Pancreatic Adenocarcinoma Pts Research Articles

Comparative analysis of the targetable landscape in KRAS-mutant and wild-type pancreatic adenocarcinoma.

4155 Background: Pancreatic adenocarcinoma (PC) is the fourth leading cause of cancer deaths, with increased incidence among patients (pts) younger than 50 years old (yo). Small cohort studies suggest early-onset PC (EOPC, diagnosis at < 50 yo) tumors may have a unique biology, harboring a higher proportion of KRAS wild-type status ( KRASWT) and enrichment of targetable mutations/fusions compared to non-EOPC. In addition, therapy targeting oncogenic fusions in KRASWT tumors has shown meaningful responses in PC pts. Here, we investigate the prevalence of fusions, mutations, and homologous recombination deficiency (HRD) in KRASWT PC to identify potential targets for therapeutic intervention, and compare EOPC to non-EOPC to better characterize EOPC differences. Methods: De-identified records from 4,956 PC pts with tumor biopsies sequenced using the Tempus xT solid tumor assay (DNA-seq of 595-648 genes at 500x coverage; full transcriptome RNA-seq) were retrospectively reviewed. Fusions were detected from RNA-seq data via the Tempus bioinformatics pipeline. Mutations identified included germline and/or somatic single-nucleotide variants and insertions/deletions. HRD status was determined from RNA-seq data by the Tempus HRD test. Significance was determined as FDR-corrected p-values < 0.05. Results: Across the entire cohort, 21% tumors were KRASWT. HRD was more frequent in KRASWT (7.9%) compared to KRASMUT (3.1%) tumors ( P< 0.001). Significant somatic mutational differences between the KRASWT and KRASMUT cohorts included BRAF (5.0% vs 0.2%) , CDKN2A (5.4% vs 25%), ARID1A (3.6% vs 8.6%), CTNNB1 (1.3% vs 0.3%), and TSC2 (1.2% vs 0.1%) (all P< 0.001). Actionable rearrangements were enriched in the KRASWTcohort (10% vs 2.1%, P< 0.001); the most common fusion partners include NRG1, MET, RAF1, BRAF, NTRK1-3, FGFR 1-4, and RET. TMB-high (10 muts/Mb) and MSI-high were more common in KRASWT (2.9% and 1.0%) compared to KRASMUT (1.7% and 0.4%) tumors ( P= 0.015). Additionally, 382 pts were classified as EOPC. KRASWT tumors comprised 30% of EOPCs, 22% of the 50-70 yo cohort, and 17% of the > 70 yo cohort. Germline (pathogenic/likely pathogenic) alterations in BRCA1 and BRCA2 were more frequent in EOPCs (2.0% and 4.5%) compared to the 50-70 yo (0.4% and 1.7%) and > 70 yo (0.4% and 0.8%) cohorts ( P= 0.015 and < 0.001, for BRCA1 and BRCA2 respectively). There were no statistical differences in actionable genes detected in EOPC vs non-EOPC. HRD was found at higher frequency in EOPC (5.4%) and decreased in age cohorts 50-70 yo (4.7%) and > 70 yo (2.5%) ( P = 0.047). Rearrangements were more common in EOPC compared to non-EOPC (10% vs 1.1%, P< 0.001). Conclusions: HRD, TMB-H/MSI-H and oncogenic rearrangements are more prevalent in KRASWT PC when compared with KRASMUT. These molecular analyses may provide additional therapeutic options for PC pts, warranting comprehensive genomic and transcriptomic profiling for this population.

A multicenter telemonitoring-telecare study with automatic assessment of physiological parameters and patient-reported outcomes in remote pancreatic cancer patients on mFOLFIRINOX: Interim technology report.

e13617 Background: There is an unmet need for improved treatment outcomes along with safer care of patients (pts) at home, at risk of severe adverse events from chemotherapy. Methods: MultiDom (NCT04263948; Ramsay Santé Research GCS) aims to reduce emergency hospitalisations of pancreatic adenocarcinoma pts on mFOLFIRINOX with a novel telemonitoring-telecare platform (Clinikali). Accelerometry and body temperature are measured q1-min using a continuously-worn telecommunicating chest surface sensor, daily body weight is self-measured with a telecommunicating balance, and 23 electronic pt-reported outcomes (ePROs) are self-rated using a tablet for 1 week before (baseline) and 6 weeks after the first course. Hidden Markov, spectrum and other algorithms automatically compute physical activity, sleep, temperature, body weight change, ePROs severity, and dichotomy index I<O of the circadian rest-activity rhythm (% activity In-bed below median activity Out-of-bed), one to four times daily. Alerts are automatically generated and sent to a coordination platform of health professionals, with trackable follow-up. Results: From 6/2021 to 2/2022, 16 of 42 planned pts (38%) from 4 centers have been included (females/males, 10/6 pts; aged 64-79 years; PS 0/1, 8/8 pts; 2 borderline/locally-advanced, and 14 metastatic (liver, 9; lymph nodes, 8; lung, 4). This technology report involves 9 pts on study for a median duration of 57 days (d) (range, 49-70d), whilst receiving 3 mFOLFIRINOX courses, with tumor control in 8 pts. Nearly 1.2x107 data were teletransmitted. At baseline, rest-activity rhythm was robust for 7 pts (I<O >97.5%) and disrupted for 2 pts (I<O <96%). Median compliance rates per pt (range) were 81% (67-92%) for rest-activity and body temperature, 96% (53-100%) for body weight, and 89% (78-100%) for self-rated e-PROMs. The median number of alerts generated for each pt (range) was 3 (0-10) for disrupted rest-activity rhythm (I<O <96%) or high temperature (2.5oC above median), and 2 (0-9) for body weight loss >5%, A median of 6 alerts per pt was generated for one or more severe ePROs (pain, 7 pts; fatigue, 6 pts; impact on activity, 5 pts; anorexia, 4 pts). Supportive treatment was mostly initiated at home, avoiding undue hospitalization. Pts rated the median global evaluation factor as 5 (4-5), i.e. the highest score, in the 23-items research participation questionnaire. Conclusions: Continuous multidimensional telemonitoring of physiology and e-PROs met excellent patient compliance, thus supporting exploration of biological mechanisms linking e-PROs to circadian rhythms and treatment activity. Study is ongoing to assess whether such telemonitoring-telecare platform may drive practice-changes in patient pathways and digitized clinical trials in the current era of precision medicine. Clinical trial information: NCT04263948.

Phase I study of the combination of alisertib (MLN8237) and gemcitabine in advanced solid tumors.

3589 Background: Aurora Kinase A (AKA) is a key mitotic regulator overexpressed in multiple solid tumors. This open-label dose escalation and expansion phase I study evaluated the safety and tolerability of alisertib (MLN8237), an oral AKA inhibitor, in combination with gemcitabine. Methods: In dose escalation, patients (pts) > 18y with refractory solid tumors received 28-day cycles of gemcitabine on days 1, 8, 15 and alisertib twice daily on days 1-3, 8-10, and 15-17. Gemcitabine was given at 1000mg/m2. Four dose levels (DL) of alisertib (20-50mg) were given per 3+3 design to investigate dose limiting toxicities (DLT) in cycle 1, to determine maximum tolerated dose (MTD) and recommended phase II dose (RP2D). In dose expansion, advanced pancreatic adenocarcinoma pts received the MTD dose twice daily on a modified dosing schedule to allow for pharmacokinetic (PK) evaluation. Anti-tumor activity was assessed by response rate (RECIST 1.1) and progression-free survival (PFS). PK evaluation of plasma gemcitabine and alisertib was performed on all pts enrolled in the dose expansion. PK sampling was performed before treatment, immediately after gemcitabine infusion, and at other pre-specified post-infusion timepoints. Results: Twenty-six pts were treated in total: 21 pts in dose escalation and 5 pts in dose expansion. Overall, median age was 57y [42-82]; 50% male; 62% PS 1 (16 pts); 2 [0-7] median prior therapies. In the dose escalation phase, 9 tumor types were included and NSCLC was most common (7 pts). Maximum administered dose (DL4) achieved 900 mg alisertib per cycle and was tolerated (1 DLT in 6 pts). The dose expansion phase enrolled 5 pts with advanced pancreatic adenocarcinoma; median age 63y [48-82]; 60% male; 60% PS 1 (3 pts); 2 [1-2] median prior therapies. Grade ≥3 TRAEs were observed in 73% of all pts and were predominantly hematologic, including neutropenia (54%), leukopenia (50%), and lymphopenia (31%). Similar TRAEs were seen at DL4; all 14 pts experienced neutropenia with 64% experiencing grade ≥3 neutropenia. Fourteen of 23 evaluable pts (61%) had stable disease and 2 pts (9%) had partial response (PR) as best overall response. Median PFS was 2.9 months (95% CI 2.0-4.2). Analysis of PK data is ongoing and will be reported. Conclusions: Alisertib can be safely administered with gemcitabine. RP2D for alisertib is 50 mg PO BID in combination with full dose gemcitabine. Best response was at least stable disease in a majority of pts with PR observed in 9% of this heavily pretreated group of patients. Most grade ≥3 TRAEs were hematologic. Results of PK studies will also be reported. Clinical trial information: NCT01924260 .

Landscape of germline mutations in 1144 patients (Pts) with gastrointestinal (GI) cancers.

e13660 Background: The efficacy of PARP inhibitors in germline BRCA-mutated pancreatic adenocarcinoma (PC) and immune checkpoint inhibitors in dMMR colorectal cancer (CRC) shows the importance of genetic testing. We aimed to characterize the frequency of pathogenic/likely pathogenic germline variants (PLPVs) in GI cancer pts. Methods: A retrospective review of pts referred to the Levine Cancer Institute Genetics Program was conducted. Genetic testing used a focused hereditary cancer 4-43 gene panel or pan-cancer 82-84 gene panel. Results: Out of 1144 GI cancer pts seen between 2010 and 2019, 869 underwent germline testing, and 199 (23%) pts had at least one PLPV in a hereditary cancer susceptibility gene, while 253 (29.3%) had a variant of uncertain significance. Of 630 CRC pts, 24% had a PLPV and 13% harbored a germline mutation in DNA MMR genes and were diagnosed with Lynch Syndrome, representing ~50% of all pts with a PLPV. Other germline PLPVs were found in APC, ATM, BRCA1, BRCA2, CHEK2, MUTYH, and PALB2. Of 163 PC pts, 16.6% had a PLPV in ATM, BRCA2, CDKN2A, and MEN1. Gastric cancer pts (17%) had germline PLPVs in APC, BRCA2, CDH1, MLH1, and MSH2; biliary cancer pts (17%) had germline PLPVs in PALB2, RAD50, and PTCH1; and GIST pts (60%) had PLPVs in SDHA or SDHB. Conclusions: Germline mutations were found in 23% of GI cancer pts, underlining the importance of multigene germline testing. Knowledge of inherited GI cancer risk helps determine the likelihood of benefit from possible specific targeted therapies. Genetic testing and counseling pose a challenge, but implications for pts with hereditary syndromes are highly significant. [Table: see text]

Homologous recombination deficiency (HRD) by BROCA-HR and survival outcomes after surgery for patients (pts) with pancreatic adenocarcinoma (PC): A single institution experience.

732 Background: 5-7% of PC pts exhibit deleterious germline mutations (MUT) in HR tumor suppressor genes BRCA1 and BRCA2. BROCA-HR is a targeted capture and massively parallel sequencing assay designed to detect all mutation classes including gene rearrangements, copy number variations, and gene aberrations within the Fanconi Anemia-BRCA HR, non-homologous end joining (NHEJ) DNA repair, and DNA mismatch repair pathways. BROCA-HR has been successfully used in breast and ovarian cancer pts for overall prognosis and prediction of response to platinum-based therapies. While BRCA1/2 MUT may confer survival advantage for PC pts if treated with platinum-chemotherapy, the survival impact of HRD is less well defined. Methods: We retrospectively identified 100 consecutive pts who underwent surgical resection for suspected PC at University of Washington Medical Center between 1999 and 2008. Formalin-fixed paraffin embedded resected tumors were sequenced using BROCA-HR. HRD was grouped based on the following deleterious genetic mutations: 1) BRCA1, BRCA2; 2) core HRD: BARD1, BRIP1, RAD51C, RAD51D, PALB2, CDK12, NBN; 3) non-core HRD: ATM, ATR, ATRX, BAP1, BLM, CHEK1/2, ERCC, FANC A/C/D2/E /F/G/L, MRE11, RAD50/51/51B, RIF1, SLX4; 4) HR proficient. Overall survival (OS) was measured from diagnosis until death or last follow-up. Results: 95 pts had histologically confirmed PC, and 81 pts had adequate tumor DNA for analysis. Six pts (7%) had BRCA1/2 MUT (n = 5), or BRCA1 methylation (n = 1), 1 pt (1%) had non-BRCA core HRD ( PALB2 MUT), 7 pts (9%) had non-core HRD: ERCC (2), CHEK2 (2), ATR, RAD51D, and FANCA MUT (1 each). Median OS was: all pts 1.93 yrs (95% C.I. 1.53, 2.16), BRCA1/2 pts 3.09 yrs (95% CI 0.41, 12.21), all core HRD pts 1.21 yrs (95% CI 0.41, 12.21), all core and non-core HRD pts 1.89 yrs (95% CI 0.57, 4.96), HR proficient pts 1.93 yrs (95% CI 1.51, 2.15). There were no OS differences between pts with HRD vs those HR proficient. Conclusions: HRD is common (17%) but does not affect OS for pts with resected PC. Prospective clinical trials should test neo/adjuvant therapies including platinum chemotherapy and PARP inhibitors for pts with HRD.

Neoadjuvant therapy (NAT) and its role for pancreatic adenocarcinoma (PC) in the current era: Institutional experience.

672 Background: PC affects 57,000 people in the U.S. annually with poor long-term outcomes. NAT for localized disease has increasingly been used but lacks robust prospective data. We investigated the disease course and outcomes for patients (pts) undergoing NAT versus upfront surgery for PC at a high-volume academic center. Methods: Utilizing our IRB-approved retrospective database of metastatic PC pts (year 2000-2017), we identified pts who presented with localized disease and were considered for surgery and present the baseline tumor and treatment characteristics here. Fisher’s exact and Wilcoxon Rank-Sum tests were used for categorical data and Kaplan Meier (KM) curves for survival data when comparing those who had upfront surgery versus surgery following NAT. Results: 352 pts with localized disease at diagnosis were included in our analysis with median age of 65 y (range 38-89) and 45% females. NAT was used in 225 (64%) pts while 109 pts (31%) had upfront surgery and 18 pts (5%) received no treatment. Adjuvant therapy was given to 77% of pts after upfront surgery and 48% of pts after surgery following NAT. NAT regimen consisted of chemotherapy (CTx) and radiation for 48%, CTx alone for 8% and radiotherapy alone for 44% of pts. Of those receiving CTx, 24% received triple agent while 51% and 25% received dual and single agent therapy. Pt factors (age, CCI, gender, BMI, smoking status, race) did not differ between those receiving upfront surgery and surgery following NAT but upfront surgery was associated with a lower stage at diagnosis (p < 0.0001). Surgical resection after NAT occurred in 79 pts (35%) with median overall survival of 26.3m vs 19.7m (p = 0.06) in those who had upfront surgery. Survival rates at year 1, 3, and 5 years were 94%, 34%, and 8% for those with NAT followed by surgery vs 76%, 17%, and 11% for those with upfront surgery (p = 0.06). Conclusions: Use of NAT is prevalent, yet only 35% of pts make it to surgical resection. Survival was improved for pts who underwent resection following NAT versus upfront, although the difference was not statistically significant. Additional research is warranted to define the optimal NAT approach for pts with borderline resectable PC.

Oncology precision medicine for hepatobiliary and pancreatic cancer: Insights and updates upon an institutional review.

570 Background: Hepatobiliary cancers - hepatocellular carcinoma (HCC), intra or extrahepatic cholangiocarcinoma (I/EC), and gallbladder carcinoma (GB) - and pancreatic adenocarcinoma (PC) do have actionable alterations (AA). The importance of testing early in a patient’s (pts) course to identify oncology precision medicine (OPM) options could be paramount for progression free survival (PFS). Methods: We identified pts with HCC, IC, EC, GB or PC in our OPM database since the centralization of our system. Pts who underwent molecular panel testing had AA’s identified and stratified by cancer type. Treatment course was analyzed using swimmers plots. Results: 456pts were diagnosed with HCC, IC, EC, GB or PC. 104/456pts (23%) were ordered for molecular testing and 88/456pts (19.3%) completed testing: 18/88pts (20.4%) I/EC, 2/88pts (2.3%) HCC, 5/88pts (5.7%) GB, and 63/88pts (71.6%) PC. Of the PC pts, 3/63 (4.8%) had a BRCA mutation. These pts did not receive targeted therapy. Overall, 5/88pts (5.7%) had a BRAF mutation (2 PC, 2 I/EC, 1GB). Thus, 8/88 (9.1%) of tested pts became eligible for some form of targeted therapy over their treatment course. Of those with a BRAF mutation, only 2/5 pts had OPM testing sent with initial diagnostic workup, and 2/5 eventually began targeted therapy. One had a progression free survival (PFS) of 2.5months while the other discontinued secondary to toxicity. Conclusions: Our data showed that we are testing a minority of pts with pancreas and hepatobiliary cancers. Of those tested, it may have occurred too late in the course of illness to improve outcomes. Given the potential utility of uncovering potential germline alterations like BRCA1/2 as well as pragmatic AAs including somatic BRCA and BRAF, we are moving to a more systematic evaluation of pts to capture and respond to these issues. [Table: see text]

Improved surgical margins with neoadjuvant versus adjuvant chemotherapy in clinical stage I resectable pancreatic adenocarcinoma: A National Cancer Database study.

651 Background: A strengthening consensus exists for neoadjuvant therapy (NAT) in borderline resectable pancreatic adenocarcinoma (PA), but the utilization of NAT in resectable stage I PA remains controversial. Many cancer centers are using NAT for these patients (pts), but others continue to offer upfront surgery and adjuvant therapy (AT). We hypothesized that NAT would improve margin negative resection in clinical stage I resectable PA. Methods: We utilized the IRB approved 2016 national cancer database for pancreas to establish a cohort of stage I PA pts. We divided this subset into pts who underwent NAT vs AT. We compared demographics. Primary endpoint was surgical margins. Results: 10,453pts from 2004 to 2016 had clinical stage I resectable PA: 8483pts (81.1%) AT and 1970pts (18.9%) total or partial NAT. There was a statistical difference in age (64.9 ± 9.9years NAT and 66.2 ± 9.9years AT, p<0.001), but no difference in Charlson comorbidity score (p=0.1693). NAT pts had significantly higher margin negative resection rates (84.5%) than AT pts (79.4%) (p<0.0001). Final pathologic staging was available for 10,237 pts: 8369 (81.8%) AT and 1868 (18.2%) NAT. Significantly fewer pts were upstaged on final pathology to stage II or greater (73.5%) in the NAT group than the AT group (84.1%) (p<0.001). Conclusions: NAT leads to significantly higher margin negative rates for resectable clinical stage I PA than surgery followed by AT. The majority of pts for both groups were upstaged suggesting that we continue to clinically understage the majority of pts. Overall, total or partial NAT for clinical stage I resectable PA provides a better chance for margin negative resection. Further study in the form of a randomized control trial is necessary. [Table: see text]

Impact of no treatment vs other nonsurgical treatments in Pancreatic Adenocarcinoma National Cancer Database: 2004-2014.

e15795 Background: Surgery remains the only curative treatment option in pancreatic adenocarcinoma, yet in more than 50% of the patients (pts) the disease is too advanced or at very high risk and considered inoperable. They are either managed with no treatment or other nonsurgical methods. Hence, we analyzed a large cohort of pancreatic adenocarcinoma pts undergoing No Treatment vs Chemotherapy, Radiation or Chemoradiation to determine their impact on survival. Methods: Only pancreatic adenocarcinoma pts who had no surgery in the National Cancer Database (NCDB) from 2004–2014 were included. Of that group, patients with unknown or missing data about chemotherapy or radiation treatment or less than 3 years of survival data were excluded. Pts were stratified into 4 groups: Chemotherapy, Radiation Therapy, Chemoradiation and No Treatment. Overall 1-, 2- and 3-year survival was calculated and the groups were compared using Pearson’s chi-squared. Results: Of the total 309,709 pancreatic cancer pts in the NCDB 2004–2014, 111,421 (36.0%) remained after application of the study criteria. Of these, 43,203 (38.8%) received chemotherapy only, 2,453 (2.2%) received radiation only, 15,764 (14.1%) received chemoradiation and 50,001 (40.0%) had no treatment. Overall survival for 1, 2, and 3 years was best in the chemoradiation group with a 1 year survival of (40.0%) compared to chemotherapy only (22.4%), radiation only (14.9%) and no treatment (9.6%). Overall, only (19.0%) of pts survived for 1 year, (5.4%) survived for 2 years and (2.3%) survived for 3 years. (Table) Conclusions: Survival in pancreatic adenocarcinoma pts remains dismal without surgery. Best survival in nonsurgical pts was seen after combination Chemoradiation therapy and worst survival in No Treatment group. Hence, whenever possible, a combination Chemoradiation should be offered even as palliation in non-surgical pancreatic adenocarcinoma pts.[Table: see text]

Outcomes in pancreatic adenocarcinoma (PDA) patients (pts) with genetic alterations in DNA damage repair (DDR) pathways: Results from the Know Your Tumor (KYT) program.

191 Background: Up to 17% of PDAs harbor mutations in the DDR pathway. However, the purely prognostic relevance of these mutations is unclear. Furthermore, outcomes in response to platinum-based therapies in PDA pts harboring mutations in a broad range of DDR genes, particularly beyond BRCA1/2 and PALB2, remain unexplored. Methods: We evaluated PDA pts enrolled in the KYT registry for whom we collected cancer related exome sequencing and clinical outcomes. Pts were categorized as resected and advanced (LAPC and metastatic pts), and tumor genomic profiles were categorized as DDR mutated (DDRmut) based on the presence of pathogenic alterations in BRCA1/2, PALB2 (Group 1), ATM, ATR, ATRX (Group 2), or BAP1, BARD1, BRIP1, CHEK1/2, RAD50/51/51B, or FANCA/C/D2/E/F/G/L (Group 3). Tumors harboring no DDR mutations were labelled DDR proficient (pDDR). Median overall survival (OS) was measured from the date of diagnosis until death. Results: The OS was similar in all resected pts, irrespective of exposure to platinum therapy (see Table). However, for the pts with advanced disease, OS was significantly longer for DDRmut vs. pDDR pts, particularly in the platinum-treated group; but no such difference was identified in the platinum-naïve pts. Detailed outcomes for the 3 Groups will be presented, but in general the OS in pts with mutations in all 3 DDRmut Groups was greater than for the pDDR pts; but again this difference was lost in the platinum-naïve pts. Conclusions: Advanced DDRmut pts have an improved OS when treated with platinums, compared to pDDR pts. But, in the absence of platinum-based therapy, there is no OS difference observed in DDRmut vs. pDDR pts, suggesting that DDR status has no pure prognostic value. These findings support the need to test all pts with advanced PDA, to ensure that DDRmut pts are treated with platinum-based therapy. [Table: see text]

Investigating disparities: The effect of social environment on pancreatic cancer survival.

208 Background: Incidence rates of pancreatic adenocarcinoma (PAC) are higher in Black compared to White patients (pts). Beyond race, exposure to poor neighborhoods or social environments also contribute to cancer disparities. However, social environmental effects on PAC are not well-studied. Social environment is defined as: 1) a neighborhood’s socioeconomic conditions (SES; i.e. a combination of education, income, poverty and employment levels); 2) racial segregation (RS) or the degree of isolation/separation of race/ethnic groups in a neighborhood. The goal of our study is to investigate whether poor social environments impact survival in a clinic population with metastatic PAC. Methods: Neighborhood SES and RS measures were derived from US census data (2011-2015) at the census tract level, which is a smaller geographic area than a county. Patient (pt) addresses were used to generate a geocode that identifies the census tract or neighborhood in which a pt lives. We joined medical records of PAC pts (n = 374; diagnosed 2010-2016 at Fox Chase Cancer Center) to neighborhood measures via the geocode. Pt variables included in the analysis were: age, sex, race, marital status, treatment, PAC family history, stage, Jewish ancestry, tobacco use and Charlson comorbidity index. Multivariable cox proportional hazards models with cluster adjustments were used and variables with p-values < 0.05 were considered significant. Results: 342 PAC deaths occurred and median survival was 12m. 81% of pts were White; < 40% resided in poor social environments (i.e. low SES or high RS). In multivariable analyses stratified by RS, median survival was lower in pts from high RS (11m) vs low RS areas (13m); however, this difference was not significant (p = 0.27). Variable effects differed by high/low RS. In high RS areas, sex, surgery, chemo, diabetes and neighborhood SES were significant predictors of survival; in low RS areas, surgery, chemo, radiation, PAC family history, tobacco use, Jewish ancestry and race were significant. Conclusions: While social environment did not appear to significantly affect survival time in metastatic pts, its potential moderating (interaction) effects on variable associations with PAC warrant further investigation.

Automation of quantifying and mapping of stromal and malignant cells in pancreatic adenocarcinoma using deep learning.

266 Background: The characteristic histological feature of pancreatic adenocarcinoma (PAD) is extensive desmoplasia alongside leukocytes and cancer-associated fibroblasts. Desmoplasia is a known barrier to the absorption and penetration of therapeutic drugs. Stromal cells are key elements for a clinical response to chemotherapy and immunotherapy, but few models exist to analyze the spatial and architectural elements that compose the complex tumor microenvironment in PAD. Methods: We created a deep learning algorithm to analyze images and quantify cells and fibrotic tissue. Histopathology slides of PAD patients (pts) were then used to automate the recognition and mapping of adenocarcinoma cells, leukocytes, fibroblasts, and degree of desmoplasia, defined as the ratio of the area of fibrosis to that of the tumor gland. This information was correlated with mutational burden, defined as mutations (mts) per megabase (mb) of each pt. Results: The histopathology slides (H&E stain) of 126 pts were obtained from The Cancer Genome Atlas (TCGA) and analyzed with the deep learning model. Pt with the largest mutational burden (733 mts/mb, n = 1 pt) showed the largest number of leukocytes (585/mm2). Those with the smallest mutational burden (0 mts/mb, n = 16 pts) showed the fewest leukocytes (median, 14/mm2). Mutational burden was linearly proportional to the number of leukocytes (R2 of 0.7772). The pt with a mutational burden of 733 was excluded as an outlier. No statistically significant difference in the number of fibroblasts, degree of desmoplasia, or thickness of the first fibrotic layer (the smooth muscle actin-rich layer outside of the tumor gland), was found among pts of varying mutational burden. The median distance from a tumor gland to a leukocyte was inversely proportional to the number of leukocytes in a box of 1 mm2 with a tumor gland at the center. Conclusions: A deep learning model enabled automated quantification and mapping of desmoplasia, stromal and malignant cells, revealing the spatial and architectural relationship of these cells in PAD pts with varying mutational burdens. Further biomarker driven studies in the context of immunotherapy and anti-fibrosis are warranted.

716P - Predictive Role of Neutrophil/Lymphocyte Ratio (Nlr) for Oxaliplatin Efficacy in Metastatic Pancreatic Cancer Patients (Pts)

ABSTRACT Aim: To investigate the prognostic and predictive value of the inflammatory index Neutrophil/Lymphocyte ratio (NLR) in two homogeneous groups of chemotherapy-naive metastatic pancreatic adenocarcinoma (PA) pts treated with either gemcitabine (GEM) or GEM + oxaliplatin (GEMOXA). Methods: Consecutive PA pts (n = 103, male:female = 53:50) treated at a single center with either GEM (n = 45) or GEMOXA (n = 58), mainly according to patient preference or pre-existent diabetic neuropathy, were included between April'08- October'12. Exclusion criteria were treatment with steroids or active infection. NLR was assessed before and during therapy and correlated with outcome together with common clinical and biochemical variables. Results: Among 17 analyzed variables NLR, Karhofsky Perfomance Status (KPS), d-dimer and erythrocyte sedimentation rate were found to be significantly associated with median Overall Survival (mOS) at the univariate analysis. Only NLR and KPS were independent prognosticator at multivariate analysis, with NLR displaying the highest statistical significance (Exp(b) = 1.09, p 0.01, meaning 9% increase in the risk of death for 1-unit increase in NLR). GEMOXA was associated with longer mOS, but this was not statistically significant (8 v 5.8 months (mo), p 0.07). NLR was also predictive of oxaliplatin activity, as only pts with NLR > 2.5 (cutoff determined upon ROC analysis) derived benefit from GEMOXA over GEM (mOS 9.7 v 3.9 mo, respectively, in pts with NLR > 2.5, p 0.005 and 11.7 vs 11.8 mo, respectively, in pts with NLR 2.5 who reported a NLR 2.5 (7.7 v 5.0 mo, p 0.003). Conclusions: We were able to confirm the independent prognostic value of NLR for metastatic PA pts treated with gemcitabine-based first-line chemotherapy. NLR had also a predictive value in that only pts with high NLR (worse outcome) seem to benefit from the addition of oxaliplatin. NLR and KPS may help select pts for whom a particularly poor prognosis might justify more intensive, yet less tolerable, combination regimens. Disclosure: All authors have declared no conflicts of interest.

Use of EGF A61G polymorphism to predict overall survival in a phase III study of gemcitabine plus cetuximab versus gemcitabine in patients with locally advanced or metastatic pancreatic adenocarcinoma (SWOG 0205).

203 Background: The SWOG 0205 phase III study failed to demonstrate an advantage from the addition of cetuximab to gemcitabine for overall survival (OS), progression-free survival (PFS) or objective response in advanced pancreatic cancer. Molecular markers may identify subgroups of pts who benefit from this regimen. This report summarizes the results of a pilot project aimed at screening germline single nucleotide polymorphisms (SNPs) in genes involved in gemcitabine metabolism pathway (CDA and RRM1) and EGFR signaling pathway (EGF, EGFR, IGF1, FCGR2A/3A, KRAS, IL8, COX-2, CCND1) to identify pts that may benefit from this treatment. Methods: Genomic DNA was extracted from 88 available serum samples for this molecular correlates pilot study. Sixteen SNPs were evaluated using PCR-based protocols. There were no differences in baseline characteristics or outcome between pts included in the analysis and the entire study population. Associations between genotype and OS were assessed using stratified Cox regression, with adjustment for treatment arm. Genotypes were parameterized as the number of dominant alleles. No p value adjustments were made for multiple comparisons. Results: Median age was 63.3 years (range: 29.6-85.0), and median OS (95% CI) was 5.7 months (mos) (range: 4.4-6.8) in this subset. There was a statistically significant association between EGF A61G SNP (rs4444903) genotype and OS (p=0.04). Among pts carrying the AA genotype, median OS was 8.4 mos (95% CI 4.9-15.0), as compared to 5.6 mos (95% CI 3.0-9.0) among those with the AG genotype and 5.3 mos. (95% CI 2.2-7.1) for the GG genotype. Evidence was weak for interaction between treatment arm and rs4444903 genotype (p=0.13). No significant associations between other SNPs and OS were observed. Conclusions: This exploratory study suggests that the EGF A61G polymorphism may be a useful molecular marker for predicting clinical outcomes in metastatic pancreatic adenocarcinoma pts treated with gemcitabine-based chemotherapy.

Maintenance therapy with capecitabine in patients with resected pancreatic adenocarcinoma after adjuvant therapy.

333 Background: The 5-year disease free survival of pancreatic adenocarcinoma (PAC) with surgery alone is below 10%, and adjuvant chemotherapy increases that to about 20%. The original GITSG adjuvant study demonstrating a survival benefit compared to surgery could be attributed to the use of 2-years of weekly IV bolus 5FU, and not chemoradiation. This hypothesis has not been tested since that trial. In theory, the prolonged exposure to therapy could maintain pressure on dormant cancer cells that may remain in G0 arrest, by attacking them as they infrequently enter G1/S phase. To evaluate this hypothesis, we retrospectively evaluated our patient (pts) who were treated with maintenance capecitabine (cape). Methods: Pts in the Georgetown University Hospital/Lombardi Cancer Center electronic medical record database since Oct 2007 were sought for PAC that were resected with curative intent, received standard adjuvant chemotherapy with or without chemoradiation, and received maintenance cape for at least 2 months. Results: Among 214 PAC pts that have been treated at our institution in this time, 21 pts met these criteria. Among the 21 pts, 1 was lost to follow up after moving overseas, and 20 pts were eligible for this analysis. 13 of the 20 pts had lymph node involvement at resection. Cape was usually given 1000mg orally twice a day, Monday through Friday following adjuvant therapy, for an indefinite period, most often two years. Pts received cape for median duration of 12.5 months (2 to 24 months), and the median followup duration was 33 months (16 to 78 months). The median overall survival (OS) for the 20 analyzed patients was 48 months. The 2 year OS was 94%, and 5 year OS was 40%. The median recurrence free survival (RFS) was 39 months. The 2 year RFS was 67%, and the 5 year RFS was 25%. Common toxicities were mild hand-and-foot syndrome and fatigue. 4 pts discontinued cape due to toxicities: febrile neutropenia, severe fatigue, weight loss and diarrhea respectively. Conclusions: Cape maintenance therapy following adjuvant chemotherapy in resected PAC is associated with a promising higher overall survival and PFS compared to historical data, and this approach should be further studied in a randomized controlled study.

A matched analysis comparing the epidemiology of intraductal papillary mucinous neoplasms to standard pancreatic adenocarcinoma and healthy controls.

173 Background: The epidemiology of intraductal papillary mucinous neoplasms (IPMNs) of the pancreas is poorly defined. Methods: An epidemiologic questionnaire was administered to patients (pts) with IPMN (n=79), pancreatic adenocarcinoma (PC) (n=689) and healthy controls (n=307). Results were adjusted for age, gender and BMI. IPMN was defined either by surgical pathology (n=62) or characteristic endoscopic ultrasound appearance and cyst fluid CEA>200 ng/ml (n=17). Results: In unadjusted analysis IPMN pts were more likely to be ≥ 70 years of age (OR 5.40 [2.88, 10.46]) when compared with PC pts (OR 2.82) and controls. After adjustment for age, gender and BMI, current tobacco smoking was associated with PC (OR 3.06 [1.78, 5.23]) but not IPMN. Pts with IPMN more often had diabetes mellitus for >3 years compared with controls (OR 3.25 [1.45, 7.00], while pts with PC (OR 1.52 (0.86, 2.67]) did not. IPMN pts were more likely to have a history of hypercholesterolemia compared with controls (OR 1.77 [1.05-2.98]); this was not seen for PC pts (OR 1.16 [0.87-1.55]). A first degree relative with PC was not associated with increased risk for IPMN (OR 0.84 [0.27, 2.62]) or PC (OR 1.48 [0.82, 2.67]). Compared to PC, pts with IPMN were more likely to have a history of an unrelated cancer (OR 1.84 [1.08, 3.14]). Conclusions: Risk factors for IPMN and PC may differ. Compared to PC and control pts, IPMN patients were older; more often had diabetes mellitus and hypercholesterolemia; and did not currently smoke. IPMN was more often associated with a prior history of cancer than PC. No significant financial relationships to disclose.