Search

The management of pancreatic ductal adenocarcinoma (PDAC) has posed a considerable challenge for decades, with incidence and mortality rates almost mirroring each other. Despite this, a deeper understanding of the complex biology inherent to PDAC has provided a roadmap for a more precise approach to treatment. PDAC deficient in homologous recombination repair and mismatch repair is a subgroup that should be identified in the clinic for a targeted approach. In addition, KRAS wild-type PDAC, occurring in approximately 10% of patients, is enriched in highly actionable alterations including fusions, underscoring the importance of integrative germline and somatic sequencing. Comprehensive sequencing efforts over the past decade have documented genomic- and transcriptomic-based classifiers, with the latter emerging as two main subtypes: the classical and basal-like, which are now being evaluated in clinical trials. Together with promising, innovative strategies to target KRAS mutations and their pleotropic effects, a new era of precision medicine in PDAC is on the horizon.

472 Background: The rates of gastric (GC) and esophageal adenocarcinoma (EA) among individuals younger than age 50 years have increased by 30% and 50%, respectively, over the past three decades in the US. There is limited information on incidence and demographics for EO esophageal and gastric cancer in Europe. Methods: We conducted a retrospective review of the prospectively maintained upper gastrointestinal (UGI) cancer database in St James’s Hospital (SJH), the National Centre for Oesophageal and Gastric Cancer in Ireland. We reviewed all cases diagnosed with esophageal cancer (EC) and gastric cancer (GC) less than or equal to 50 years of age from 2000 to 2021. Results: From 2000 to 2021, 374 EO EA & GC cases were identified, 8% of the 4663 total EC & GC cases. The results are presented in the table. Conclusions: The major presenting symptoms for EO-UGI cancer were dysphagia, weight loss and dyspepsia, further investigation is warranted in young patients presenting with these symptoms, even in the absence of elevated BMI or other known risk factors for malignancy. A high proportion have advanced disease at diagnosis despite good PS and clinical outcomes are poor. There is a critical need to understand the pathogenesis of EO-UGI cancers to determine optimal strategies for prevention and management.[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

105 Background: NRG1 fusions are rare oncogenic drivers that have been identified in a variety of solid tumors. These proteins bind HER3, leading to HER2/HER3 heterodimerization and oncogenic transformation. Zenocutuzumab (MCLA-128; Zeno) is a Biclonics antibody that overcomes HER3 mediated NRG1 (or NRG1 fusion) signaling in tumor cells. Zeno docks on HER2, then binds to and blocks the NRG1 fusion-HER3 interaction and HER3 heterodimerization with HER2. Zeno is being evaluated in patients (pts) with NRG1+ cancer in the ongoing pivotal phase 2 part of the eNRGy study and early access program (EAP). Methods: Pts with NRG1 + solid tumors previously treated with or not candidates for standard therapy, aged ≥ 18 years, with ECOG PS ≤ 2, and measurable (RECIST 1.1) or evaluable disease, were enrolled. NRG1 fusions were determined by next generation sequencing (NGS) before enrollment. Zeno (750 mg IV Q2W) was administered until disease progression or unacceptable toxicity. Tumor imaging was conducted every 8 weeks. The primary endpoint is investigator (INV)-assessed objective response rate (ORR) and secondary endpoints include duration of response (DOR) and safety. Results: As of 12 Jan 2022, 99 pts with NRG1 + cancer (85 eNRGy, 14 EAP) were enrolled. Efficacy was assessed in 73 pts who received ≥ 1 dose of Zeno and who were enrolled as of 12 Jul 2021 to allow for the opportunity to have ≥ 6 months (mo) follow-up and met the criteria for the primary efficacy population. Median age was 59 y (range 22–84), 58% were female, 47%/53% pts had ECOG PS 0/1. Tumor types were non-small cell lung cancer (NSCLC; 41 pts), pancreas cancer (18 pts), breast cancer (5 pts), cholangiocarcinoma (3 pts), colorectal cancer (2 pts), and 4 other tumor types (1 pt each), with a median 2 prior systemic therapies (range 0-9). The most frequent fusion partners were CD74 (27%), SLC3A2 (18%), and ATP1B1 (15%). Among the 71 pts with measurable disease, the INV-assessed confirmed ORR was 34% (90% CI, 25;44), including responses in NSCLC (35%; 14/40 pts), pancreas cancer (39%; 7/18 pts), breast cancer (2/4 pts), and cholangiocarcinoma (1/3 pts). Responses occurred at the first tumor assessment in 20/24 responders, and are ongoing in 13 pts. Treatment is ongoing in 22 pts (13 NSCLC, 6 pancreas, 3 other solid tumors). Median DOR was 9.1 mo (95% CI, 5.2-12.0). Kaplan-Meier estimate of DOR rate at 6 mo was 70%. Among the 208 pts treated with Zeno monotherapy across all dosing schedules in the phase 2 setting, for individual adverse events irrespective of causality, grade ≥ 3 events were reported in

4149 Background: Systemic inflammatory scores have been developed as tools to aid clinicians in prognostication and patient (pt) selection for clinical trials. We compared the accuracy of five prognostic scores to predict overall survival (OS) in pts with advanced pancreatic adenocarcinoma (PDAC). Methods: Pts with advanced PDAC enrolled on the COMPASS trial (NCT02750657) from 2015 to 2020 were included. All pts had biopsies for whole genome and RNA sequencing prior to standard first-line chemotherapy in the advanced setting. Prognostic risk was calculated using: neutrophil-to-lymphocyte ratio (NLR; >5 = high), platelet-to-lymphocyte ratio (PLR; > 150 = high), Prognostic Nutritional Index (PNI = albumin + 5 x lymphocytes. PNI < 45 = high risk), Gustave Roussy Immune Score (GRIm-S; NLR>6 = 1 point, albumin upper limit of normal [ULN] = 1 point. GRIm-S ≥2 = high risk), and Memorial Sloan Kettering Prognostic Score (MPS; NLR >4 and albumin < 40 = high risk). OS was estimated using the Kaplan-Meier method and compared between risk groups (high vs. not-high) for each scoring system using the log-rank test. Cox proportional hazards models were used to analyze the association between each prognostic score and OS, adjusting for baseline clinical and genomic factors. Results: In total, 263 pts with advanced PDAC cancer were included, with median follow up of 32.9 (95% CI 15.9-64.2) months. Median OS in the intention to treat population was 9.3 months (95% CI 8-10.2). PLR and PNI were not prognostic. High risk NLR (N=85, 32%), GRIm-S (N=47, 18%) and MPS (N=46, 17%) identified pts with poor prognosis. The GRIm-S and MPS were most significant: median OS in high vs low risk pts 6.4 vs. 10 months p

2558 Background: Immune checkpoint inhibitors (ICI) are highly effective in select cancers. Novel predictors of T cell-inflammation may identify a broader subset of tumors with ICI responsiveness. Our group has identified four chemokines (CCL4, CCL5, CXCL9, CXLC10) able to predict a T cell-inflamed phenotype in primary and metastatic pancreatic tumors. Here, we test whether this 4-chemokine signature can predict T cell-inflammation across additional tumor types and response to ICI. Methods: Using matched genomic and transcriptomic data from 6,455 patients spanning 25 tumor types from The Cancer Genome Atlas, we searched for associations between the 4-chemokine signature and metrics of antitumor immunity. Further, we tested the association of this signature with markers of DNA damage repair deficiency. We also investigated the ability of this signature to predict response to immunotherapy using real-world data from a pan-cancer cohort of 82 patients in the Personalized OncoGenomics Program who had received ICI. Results: The majority of tumor types displayed sub-populations with high expression of the 4-chemokines (4-chemokinehi) and transcriptional hallmarks of the cancer-immunity cycle. Testicular germ cell tumors, cervical squamous cell carcinomas, and head and neck squamous cell carcinomas were the strongest expressors of the signature. Immunomodulatory genes, including PD-L1, PD-1, TIM3, LAG3, TIGIT, CTLA-4, and FASLG, were significantly overexpressed (phi cohorts. Genesets of processes involved in the cancer-immunity cycle, including MHC I expression and cytolytic activity, were upregulated in the 4-chemokinehi cohorts (phi tumors. Among patients treated with ICIs, those with 4-chemokinehi tumors had a longer median time to progression (104 versus 71 days, p=0.013) and overall survival (391 versus 195 days, p=0.016). The 4-chemokine signature outperformed TMB for overall survival prediction. Conclusions: Sub-populations of T cell-inflamed patients exist across tumor types and may therefore respond favourably to ICI. The 4-chemokine signature has the potential to select a wider spectrum of patients that may benefit from ICIs. [Table: see text]

3049 Background: Mutations in IDH1/2 genes are common in low-grade glioma (GM) and occur in approximately 20% of intrahepatic cholangiocarcinoma (ICCA). Mutant IDHs leads to preferential accumulation of the R-relative (R2HG) to the S-enantiomer (S2HG) of 2HG. We investigated the utility of circulating R2HG, total R2HG+S2HG (tRS) and ratio of R2HG/S2HG (rRS) as biomarkers in GM and ICCA patients (pts). Methods: Blood and tumor tissues samples were obtained from pts with IDH1/2-mutant GM and ICCA. IDH mutation was confirmed by either immunohistochemistry (GM pts) or next generation sequencing (ICCA pts). Samples were analyzed for S2HG and R2HG using a validated HPLC tandem mass spectrometry method. Tissue and serum R2HG, tRS and rRS were compared with paired t-tests for GM pts. Serum S2HG, R2HG, tRS and rRS were compared with unpaired t-tests between GM and ICCA pts. A p

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.

TPS635 Background: Over 70% of patients with pancreatic ductal adenocarcinoma (PDAC) present with metastatic disease where the mainstay of treatment is combination chemotherapy. Two pivotal phase III trials showed survival benefit of mFOLFIRINOX (mFFX) and gemcitabine/nab-paclitaxel (GnP), respectively, compared to gemcitabine alone. Both are considered standard 1st line treatment options but have not been compared prospectively. Other than the BRCA phenotype there are no predictive molecular markers to identify which patients will benefit from mFFX versus GnP. Growing data suggests that RNA signatures and GATA6 expression may predict response to chemotherapy. Genomic platforms do identify small subsets of patients who may benefit from a targeted approach however, impact has been small. Patient-derived organoids (PDOs) are now feasible to passage for drug pharmacotyping that could inform drug therapy approaches. Combining all molecular strategies in real time including genomics, RNA signatures and adding PDO drug sensitivities could enable better precision choices for more patients with metastatic PDAC. Methods: PASS-01 is a multi-institutional randomized phase II trial evaluating the benefit of 1st line mFFX vs GnP in de novo metastatic PDAC patients with good PS who have undergone baseline tumor biopsies with tissue prepared for whole genome (WGS) and RNA sequencing and PDO generation/pharmacotyping using standard and novel drugs. The 10 objective is to determine the PFS benefit of mFFX compared to GnP as 1st line treatment with 80% power to detect a median PFS of 7 vs 5 months, favoring mFFX. 27 of a planned 150 patients have been accrued to date. Secondary endpoints include ORR (RECIST), DOR, OS by chemotherapy and biomarkers of therapy response including GATA-6 as a surrogate biomarker for the Moffit RNA classifier. Exploratory objectives include: to evaluate if each PDO DNA/RNA signature matches the patient and if the PDO chemotherapy sensitivities correlate to the patient’s 1st line response; to evaluate the benefit in switching patients to 2nd line treatment based on PDO drug sensitivity; to evaluate novel agents derived from PDO pharmacotyping and potential findings from profiling in 2nd/3rd line treatment; to explore retrospectively whether serial cell-free circulating tumor DNA analysis, circulating tumor cells and CA19.9 could reflect potential early predictors of emerging or de novo resistance and explore biomarkers of immune-oncologic sensitivity with multiplex immunohistochemistry. Each patient’s WGS and PDO data is discussed at a combined tumor board with study investigators immediately following their 1st 8-week CT and ongoing as data develops with the goal of recommending precision treatment choices back to their treating investigator. References: Conroy T et al. NEJM, 2011.; Von Hoff DD et al. NEJM,2013; Aung KL et al. CCR 2017; O’Kane G et al. CCR 2019; Tiriac H et al. Can Discov, 2018. Clinical trial information: NCT04469556.

469 Background: The Gustave Roussy Immune Score (GRIm-S) considers a composite of neutrophil to lymphocyte ratio (> 6 = 1), albumin (< 35 = 1) and LDH (> ULN = 1) and has been established as a prognostic score and may in aid in the selection of patients for phase 1 trials of immune checkpoint inhibitors. Methods: We explored the prognostic impact of the GRIm-S (high > 1) in patients enrolled on the COMPASS trial and correlated the score with genomic and clinical characteristics. Patients in this trial had biopsies for whole genomic and RNA sequencing prior to standard chemotherapy regimens in the advanced setting. Results: 252 patients were included in the analyses with a median follow-up time of 28 months. 16% of patients had a high GRIm-S with significantly shorter median overall survival (OS) of 4.1 months versus 10.0 months in those with a low score (HR 2.18, 95% CI 1.4-3.4, p < 0.0001). In the GRIm-S-high cohort, early progression with non-evaluable disease and disease progression were more common than in the GRIm-S low cohort (56% vs 31%, p = 0.003). In a multivariable analysis, a high GRIm-S was poorly prognostic (HR 1.6 95% CI 1.3-1.9, p < 0.001), whereas the classical RNA subtype (vs. basal-like) (HR 0.41, 95% CI 0.3-0.6, p < 0.001) and a high HRDetect score (HR 0.47 95% CI 0.3-0.7, p < 0.001) associated with superior OS. The GRIm-S did not correlate with RNA subtypes or with specific KRAS mutations. There were no differences in structural variant load or tumour mutational burden between groups. However those with a high GRIm-S did have a higher total target lesion diameter at baseline (p < 0.001). Conclusions: The GRIm-S identifies a subset of patients who have aggressive pancreas cancer and short life expectancy. This information may help clinicians in treatment decision making and selection for clinical trials.

3003 Background: NRG1 fusion proteins are oncogenic drivers in pancreas cancer and other solid tumors. They bind HER3, leading to HER2/HER3 heterodimerization and oncogenic transformation. The activity of zenocutuzumab (MCLA-128; zeno), a bispecific antibody targeting NRG1 fusion signaling in NRG1 fusion positive ( NRG1+) cancers, is being evaluated in the ongoing global multicenter phase 2 part of the eNRGy study and a global early access program (EAP). Methods: Enrolled patients (pts) have advanced NRG1+ pancreas cancer, non-small cell lung cancer (NSCLC), and other solid tumors previously treated with standard therapy, are ≥ 18 years-old, have ECOG ≤1, adequate organ function, and measurable disease (RECIST v1.1). Zeno dosing: 750 mg IV every 2 weeks until progression or unacceptable toxicity. Primary endpoint: investigator (INV)-assessed objective response rate (ORR). Secondary endpoints: ORR per central independent radiologist review, duration of response (DOR), and safety. Tumor imaging is conducted every 8 weeks. Results: 51 pts with NRG1+ cancer have received zeno, 37 in the eNRGy study and 14 pts in the EAP. As of 12 Jan 2021, treatment is ongoing in 27/51 pts (8/13 pancreas, 10/25 NSCLC, 9/13 other solid tumors). Among the 51 pts, 10 pts with pancreas cancer, 18 pts with NSCLC, and 5 pts with other solid tumors had measurable disease and had the opportunity for ≥1 tumor assessment (TA) and are included in this analysis. Among the 10 pts with pancreas cancer, median age was 49 y (range 34-72), 50% were male, 6/4 pts had ECOG 0/1, and all had metastatic disease and were KRAS wild-type. The median number of prior therapies was 3 (range 1-6). The INV-assessed confirmed ORR was 40% (4/10; 90% CI, 15;70), and for this cohort of pts, responses occurred at the first TA. Tumor regression was seen in 7/10 pts, and the disease control rate was 90% (90% CI, 61-100). A CA 19-9 decline of ≥ 50% was observed in 9/9 (100%) pts. DOR is pending. In the overall NRG1+ population, tumor regression was observed in 25 of 33 pts and confirmed INV-assessed responses were seen in 9 of 33 pts (ORR 27%; 90% CI, 15;43), including in pts who previously received afatinib. Zeno was well tolerated with no pts requiring dose reduction for toxicity. Across all cohorts, for individual AEs, grade 3 events were reported in ≤5% of pts, and there was a notable lack of cardiotoxicity and severe gastrointestinal or skin toxicity. Updated data from all cohorts (pancreas, NSCLC, other solid tumors) will be presented. Conclusions: Zeno induces rapid and major radiologic tumor regression and biomarker responses in heavily-pretreated metastatic KRAS wild-type NRG1+ pancreas cancer, with minimal toxicity. Zeno is a promising novel targeted therapeutic option for pts with NRG1+ cancers. Clinical trial information: NCT02912949.

4142 Background: KRAS mutations (m) (KRASm) are present in over 90% of pancreatic adenocarcinomas (PDAC) with a predominance of G12 substitutions. KRAS wildtype (WT) PDAC relies on alternate oncogenic drivers, and the prognostic impact of these remains unknown. We evaluated alterations in WT PDAC and explored the impact of specific KRASm and WT status on survival. Methods: WGS and RNAseq were performed on 570 patients (pts) ascertained through our translational research program from 2012-2021, of which 443 were included for overall survival (OS) analyses. This included 176 pts with resected and 267 pts with advanced PDAC enrolled on the COMPASS trial (NCT02750657). The latter cohort underwent biopsies prior to treatment with first line gemcitabine-nab-paclitaxel or mFOLFIRINOX as per physician choice. The Kaplan-Meier and Cox proportional hazards methods were used to estimate OS. Results: KRAS WT PDAC (n = 52) represented 9% of pts, and these cases trended to be younger than pts with KRASm (median age 61 vs 65 years p = 0.1). In resected cases, the most common alterations in WT PDAC (n = 23) included GNASm (n = 6) and BRAFm/fusions (n = 5). In advanced WT PDAC (n = 27), alterations in BRAF (n = 11) and ERBB2/3/4 (n = 6) were most prevalent. Oncogenic fusions (NTRK, NRG1, BRAF/RAF, ROS1, others) were identified in 9 pts. The BRAF in-frame deletion p.486_491del represented the most common single variant in WT PDAC, with organoid profiling revealing sensitivity to both 3rd generation BRAF inhibitors and MEK inhibition. In resected PDAC, multivariable analyses documented higher stage (p = 0.043), lack of adjuvant chemotherapy (p < 0.001), and the KRAS G12D variant (p = 0.004) as poor prognostic variables. In advanced disease, neither WT PDAC nor KRAS specific alleles had an impact on prognosis (median OS WT = 8.5 mths, G12D = 8.2, G12V = 10.0, G12R = 12.0, others = 9.2, p = 0.73); the basal-like RNA subtype conferred inferior OS (p < 0.001). A targeted therapeutic approach following first line chemotherapy was undertaken in 10% of pts with advanced PDAC: MMRd (n = 1), homologous recombination deficiency (HRD) (n = 19), KRASG12C (n = 1), CDK4/6 amplification (n = 3), ERBB family alterations (n = 2), BRAF variants (n = 2). OS in this group was superior (14.7 vs 8.8 mths, p = 0.04), mainly driven by HRD-PDAC where KRASm were present in 89%. Conclusions: In our dataset, KRAS G12D is associated with inferior OS in resected PDAC, however KRAS mutational status was not prognostic in advanced disease. This suggests that improved OS in the WT PDAC population can only be achieved if there is accelerated access to targeted drugs for pts.

4011 Background: Human equilibrative nucleoside transporter 1 (hENT1) belongs to a family of nucleoside transporters critical to entry of gemcitabine into cells. The prognostic and predictive characteristics of this biomarker in pancreatic ductal adenocarcinoma (PDAC) have primarily been evaluated by immunohistochemistry, with conflicting results. We explored the impact of hENT1 gene expression in the Comprehensive Molecular Characterization of Advanced Ductal Pancreas Adenocarcinoma for Better Treatment Selection (COMPASS) trial. Methods: Patients were enrolled on COMPASS from December 2015 to June 2020 and underwent a biopsy for whole genome sequencing (WGS) and RNA sequencing prior to first chemotherapy in the advanced setting. Biopsies underwent laser capture microdissection to enrich for tumour epithelium. Chemotherapy regimen was determined based on clinician preference. The cut-off thresholds for hENT1 expression were determined using the maximal chi-squared statistic. Response rates and overall survival (OS) were computed based on hENT1 expression and chemotherapy regimen. Results: 254 patients were included in the analyses with a median follow-up time of 18 months. 146 patients were treated with modified FOLFIRINOX (FFX), 104 with gemcitabine and nab-paclitaxel (GnP), and 16 received no systemic therapy. Based on gene expression levels, 133 patients were classified as hENT1 high and 121 as hENT1 low. hENT1 expression was significantly associated with the modified Moffitt classifier with higher expression seen in classical tumours (p < 0.001). In the entire cohort, median OS was 10.0 months in hENT1 high vs. 8.3 months in hENT1 low (adjusted HR 0.78, 95% confidence interval 0.59 - 1.03, p = 0.08). In patients receiving modified FFX, there was no difference in response rates (32% vs. 31%, p = 1.00) or OS (10.6 vs. 10.6 months, p = 0.94) between the hENT1 high and hENT1 low groups, respectively. However, in patient treated with GnP, response rates were significantly higher in hENT1 high patients compared to those with hENT1 low tumors (45% vs. 21%, p = 0.035). Median OS in this GnP treated cohort was 9.8 months in hENT1 high vs. 6.1 months hENT1 low (p = 0.003). In an interaction analysis, hENT was predictive of treatment response to GnP (p = 0.0002). Conclusions: Biomarkers predictive of response to GnP and FFX are urgently needed. Here we demonstrate that hENT1 gene expression is a predictor of response to GnP in advanced PDAC.

396 Background: BRCA1/2 and PALB2 are genes critical to the faithful repair of double strand breaks through the homologous recombination repair (HRR) pathway. Alterations in these genes serve as predictive biomarkers to both platinum and PARP inhibitors. Ataxia-telangiectasia mutated ( ATM) is also indirectly involved in HRR; however, its role as a predictive biomarker to DNA damage response agents is debated. Herein we evaluated the genomic characteristics and clinical outcomes of patients with ATM alterations on the Comprehensive Molecular Characterization of Advanced Ductal Pancreas Adenocarcinoma for Better Treatment Selection (COMPASS) trial. Methods: Patients on this study undergo a biopsy for whole genome sequencing (WGS) and RNA sequencing prior to chemotherapy; those with germline variants in ATM were reviewed by a genetics counsellor and defined as pathogenic, likely pathogenic, variant of unknown significance (VUS) or benign/likely benign. Genomic characteristics were reviewed and published classifiers of homologous recombination deficiency (HRD) were applied to all cases and included the percentage of substitution base signature (SBS) 3, the HRDetect score, the computed algorithm of large scale transitions, telomeric allelic imbalances and loss of heterozygosity (LOH), otherwise known as the genomic instability score (GIS). Results: As of January 2020, 304 patients were enrolled and 245 patients had both WGS and clinical data available. 86 germline variants in ATM were present in 70 patients. The majority of these (80%) were classified as benign or likely benign. 10 VUS were detected and 4 patients (2%) had pathogenic/likely pathogenic variants (PV). Of these 4 patients, LOH or a second somatic hit was evident in 1 case. Upon review of the PVs and VUS, SBS were consistent with typical PDAC and tumour mutational burden was low. HRDetect scores were low ( < 0.1) for 13/14 cases with either a VUS or PV; one VUS without biallelic loss, had a high HRDetect score, with presence of SBS 3 and a high GIS. This particular case was also found to have a tandem duplicator phenotype. None of the 4 cases with PV had evidence of HRD. Furthermore all four were treated with platinum based regimens without evidence of response. Conclusions: In a large series of sequenced pancreatic cancers, the presence of pathogenic germline variants in ATM was rare, with none of the cases demonstrating evidence of HRD. This suggests that this population is unlikely to benefit from PARP inhibition.

430 Background: Familial pancreatic cancer (FPC) is broadly defined as kindreds with at least a pair of first-degree relatives with pancreatic ductal adenocarcinoma (PDA). The role of DNA damage response agents, including platinum has not been well studied in this patient population. Methods: In this retrospective analysis, treatment details and clinical outcomes were analyzed in pts with FPC with advanced, unresectable or recurrent disease enrolled in the Ontario Pancreatic Cancer Study database. The primary outcome, overall survival (OS) was calculated from the initial diagnosis of advanced disease until death. 179 non-FPC patients from the COMPASS trial [NCT02750657] served as a control cohort all of whom had full molecular profiling and family history documented. OS between pts that received platinum-based therapy, and those that did not was compared using multivariable Cox proportional hazards model adjusting for age, sex, diagnosis year and FPC status. Interaction between FPC status and platinum was evaluated. Results: A total of 205 FPC pts were identified, 71% of pts had full germline testing and 16 (8%) had germline pathogenic variants in BRCA1/2. 104 (51%) were female and 101 (49%) male. Median age was 63 years (20-93) and 58 (28%) received platinum-based chemotherapy. Within the control arm (n=179), 71 (40%) were female, and 108 (60%) male; the median age was 64 years (29-84) and 106 (59%) received platinum-based therapy. In univariable analysis, median OS in pts with FPC was 16.9 months compared to 9.6 months (HR 0.46 [95% CI 0.37-0.58]). FPC patients receiving platinum had a superior median OS of 19 months compared to 15.5 months without platinum. In a multivariable analysis, both FPC (HR 0.33 [95% CI 0.21-0.51]) and receipt of platinum HR 0.53 [95% CI 0.38-0.73]) were prognostic. No interaction was seen with FPC and receipt of platinum (p=0.15). Conclusions: FPC status is prognostic but not predictive of platinum response in this study. Further molecular profiling of this unique cohort of patients will provide insights into putative predisposing germline alterations, and novel treatment strategies.

340 Background: Cholangiocarcinoma (CCA) is a lethal malignancy with limited treatment options. Molecular profiling of these tumours has revealed a number of actionable mutations and four clusters have been defined through integrative genomic analysis. We sought to explore whole genome sequencing (WGS) data in 20 patients enriched for CCA arising in the setting of cirrhosis or inflammatory disorders. Methods: A previously established Biliary Tract Cancer (BTC) database at the Princess Margaret Cancer Centre (PMCC)/University Health Network (UHN) was used to identify patients of clinical interest including long-term survivors, those with germ-line mutations, and those with chronic inflammatory disorders. WGS and bioinformatic analyses were performed at the Ontario Institute for Cancer Research. Results: The 20 resected samples included 12 patients (pts) with intrahepatic CCA, 7 perihilar CCA, and 1 distal CCA. 8 pts were alive > 8 years post resection and one patient harboured a germline MLH1 pathogenic variant (MLH1 G67R). 3 pts had documented cirrhosis (hepatitis B n = 2, haemochromatosis n = 1); other inflammatory disorders (n = 4) included ulcerative colitis (UC) without documented PSC (n = 1), PSC alone (n = 1), UC with PSC (n = 1), and ankylosing spondylitis (n = 1). The remaining cases were randomly selected. The predominant COSMIC single base substitution (SBS) mutational signatures were 1, 8 and 5, and the median TMB was 1.75 mutations per MB (0.73-33.23). The pt with an MLH1 mutation exhibited a TMB of 33.23 with predominance of SBS26. Actionable variants were enriched in the 8 pts who are alive and disease free including 2 predicted FGFR fusions and mutations in IDH1 (n = 1), BRAF V600E (n = 1), and BAP1 (n = 1). TP53 mutations (n = 7) were present exclusively in patients with cirrhosis or inflammatory disorders and the median survival in the group was 12 months (3 -21 months). Although SBS signatures were similar in this group, in the two cases of UC, SBS17 (unknown etiology) was evident. SBS17 was also dominant in one case where the TMB was 13 mutations per MB and the patient died within 3 months of diagnosis. A somatic case of homologous recombination deficiency, with no causative genetic alteration identified, was evident in a further patient receiving maintenance rituximab for a coexisting stage 4 mantle cell lymphoma. Conclusions: WGS may provide additional biological information in CCA particularly in patients with underlying inflammatory disorders, where TP53 mutations are prevalent and mutational signatures are distinct.

TPS145 Background: Although most cancers are initially susceptible to existing anticancer therapies, over time cancer cells develop resistance. BOLD-100 is a first-in-class therapy that targets the GRP78 pathway, a major regulator of cellular stress and resistance. This therapy suppresses drug resistance, survival and proliferation by restraining stress-induced upregulation of GRP78 in tumor cells, leading to inhibition of the cell survival response. BOLD-100 successfully completed a Phase 1 monotherapy trial, with a manageable safety profile; it has demonstrated synergy in established preclinical models in combination with a wide variety of anticancer therapies and shown to restore sensitivity of drug-resistant cell lines. Methods: BOLD-100-001 is a multicenter, prospective, two-stage, non-randomized Phase 1b dose escalation and expansion study of BOLD-100 in combination with FOLFOX chemotherapy for selected advanced solid gastrointestinal (GI) tumors: colorectal (CRC), pancreatic (PANC), gastric (GC) and bile duct cancers (BDC). Each patient will receive BOLD-100 along with FOLFOX chemotherapy on Day 1 of each 14-day cycle. In the dose-escalation phase (Part A), patients will be enrolled to determine the combination recommended dose using a standard 3+3 design. In the dose-expansion phase (Part B), up to 80 patients with selected GI tumors will be enrolled in 5 cohorts (comprising the 4 GI cancers referred) and treated with BOLD-100 at the recommended dose and FOLFOX, until progressive disease or unacceptable toxicity. In Part A dose-escalation, the primary objective is to assess the safety, tolerability and maximum tolerated dose. The Part B dose-expansion will assess the efficacy (progression free survival, overall survival and response rate), of FOLFOX chemotherapy plus BOLD-100. Secondary objectives in Part A and B include the assessment of pharmacokinetic and pharmacodynamic parameters and potential biomarkers predictive of efficacy. Eligible patients will have previously treated histologically/cytologically confirmed metastatic/unresectable GI tumor (CRC, PANC, GC or BDC), measurable disease per RECIST criteria, an ECOG performance score of 0 or 1, and adequate organ function. In Part A, no formal statistical hypotheses will be tested; analysis of the data will focus on comparisons of safety and dose-limiting toxicities between cohorts and only descriptive methods will be used. In Part B, Bayesian modelling will be used to continually reassess the efficacy endpoints: progression free survival, overall survival and objective response rates. The study was opened for enrollment in August 2020; approximately 25-30 patients will be screened to achieve up to 20 patients in Part A and up to 80 patients will be enrolled in Part B with a maximum of 25 patients per arm. Clinical trial information: NCT04421820.

e19306 Background: Recent advances in small cell lung cancer (SCLC) treatments necessitate a better understanding of health utility scores (HUS) of patients treated under standard regimens to facilitate robust pharmaco-economic assessments. HUS collected in clinical trials may be inherently skewed due to restrictive eligibility criteria, highlighting the need for real-world data. Methods: In this cohort observational study, HUS were evaluated in SCLC patients through EQ-5D surveys. We also evaluated patient reported (pr) ECOG performance status (PS)), treatment toxicities (modified patient reported (mpr) CTCAE) and symptoms (Edmonton Symptom Assessment System, ESAS). Clinical data were abstracted from electronic medical records. The impact of these variables on HUS was explored using regression. Results: Of 282 clinical encounters (12% newly diagnosed; 37% stable on treatment; 22% progressing; 29% stable off therapy/other) in 111 SCLC patients (58% male; 64% extensive stage), 29% had pr-ECOG PS ≥ 2 at the first encounter. Mean HUS in treatment naïve patients with limited disease was 0.848 (SEM = 0.028); for extensive stage, mean HUS = 0.715 (SEM = 0.046). Extensive stage (β = -0.12; p = 0.03), bone metastases (β = -0.10; p = 0.04), females (β = -0.07; p = 0.006), and pr-ECOG PS ≥ 2 (β = -0.21; p < 0.001) were each associated with decreased HUS in multivariable analyses. When excluding pr-ECOG PS from the model (highly correlated with HUS), progressive disease (β = -0.07; p = 0.02) was also associated with decreased HUS. Longitudinally, in patients with disease stability, HUS were unchanged in limited disease, but slowly decreased for extensive stage over time. HUS were inversely associated with increasing severity of most measured toxicities (mpr-CTCAE rho values ranged from -0.34 to -0.47) and symptoms (rho values -0.27 to -0.54). Conclusions: HUS in SCLC is impacted by the presence of extensive stage and bone metastases, in addition to cancer symptoms and treatment toxicities. The values reported from this real-world sample provides a basis in which to compare with new SCLC therapies in health technology assessments. [Table: see text]

4630 Background: The HRDetect score uses whole genome sequencing (WGS) to incorporate patterns of substitution base signatures and structural variation to identify tumours deficient in homologous recombination repair (HRD). HRD-tumours, with a higher mutational burden, may be more immunogenic. Methods: We applied HRDetect to 182 resected pancreatic cancers (PDA) and 233 advanced PDA enrolled on the COMPASS trial; both cohorts underwent WGS after tumour enrichment. Patients were classified as high(hi) or low(lo) according to the published score threshold of 0.7; clinical characteristics and survival outcomes were determined. Immunogenicity of the cohorts was explored by analyzing cytolytic activity (CYT) as measured by RNA expression of perforin and granzyme A. Results: 14% of resected (25/182) and 14% of advanced cases (32/233) were considered HRDetecthi . The median age at PDA diagnosis was younger in HRDetecthi vs HRDetectlo (61 vs 66 years, p = 0.005), with no difference in sex between groups. Of the 57 cases identified, 37 (65%) were considered true HRD-PDA with inactivation of BRCA1, BRCA2, PALB2, RAD51C and XRCC2. The remaining 20 cases, were considered false positives for HRD; of these 7 had evidence of a tandem duplicator phenotype with duplications ranging from 10Kbp to 1Mbp in size and 13 had no defining genomic characteristics of the HRD-subtype. In resected PDA, the HRDetect score after adjusting for stage, was not prognostic. In contrast in a multivariable analysis of advanced cases, both HRDetect (HR 0.51, 95% CI 0.30-0.87, p = 0.01) and the Moffitt RNA classifier were highly prognostic (HR 1.99, 95% CI 1.32-3.00, p = 0.0001) with improved survival in HRDetecthi and classical PDA. Of patients receiving platinum in advanced disease (n = 128) HRDetecthi PDA had longer survival compared to the HRDetectlo (15.6 vs. 9.9 months, p = 0.02) although the interaction term between chemotherapy regimen (gemcitabine vs. platinum) and HRDetect score was not significant in this cohort. HRDetecthi tumours had increased cytolytic activity than HRDetectlo PDA; furthermore, within the cohort of HRDetecthi PDA, higher CYT scores were evident in primary lesions compared to metastatic sites sequenced. Conclusions: A high HRDetect score is prognostic in advanced PDA where patients treated with platinum have longest survival. HRDetecthi tumours have increased cytolytic activity with differences observed between primary and metastatic lesions.

TPS3167 Background: Somatic IDH mutations are common in low grade gliomas, and rare in other solid tumors with the exception of intrahepatic cholangiocarcinoma (ICA) and certain subtypes of chondrosarcoma. IDH mutations confer a gain-of-function neomorphic activity, such that mutant IDH enzymes preferentially convert αKG to 2-hydroxyglutarate (2HG), resulting in abnormal accumulation of 2HG. 2HG competitively inhibits αKG-dependent dioxygenases, many of which are involved in DNA repair. Preclinical studies show that IDH mutated cancer cells have defective homologous recombination repair and are exquisitely sensitive to poly (adenosine 5’-diphophate-ribose) polymerase (PARP) inhibition. Methods: This is a single arm phase II basket study (NCT03991832). Pts with IDH mutated solid tumors are divided into three cohorts; A: low-grade glioma; B: cholangiocarinoma; C: all other solid tumors. Pts are treated with olaparib 300 mg twice daily continuously and durvalumab 1500 mg every 4 weeks until progression, intolerable toxicity or consent withdrawal. Radiological assessment is performed after every 2 cycles of study treatments. Major eligibility criteria include IDH mutation by immunohistochemistry or sequencing, up to 2 lines of systemic therapy for advanced disease, performance status 0 – 2 and adequate organ function. Pts are excluded if they received prior PARP inhibitors and anti-PD-1/PD-L1 antibody. The Simon’s optimal 2-stage design is applied for Cohorts A and B. 10 pts will be enrolled in each of Cohort A and B initially. If 2 or more partial responses (PR) are seen in these 10 pts, additional 19 pts will be enrolled for a total of 29 pts in that cohort. The combination is considered to be of clinical interest for further development if ≥ 6 PRs are seen in each cohort. Cohort C will enroll 20 pts. Archival tumor tissues and serial blood samples will be collected on study. 2HG levels will be measured and correlated with responses. The study was activated in December 2019. 3 pts have been enrolled into Cohort B. Clinical trial information: NCT03991832 .

741 Background: Whole genome sequencing (WGS) can reveal patterns of substitution base signatures and structural variation consistent with tumours deficient in homologous recombination repair. We evaluated the published HRDetect score and a novel HRD hallmark score (HS) in patients receiving combination chemotherapy (cCT) on the COMPASS trial for advanced PDAC. Methods: The HRD-HS incorporates 10 genomic characteristics of HRD-PDAC with a score ≥ 4 defining HRD. HRD-HS and an HRDetect score ≥0.7 were applied to WGS data and overall survival (OS) and response (ORR) evaluated. Sensitivity and specificity were ascertained. Results: As of 05/19, 205 eligible patients (pts) were enrolled and 186 received cCT including modified FOLFIRINOX n = 108 (58%) or cisplatin/gemcitabine n = 2 (1%) and gemcitabine/nab-paclitaxel n = 76 (41%). HRD-HS had a sensitivity of 87.5% and specificity of 100% in detecting HRD-PDAC. In contrast, HRDetect (≥0.7) had sensitivity of 51.9% and specificity of 100%; sensitivity increased to 73.7% when using a cutoff score of ≥0.99. 23/186 (12%) pts were classified as HRdetecthi and median OS was 15.3months (mo) vs 8.7mo in HRDetectlo pts (HR 0.44 95% CI 0.27-.70, p = 0.009). In platinum treated pts, median OS was 18.1mo (HRDetecthi) vs 9.3mo (HRDetectlo) (HR 0.38 95%CI 0.21-0.69, p = 0.02). HRD-HS predicted the longest median OS for platinum of 21.0mths. ORR in HRDetecthi was not different to HRDetectlo pts treated with cCT, however in those receiving platinum the ORR was 50% vs 19% respectively (p < 0.001). Of the false positives by HRDetect, 46% had a non-BRCA1 tandem duplicator phenotype (TDP). The TDP group comprised 8% of all patients enrolled. HRD-PDAC was caused by inactivation of BRCA1/2, PALB2, RAD51C and XRCC2; all germline variants were pathogenic. Pathogenic ATM and CHEK2 germline variants were present in 3 pts with evidence of a second somatic hit or LOH, none of these identified as HRD by either classifier nor considered a TDP. Conclusions: HRD-HS most correctly identified HRD-PDAC however the HRDetect score classifies additional patients sensitive to cCT, especially platinum. The TDP cohort may be responsive to DNA damaging agents warranting further evaluation. Clinical trial information: NCT02750657.

747 Background: It is unclear whether locally advanced pancreatic cancer (LAPC) should be treated similarly to metastatic pancreatic cancer (MPC). Clinical trials often exclude LAPC. We compare clinical and genomic characteristics of LAPC and MPC. Methods: Patients with LAPC and MPC were enrolled in the COMPASS trial (NCT02750657). Clinical, demographic and survival data was collected (cut off 8/31/19). WGS, RNAseq and modified Moffitt classification was performed. Results: Patients with LAPC (n = 28) and MPC (n = 180) did not differ in terms of age, gender, smoking or diabetes history. Patients with LAPC had lower BMI (p = 0.005) and lower baseline Ca19-9 (p = 0.02) than those with MPC. LAPC/MPC tumors had similar rates of KRAS, p53, CDKN2A and SMAD4 mutations and similar levels of ploidy, indels and neoantigens. There were increases in single nucleotide variants (p = 0.026) and structural variants (p = 0.04) in MPC compared with LAPC. No LAPC tumors were homologous recombination deficient (HRD) or KRAS wild type (WT), compared with 14 (8%) HRD and 16 (9%) KRAS-WT MPC. All LAPC were classical subtype compared with 77% MPC (p = 0.0052). OS data is shown in the table. MPC classical subtype tumors had improved OS compared with basal-like tumors (p = 0.008). Patients with MPC and p53 mutation trended towards worse OS compared with those without p53 mutation (p = 0.07); this was not seen in LAPC. There was a significant correlation between time on chemotherapy and OS in LAPC (p = 0.002) and MPC (p < 0.0001). Conclusions: Patients with LAPC have similar molecular profiles to those with MPC with similar rates of altered drivers. LAPC tumors are more likely to be Moffitt classical subtype and have similar OS to classical subtype MPC. LAPC patients benefit from local therapy; all patients benefit from increased time on chemotherapy. These data suggest that patients with LAPC should be treated similarly to those with classical subtype MPC but should be offered local therapy when possible. Clinical trial information: NCT02750657 . [Table: see text]

188 Background: Predictive mutational and transcriptional features in advanced PDAC are needed for improved patient stratification and treatment selection. Methods: Patients (pts) on the COMPASS trial with advanced PDAC are prospectively recruited prior to first-line chemotherapy for WGS and RNAseq. Tumor tissue undergoes enrichment by laser capture microdissection with genomic analyses available within eight wks. Tumor responses and clinical outcomes in this maturing cohort were correlated with molecular characteristics. Results: 157 pts underwent a tumor biopsy between Dec 2015 and Jul 2018 with over 95% success in achieving results. 141 genomes have been reported in pts planned to receive chemotherapy; the median time from biopsy to report was 35.5 days. In the ITT population,118 pts are summarised for first-line response. The median age was 63 yrs (29-81), 55% were male, and 16% had locally advanced disease. 66 (56%) received modified FFX as first-line treatment. 25 (21%) tumors displayed the Moffitt basal-like RNA signature which associated with chemotherapy resistance, with tumor shrinkage mainly observed in the classical RNA subtype (p = 0.002). GATA6 expression (log10 scale) clearly separated Moffitt subgroups with classical tumors exhibiting high expression (p < 0.0001). GATA6 in situ hybridization strongly correlated with RNAseq, (r = 0.89, p < 0.001) with strong correlation also seen with GATA6 IHC. The median overall survival (OS) in the classical group was 8.5 mths vs. 6.6 mths in the basal-like group (HR 0.53, 95% CI 0.32-0.89 p = 0.015). In those treated with mFFX, median OS was 10.1mths in classical versus 6.6mths in the basal-like subtypes, (HR 0.32, 95% CI 0.16-0.63, p=0.001). 30% of pts had potentially actionable genetic alterations including BRAF variants (n = 4) and an NTRK3-EML4 fusion in KRAS WT tumors (8%). Signatures of homologous recombination deficiency (HRD) were found in seven pts; two additional pts with BRCA germline variants did not exhibit somatic HRD hallmarks. Conclusions: Subsets of pts with advanced PDAC have actionable variants including those with HRD signatures and patients with KRAS WT tumors. GATA6 is a putative predictive biomarker of transcriptomic subgroups which should be incorporated in clinical trials. Clinical trial information: NCT02750657.

200 Background: Brain metastasis is a common occurrence in many cancers. However, with new therapies (stereotactic radiotherapy) and longer survival due to improved systemic therapies, more contemporary estimates of health utility scores (HUS) are required for this specialized cancer population. Further, recent efforts have tried to incorporate routine collection of such data, especially in an era when new radiation and new systemic therapies, especially targeted therapies require such data when undergoing health technology assessments. Methods: In a cross-sectional study design, patients in the specialized brain metastases clinic at Princess Margaret Cancer Centre were approached to complete the Health-related quality of life (HRQoL) tool, FACT-Br, and the EQ-5D-5L (to derived HUS using Canadian reference values) on iPad or paper. In addition, consent was obtained to collect clinico-demographic data from the patient and chart. In addition to descriptive analysis and participation rates, HUS were correlated to FACT-Br and its subscales. Results: Of 204 eligible patients, 134 were recruited (66% participation rate) from May 2017- Feb 2018. Of 105 patients in this preliminary analysis, the median age was 60 (range: 25-94) years; 73% were female; 64% were Caucasian; 20% were Asian; 49% had lung, 15% had breast, 37% had other primary cancer; 81% received some form of radiotherapy. Median time from first treatment of brain metastasis to survey was 12 (range 0.03-100) months. There were correlations with the following FACT-Br subscales: physical well-being (WB) (rho = 0.70), emotional WB (rho = 0.42), social WB (rho = 0.25), functional WB (rho = 0.59), and brain-specific subscale (rho = 0.67); overall FACT-Br (rho = 0.73); all correlations were p < 0.001. Conclusions: Patients with brain metastasis were generally willing to complete EQ-5D-5L, and patients had good HUS. HUS as measured indirectly by EQ-5D-5L were correlated to HRQoL as measured by FACT-Br or its subscales, with high correlation to the overall FACT-Br. EQ-5D-5L is a practical and useful tool to assess HUS routinely in brain metastases patients.

185 Background: The Patient Reported Outcome version of Common-Terminology-Criteria-for-Adverse-Events (PRO-CTCAE) and Edmonton Symptom Assessment System (ESAS) are validated tools that measure toxicities and symptoms in cancer patients. We compared the extent in which the presence and severity of toxicities and symptoms affected HUS in SCLC and NSCLC patients. Methods: Adult SCLC and NSCLC patients were recruited from the Princess Margaret Cancer Centre and surveyed cross-sectionally for clinico-demographic variables, EQ5D-5L, PRO-CTCAE and ESAS. HUS were estimated using EQ5D-5L. PRO-CTCAE toxicities include diarrhea, constipation, decreased appetite, nausea, vomiting, fatigue, neuropathy and rash. ESAS symptoms include pain, tiredness, drowsiness, appetite loss, nausea, shortness of breath, depression, anxiety and lack of well-being. These were combined to show frequency and average severity of toxicities/symptoms per patient. Univariable and multivariable linear regression analyses identified toxicity/symptom influencing HUS. Results: Of 75 SCLC and 150 NSCLC, 52% were male with median age of 65 years. The mean HUS was 0.76 (SCLC = 0.69; NSCLC = 0.79; p = 0.001). Compared to NSCLC, SCLC patients had a significantly higher number of toxicities (3.14 versus 1.33 using PRO-CTCAE, p < 0.0001); symptoms (6.75 vs 5.45 using ESAS, p = 0.0003) and severity of toxicities/symptoms (PRO-CTCAE 0-4: 0.95 versus 0.35, p < 0.0001; ESAS 0-10: 3.37 versus 2.04. p < 0.0001). There were significant correlations between the average severity of toxicities/symptoms and HUS (p < 0.0001) adjusted for age, histology, smoking pack years and performance status. For each increase in the average severity of toxicities, there was a corresponding mean drop of 0.03 in the HUS; for every increase in the average severity of symptoms, the drop was 0.04. These relationships were similar for both SCLC and NSCLC patients. Conclusions: Patient reported toxicities and symptoms have a significant impact on HUS in both SCLC and NSCLC patients. Early and aggressive management of such adverse events may be necessary to improve patients’ HRQoL.

8573Background: 5-10% of EGFR-mutant ACs undergo SCLC/neuroendocrine transformation (SCLC-T) but the clinical course of such pts is poorly characterized. Methods: We identified 50 pts with EGFR-mut...

e21072Background: Peripheral blood sampling for T790M in patients (pts) failing initial EGFR-TKIs is now standard practice. The value of longitudinal sampling is unknown. Methods: A study of cell f...

4084Background: Sel (AZD6244, ARR142886) is an oral MEK inhibitor, with preclinical evidence of synergy with Gem in BTC. CisGem is standard first-line treatment for advanced BTC. This trial assesse...

211 Background: COMPASS is a prospective study with the primary aim to identify predictive mutational and transcriptional features in advanced PDAC for improved patient stratification and treatment selection. Methods: Patients with advanced PDAC were prospectively recruited prior to first-line combination chemotherapy for whole genome sequencing (WGS) and RNA sequencing (RNASeq). Fresh tumor tissue was acquired by image guided percutaneous core needle biopsy of locally advanced primary or metastatic tumors. Laser capture microdissection was performed for all cases to ensure high-resolution genomic analyses. Primary endpoint was feasibility to report WGS results prior to first disease assessment CT scan at 8 weeks. The main secondary endpoint was discovery of patient subsets with predictive mutational and transcriptional signatures. Results: Of 63 patients who underwent a tumor biopsy between December 2015 and June 2017, WGS and RNASeq were successful in 62 (98%) and 60 (95%), respectively. Genomic results were reported at a median of 35 days (range 19-52 days) from biopsy, meeting the primary feasibility endpoint. Three patients with an ‘unstable’ genomic subtype, including two with a novel ‘duplicator’ phenotype, responded well to m-FOLFIRINOX. Of two cases with the same germline BRCA2 mutation, only the chemotherapy responder had loss of heterozygosity and genomic hallmarks of double stranded break repair deficiency. Approximately 25% of tumors displayed the basal-like RNA expression signature and these were chemotherapy resistant, with tumor shrinkage mainly observed in those with the classical RNA subtype (P = 0.003). Thirty percent of patients had potentially actionable genetic alterations. Conclusions: Prospective comprehensive genomic profiling of advanced PDAC is feasible and our early data indicate that chemotherapy response differs among patients with different genomic/transcriptomic subtypes providing the impetus for further studies. Clinical trial information: NCT02750657.

426 Background: The prognostic significance of anemia in BTC is unknown, but is common and may be multifactorial; data regarding its causes are limited. This study interrogated a large institutional database to examine the effect of anemia on overall survival (OS) in BTC, and explore factors associated with anemia. Methods: This Princess Margaret Cancer Centre cohort study included patients with BTC with available baseline hemoglobin (Hb) (1987-2016). Anemia was defined as Hb < 132 mg/dL for men > 60 years, < 137mg/dL for men aged 20-59, and < 122mg/dL in women. Additional relevant covariates were included in multivariable Cox regression for OS, and linear regression for association with Hb. Results: Of 1398 patients included, 711 (51%) were anemic at baseline (mean Hb 112 mg/dL). Anemic versus non-anemic patients were older (median age 66 vs 64 yrs, p = 0.006), had worse ECOG PS (12% ECOG 2-3 vs 6%, p < .001), BMI < 20 (31% vs 27%, p = 0.006) and elevated neutrophil:lymphocyte ratio (NLR), (64% vs 47%, p < .001), but cancer staging was not significantly different. Anemia was associated with shorter OS on univariate (HR 1.35, p < .001) and multivariable (HR 1.39, p < .001) regression (Table). Factors associated with lower Hb included older age, male gender, worse ECOG PS, tumor site, thrombocytopenia, elevated NLR. Conclusions: Baseline anemia was associated with shorter survival following diagnosis of BTC, independent of tumor stage or ECOG PS. Clinicians should be aware of this prognostic marker; validation in prospective datasets is warranted. [Table: see text]

494 Background: Pancreatic ductal adenocarcinoma (PDAC) remains a highly fatal disease inherently resistant to cytotoxic treatment. Despite the prevalence of liver and peritoneal metastases, subsets of patients also develop lung and bone metastases highlighting phenotypic heterogeneity. We reviewed the prognostic implications of metastatic sites on survival. Methods: This retrospective cohort study included all patients with PDAC who received surgical or oncological treatment at the University Health Network from September 2012 until December 2016. Clinical and pathological variables were obtained from patient electronic records. Radiological images and pathology reports were reviewed to ascertain sites of metastatic disease. The Kaplan-Meier method for survival and multivariable cox regression analysis identified prognostic factors. Results: 1153 patients were reviewed and 985 were included. 55% were male and the median age at diagnosis was 67 years. 287 (29%) completed curative surgery and 698 (71%) had locally advanced or metastatic disease; the median survival was 22 months and 9 months respectively. Lung and bone metastases were present in 18% (N = 180) and 6% (N = 58) of patients. In multivariable analysis increasing age and stage at diagnosis correlated with inferior survival (p < 0.0001) and the presence of any lung (HR 0.77; 95% CI 0.63-0.94, p = 0.01) or bone metastases (HR 0.74; 95% CI 0.54-1.0, p = 0.05) resulted in improved outcomes. Liver and peritoneal disease were not prognostic. Sex and family history of PDAC did not associate with survival. There was no association between site of metastases and sex however patients with bone metastases were significantly younger at first diagnosis (median age 63yrs, p < 0.01). Conclusions: Patients with advanced PDAC and metastases to lung or bone may represent distinct biological subtypes of PDAC. Molecular profiling of available tissue is ongoing.

164 Background: Dysphagia as a cancer symptom can be associated with significant morbidity. We developed dysphagia screener tools for use in observational studies (Phase 1) and routine symptom monitoring in clinical care (Phase 2). Methods: Various dysphagia or odynophagia screening questions, selected after an expert panel reviewed content, criterion, and construct validity, were compared to either FACT-E Swallowing Index Cut-Off Values (SICV) or to questions adapted from the Patient Reported Outcomes for Common Terminology Criteria for Adverse Events (PRO-CTCAE). Sensitivity, specificity and patient acceptability were assessed. Results: In developing a tool for observational studies (Phase 1; n = 178 esophageal cancer patients), the screening question, “How are you currently eating?” had the highest sensitivities and specificities against various SICV cut-offs, with the best optimal cut-off associated with the clinical outcome of weight loss (80% sensitivity, 75% specificity). When developing a rapid screening tool for routine symptom monitoring (Phase 2; 255 head and neck, gastro-esophageal, and patients undergoing thoracic radiation), a single question screener (“Do you experience any difficulty or pain upon swallowing?”) versus a PRO-CTCAE-like gold standard generated sensitivities between 86-94% and specificities between 93-100%. The screening question (+/- follow-up questions where indicated) had a median completion time of under 2 minutes, and > 90% of patients were happy to complete the survey on an electronic tablet, did not feel that survey completion made their clinic visit more difficult, and did not find the questions upsetting or distressful. Conclusions: Two screener tools (for prospective observational studies “How are you currently eating?”, and for routine clinical monitoring “Do you experience any difficulty or pain upon swallowing?”) can effectively screen dysphagia symptoms without increasing cancer outpatient clinic burden.

e15715 Background: The standard second line treatment in advanced PDAC pts after m-FFX first line palliative chemotherapy is not clearly defined. G is the standard 2nd line treatment in Canada and GA can only be prescribed for pts with a private fund or private drug plan that covers the cost of Abraxane. More data is needed to establish OS achieved with 2ndline G or GA. Methods: The OS of PDAC pts treated with G or GA after m-FFX palliative chemotherapy given between 01-Dec-2011 and 30-Nov-2015 at the Princess Margaret Cancer Centre (PM), Toronto, Canada were retrospectively reviewed. OS was calculated from date of commencement of 2ndline treatment until death or date of last oncology follow up. Results: Over the 4-year study period, 132 pts were treated with 1st line palliative m-FFX at PM. Of them, at disease progression, 78 (59%) pts received a 2ndline therapy (50 G, 17 GA, 9 clinical trials, and 2 other). The results including demographics, treatment details and survival outcomes of patients treated with G and GA are summarized in the Table 1. Conclusions: Our results do not support single agent G after m-FFX 1st line palliative chemotherapy as median duration of treatment was too short and OS observed with G in this setting is very much limited. The median duration of treatment on GA and OS achieved with GA seem superior but sample size was too small to make a conclusion. [Table: see text]

2073 Background: Survival in stage IV lung cancer (S4LC) patients (pts) continues to improve, highlighting the importance of assessing health related quality of life (HRQoL). Lung cancer, involvement with brain metastases (BM), and systemic or brain-specific treatment can all impact neurocognitive function (NCF) and HRQoL. We evaluated the relationship between NCF and HRQoL in S4LC pts by BM status. Methods: S4LC pts with BM (BM+) were frequency distributionally-matched to pts without BM (NBM). NCF was measured using the Hopkins Verbal Learning Test – Revised (HVLT-R), the Controlled Oral Word Association Test (COWAT) and Trail Making Tests (TMT-A/B); scores were correlated with health utility score (HUS) data from EQ5D-3L surveys (Pearson Coefficient, R). Results: BM+ (n = 54) and matched NBM (n = 40) pts had similar demographics. The overall median age was 61years; 59% were female; of 89% that were adenocarcinomas, half had EGFR/ALK alterations; mean time since diagnosis was 2.6 years; mean time since BM were diagnosed in BM+ pts was 0.5 years. Mean HUS (mHUS) were similar between groups: 0.77 for BM+ vs. 0.78 for NBM; p = 0.86. However, pts with stable BM had higher HUS than those with progressive BM (mHUS: 0.80 vs 0.69; p = 0.045). Of BM+ pts, 44% had received whole brain radiation (WBRT). Correlations of NCF and HUS specific for BM+ pts were observed for several HLVT scores, including Total Recall (TR), which was correlated with HUS in BM+ (R = 0.35, p = 0.01) but not in NBM (R = 0.04, p = 0.84) and Recognition Discrimination Index (BM+: R = 0.32, p = 0.03 vs NBM: R = 0.13, p = 0.51). In contrast, TMT-A/B NCF test results had slightly stronger associations with HUS in NBM pts. COWAT was least associated with HUS in BM+ or NBM pts. In BM+ pts treated vs untreated with WBRT, HLVT scores were better in untreated patients (TR, p < 0.0001; delayed recall, p = 0.006; retention, p = 0.089), associations not seen with either TMT-A/B or COWAT. Mutation status had no bearing on these associations. Conclusions: NCF impacts HUS in S4LC pts and should be considered in treatment planning. HVLT scores are useful to assess specifically the impact of BM and WBRT in S4LC pts, and is reflected in associations with HRQoL.

e15776 Background: Prospective characterization of advanced PDAC using whole genome sequencing (WGS) and whole transcriptome sequencing (WTS) may identify patients (pts) who will benefit from a personalized treatment strategy. The feasibility of this approach has not previously been established in PDAC. Methods: Advanced PDAC pts with ECOG PS 0-1 and accessible metastatic or primary tumors, were recruited. A fresh tumor sample for WGS and WTS was obtained by percutaneous biopsy and a reference whole blood sample for germline DNA analysis collected before starting 1st line palliative chemotherapy. Tumor biospecimens were enriched by laser capture microdissection before genomic analysis. The primary endpoint was feasibility (to report WGS results within 56 days in 40 of the first 50 recruited pts.) with a plan to accrue 200 pts. Results: From Dec 2015 – Jan 2017, 42 pts with advanced PDAC safely underwent tumor biopsy (32 liver, 1 omentum, 1 adrenal, 8 primary tumor). Adequate tumor biospecimens were obtained in all but one pt. The median number of tissue cores was 5 (range 3-6). The median post-enrichment tumor cellularity was 77% (range 37-93). WGS was successful for all samples analyzed. In 41 of 42 pts (97.6%), WGS was feasible and results were reported within 56 days (Median 37 days; range 19 to 52 days) meeting the primary endpoint. Potentially actionable somatic genetic aberrations were found in 12 pts (29%) involving ARID1A (N = 4), PIK3CA (N = 1), PDGFRB (N = 1), ERBB4 (N = 1), ATM (N = 2), CDK4 (N = 1) and CDK6 (N = 2). One pt with a germline BRCA2 mutation and another with somatic DSBR mutation signature achieved partial response with FOLFIRINOX (FFX). WTS results were available for 35 pts (85%). Of those, 19 were evaluable for FFX response. Eight of 9 pts (88%) with ‘classic’ signature and 4 of 10 pts (40%) with ‘basal like’ signature had tumor shrinkage with FFX (P = 0.08). Conclusions: Prospective comprehensive genomic profiling of advanced PDAC is feasible with an acceptable turnaround time. Approximately 30% of advanced PDAC pts have targetable genomic signatures and aberrations. RNA signature may predict FFX response.

24 Background: Novel therapies have improved the survival of S4LC patients, particularly those with EGFR or ALK alterations. As BM are common among such patients, determining whether HUS appropriately capture BM-specific HRQoL is crucial in economic analyses. We evaluated the relationship between the FACT-Brain (Br) HRQoL, presence of BM, and HUS. Methods: This cross-sectional study of S4LC outpatients at a comprehensive cancer centre assessed FACT-Br and EQ-5D-3L-derived HUS. Correlational analyses, stratified by BM status, were performed between HUS and each of FACT-Br and its subscales: FACT-General (G), physical well-being (PWB), social well-being (SWB), emotional well-being (EWB), functional well-being (FWB), and brain cancer (BrC). Linear regression interaction models assessed whether BM modified the associations between FACT-Br or its subscales and HUS. Results: The 65 BM and 42 non-BM patients had similar demographics: median age (range) was 62 (30-83) years, 61% were female, 55% were Caucasian, and 54% were EGFR/ ALK+. Mean± standard error of the mean (SEM) values of HUS were similar between BM and non-BM groups (0.77±0.02 vs. 0.78±0.02; p=0.49). However, 40 BM patients with stable brain disease had higher HUS than the 14 with progressive disease (0.81 vs. 0.69; p=0.007). Mean±SEM values for FACT-Br, FACT-G, and BrC were 148±2.6, 79±1.5, and 69±1.3, respectively, with no differences between BM and non-BM groups. With the exception of SWB, FACT-Br and its subscales were each individually correlated with HUS (all p0.35). Conclusions: The FACT-Br HRQoL measures and the majority of its subcomponents moderately correlate with HUS but are not specific to patients with BM. Progression of CNS disease greatly alters HUS. EQ-5D-3L-derived HUS are a useful index of HRQoL in S4LC.

225 Background: As targeted therapies for S4LC pts have improved survival, the impact of health-related quality of life (HRQoL) in these pts grows in importance. In pts with BrM, loss of NCF may affect HRQoL significantly. We evaluated the relationship between NCF and HUS as a measure of HRQoL in S4LC pts. Methods: Self-reported HUS data from EQ5D-3L were obtained cross-sectionally from S4LC pts with (BrM) or without BrM (non-BrM). NCF was measured using the Hopkins Verbal Learning Test – Revised (HVLT-R), the Controlled Oral Word Association Test (COWAT) and Trail Making Tests (TMT-A/B). NCF scores were correlated with HUS (Pearson Coefficient, R). Results: BrM (n = 54) and non-BrM (n = 40) patients had similar demographics- overall median age was 61 (range 33-89) years; 59% were female; 45% had EGFR/ALK alterations; BrM pts were treated with whole brain (n = 20), stereotactic radiation (n = 19), both (n = 6), or sole use of systemic agents (n = 6); 3 were observed; 7 had BrM resection. Overall HUS were similar between BrM and non-BrM groups (mean HUS (mHUS): 0.77 vs. 0.78; p = 0.86). However, pts with stable brain disease had higher HUS than those with progressive brain disease (mHUS: 0.80 (n = 36) vs 0.69 (n = 17); p = 0.045). There was a trend towards lower HUS in symptomatic vs. asymptomatic BrM (mHUS: 0.70 (n = 10) vs. 0.78 (n = 43); p = 0.07). Multiple correlations between NCF scores and HUS were found. HLVT-Total Recall correlated with HUS in BrM but not non-BrM (R=0.35, p = 0.01; vs. non- R=0.04, p = 0.84 respectively) as did the HLVT-Recognition (BrM: R=0.32, p = 0.03 vs. non-BrM: R=0.13, p = 0.51). In contrast, TMT-A/B were associated with HUS in both BrM (p = 0.03, 0.06) and non-BrM pts (p = 0.001, 0.03). COWAT was associated with HUS only when all pts were analyzed together (p = 0.04). Conclusions: In S4LC pts, HUS were correlated with multiple measures of NCF. The impact of uncontrolled brain disease and poor NCF on HUS demonstrates that this measure has clinical utility, with important HRQoL implications as metastatic lung cancer pts live longer.

582 Background: Lynch Syndrome (LS) is caused by germline mutations in mismatch repair genes (MMR) genes which are critical in maintaining cellular integrity. Failure of the MMR pathway in LS culminates in the hypermutable phenotype of Microsatellite Instability. LS confers an increased risk of malignancy of which colorectal cancer (CRC) is most common. Carriers exhibit significant phenotypic variation in the age of onset of malignancy which cannot be predicted. Telomere length attrition is considered an early step in carcinogenesis and may be accelerated by oxidative stress. We investigated an association between relative telomere length (RTL) and levels of DNA oxidative damage in LS affected carriers (AC), unaffected carriers (UAC) and in patients with MMR- proficient colorectal cancer (MPC). Methods: Peripheral blood mononuclear cells were isolated from patients within each group. DNA was extracted and RTL measured by quantitative polymerase chain reaction (PCR). Real-Time PCR was used to quantitate expression levels of TERT, TERC and DKC1 (telomerase components) from RNA. Serum levels of 8-hydroxydeguanosine (8-OHdG) were measured from patients using the ELISA technique. Pearson’s correlation was used to compare mean RTL, telomerase levels and 8-OHdG between groups. Results: RTL and telomerase components were measured in 27 AC (median age 50yrs) 27 UAC (median age 40yrs) and 27 MPC (median age 66yrs). Corresponding RTLs were 0.89, 0.91 and 1.69 respectively. AC had significantly shorter RTL compared to UAC (p = 0.03) and MPC (p < 0.0001). There we no differences in the mean expression of TERT, TERC or DKC between groups. Younger age of tumour onset was associated with shorter telomere length in both AC (p = 0.0006) and MPC (p < 0.0001). 8-OHdG levels have been measured in 17 AC, 19 UAC and 14 MPC. The mean levels of AC and UAC were not statistically different. However the mean MPC level was significantly less than UAC (p = 0.03) and AC (p < 0.0001). Conclusions: Shortened telomere length is an important step in carcinogenesis. Affected LS patients have shorter telomeres and evidence of higher levels of DNA oxidative stress than patients with MMR-proficient CRC.

3574 Background: Colorectal cancers that show evidence of microsatellite instability (MSI-H) are marked by a high tumour infiltrating lymphocyte population (TiL) which is thought to be prognostic. ...

3571 Background: Lynch Syndrome (LS) accounts for 2-4% of all colorectal cancers (CRC) and is caused by germline mutations in DNA mismatch repair (MMR) genes. Increasing literature supports routine...

e12527 Background: Anticipation has been reported in Lynch Syndrome (LS) and linked to telomere length shortening. We recently investigated affected carriers (AC) and unaffected carriers (UAC) and ...

525 Background: Lynch Syndrome (LS) accounts for approximately 3% of all colorectal cancers (CRC) and is caused by germline mutations in DNA mismatch repair (MMR) genes. Increasing literature supports routine screening for LS using immunohistochemistry (IHC) to detect loss of MMR protein expression on tumour samples. We evaluated genetic clinic referrals and consequent LS diagnoses at our institution pre and post the adoption of universal screening. Methods: Reflex immunohistochemistry (rIHC) on all colorectal adenocarcinomas was implemented in November 2008. Prior to this IHC was performed at the pathologist’s discretion or upon request. Tumour specimens reviewed from November 2004 to October 2008 (pre rIHC) and from November 2008 to October 2012 (post rIHC) were evaluated and patients with evidence of MMR deficiency (MMR-D) were identified. The number of genetic referrals and number of patients with confirmed LS were determined. Results: During an 8-year period, 1,131 pathology specimens were reviewed in 1,103 patients (Table). Completion of IHC increased from 19% to 75% post implementation of rIHC. Ninety-two percent of incomplete samples in the post rIHC period (133/145) had insufficient tumour material for analysis. In patients with available IHC, 18/96 (20%) in the pre rIHC era were MMR-D of which 44% were referred to genetics, resulting in the diagnosis of LS in 4 patients. In the post rIHC era 30/429 (7%) of patients had evidence of MMR-D of which 20% were referred for genetic testing resulting in the detection of 3 LS patients. Conclusions: Universal testing for MMR protein expression is difficult to achieve, and does not necessarily result in increased genetic testing or Lynch Syndrome diagnosis. Systematic work-up of test results is essential once mismatch repair evaluation is initiated. [Table: see text]

1554 Background: Lynch Syndrome (LS) is an autosomal dominant cancer predisposition syndrome conferring a 60-80% lifetime risk of developing colorectal cancer in addition to a spectrum of other mal...

10530 Background: Chemotherapy in the treatment of osteosarcoma has improved 5 year overall survival (OS) from 20% with surgery alone to 60-70%. However, poor tumor necrosis following neoadjuvant chemotherapy (NAC) is associated with decreased survival, therefore strategies to improve outcomes are required for these patients. Methods: Records from all adult patients diagnosed with osteosarcoma between 1986 and 2012 were retrospectively reviewed. Patients were stratified according to age at diagnosis (40yrs), stage (localised or metastatic) and tumor necrosis post NAC (90%). All patients received 2 cycles of methotrexate alternating with cisplatin/doxorubicin (MAP) preoperatively. Following surgery, patients with >90% tumor necrosis continued MAP whilst those with 40 yrs (N=12) (p=0.05). 2/13 patients with metastatic disease at diagnosis are disease free >10 years post diagnosis. 65 of 73 patients with localised disease < 40 yrs had histology reviewed post neoadjuvant MAP. 34/65 (52%) had >90% tumor necrosis and continued on MAP, 5 yr OS 79%, 31 patients (48%) had

e12533 Background: Cancer in pregnancy accounts for ~1 in 1,000 pregnancies. Studies show that cytotoxic agents are safe from the second trimester. Long-term follow up has not shown increased malformations or malignancies in children exposed to chemotherapy in utero. There is no evidence of worse outcomes among women diagnosed in pregnancy. Methods: We retrospectively identified women diagnosed with cancer in pregnancy over a 25-year period. Medical records were reviewed for demographics, diagnosis, gestation, timing of treatment and outcomes. We assessed if all cancers need to be treated in pregnancy or if treatment could be safely deferred to allow normal delivery. Results: Twenty-five women were diagnosed with cancer in pregnancy and referred to medical oncology. Of 25 women, 16 (64%) received chemotherapy during pregnancy. These included 13 cases of breast cancer, one Ewing’s sarcoma, one ovarian cancer, and one small cell of cervix. All 16 women received doxorubicin and cyclophosphamide. There were 15 live births and no abnormalities seen in children who received chemotherapy in utero. At a median follow-up of 6 years 11 mothers (69%) are disease free and 4 (25%) have recurrent disease. Of nine mothers who did not receive chemotherapy in pregnancy, seven received chemotherapy immediately post-partum. Six (86%) were diagnosed in early pregnancy (median gestation 13 weeks). There were three cases of Hodgkin lymphoma, two breast cancers, and one ovarian cancer. At a median follow-up of 12 years, all mothers remain disease free. There were no abnormalities seen in these children. Conclusions: We did not identify any adverse outcomes in mothers or infants exposed to chemotherapy during pregnancy. We identified a cohort of patients that do not need immediate treatment during pregnancy. In selected cases, it is safe and appropriate to delay chemotherapy until delivery of the baby. There were no adverse outcomes to mothers due to delayed treatment and no adverse outcomes to babies not exposed to chemotherapy in utero. A multi-disciplinary team is essential to individualize treatment planning. [Table: see text]