Search

PURPOSE Anti–epidermal growth factor receptor (EGFR) antibodies are effective treatments for metastatic colorectal cancer. Improved understanding of acquired resistance mechanisms may facilitate circulating tumor DNA (ctDNA) monitoring, anti-EGFR rechallenge, and combinatorial strategies to delay resistance. METHODS Patients with treatment-refractory metastatic colorectal cancer (n = 169) enrolled on the CO.26 trial had pre–anti-EGFR tissue whole-exome sequencing (WES) compared with baseline and week 8 ctDNA assessments with the GuardantOMNI assay. Acquired alterations were compared between patients with prior anti-EGFR therapy (n = 66) and those without. Anti-EGFR therapy occurred a median of 111 days before ctDNA assessment. RESULTS ctDNA identified 12 genes with increased mutation frequency after anti-EGFR therapy, including EGFR ( P = .0007), KRAS ( P = .0017), LRP1B ( P = .0046), ZNF217 ( P = .0086), MAP2K1 ( P = .018), PIK3CG ( P = .018), BRAF ( P = .048), and NRAS ( P = .048). Acquired mutations appeared as multiple concurrent subclonal alterations, with most showing decay over time. Significant increases in copy-gain frequency were noted in 29 genes after anti-EGFR exposure, with notable alterations including EGFR ( P < .0001), SMO ( P < .0001), BRAF ( P < .0001), MET ( P = .0002), FLT3 ( P = .0002), NOTCH4 ( P = .0006), ERBB2 ( P = .004), and FGFR1 ( P = .006). Copy gains appeared stable without decay 8 weeks later. There were 13 gene fusions noted among 11 patients, all but one of which was associated with prior anti-EGFR therapy. Polyclonal resistance was common with acquisition of ≥ 10 resistance related alterations noted in 21% of patients with previous anti-EGFR therapy compared with 5% in those without ( P = .010). Although tumor mutation burden (TMB) did not differ pretreatment ( P = .63), anti-EGFR exposure increased TMB ( P = .028), whereas lack of anti-EGFR exposure resulted in declining TMB ( P = .014). CONCLUSION Paired tissue and ctDNA sequencing identified multiple novel mutations, copy gains, and fusions associated with anti-EGFR therapy that frequently co-occur as subclonal alterations in the same patient.

Tumor mutation burden (TMB) is a measure to predict patient responsiveness to immune checkpoint immunotherapy because with increased mutation frequency, the likelihood of a greater neoantigen burden is increased. Although neoantigen prediction tools exist, tumor neoantigen burden has not been adopted as a measure to predict immunotherapy response. With both measures, current guidelines are limited to the coding regions, but ectopic expression of sequences in the noncoding space may potentially be a source of neoantigens. A pan-cancer cohort of 574 advanced disease stage patients with whole genome and transcriptome sequencing was analyzed to report mutation burden and neoantigen counts within the coding and noncoding regions. The efficacy of tumor neoantigen burden, reported as tumor neoantigen count (TNC), including neoantigens derived from the expression of noncoding regions, compared with TMB as a predictor of response to immunotherapy for 80 patients who had received treatment, was evaluated. TMB was found to be the best predictor of response to immunotherapy, whereas expression-derived TNC from the noncoding regions did not improve prediction of response. Therefore, there is minimal benefit in extending the calculation of TNC to the noncoding space for the purposes of predicting response. However, it is likely that there is a wealth of neoantigens derived from the noncoding space that may impact patient outcomes and treatments.

Real-world evidence regarding the value of integrating genomic profiling (GP) in managing cancer of unknown primary (CUP) is limited. We assessed this clinical utility using a prospective trial of 158 patients with CUP (10/2016-09/2019) who underwent GP using next-generation sequencing designed to identify genomic alterations (GAs). Only 61 (38.6%) patients had sufficient tissue for successful profiling. GAs were seen in 55 (90.2%) patients; of which GAs with FDA approved genomically-matched therapy were seen in 25 (40.9%) cases. A change in therapy was recommended and implemented (primary endpoint of the study) in 25 (10.1%) and 4 (2.5%) patients of the entire study cohort, respectively. Most common reason for inability to implement the profiling-guided therapy was worsening of performance status (56.3%). Integrating GP in management of CUP is feasible but challenging due to paucity of tissue and aggressive natural history of the disease and requires innovative precision strategies.

Supplemental Table S1 lists clinical characteristics for the study cohort. Figure S1 depicts results of batch correction. Figures S2 and S3 show comparisons between ERV levels and tumor content. Figure S4 shows ERV levels across different biopsy sites.

Supplemental Table S3 lists gene sets significantly enriched among genes up-regulated in VMP samples.

Next-generation sequencing of solid tumors has revealed variable signatures of immunogenicity across tumors, but underlying molecular characteristics driving such variation are not fully understood. Although expression of endogenous retrovirus (ERV)-containing transcripts can provide a source of tumor-specific neoantigen in some cancer models, associations between ERV levels and immunogenicity across different types of metastatic cancer are not well established. We performed bioinformatics analysis of genomic, transcriptomic, and clinical data across an integrated cohort of 199 patients with metastatic breast, colorectal, and pancreatic ductal adenocarcinoma tumors. Within each cancer type, we identified a subgroup of viral mimicry tumors in which increased ERV levels were coupled with transcriptional signatures of autonomous antiviral response and immunogenicity. In addition, viral mimicry colorectal and pancreatic tumors showed increased expression of DNA demethylation gene TET2. Taken together, these data demonstrate the existence of an ERV-associated viral mimicry phenotype across three distinct metastatic cancer types, while indicating links between ERV abundance, epigenetic dysregulation, and immunogenicity.

Supplemental Table S4 lists gene sets significantly enriched among genes down-regulated in VMP samples.

Supplemental Table S5 lists EGA accession numbers for each sample included in the study.

Supplemental Table S2 lists all genes differentially expressed in VMP samples.

Supplemental Data

Heatmap showing median gene expression levels (z-score) of genes involved in various pathways related to cellular metabolism. For each gene, within each of the four metabolic subgroups, expression values were assessed for significant deviation from zero using Wilcoxon signed rank test (mu = 0).

Supplementary Figures S1-7

Purpose:RNA-sequencing–based subtyping of pancreatic ductal adenocarcinoma (PDAC) has been reported by multiple research groups, each using different methodologies and patient cohorts. “Classical” and “basal-like” PDAC subtypes are associated with survival differences, with basal-like tumors associated with worse prognosis. We amalgamated various PDAC subtyping tools to evaluate the potential of such tools to be reliable in clinical practice.Experimental Design:Sequencing data for 574 PDAC tumors was obtained from prospective trials and retrospective public databases. Six published PDAC subtyping strategies (Moffitt regression tools, clustering-based Moffitt, Collisson, Bailey, and Karasinska subtypes) were used on each sample, and results were tested for subtype call consistency and association with survival.Results:Basal-like and classical subtype calls were concordant in 88% of patient samples, and survival outcomes were significantly different (P < 0.05) between prognostic subtypes. Twelve percent of tumors had subtype-discordant calls across the different methods, showing intermediate survival in univariate and multivariate survival analyses. Transcriptional profiles compatible with that of a hybrid subtype signature were observed for subtype-discordant tumors, in which classical and basal-like genes were concomitantly expressed. Subtype-discordant tumors showed intermediate molecular characteristics, including subtyping gene expression (P < 0.0001) and mutant KRAS allelic imbalance (P < 0.001).Conclusions:Nearly 1 in 6 patients with PDAC have tumors that fail to reliably fall into the classical or basal-like PDAC subtype categories, based on two regression tools aimed toward clinical practice. Rather, these patient tumors show intermediate prognostic and molecular traits. We propose close consideration of the non-binary nature of PDAC subtypes for future incorporation of subtyping into clinical practice.

Supplemental Table 2

Supplemental Figures and Tables

Supplemental Table 1

Purpose:Gene fusions are important oncogenic drivers and many are actionable. Whole-genome and transcriptome (WGS and RNA-seq, respectively) sequencing can discover novel clinically relevant fusions.Experimental Design:Using WGS and RNA-seq, we reviewed the prevalence of fusions in a cohort of 570 patients with cancer, and compared prevalence to that predicted with commercially available panels. Fusions were annotated using a consensus variant calling pipeline (MAVIS) and required that a contig of the breakpoint could be constructed and supported from ≥2 structural variant detection approaches.Results:In 570 patients with advanced cancer, MAVIS identified 81 recurrent fusions by WGS and 111 by RNA-seq, of which 18 fusions by WGS and 19 by RNA-seq were noted in at least 3 separate patients. The most common fusions were EML4-ALK in thoracic malignancies (9/69, 13%), and CMTM8-CMTM7 in colorectal cancer (4/73, 5.5%). Combined genomic and transcriptomic analysis identified novel fusion partners for clinically relevant genes, such as NTRK2 (novel partners: SHC3, DAPK1), and NTRK3 (novel partners: POLG, PIBF1).Conclusions:Utilizing WGS/RNA-seq facilitates identification of novel fusions in clinically relevant genes, and detected a greater proportion than commercially available panels are expected to find. A significant benefit of WGS and RNA-seq is the innate ability to retrospectively identify variants that becomes clinically relevant over time, without the need for additional testing, which is not possible with panel-based approaches.

Purpose:Identification of clinically actionable molecular subtypes of pancreatic ductal adenocarcinoma (PDAC) is key to improving patient outcome. Intertumoral metabolic heterogeneity contributes to cancer survival and the balance between distinct metabolic pathways may influence PDAC outcome. We hypothesized that PDAC can be stratified into prognostic metabolic subgroups based on alterations in the expression of genes involved in glycolysis and cholesterol synthesis.Experimental Design:We performed bioinformatics analysis of genomic, transcriptomic, and clinical data in an integrated cohort of 325 resectable and nonresectable PDAC. The resectable datasets included retrospective The Cancer Genome Atlas (TCGA) and the International Cancer Genome Consortium (ICGC) cohorts. The nonresectable PDAC cohort studies included prospective COMPASS, PanGen, and BC Cancer Personalized OncoGenomics program (POG).Results:On the basis of the median normalized expression of glycolytic and cholesterogenic genes, four subgroups were identified: quiescent, glycolytic, cholesterogenic, and mixed. Glycolytic tumors were associated with the shortest median survival in resectable (log-rank test P = 0.018) and metastatic settings (log-rank test P = 0.027). Patients with cholesterogenic tumors had the longest median survival. KRAS and MYC-amplified tumors had higher expression of glycolytic genes than tumors with normal or lost copies of the oncogenes (Wilcoxon rank sum test P = 0.015). Glycolytic tumors had the lowest expression of mitochondrial pyruvate carriers MPC1 and MPC2. Glycolytic and cholesterogenic gene expression correlated with the expression of prognostic PDAC subtype classifier genes.Conclusions:Metabolic classification specific to glycolytic and cholesterogenic pathways provides novel biological insight into previously established PDAC subtypes and may help develop personalized therapies targeting unique tumor metabolic profiles.See related commentary by Mehla and Singh, p. 6

Purpose:Gene fusions are important oncogenic drivers and many are actionable. Whole-genome and transcriptome (WGS and RNA-seq, respectively) sequencing can discover novel clinically relevant fusions.Experimental Design:Using WGS and RNA-seq, we reviewed the prevalence of fusions in a cohort of 570 patients with cancer, and compared prevalence to that predicted with commercially available panels. Fusions were annotated using a consensus variant calling pipeline (MAVIS) and required that a contig of the breakpoint could be constructed and supported from ≥2 structural variant detection approaches.Results:In 570 patients with advanced cancer, MAVIS identified 81 recurrent fusions by WGS and 111 by RNA-seq, of which 18 fusions by WGS and 19 by RNA-seq were noted in at least 3 separate patients. The most common fusions were EML4-ALK in thoracic malignancies (9/69, 13%), and CMTM8-CMTM7 in colorectal cancer (4/73, 5.5%). Combined genomic and transcriptomic analysis identified novel fusion partners for clinically relevant genes, such as NTRK2 (novel partners: SHC3, DAPK1), and NTRK3 (novel partners: POLG, PIBF1).Conclusions:Utilizing WGS/RNA-seq facilitates identification of novel fusions in clinically relevant genes, and detected a greater proportion than commercially available panels are expected to find. A significant benefit of WGS and RNA-seq is the innate ability to retrospectively identify variants that becomes clinically relevant over time, without the need for additional testing, which is not possible with panel-based approaches.

Purpose: With the rising incidence of early-onset pancreatic cancer (EOPC), molecular characteristics that distinguish early-onset pancreatic ductal adenocarcinoma (PDAC) tumors from those arising at a later age are not well understood. Experimental Design: We performed bioinformatic analysis of genomic and transcriptomic data generated from 269 advanced (metastatic or locally advanced) and 277 resectable PDAC tumor samples. Patient samples were stratified into EOPC (age of onset ≤55 years; n = 117), intermediate (age of onset 55–70 years; n = 264), and average (age of onset ≥70 years; n = 165) groups. Frequency of somatic mutations affecting genes commonly implicated in PDAC, as well as gene expression patterns, were compared between EOPC and all other groups. Results: EOPC tumors showed significantly lower frequency of somatic single-nucleotide variant (SNV)/insertions/deletions (indel) in CDKN2A (P = 0.0017), and were more likely to achieve biallelic mutation of CDKN2A through homozygous copy loss as opposed to heterozygous copy loss coupled with a loss-of-function SNV/indel mutation, the latter of which was more common for tumors with later ages of onset (P = 1.5e-4). Transcription factor forkhead box protein C2 (FOXC2) was significantly upregulated in EOPC tumors (P = 0.032). Genes significantly correlated with FOXC2 in PDAC samples were enriched for gene sets related to epithelial-to-mesenchymal transition (EMT) and included VIM (P = 1.8e-8), CDH11 (P = 6.5e-5), and CDH2 (P = 2.4e-2). Conclusions: Our comprehensive analysis of sequencing data generated from a large cohort of PDAC patient samples highlights a distinctive pattern of biallelic CDKN2A mutation in EOPC tumors. Increased expression of FOXC2 in EOPC, with the correlation between FOXC2 and EMT pathways, represents novel molecular characteristics of EOPC. See related commentary by Lou, p. 8

Purpose: RNA-sequencing–based subtyping of pancreatic ductal adenocarcinoma (PDAC) has been reported by multiple research groups, each using different methodologies and patient cohorts. “Classical” and “basal-like” PDAC subtypes are associated with survival differences, with basal-like tumors associated with worse prognosis. We amalgamated various PDAC subtyping tools to evaluate the potential of such tools to be reliable in clinical practice. Experimental Design: Sequencing data for 574 PDAC tumors was obtained from prospective trials and retrospective public databases. Six published PDAC subtyping strategies (Moffitt regression tools, clustering-based Moffitt, Collisson, Bailey, and Karasinska subtypes) were used on each sample, and results were tested for subtype call consistency and association with survival. Results: Basal-like and classical subtype calls were concordant in 88% of patient samples, and survival outcomes were significantly different (P < 0.05) between prognostic subtypes. Twelve percent of tumors had subtype-discordant calls across the different methods, showing intermediate survival in univariate and multivariate survival analyses. Transcriptional profiles compatible with that of a hybrid subtype signature were observed for subtype-discordant tumors, in which classical and basal-like genes were concomitantly expressed. Subtype-discordant tumors showed intermediate molecular characteristics, including subtyping gene expression (P < 0.0001) and mutant KRAS allelic imbalance (P < 0.001). Conclusions: Nearly 1 in 6 patients with PDAC have tumors that fail to reliably fall into the classical or basal-like PDAC subtype categories, based on two regression tools aimed toward clinical practice. Rather, these patient tumors show intermediate prognostic and molecular traits. We propose close consideration of the non-binary nature of PDAC subtypes for future incorporation of subtyping into clinical practice.

The mixture of epithelial and stromal components in pancreatic ductal adenocarcinoma (PDAC) may confound sequencing-based studies of tumor gene expression. Virtual microdissection has been suggested as a bioinformatics approach to segment the aforementioned components, and subsequent prognostic gene sets have emerged from this research. We examined the prognostic signature from the epithelial gene set of one such study using laser capture microdissected (LCM) epithelial samples. We also examined this gene set in matched stromal samples to determine whether prognostic findings were specific to the epithelium. LCM samples from 48 long-term and 48 short-term PDAC survivors were obtained. The resultant epithelial and stromal components were subjected to direct mRNA quantification using a 49 gene published PDAC classifier. Component-specific unsupervised hierarchical clustering was used to derive groups and survival differences were quantified. Immunohistochemical validation of particular genes was performed in an independent cohort. Clustering in the epithelial component yielded prognostic differences in univariable analysis (P = .02), but those differences were not significant when controlled for other clinicopathologic covariates (P = .06). Clustering in the stromal component yielded prognostic differences that persisted in the presence of other clinicopathologic covariates (P = .0005). Validation of selected genes in the epithelium (KRT6A-negative prognostic [P = .004]) and stroma (LY6D-improved prognostic [P = .01] and CTSV-negative prognostic [P = .0002]) demonstrated statistical independence in multivariable analysis. Although the genes used in this study were originally identified as being representative of the epithelial component of PDAC, their expression in the stroma appears to provide additional information that may aid in improved prognostication.

ObjectiveTo describe the clinical, pathological and genomic characteristics of pancreatic cancer with DNA mismatch repair deficiency (MMRD) and proficiency (MMRP).DesignWe identified patients with MMRD and MMRP pancreatic cancer in a clinical cohort (N=1213, 519 with genetic testing, 53 with immunohistochemistry (IHC)) and a genomic cohort (N=288 with whole-genome sequencing (WGS)).Results12 out of 1213 (1.0%) in the clinical cohort were MMRD by IHC or WGS. Of the 14 patients with Lynch syndrome, 3 (21.4%) had an MMRP pancreatic cancer by IHC, and 4 (28.6%) were excluded because tissue was unavailable for testing. MMRD cancers had longer overall survival after surgery (weighted HR after coarsened exact matching 0.11, 95% CI 0.02 to 0.78, p=0.001). One patient with an unresectable MMRD cancer has an ongoing partial response 3 years after starting treatment with PD-L1/CTLA-4 inhibition. This tumour showed none of the classical histopathological features of MMRD. 9 out of 288 (3.1%) tumours with WGS were MMRD. Despite markedly higher tumour mutational burden and neoantigen loads, MMRD cancers were significantly less likely to have mutations in usual pancreatic cancer driver genes like KRAS and SMAD4, but more likely to have mutations in genes that drive cancers with microsatellite instability like ACV2RA and JAK1. MMRD tumours were significantly more likely to have a basal-like transcriptional programme and elevated transcriptional markers of immunogenicity.ConclusionsMMRD pancreatic cancers have distinct clinical, pathological and genomic profiles. Patients with MMRD pancreatic cancer should be considered for basket trials targeting enhanced immunogenicity or the unique genomic drivers in these malignancies.

Next-generation sequencing of solid tumors has revealed variable signatures of immunogenicity across tumors, but underlying molecular characteristics driving such variation are not fully understood. Although expression of endogenous retrovirus (ERV)-containing transcripts can provide a source of tumor-specific neoantigen in some cancer models, associations between ERV levels and immunogenicity across different types of metastatic cancer are not well established. We performed bioinformatics analysis of genomic, transcriptomic, and clinical data across an integrated cohort of 199 patients with metastatic breast, colorectal, and pancreatic ductal adenocarcinoma tumors. Within each cancer type, we identified a subgroup of viral mimicry tumors in which increased ERV levels were coupled with transcriptional signatures of autonomous antiviral response and immunogenicity. In addition, viral mimicry colorectal and pancreatic tumors showed increased expression of DNA demethylation gene TET2. Taken together, these data demonstrate the existence of an ERV-associated viral mimicry phenotype across three distinct metastatic cancer types, while indicating links between ERV abundance, epigenetic dysregulation, and immunogenicity.

Purpose: Identification of clinically actionable molecular subtypes of pancreatic ductal adenocarcinoma (PDAC) is key to improving patient outcome. Intertumoral metabolic heterogeneity contributes to cancer survival and the balance between distinct metabolic pathways may influence PDAC outcome. We hypothesized that PDAC can be stratified into prognostic metabolic subgroups based on alterations in the expression of genes involved in glycolysis and cholesterol synthesis. Experimental Design: We performed bioinformatics analysis of genomic, transcriptomic, and clinical data in an integrated cohort of 325 resectable and nonresectable PDAC. The resectable datasets included retrospective The Cancer Genome Atlas (TCGA) and the International Cancer Genome Consortium (ICGC) cohorts. The nonresectable PDAC cohort studies included prospective COMPASS, PanGen, and BC Cancer Personalized OncoGenomics program (POG). Results: On the basis of the median normalized expression of glycolytic and cholesterogenic genes, four subgroups were identified: quiescent, glycolytic, cholesterogenic, and mixed. Glycolytic tumors were associated with the shortest median survival in resectable (log-rank test P = 0.018) and metastatic settings (log-rank test P = 0.027). Patients with cholesterogenic tumors had the longest median survival. KRAS and MYC-amplified tumors had higher expression of glycolytic genes than tumors with normal or lost copies of the oncogenes (Wilcoxon rank sum test P = 0.015). Glycolytic tumors had the lowest expression of mitochondrial pyruvate carriers MPC1 and MPC2. Glycolytic and cholesterogenic gene expression correlated with the expression of prognostic PDAC subtype classifier genes. Conclusions: Metabolic classification specific to glycolytic and cholesterogenic pathways provides novel biological insight into previously established PDAC subtypes and may help develop personalized therapies targeting unique tumor metabolic profiles. See related commentary by Mehla and Singh, p. 6

Global initiatives focused on whole genome and transcriptome analysis (WGTA) of metastatic pancreatic ductal adenocarcinoma (PDAC) tumor cohorts are driving progress in understanding the clinical impact of PDAC heterogeneity. This knowledge is key to continue the discovery of subsets of patients who could benefit from biomarker-informed targeted therapy and expand the currently limited PDAC treatment options. To identify clinically actionable subtypes in metastatic PDAC (mPDAC), our team analyzes prospectively collected WGTA data from patients enrolled in the PanGen trial (NCT02869802), in parallel with analysis of a set of patient-derived organoids (PDOs). In a cohort of 69 sequenced tumors, four cases (5.8%) had an amplification of chromosome 7q22, which included copy gains of transcriptional co-factor TRRAP, drug metabolizing cytochrome P450 genes CYP3A4 and CYP3A5, and SMURF1. SMURF1 is a ubiquitin-protein ligase that regulates TGFβ receptor signaling, in part via its interaction with SMAD7, and has been implicated in the epithelial-to-mesenchymal program and tumor invasiveness in PDAC. The proteasome inhibitor (PI) bortezomib has been shown to attenuate SMURF1 levels. We investigated whether SMURF1 signaling axis is associated with sensitivity to PIs in mPDAC, by analyzing the cytotoxic effects of the PIs bortezomib, carfilzomib and ixazomib in PDOs. Eight PDOs were treated with PIs at concentrations between 0.1pM-1mM. The presence of live and dead cells was quantified using the IN Cell Analyzer, and cell toxicity was analyzed using GRtoxic metrics (grcalculator.org). The PIs bind catalytic subunits of the 26S proteasome, primarily proteasome subunit beta-5, encoded by PSMB5. The cytotoxic activity of all three PIs negatively correlated with tumor biopsy PSMB5 expression (p75th percentile) tumor SMURF1 expression showed shorter overall survival (OS) compared to the rest of the cohort (median OS 9 vs. 14 months, respectively; p=0.004). Immunohistochemistry analysis of a tumor tissue microarray comprising 175 resected PDAC cases detected an association of concomitant low SMURF1 and high SMAD7 protein levels with adjuvant therapy response (median OS: 40 vs. 12 months with no adjuvant treatment, p=0.002), suggesting that attenuation of SMURF1 in the presence of SMAD7 may improve chemotherapy response in early disease stages. In summary, we present data indicating increased sensitivity to PIs in a subset of tumors with low PSMB5 and high SMURF1 expression, and highlight the translational utility of the investigation of genomic and clinically annotated pre-clinical models in PDAC predictive biomarker discovery. Citation Format: Andrew Metcalfe, Joanna M. Karasinska, James T. Topham, Steve E. Kalloger, Hassan Ali, Dawn Ashforth, Marco A. Marra, Janessa Laskin, Patricia A. Tang, Rachel Goodwin, Oliver F. Bathe, Daniel J. Renouf, David F. Schaeffer. Targeting SMURF1 with low-dose proteasome inhibitors in pancreatic cancer organoids [abstract]. In: Proceedings of the AACR Special Conference on Pancreatic Cancer; 2022 Sep 13-16; Boston, MA. Philadelphia (PA): AACR; Cancer Res 2022;82(22 Suppl):Abstract nr B053.

Background & Aims Most pancreatic ductal adenocarcinomas (PDACs) express an activated form of KRAS, become hypoxic and dysplastic, and are refractory to chemo and radiation therapies. To survive in the hypoxic environment, PDAC cells upregulate enzymes and transporters involved in pH regulation, including the extracellular facing carbonic anhydrase 9 (CA9). We evaluated the effect of blocking CA9, in combination with administration of gemcitabine, in mouse models of pancreatic cancer. Methods We knocked down expression of KRAS in human (PK-8 and PK-1) PDAC cells with small hairpin RNAs. Human and mouse (KrasG12D/Pdx1-Cre/Tp53/RosaYFP) PDAC cells were incubated with inhibitors of MEK (trametinib) or extracellular signal-regulated kinase (ERK), and some cells were cultured under hypoxic conditions. We measured levels and stability of the hypoxia-inducible factor 1 subunit alpha (HIF1A), endothelial PAS domain 1 protein (EPAS1, also called HIF2A), CA9, solute carrier family 16 member 4 (SLC16A4, also called MCT4), and SLC2A1 (also called GLUT1) by immunoblot analyses. We analyzed intracellular pH (pHi) and extracellular metabolic flux. We knocked down expression of CA9 in PDAC cells, or inhibited CA9 with SLC-0111, incubated them with gemcitabine, and assessed pHi, metabolic flux, and cytotoxicity under normoxic and hypoxic conditions. Cells were also injected into either immune-compromised or immune-competent mice and growth of xenograft tumors was assessed. Tumor fragments derived from patients with PDAC were surgically ligated to the pancreas of mice and the growth of tumors was assessed. We performed tissue microarray analyses of 205 human PDAC samples to measure levels of CA9 and associated expression of genes that regulate hypoxia with outcomes of patients using the Cancer Genome Atlas database. Results Under hypoxic conditions, PDAC cells had increased levels of HIF1A and HIF2A, upregulated expression of CA9, and activated glycolysis. Knockdown of KRAS in PDAC cells, or incubation with trametinib, reduced the posttranscriptional stabilization of HIF1A and HIF2A, upregulation of CA9, pHi, and glycolysis in response to hypoxia. CA9 was expressed by 66% of PDAC samples analyzed; high expression of genes associated with metabolic adaptation to hypoxia, including CA9, correlated with significantly reduced survival times of patients. Knockdown or pharmacologic inhibition of CA9 in PDAC cells significantly reduced pHi in cells under hypoxic conditions, decreased gemcitabine-induced glycolysis, and increased their sensitivity to gemcitabine. PDAC cells with knockdown of CA9 formed smaller xenograft tumors in mice, and injection of gemcitabine inhibited tumor growth and significantly increased survival times of mice. In mice with xenograft tumors grown from human PDAC cells, oral administration of SLC-0111 and injection of gemcitabine increased intratumor acidosis and increased cell death. These tumors, and tumors grown from PDAC patient-derived tumor fragments, grew more slowly than xenograft tumors in mice given control agents, resulting in longer survival times. In KrasG12D/Pdx1-Cre/Tp53/RosaYFP genetically modified mice, oral administration of SLC-0111 and injection of gemcitabine reduced numbers of B cells in tumors. Conclusions In response to hypoxia, PDAC cells that express activated KRAS increase expression of CA9, via stabilization of HIF1A and HIF2A, to regulate pH and glycolysis. Disruption of this pathway slows growth of PDAC xenograft tumors in mice and might be developed for treatment of pancreatic cancer.

Purpose: Gene fusions involving neuregulin 1 (NRG1) have been noted in multiple cancer types and have potential therapeutic implications. Although varying results have been reported in other cancer types, the efficacy of the HER-family kinase inhibitor afatinib in the treatment of NRG1 fusion–positive pancreatic ductal adenocarcinoma is not fully understood. Experimental Design: Forty-seven patients with pancreatic ductal adenocarcinoma received comprehensive whole-genome and transcriptome sequencing and analysis. Two patients with gene fusions involving NRG1 received afatinib treatment, with response measured by pretreatment and posttreatment PET/CT imaging. Results: Three of 47 (6%) patients with advanced pancreatic ductal adenocarcinoma were identified as KRAS wild type by whole-genome sequencing. All KRAS wild-type tumors were positive for gene fusions involving the ERBB3 ligand NRG1. Two of 3 patients with NRG1 fusion–positive tumors were treated with afatinib and demonstrated a significant and rapid response while on therapy. Conclusions: This work adds to a growing body of evidence that NRG1 gene fusions are recurrent, therapeutically actionable genomic events in pancreatic cancers. Based on the clinical outcomes described here, patients with KRAS wild-type tumors harboring NRG1 gene fusions may benefit from treatment with afatinib. See related commentary by Aguirre, p. 4589

3551 Background: CO.26 was a phase 2 trial (2-sided α=0.1 and 80% power) that randomized 180 patients with refractory mCRC 2:1 to durvalumab + tremelimumab vs BSC with improved overall survival (OS) (HR 0.73, 90%CI 0.55-0.97, P=0.07). A Nanostring assay validated for use with FFPE was used to determine CMS for correlation with outcome. Methods: Archival FFPE from 163/180 (91%) of patients (pts) underwent RNA extraction and CMS subtyping. Cox proportional hazard models evaluated the prognostic and predictive impact of CMS on overall survival. Results: CMS distribution was skewed towards CMS4 (76%), with lower prevalence of CMS1 (2%), CMS2 (16%) and CMS3 (2%). There were 7/163 cases of indeterminate CMS (4%). Subgroup analysis was restricted to CMS2 and CMS4 based on sample size. With BSC alone, CMS2 showed trends to worse OS compared to all other patients pooled (HR 1.93, 90% CI 1.03-3.61, P=0.085), while CMS4 did not (HR 0.86, 90% CI 0.50-1.48, P=0.64). OS but not progression free survival (PFS) was improved with durvalumab + tremelimumab in the overall population. OS was improved with durvalumab + tremelimumab among patients with CMS2 tumors (HR 0.39, 90% CI 0.19-0.82, P=0.035) but not in patients with CMS4 tumors (HR 0.73, 90% CI 0.52-1.02, P=0.12) compared to BSC. Neither CMS2 (P-interaction=0.37) nor CMS4 (P-interaction=0.91) were predictive of OS benefit from durvalumab + tremelimumab compared to BSC. Disease control rate (DCR) trended to being better among CMS4 (24/85) than CMS2 cancers (1/15, OR 5.51, 90% CI 1.10-29.88, P=0.11) or CMS4 vs all non CMS4 cancers (1/21, OR 7.87, 90% CI 1.65-41.98, P=0.023) for patients on durvalumab + tremelimumab. PFS was not improved with durvalumab + tremelimumab in CMS2 (P=0.19) or CMS4 (P=0.29) cancers relative to BSC. Conclusions: In this trial of refractory colorectal cancer, we saw a shift in CMS subtype with more CMS4 than expected. Compared to CMS4, CMS2 showed stronger signals towards improved OS with durvalumab + tremelimumab but had a lower disease control rate. Differences in immune signaling by CMS may be important determinants of which component of immune regulation needs to be targeted in mCRC to improve outcomes. Clinical trial information: NCT02870920. [Table: see text]

4024 Background: Median overall survival (mOS) for advanced pancreatic ductal adenocarcinoma (PDAC) is ≤1 year. However, there are patients (pts) who live for >2 years on chemotherapy and those who rapidly progress prior to a first scan, underscoring biological heterogeneity in PDAC subgroups. Understanding these differences is important in clinical trial design and provides prognostic information for pts. Methods: Clinical and molecular data (WGS and RNAseq) from pts with a diagnosis of locally advanced or metastatic PDAC enrolled in the COMPASS (NCT02750657) and POG/PanGen (NCT02155621, NCT02869802) studies were available for analysis as part of the TFRI’s EPPIC program and Marathon of Hope Cancer Centres Network, a multi-institutional collaborative network aiming to implement precision medicine in Canada. Clinical data were collected prospectively from both cohorts. Pts had an ECOG PS 0-1 and chemotherapy regimen was based on clinician’s preference. Profiling was performed on fresh biopsies and homologous recombination deficiency was identified with the HRDetect assay. We compared pts with very short OS (STS; ≤3 months, mo) vs long OS (LTS; ≥24 mo). Results: 341 pts were included in the analysis of which 47 were STS (mOS 1.6 mo) and 42 LTS (mOS 29.7 mo). There was no difference in median age, BMI or sex (p>0.1) between cohorts. STS were more likely to have an ECOG PS 1 vs 0 (p=0.03), higher tumor burden (RECIST) (p

Background RNA‐sequencing‐based classifiers can stratify pancreatic ductal adenocarcinoma (PDAC) into prognostically significant subgroups but are not practical for use in clinical workflows. Here, we assess whether histomorphological features may be used as surrogate markers for predicting molecular subgroup and overall survival in PDAC. Methods Ninety‐six tissue samples from 50 patients with non‐resectable PDAC were scored for gland formation, stromal maturity, mucin, necrosis, and neutrophil infiltration. Prognostic PDAC gene expression classifiers were run on all tumors using whole transcriptome sequencing data from the POG trial (NCT02155621). Findings were validated using digital TCGA slides (n = 50). Survival analysis used multivariate Cox proportional‐hazards tests and log‐rank tests. Results The combination of low gland formation and low neutrophil infiltration was significantly associated with the poor prognosis PDAC molecular subgroup (basal‐like or squamous) and was an independent predictor of shorter overall survival, in both frozen section (n = 47) and formalin‐fixed paraffin‐embedded (n = 49) tissue samples from POG patients, and in the TCGA samples. This finding held true in the subgroup analysis of primary (n = 17) and metastatic samples (n = 79). The combination of high gland formation and high neutrophils had low sensitivity but high specificity for favorable prognosis subgroups. Conclusions The assessment of gland formation and neutrophil infiltration on routine histological sections can aid in prognostication and allow inferences to be made about molecular subtype, which may help guide patient management decisions and contribute to our understanding of heterogeneity in treatment response., The combination of low gland formation and low neutrophils is significantly associated with poor prognosis molecular subgroups of pancreatic ductal adenocarcinoma, and is an independent predictor of shorter overall survival. These features can be readily assessed in archival tissue slides, allowing inferences regarding molecular subtype and prognosis to be made as part of routine patient care.

Purpose: Immune checkpoint inhibitors (ICI) have revolutionized the treatment of solid tumors with dramatic and durable responses seen across multiple tumor types. However, identifying patients who will respond to these drugs remains challenging, particularly in the context of advanced and previously treated cancers. Experimental Design: We characterized fresh tumor biopsies from a heterogeneous pan-cancer cohort of 98 patients with metastatic predominantly pretreated disease through the Personalized OncoGenomics program at BC Cancer (Vancouver, Canada) using whole genome and transcriptome analysis (WGTA). Baseline characteristics and follow-up data were collected retrospectively. Results: We found that tumor mutation burden, independent of mismatch repair status, was the most predictive marker of time to progression (P = 0.007), but immune-related CD8+ T-cell and M1-M2 macrophage ratio scores were more predictive for overall survival (OS; P = 0.0014 and 0.0012, respectively). While CD274 [programmed death-ligand 1 (PD-L1)] gene expression is comparable with protein levels detected by IHC, we did not observe a clinical benefit for patients with this marker. We demonstrate that a combination of markers based on WGTA provides the best stratification of patients (P = 0.00071, OS), and also present a case study of possible acquired resistance to pembrolizumab in a patient with non–small cell lung cancer. Conclusions: Interpreting the tumor-immune interface to predict ICI efficacy remains challenging. WGTA allows for identification of multiple biomarkers simultaneously that in combination may help to identify responders, particularly in the context of a heterogeneous population of advanced and previously treated cancers, thus precluding tumor type–specific testing.

587 Background: Oncogenic driver mutations in KRAS represent a hallmark genomic event in approximately 90% of pancreatic adenocarcinoma (PDAC). For the remaining 10% of patients with KRAS wildtype (wt) PDAC, distinct driver mutations have been described, but their transcriptional landscape has not been reported. Here, we leverage sequencing data from the PanGen trial to provide a comprehensive characterization of advanced KRASwt PDAC. Methods: 63 patients with advanced PDAC received whole genome and transcriptome sequencing prior to treatment for metastatic disease as part of the PanGen trial (NCT02869802). Clinical features, somatic mutation data and gene expression patterns were compared between KRASwt and mutant groups. PDAC samples were contrasted with 77 other metastatic carcinoma (colorectal and cholangiocarcinoma) samples from the Personalized OncoGenomics trial (NCT0215562). KRAS wt-associated genes were further investigated using 3 additional PDAC cohorts (COMPASS NCT02750657, TCGA, and ICGC). Results: 9 of 63 (14%) samples were KRASwt, with an earlier median age at diagnosis (51.4 vs. 60.9 years; p=0.03). Clinical features, including diabetes, family history of malignancy, and location of primary tumor, were comparable. CA 19-9 at baseline was lower in the KRASwt group, with median 58 vs. 4900 U/mL in the KRAS-mutant group ( p=0.03). Patients with KRASwt PDAC showed increased overall survival in univariable ( p=0.0024) and multivariable ( p=0.0089) analyses. 6 of 9 (67%) KRASwt tumors had fusions involving NRG1 (n = 3), FGFR2 (n = 1), BRAF (n = 1) or NTRK2 (n = 1), while known actionable fusions were not observed in KRAS mutant patients. KRASwt tumors showed increased expression of genes associated with cholangiocytes and grouped with cholangiocarcinoma samples in unsupervised clustering analysis. Validation using three independent PDAC cohorts revealed a core set of 70 KRAS wt-associated genes that converge on keratinization, ion transport, and hormone metabolism pathways. Conclusions: Patients with KRASwt PDAC show potentially targetable molecular traits with actionable fusions. We also highlight novel mutation and expression-based similarities between KRASwt PDAC and cholangiocarcinoma samples. Recurrent dysregulation of genes involved in cellular structure and metastasis provide impetus for further investigation into the developmental trajectory and potential therapeutic vulnerabilities of KRASwt PDAC.

Purpose Children with acute myeloid leukemia (AML) whose disease is refractory to standard induction chemotherapy therapy or who experience relapse after initial response have dismal outcomes. We sought to comprehensively profile pediatric AML microRNA (miRNA) samples to identify dysregulated genes and assess the utility of miRNAs for improved outcome prediction. Patients and Methods To identify miRNA biomarkers that are associated with treatment failure, we performed a comprehensive sequence-based characterization of the pediatric AML miRNA landscape. miRNA sequencing was performed on 1,362 samples—1,303 primary, 22 refractory, and 37 relapse samples. One hundred sixty-four matched samples—127 primary and 37 relapse samples—were analyzed by using RNA sequencing. Results By using penalized lasso Cox proportional hazards regression, we identified 36 miRNAs the expression levels at diagnosis of which were highly associated with event-free survival. Combined expression of the 36 miRNAs was used to create a novel miRNA-based risk classification scheme (AMLmiR36). This new miRNA-based risk classifier identifies those patients who are at high risk (hazard ratio, 2.830; P ≤ .001) or low risk (hazard ratio, 0.323; P ≤ .001) of experiencing treatment failure, independent of conventional karyotype or mutation status. The performance of AMLmiR36 was independently assessed by using 878 patients from two different clinical trials (AAML0531 and AAML1031). Our analysis also revealed that miR-106a-363 was abundantly expressed in relapse and refractory samples, and several candidate targets of miR-106a-5p were involved in oxidative phosphorylation, a process that is suppressed in treatment-resistant leukemic cells. Conclusion To assess the utility of miRNAs for outcome prediction in patients with pediatric AML, we designed and validated a miRNA-based risk classification scheme. We also hypothesized that the abundant expression of miR-106a could increase treatment resistance via modulation of genes that are involved in oxidative phosphorylation.

Pancreatic ductal adenocarcinoma (PDAC) can be stratified into distinct transcriptome subtypes, with the ‘basal-like’ or ‘squamous’ subtype being associated with worse prognosis, compared to the ‘classical’ subtype. Our group recently demonstrated that PDAC tumors have unique metabolic transcriptome profiles, and that genes involved in glycolysis and cholesterol synthesis pathways are positively correlated with basal-like and classical gene expression patterns, respectively. The mitochondrial pyruvate complex (MPC) mediates the transport of pyruvate into the mitochondria which attenuates the effect of glycolysis on tumor progression. The mitochondrial pyruvate carrier 1 (MPC1) gene, which encodes one of two subunits of MPC, is deleted in over 60% of metastatic PDAC and PDAC glycolytic tumors have lowest levels of MPC1 expression. Using PDAC tissue microarrays, we also found that reduced MPC1 protein expression correlates with reduced survival in patients. We hypothesized that targeting MPC1 will alter metabolic reprogramming and may modulate tumor aggressiveness and therapeutic vulnerability in PDAC tumor cells. Genomically and clinically annotated patient-derived tumor organoids (PDOs) were generated from metastatic biopsies from patients enrolled in the PanGen study (NCT02869802). PDOs from both basal and classical tumors were used in the study. In order to investigate glycolysis in PDOs, we adapted the Seahorse Glycolytic Stress Test. Glycolysis, glycolytic capacity and reserve were analyzed in PDOs under basal and treated conditions. To alter MPC1 activity, PDOs were treated for 48 hours with 5uM of UK-5099, an MPC1 inhibitor, or 2.5-5uM SRT1720. SRT1720 is an activator of sirtuin 1 (SIRT1) and the transcriptional coactivator peroxisome proliferator-activated receptor γ coactivator-1α (PGC1-α), which regulates the expression of MPC1. An unpaired t-test with an alpha of 0.05 was used for all statistical analysis. Glycolysis analysis revealed distinct glycolytic profiles in PDOs with differences in glycolytic capacity and reserves trending with different tumor subtypes. Treatment with UK-5099 resulted in an increase in both glycolytic rate and reserve in PDOs from basal and classical tumors. Treatment with SRT1720 resulted in significantly reduced glycolytic rate and capacity. These data suggest that PDAC PDOs exhibit distinct metabolic profiles and that targeting MPC1 can modulate glycolysis in PDOs. Our ongoing efforts aim to further characterize the subtype-specific effect of MPC1 modulators on glycolysis and chemotherapy response in PDAC PDOs. Citation Format: Hassan A. Ali, Andrew Metcalfe, James T. Topham, Cassia S. Warren, Joanna M. Karasinska, David F. Schaeffer, Daniel J. Renouf. Targeting the mitochondrial pyruvate complex to alter metabolic programming in pancreatic cancer [abstract]. In: Proceedings of the AACR Virtual Special Conference on Pancreatic Cancer; 2021 Sep 29-30. Philadelphia (PA): AACR; Cancer Res 2021;81(22 Suppl):Abstract nr PO-021.

Dramatic and durable responses to immune checkpoint inhibitors (ICIs) have been observed across multiple tumor types, but identifying the patients most likely to respond to these drugs remains challenging, particularly in the context of metastatic and pre-treated disease. Recent clinically approved biomarkers for patient selection are tumor agnostic, but currently all approved markers are evaluated independently. Through the Personalized Onco Genomics (POG) program at BC Cancer, we aimed to study the impact of evaluating multiple biomarkers in a tumor agnostic cohort. We performed whole genome and transcriptome analysis (WGTA) on fresh tumor biopsies from a heterogeneous pan-cancer cohort of 82 patients with advanced metastatic disease subsequently treated with ICIs. Established biomarkers, including tumor mutation burden (TMB) and CD8+ T cell scores, were able to distinguish responders in our advanced and pre-treated cohort. Additionally, we discovered that combining multiple biomarkers provided the best stratification of patients, suggesting a multifaceted approach, such as WGTA, may be suitable for more accurate identification of patients that may benefit from ICIs. As such, we have initiated a Phase II clinical trial, CAPTIV-8 (NCT04273061), which is distinctive in its use of WGTA to evaluate multiple markers including TMB, CD8+ T cell scores, an M1-M2 macrophage score and viral integration to select patients most likely to respond to atezolizumab. Clinical and genomic data prospectively collected from two hundred patients will be evaluated to test the efficacy of combining these biomarkers and identify additional biomarkers of response which can be used to guide treatment with ICIs. Citation Format: Emma Titmuss, Alexandra Pender, Erin Pleasance, Scott Brown, Cameron J. Grisdale, James Topham, Yaoqing Shen, Melika Bonakdar, Gregory A. Taylor, Laura Williamson, Karen Mungall, Eric Chuah, Andrew J. Mungall, Richard A. Moore, Jean-Michel Lavoie, Stephen Yip, Howard Lim, Daniel J. Renouf, Sophie Sun, Steven J. Jones, Robert Holt, Marco A. Marra, Janessa Laskin. CAPTIV-8: A prospective trial of atezolizumab using a multivariate model incorporating whole genome and transcriptome analysis [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2021; 2021 Apr 10-15 and May 17-21. Philadelphia (PA): AACR; Cancer Res 2021;81(13_Suppl):Abstract nr 1647.

4135 Background: PA.7 evaluated whether combining PD-L1 and CTLA-4 inhibition with GEM and Nab-P increases efficacy. A previous analysis of the PA.7 data demonstrated high plasma based TMB (≥9 mut/Mb) was associated with improved OS in the Gem, Nab-P, D+T arm. DNA repair pathway aberrations beyond mismatch repair have been associated with potential immune sensitivity. We assessed the predictive value of germline ATM mutations in the PA.7 trial. Methods: This randomized phase II study (ClinicalTrials.gov NCT02879318) assessed the efficacy and safety of GEM, Nab-P, D, and T (arm A) vs. GEM and Nab-P (arm B) in patients (pts) with mPDAC (n = 180). The primary endpoint was overall survival (OS). Pre-treatment plasma was sequenced with the Predicine ATLAS next generation assay (600 gene, 2.4 Mb panel). 2-sided alpha set at 0.1. Results: 180 pts were randomized (119 to arm A and 61 to arm B) There was no significant difference in OS (9.8 months in arm A vs. 8.8 months in arm B, p-value 0.72) or PFS (5.5 months and 5.4 months respectively, HR 0.98, p-value 0.91). Plasma analysis was performed on 174/180 pts with available samples. 16/174 (9.2%) pts had germline ATM mutations, 12 in arm A and 4 in arm B. GEM, Nab-P, D+T was associated with improved OS in patients with ATM mutations (HR 0.10, 90% CI 0.03-0.37; median OS 13.9 months vs. 4.9 months) while no activity was seen in pts with ATM Wild Type (HR 0.99, 90% CI 0.73-1.33; median OS 9.79 months vs. 10.2 months); interaction p = 0.014. Germline ATM mutation status was independent of plasma TMB levels (Wilcoxon p = 0.76). Conclusions: Germline ATM mutation appeared predictive of benefit from the addition of dual immune checkpoint inhibitors (D and T) to Gem and Nab-P, with a significant interaction p-value. In addition to previous data from this trial regarding the predictive value of high plasma TMB (≥9 mut/Mb), this data further supports that there may be independent subgroups of PDAC, beyond MSI-H, that may benefit from immunotherapy, and trials evaluating immunotherapy in subgroups of PDAC with these profiles are warranted. Clinical trial information: NCT02879318.

411 Background: PA.7 evaluated whether combining PD-L1 and CTLA-4 inhibition with GEM and Nab-P increases efficacy as first line therapy in mPDAC. High TMB is associated with immunotherapy sensitivity, with a threshold of ≥10 mut/Mb receiving FDA accelerated approved for pembrolizumab in a tissue agnostic setting. We assessed the predictive value of plasma TMB in the PA.7 trial. Methods: This randomized phase II study (ClinicalTrials.gov NCT02879318) assessed the efficacy and safety of GEM, Nab-P, D, and T (arm A) vs. GEM and Nab-P (arm B) in patients (pts) with mPDAC (n = 180). The primary endpoint was overall survival (OS). Pre-treatment plasma was sequenced with the CLIA-certified PredicineATLAS cfDNA next generation assay (600 genes, 2.4 Mb panel). A pre-specified cut point of 5 mut/MB was selected based on distribution of TMB in the trial. 2-sided alpha set at 0.1. Results: 180 pts were randomized (119 to arm A and 61 to arm B). There was no significant difference in OS (9.8 months in arm A vs. 8.8 months in arm B, p-value 0.72) or PFS (5.5 months and 5.4 months respectively, HR 0.98, p-value 0.91). Plasma TMB analysis was performed on 174/180 pts with available samples, and tumor derived variants were detected in 173/174 pts (99.4%). 172 pts were microsatellite stable and 1 pt was microsatellite high (MSI-H) (plasma TMB 52.9 muts/Mb). Using the pre-specified cut-point of 5 mut/Mb there was no significant predictive value from plasma TMB (interaction p = 0.91). Using a minimum p-value approach, a cut-point of 9 mut/MB appeared predictive (p-interaction = 0.064; significant at pre-specified p = 0.1). 8/174 (4.6%) pts had a plasma TMB ≥9 mut/Mb (5/115 (4.4%) in arm A and 3/59 (5%) in arm B). GEM, Nab-P, D+T was associated with improved OS in patients with plasma TMB ≥9 mut/Mb (HR 0.30, 90% CI 0.06-1.37) while no activity was seen in pts with < 9 mut/Mb, (HR 0.97, 90% CI 0.73-1.29). TMB cut-point analysis revealed a clear trend for a decreasing HR favoring the GEM, Nab-P, D and T arm above the selected cut point, with no benefit in the low TMB group. Conclusions: Plasma TMB analysis was successful in over 99% of pts with available samples. Plasma TMB ≥9 mut/Mb was predictive of benefit from the addition of dual immune checkpoint inhibitors (D and T) to Gem and Nab-P, with a significant interaction p-value. While only present in a subgroup of pts (4.6%), this data defines a group beyond MSI-H PDAC that may benefit from immunotherapy. The optimal cut-point for high TMB in this setting requires validation. A clinical trial specifically assessing the role of chemotherapy combined with immune checkpoint inhibition in high TMB mPDAC is warranted. Clinical trial information: NCT02879318.

61 Background: Pembrolizumab was recently granted tissue agnostic FDA accelerated approval for metastatic cancers with TMB≥10 mut/Mb. However, limited data supports immunotherapy in microsatellite stable (MSS) mCRC with TMB≥10 mut/Mb. We assessed tissue TMB and contrasted it to plasma derived TMB in the CO.26 trial. Methods: CO.26 was a phase 2 trial (2-sided ⍺ = 0.1 and 80% power) that randomized 180 patients (pts) 2:1 to D+T or BSC in refractory mCRC. Pre-treatment plasma was sequenced with the GuardantOMNI assay and archival tissue underwent exome sequencing with TMB assessed per the TMB harmonization project. MSI-H cases were excluded. For plasma TMB, we used a previously published cut point (≥28). Results: Overall survival (OS) but not progression free survival (PFS) was improved with D+T in the entire population. Of 180 pts, 163 were evaluable for plasma and 110 for tissue TMB. Median time between archival tissue and plasma collection was 3.1 yrs (IQR 1.9-5.1). Median tissue TMB was 6.6 muts/Mb (IQR 4.1-12.0), while median plasma TMB was 16.3 muts/Mb (IQR 9.4-25.9). Tissue and plasma TMB (r = -0.039, P = 0.69) were not correlated. Tissue TMB≥10 was not prognostic in the BSC arm (HR 1.01, 90%CI 0.52-1.92, P = 0.99) and OS was not improved in pts with tissue TMB≥10 (32/110 pts) following D+T vs BSC. A test of interaction suggested this threshold was not predictive (P = 0.85). Using a minimum P-value approach, no threshold supported high tissue TMB as predictive in MSS mCRC. In fact, the optimal cut point suggested low tissue TMB ( < 4.1 muts/Mb) had the greatest benefit from D+T (P-interaction = 0.048) and pts with TMB ≥4.1 mut/Mb (HR 0.50, 90%CI 0.26-0.96, P = 0.083) trended to better OS in the BSC arm. In contrast, 35/163 pts (21%) were identified in a high plasma TMB group associated with worse OS (HR 2.56, 90%CI 1.45-4.54, P = 0.007) in the BSC arm but improved OS following D+T compared to BSC with P-interaction = 0.082. Only 1 response was noted following D+T in a pt with tissue TMB = 16 mut/Mb and plasma TMB = 13 mut/Mb. Conclusions: Archival tissue TMB≥10 mut/Mb does not appear predictive of D+T benefit in MSS mCRC. Plasma derived TMB may better reflect evolutionary changes following intervening therapy than archival tissue. Clinical trial information: NCT02870920. [Table: see text]

117 Background: Previous studies have identified MAPK and PIK3CA/AKT/mTOR pathways as common mechanisms of acquired resistance to anti-EGFR antibodies (EGFRab) in mCRC. However, such alterations do not account for all patients that become treatment resistant. Using paired whole-exome (WES; tissue) and circulating tumor DNA (ctDNA; plasma) sequencing, we performed characterization of the EGFRab resistance landscape in patients with mCRC. Methods: Post-treatment (ctDNA; plasma) sequencing was performed for 169 patients with mCRC, with 112 (66%) patients also receiving pre-treatment (WES; primary tumor) sequencing as part of the CO.26 trial. 66 (39%) patients received EGFRab previously at a median of 132.0 days prior to plasma sampling. Using bespoke bioinformatics pipelines (WES) coupled with the GuardantOMNI assay (plasma), we compared somatic mutation (SNV/indel, CNV and fusion) frequencies between pre- and post-EGFRab timepoints, and contrasted results between the two groups. Results: Significant increases in de novo acquisition of EGFR (p = 5.6e-4), KRAS (p = 0.011), ZNF217 (p = 0.0022), MAP2K1 (p = 0.0078) and LRP1B (p = 0.017) SNV/indels were unique to the EGFRab group and often occurred as multiple, low allele frequency events in the same patient. De novo copy number amplification of known resistance genes EGFR/ BRAF/ MET were observed in EGFRab-treated patients (p < 0.05), along with SMO (p = 6.8e-7), PTEN inhibitory gene PREX2 (p = 5.6e-4), FLT3 (p = 2.0e-5), NOTCH4 (p = 6.3e-5), ERBB2 (7.4e-4), KMT2A (p = 3.7e-4) and ARID1B (p = 0.0014). Genes impacted by fusion events in EGFRab-treated patients included BRAF-KIAA1549 (1 patient) and MET-CAV1 (1 patient), and these events were not detected in matched pre-treatment samples. EGFRab-treated patients were found to acquire a combination of multiple (≥5) mutation events (SNV/indel, CNV or fusion) at much higher frequency compared to non-EGFRab-treated patients (67% versus 25% of patients, p = 8.7e-8). Tumor mutation burden (TMB) was not significantly different (p = 0.71) between treatment groups prior to therapy initiation, while post-treatment TMB was significantly higher (p = 1.8e-7) in EGFRab-treated patients (median 25.4 versus 13.1 mut/mb). Conclusions: In addition to previously established resistance pathways, we identified acquired alterations in additional genes such as SMO, PREX2 and epigenetic modifiers KMT2A/ARID1B in EGFRab-treated patients . Moreover, we highlight the phenomenon by which EGFRab-treated tumors acquire multiple concurrent resistance mutations and heightened TMB. Our analysis provides novel insight into the landscape of resistance mechanisms to EGFRab in mCRC while highlighting the potential role for immunotherapy post-EGFRab. Clinical trial information: NCT02870920.

472 Background: There is emerging evidence about the predictive role of homologous recombination deficiency (HRD) in multiple cancers. The clinical utility of HRD is less well defined in gastrointestinal (GI) malignancies. Methods: We reviewed the whole genome (WGS) and transcriptomic (RNA-Seq) data of patients with advanced GI cancers between 2012-2018 in the Personalized Oncogenomics trial (NCT02155621). Scores were calculated as the sum of loss of heterozygosity, telomeric allelic imbalance, and large-scale state transitions scores. HRD was defined as a score ≥34. Mutational analysis was performed to determine the presence of mutational signature 3, which is usually strongly associated with BRCA status. Retrospective chart review was conducted to extract treatment and survival outcomes. Overall survival (OS) from initiation of first-line systemic therapy and time to progression on platinum therapy (TTPp) were calculated. Linear and multivariable regression analyses were conducted. Results: Of 154 patients with GI primaries, 56% were male and 105 (68%) were exposed to a platinum agent in the metastatic setting. Primary sites included upper GI (N=20, 9%), pancreas (N=35, 16%), colorectal (N=74, 33%), and other GI primary (N=25, 11%). Ten patients (6%) had a BRCA1/2 mutation, 20 (13%) had a high HRD score, and 11 (7%) had a high signature 3 score (>0.05). Six patients had both high HRD and high signature 3 scores (Table). On linear regression, high HRD scores and mutational signature 3 were independently associated with longer TTPp (β=4.17, 95% CI 0.15-8.19, p=0.04; β=8.03, 95% CI 2.87-13.18, p