20 v EGFR alt other: 0% v 100%, p = 5.1E-5; BUC EGFRexon20 v EGFR alt other: 0% v 86%, p = 2.2E-7) , PIK3CA (14% v 10%; 0% v 19%) , RAF1 (0% v 10%; 0% v 16%) , or FGFR3 (0% v 10%; 0% v 6.3%) alterations. Only 2.2% (2/93) of BUC EGFRalt were L858R mutations and none were ex19del (0/93), while neither mutation was detected in UTUC. Conclusions: EGFRexon20 defines a subset of UC cases, and is the most common non-amplification GA seen in EGFR in UC, with some enrichment in UTUC. Consideration should be given to developing a trial for EGFR exon20 UC patients given the recent investigational work on inhibitors with activity against EGFRexon20, such as poziotinib and TAK-788.

9591 Background: EMPD is an intraepidermal adenocarcinoma which frequently progresses into invasive/metastatic disease. Methods: 29 EMPD underwent hybrid-capture based CGP to evaluate all classes of genomic alterations (GA). Tumor mutational burden (TMB) was determined on 1.1 Mbp of sequenced DNA and reported as mut/Mb. Microsatellite instability (MSI) was determined on 114 loci. PD-L1 expression was determined by IHC (Dako 22C3 antibody). Results: All patients had locally recurrent and/or metastatic disease. The 20 (69%) male and 9 (31%) female patients had a median age of 70 years (range 45 to 101 years). The primary EMPD site, known in 90% of cases and involved the scrotum (6 cases), groin (6 cases), vulva (5 cases), anus (3 cases), perineum (2 cases), abdomen (2 cases), penis (1 case) and buttock (1 case). was used for sequencing in 15 (52%) cases and a metastatic site 14 (48%) cases. The GA/tumor frequency was 5.0. The most frequent currently untargetable GA involved TP53 (48%), CDKN2A (38%), CDKN2B (31%), KMT2C (28%), and MYC (14%). The most frequent potentially targetable GA were identified in ERBB2 (35%), ERBB3 (17%) and PTEN (14%) and rarely in PDGFRA, CDK4/6, NF1/2, and ROS1 (fusion) all at 3%. The ERBB2 GA included short variant (SV) activating mutations (21%), amplifications (10%) and multiple ERBB2 GA in 3%. This finding is in contrast with mammary Paget’s disease which classically features > 90% of cases with ERBB2 amplification. No EMPD cases were either MSI-High or stained positively for PD-L1 expression. The median and mean TMB were 5.2 mut/Mb and 6.3 mut/Mb, respectively; 4 (14%) had TMB ≥ 10 mut/Mb and 1 (3%) had TMB ≥ 20 mut/Mb. Conclusions: EMPD is a rare source of relapsed/metastatic adenocarcinoma which features GA that are distinct from the more common disease originating on the nipple of the breast including a lower ERBB2 GA frequency and lower relative frequency of ERBB2 amplification versus SV mutations. A variety of MTOR pathway, cell cycle and kinase targets are also identified when EMPD undergoes CGP. However, the low TMB and absence of both MSI-High status and PD-L1 expression in EMPD cohort indicates a likely lack of benefit for immunotherapies for patients suffering from this rare form of malignancy.

3533 Background: Pre-clinical models implicate RAS amplifications ( RASa) as a mechanism of resistance to anti-Epidermal Growth Factor Receptor (EGFR) therapies. Yet, there is little guidance on the impact of RASa, as identified by next generation sequencing (NGS), on anti-EGFR response. Methods: We investigated the Foundation Medicine (FM) database for RASa in CRC and characterized this population based on patient (pt) characteristics and other concurrent genomic alterations. We subsequently investigated City of Hope (COH) mCRC molecular data set (using FoundationOne) and described the RASa population characteristics and response to anti-EGFR. Results: FM cohort included 21,315 CRC unique pt cases. 365 (1.5%) pts had RASa, of which 123 (0.6%) had ≥ 20 copy number (CN) ( RASa≥20). The incidence of MSI-H, RAS and BRAF short variant mutations in the overall, RASa, and RASa≥20 populations were (MSI-H: 5%, 0%, 0%), ( RAS: 54%, 32%, 2%), and ( BRAF: 6%, 1%, 0%), respectively. Copy number variation inversely correlated with likelihood of RAS mutation. COH cohort included 396 mCRC. 13 pts (3.3%) had RASa. The median RAS CN was 25 (7 - 79); 8/13 pts had RASa≥20. All but 3 COH pts with RASa had a RAS/BRAF-WT tumors. Pts with concurrent RAS mutations had relatively low RAS CN (7, 8, &13). Tumors with RASa were predominantly left-sided (12/13). 7/13 pts were treated with anti-EGFR therapy. All 7 pts were RAS/ BRAF wild-type: 6 left-sided; 3 pts 1st/2nd lines (1 pt oxaliplatin-based and 2 pts irinotecan-based) and 4 pts chemo-refractory (all irinotecan-based). All 4 chemo-refractory pts had RASa≥20and progressed on anti-EGFR on 1st post-treatment CT (PFS in all pts ≤ 3 mo). One 1st line pt had SD (PFS = 4 mo). One 2nd line pt had SD (PFS = 7 mo) and one had PD as best response. Conclusions: NGS identifies RASa and RASa≥20in 1.5% and 0.6% of CRC, respectively. RASa≥20 tumors are MSS and, generally lack RAS/ BRAF mutations, and predict for resistance to anti-EGFR. This supports the potential relevance of 20-CN cut-point for the exclusion of pts from anti-EGFR. These findings would benefit from additional validation from retrospective analyses of completed prospective anti-EGFR clinical trials.

4533 Background: We performed a CGP to compare the genomic alterations (GA) in ABC, UBC and SCCB. Methods: 143 cases of ACB, 2,142 cases of UCB and 83 cases of SCCB were subjected to CGP using a hybrid-capture based assay. Tumor mutational burden (TMB) was determined on 1.1 Mbp of sequenced DNA and microsatellite instability (MSI) was determined on 114 loci. PD-L1 expression was determined by IHC. Results: ABC patients were younger and more often female than UBC and SCCB (P < 0.0001). UBC and SCCB had a higher GA/tumor than ABC (P = 0.01). Un-targetable GA were similar in all 3 groups involving TP53 and KRAS. APC GA were more frequent in ABC whereas TERT, CDKN2A/B and DNA-repair genes ( ARID1A and KDM6A) more frequently altered in UBC and SCCB. Targetable MTOR pathway GA ( PIK3CA, TSC1, PTEN) were more frequent in UBC and SCCB as were targetable kinase alterations ( FGFR3 and ERBB2). The UBC and SCCB had a significantly higher TMB than ABC (P < 0.0001) including mean TMB and TMB > 20 mut/Mb (P < 0.0001). CD274 (PD-L1) was amplified more frequently in SCCB than ACB or UBC (P < 0.0001). MSI high status was very uncommon in all tumor types. Conclusions: Deep sequencing reveals that ABC features a widely different genomic profile from UBC and SCCB. UBC has the highest frequencies of targetable kinase GA and high TMB. SCCB has the highest frequencies of IO efficacy predicting biomarkers including mean TMB and PD-L1 amp. Nonetheless, ABC does feature potential kinase targets such as FGFR3 and ERBB2. [Table: see text]

4585 Background: We compared the genomic alteration (GA) profiles of mCSCC, mPSCC and mSSCC to study impact on the targeted and immunotherapy options for the men and women suffering from these refractory cancers. Methods: 78 mPSCC, 604 mCSCC and 338 mSSCC underwent CGP using a hybrid-capture based assay. Tumor mutational burden (TMB) was determined on 1.1 Mbp of sequenced DNA and and microsatellite instability (MSI) was determined on 114 loci. Results: The HPV+/CDKN2A- status was significantly more frequent in the mCSCC than mPSCC or mSSCC (P < 0.0001). The GA/tumor frequencies were similar for mCSCC and mPSCC, but significantly higher in mSSCC (P < 0.0001). TP53 mutations were more common in mSSCC (UV light exposure) and mPSCC (likely due to loss of an original HPV+ status). TERT, NOTCH1 and FAT1 GA were more frequent in mPSCC and mSSCC whereas PIK3CA GA were more common in mCSCC. MTOR pathway targets (GA in STK11, FBXW7 and PTEN) were more common in mCSCC than mPSCC or mSSCC. MSI high status was extremely rare in all cases. TMB ≥ 10/20 mut/Mb frequencies were noteworthy in mPSCC and mCSCC but extraordinarily high in mSSCC. Examples of patients with mCSCC, mPSCC and mSSCC responding to targeted and immunotherapies will be presented. Conclusions: Although mCSCC, mPSCC and mSSCC share a variety of clinicopathologic features, the 3 tumor types can be sharply differentiated on CGP. The TP53, CDKN2A and HPV status of the tumor types differ significantly with HPV+ higher in the mCSCC group. There are opportunities for targeted therapies in all groups predominantly identified in the MTOR pathway. The relatively high numbers of cases with significantly elevated TMB in all 3 tumor types suggest that immunotherapies would be beneficial in a large subset of patients. [Table: see text]