Berger, Adam C.

Objective: This study aims to present a full spectrum of individual patient presentations of pancreatic fistula risk, and to define the utility of mitigation strategies amongst some of the most prevalent, and vulnerable scenarios surgeons encounter. Background: The FRS has been utilized to identify technical strategies associated with reduced CR-POPF incidence across various risk strata. However, risk-stratification using the FRS has never been investigated with greater granularity. By deriving all possible combinations of FRS elements, individualized risk assessment could be utilized for precision medicine purposes. Methods: FRS profiles and outcomes of 5533 PDs were accrued from 17 international institutions (2003–2019). The FRS was used to derive 80 unique combinations of patient “scenarios.” Risk-matched analyses were conducted using a Bonferroni adjustment to identify scenarios with increased vulnerability for CR-POPF occurrence. Subsequently, these scenarios were analyzed using multivariable regression to explore optimal mitigation approaches. Results: The overall CR-POPF rate was 13.6%. All 80 possible scenarios were encountered, with the most frequent being scenario #1 (8.1%) – the only negligible-risk scenario (CR-POPF rate = 0.7%). The moderate-risk zone had the most scenarios (50), patients (N = 3246), CR-POPFs (65.2%), and greatest non-zero discrepancy in CR-POPF rates between scenarios (18-fold). In the risk-matched analysis, 2 scenarios (#59 and 60) displayed increased vulnerability for CR-POPF relative to the moderate-risk zone (both P < 0.001). Multivariable analysis revealed factors associated with CR-POPF in these scenarios: pancreaticogastrostomy reconstruction [odds ratio (OR) 4.67], omission of drain placement (OR 5.51), and prophylactic octreotide (OR 3.09). When comparing the utilization of best practice strategies to patients who did not have these conjointly utilized, there was a significant decrease in CR-POPF (10.7% vs 35.5%, P < 0.001; OR 0.20, 95% confidence interval 0.12–0.33). Conclusion: Through this data, a comprehensive fistula risk catalog has been created and the most clinically-impactful scenarios have been discerned. Focusing on individual scenarios provides a practical way to approach precision medicine, allowing for more directed and efficient management of CR-POPF.

63 Background: A novel vaccine strategy may prevent recurrence in high-risk melanoma patients (pts). The TLPLDC vaccine uses yeast cell wall particles (YCWP) to load tumor lysate into autologous dendritic cells (DC). In this phase IIb trial of TLPLDC vs. placebo in resected stage III/IV pts, TLPLDC increased 24 month (mo) disease free survival (DFS) in the per treatment (PT) population. Here, we present a 24mo DFS subgroup analysis and estimated overall 36mo DFS. Methods: Disease-free pts were randomized 2:1 to the TLPLDC vaccine vs. unloaded YCWP+DC at 0, 1, 2, 6, 12, and 18mo. The protocol was amended to allow concurrent adjuvant checkpoint inhibitor (CPI) therapy once approved. The pre-specified PT population included only pts completing the primary vaccine/placebo series (PVS) at 6 mo. Kaplan-Meier estimates of DFS were used to compare treatment arms by stage (III or IV) and CPI therapy (yes/no) in the ITT and PT populations. Results: 144 pts were randomized (103 TLPLDC, 41 placebo); 98 pts (66 TLPLDC, 32 placebo) completed the PVS. There were no clinicopathologic differences between treatment groups. There was no difference in 24mo DFS in stage III pts (n = 112), but in stage IV pts (n = 32), the 24mo DFS was 44% vs 0% (TLPLDC vs placebo) (p = 0.41) in ITT and 73.3% vs. 0% (HR 0.14, p = 0.002) in PT. Stage IV pts were more likely to receive CPI than stage III pts (50% vs. 30%, p = 0.003). There was no difference in 24mo DFS for pts who did not receive CPI (n = 102), but in pts who received CPI (n = 42), the 24mo DFS was 49.3% vs. 31.3% (p = 0.71) in ITT and 68.8% vs. 41.7% (HR 0.46, p = 0.28) in PT, showing a trend toward improved DFS in pts who completed the PVS and received CPI (n = 31). Overall, the 36mo estimated DFS was 34.2% vs. 21.6% (p = 0.89) for ITT and 56.9% vs. 27.9% (p = 0.021) for PT. Conclusions: The TLPLDC vaccine improved DFS in patients completing the PVS at 24 and 36 mos, particularly in the resected stage IV subset. The apparent synergistic effect with TLPLDC + CPI will be confirmed in a phase III study evaluating adjuvant TLPLDC + CPI vs. CPI alone in resected stage IV melanoma pts. Clinical trial information: NCT02301611.

4007 Background: NRG/RTOG 0848 is a 2-step study designed to determine whether erlotinib (E) added to gemcitabine (G) (randomized Ph II) &/or adjuvant radiation with concurrent 5-FU or capecitabine following 6 months of systemic chemotherapy (Ph III), improve survival in patients (pts) with resected pancreatic head adenocarcinoma. The erlotinib results are reported here. Methods: Eligible pts include those with resected pancreatic head adenocarcinoma, pathologic stage T1-T3, N0-1, M0; PS 0-1, & CA19-9 ≤ 180 IU/L. Pts in Arms 1 & 2 received G 1 gm/m2 weekly for 3 weeks in a 28-day cycle for 6 cycles. Pts in Arm 2 also received E 100 mg/day. The primary hypothesis for the E portion was that G+E would increase overall survival (OS) compared to G alone. With a 1-sided alpha of 0.15, 200 OS events provide 80%/90% power to detect a signal for an increase in median OS from 22 to 28.8/30.6 months (mos). OS was estimated by the Kaplan-Meier method & arms compared using the log rank test. The Cox proportional hazards model was used to analyze treatment effect. Results: 336 pts were randomized from 11/17/2009 to 2/28/2014, with 163 pts evaluable for G and 159 for G+E. Median age was 63 years (39-86). Most pts had pathologic T3 disease (78%) & CA19-9 ≤ 90 (93%). There are 32 pts (20%) with grade 4 adverse events (AEs) & 2 pts (1%) with grade 5 AEs on G and 27 (17%) & 3 (2%) on G+E arm, respectively. There are fewer grade ≥ 3 GI AEs on the G arm (22%) as compared to the G+E arm (28%), and 110 (69.2%) & 93 (59.6%) pts received at least 85% of planned G dose for the G & G+E arms, respectively. 58% of E pts received at least 85% of planned E dose. The median follow-up for alive pts is 42.5 mos (min-max: < 1-75). With 203 deaths, median & 3-yr OS (95% CI) are 29.9 mos (21.7-33.4) & 39% (30, 45) for G and 28.1 mos (20.7-30.9) & 39% (31, 47) for G+E; log-rank p = 0.62. The hazard ratio (95% CI) comparing OS of G+E to G is 1.04 (0.79- 1.38). Conclusions: The addition of adjuvant E to G did not provide a signal for increased OS in pts with resected pancreatic head cancer compared to G alone. Accrual to the trial is continuing to answer the Ph III radiation question. Clinical trial information: NCT01013649.