Ahn, Daniel H.

399 Background: The standard of care for resectable GEA adenocarcinomas is perioperative chemotherapy with the FLOT (fluorouracil, leucovorin, oxaliplatin, docetaxel) regimen. Recent studies investigating the addition of checkpoint inhibitors (CPIs) to perioperative chemotherapy have shown mixed results. An NMA can leverage both direct and indirect comparisons to evaluate the relative efficacies of available regimens in this setting. Methods: MEDLINE, EMBASE, Scopus, Web of Science and CENTRAL were searched until August 27th, 2025. Eligible studies included phase 3 randomized trials evaluating perioperative or neoadjuvant chemotherapy +/- CPIs ± radiation in resectable GEA. The primary outcomes were disease-free survival (DFS) and overall survival (OS). Hazard ratios and 95% confidence intervals (CIs) were estimated using a frequentist NMA framework. Results: Fifteen trials including 8072 patients were analyzed. For DFS, perioperative durvalumab + FLOT (dFLOT) ranked highest (P-score = 0.99), followed by FLOT (P-score = 0.84) (HR for dFLOT vs. FLOT was 0.71, 95% CI 0.58-0.86). dFLOT also ranked higher than perioperative pembrolizumab + cisplatin and fluorouracil/capecitabine (KN-585), though this comparison was not statistically significant (HR 0.65, 95% CI 0.40-1.04). Compared with CROSS, dFLOT showed superior DFS (HR of 0.48, 95% CI 0.36-0.63). For OS, dFLOT ranked the highest (P-score=0.95) and was superior to CROSS (HR 0.57, 95% CI 0.42-0.77). While dFLOT appeared favorable versus FLOT (HR 0.78, 95% CI 0.62-0.98), this finding was primarily driven by MATTERHORN, which did not meet its prespecified OS significance boundary (P = 0.03 vs threshold P < 0.0001), and the proportional hazards assumption was violated. Similarly, no significant OS difference was observed between dFLOT and the KN-585 arms (HR 0.75, 95% CI 0.45-1.23). Conclusions: In this NMA, perioperative FLOT remained the benchmark regimen for resectable GEA. dFLOT has shown promising early results, ranking the highest for DFS and showing favorable, but not statistically definitive OS trends in comparison to FLOT. Importantly, the network enabled indirect comparisons not tested in head-to-head trials, demonstrating that dFLOT ranked above pembrolizumab+CF/CX (KN-585) without reaching statistical significance, and also outperformed CROSS. However, these findings should be interpreted with caution, as they are primarily driven MATTERHORN data in which OS did not cross the prespecified threshold and proportional hazards assumptions were violated. Perioperative immunotherapy combinations appear promising, but final MATTERHORN results and further confirmatory studies are needed before dFLOT can be considered as a new standard of care.

103 Background: Clinicians spend significant time extracting and synthesizing information to risk-stratify and make adjuvant chemotherapy (AC) recommendations for CRC. By automating parts of this process, LLMs could help reduce documentation burden and variability in decision-making. Methods: We selected 144 stage III, pMMR CRC patients treated at Mayo Clinic (2017–2021). HIPAA-compliant Gemini-2.0-Flash-001 and GPT-4o extracted clinical variables (age, neuropathy grade, pathological stage) from EHRs and generated AC recommendations. Two approaches were compared: (1) a rule-based algorithm with nested if/else statements translating variables into NCCN-based recommendations, and (2) a dynamic LLM agent generating recommendations using a structured prompt informed by NCCN guidelines. Concordance with treating physicians was assessed, with disagreements reviewed by a blinded GI oncologist. Results: The dataset included 144 patients with median age 67 (IQR 66–68), 48.6% female. The LLM pipeline extracted neuropathy grade with 100% accuracy and pathological T/N stages with 99.3% accuracy. Concordance with physician recommendations was 93.1% for the rule-based algorithm and 62.5% for the LLM agent. In discordant cases, blinded adjudication by a GI oncologist favored the algorithm and LLM agent over physicians in most cases (Table 1). Conclusions: This exploratory work shows LLM agents can reliably extract variables and generate evidence-based treatment recommendations for stage III CRC. The high rate of expert preference for LLM recommendations in discordant cases is intriguing but limited by small sample size and possible adjudicator bias. Discordant physician recommendations reflect the complexity of decision-making, including patient preferences, performance status, comorbidities, and subtle findings not fully captured in EHRs. Rather than replacing judgment, we envision LLMs as decision support tools—analogous to an advanced trainee reviewing charts and presenting evidence-based recommendations to the attending physician with ultimate authority. Larger multi-institutional studies with multiple adjudicators are needed to validate these findings and clarify AI’s role in oncology. Patient characteristics, LLM extraction accuracy, and concordance with physician recommendations. Metric Result Total patients 144 Median age (IQR) 67 (66-68) Female (%) 48.6% LLM extraction accuracy – Neuropathy grade 100% LLM extraction accuracy – Pathological T and N stages 99.3% Primary Concordance – Rule-based algorithm vs. physician 93.1% Adjudicator preference of rule-based algorithm over treating physician for discordant cases 100% Concordance – LLM agent vs. physician 62.5% Adjudicator preference of LLM agent over treating physician for discordant cases 82.0%

Abstract Background This phase 1 study aimed to determine the maximum tolerated dose (MTD) and evaluate the safety and preliminary efficacy of rucaparib (RUB), a poly(adenosine diphosphate ribose) polymerase (PARP) inhibitor, combined with liposomal irinotecan (nal‐IRI) and 5‐fluorouracil (5‐FU) in metastatic gastrointestinal (GI) cancers. RUB targets DNA repair pathways, showing efficacy in tumors with homologous recombination deficiency, such as BRCA mutations. Preclinical data suggest synergy with irinotecan, but overlapping toxicities pose challenges. Methods Eighteen patients with metastatic GI cancers, who previously progressed on at least one systemic therapy, were enrolled. A novel sequential dosing regimen, informed by nal‐IRI pharmacokinetic analysis, was used. Twelve patients were evaluable for dose‐limiting toxicity (DLT) and 12 for response per Response Evaluation Criteria in Solid Tumors v1.1 criteria. Results The MTD was established as RUB 600 mg twice daily, nal‐IRI 50 mg/m 2 , and 5‐FU 2400 mg/m 2 over 46 hours. The objective response rate (ORR) was 33% (4 of 12), and the disease control rate (DCR) was 75% (9 of 12). Common grade 3 adverse events included diarrhea (33%) and neutropenia (25%), with no grade 4/5 events. Responses were notable in patients with somatic ATM and BRCA mutations, especially those with prior platinum exposure. No DLTs occurred at the recommended phase 2 dose. Conclusions This optimized dosing schedule successfully established the MTD for RUB with nal‐IRI and 5‐FU, overcoming prior challenges with PARP inhibitor and irinotecan combinations. The promising ORR and DCR support further evaluation of this regimen in advanced GI malignancies.

PURPOSE Neoadjuvant immune checkpoint inhibitors (nICIs) have demonstrated high response rates in deficient mismatch repair/microsatellite instability–high (dMMR/MSI-H) gastroesophageal adenocarcinoma (GEA). The NEONIPIGA and INFINITY trials demonstrated high rates of pathologic complete response (pCR) in this patient population. Furthermore, the INFINITY trial explored the feasibility of managing these patients nonoperatively, demonstrating promising results. This study aimed to evaluate clinical outcomes of nICIs in resectable dMMR/MSI-H GEA, with a focus on the feasibility of nonoperative management (NOM). MATERIALS AND METHODS This retrospective cohort study included patients with resectable dMMR/MSI-H GEA and treated with nICIs ± surgery at the Mayo Clinic. Patients were identified from institutional records, and clinical data were retrospectively reviewed. Primary outcomes were clinical complete response (cCR) and pCR. Secondary outcomes included event-free survival (EFS), radiologic complete response (rCR), and immune-related adverse events (irAEs). RESULTS A total of 26 patients treated between April 1, 2017, and July 30, 2025, were identified. Nine patients (34.6%) underwent surgery, of whom six (66.7%) achieved pCR. Seventeen patients (65.4%) pursued NOM, with 10 (71.4%) of 14 evaluable patients achieving cCR and 14 (82.4%) of 17 evaluable achieving rCR. One patient who initially achieved cCR had a local recurrence on surveillance endoscopy and underwent salvage endoscopic resection. At a median follow-up of 19.3 months, 15 (88.2%) of 17 patients in the NOM cohort were alive and metastasis-free, with EFS rates of 87.3% at 12 and 24 months for all patients. irAEs occurred in nine patients (34.6%), with no grade ≥3 toxicities. CONCLUSION In this retrospective cohort study, nICIs led to high cCR and pCR rates in resectable dMMR/MSI-H GEA, supporting the use of immune checkpoint inhibitors in this setting and the feasibility of NOM in select patients.

Abstract Molecular profiling reveals that up to 40% of advanced biliary tract cancers (BTCs) harbor actionable genomic alterations amenable to matched targeted therapies. However, conventional trial designs typically evaluate these agents in pretreated populations, resulting in approvals for later-line use. Notably, 25–44% of patients with advanced BTC may become ineligible for treatment after first-line therapy due to disease progression or functional decline, thus the real-world survival outcomes with matched therapies remains poorly characterized. This international collaborative study includes data from Mayo Clinic, Duke Cancer Institute, and National Cancer Center East (Japan), as well as a combined dataset from the SCRUM-Japan GOZILA, MONSTAR-SCREEN-1, and MONSTAR-SCREEN-2 projects. We assessed overall survival (OS) of patients based on the presence of actionable alterations and the receipt of matched therapy. To adjust for confounding, we applied propensity score weighting with inverse probability of treatment and multivariate Cox regression modeling, controlling for age, sex, tumor location, prior resection, and mutational subgroup. Among 1,049 patients that were analyzed, 358 (34.1%) harbored actionable alterations, with 160 (44.7%) receiving matched targeted therapy. The matched group (patients with actionable tumors who received a matched therapy) were significantly associated with longer median OS (23.3 months; 95% CI: 19.5–30.4) compared to the unmatched group (14.7 months; 95% CI: 12.8–17.9) and those with non-actionable tumors (17.1 months; 95% CI: 15.5–18.7; HR 0.57, 95% CI: 0.46–0.71; p < 0.0001). The unmatched group also was associated with the lowest three-year OS rate (13% vs. 35% in matched and 19% in actionable groups, respectively; p<0.0001). On multivariable analysis with patients who received ≥2 lines of therapy, receipt of matched therapy remained an independent predictor of improved OS (HR 0.62; 95% CI 0.43–0.90). These findings support the early integration of targeted therapies in BTC management and underscore the need to overcome systemic barriers to treatment access.

Abstract Background Unresectable hepatocellular carcinoma (uHCC) is a leading cause of cancer death. FDA-approved first-line systemic therapies include atezolizumab/bevacizumab (atezo/bev) and durvalumab/tremelimumab (durva/treme); however, there is a lack of comparative data. Methods We reviewed outcomes of patients with uHCC who initiated atezo/bev or durva/treme between 2017 and 2024, across six institutions. Overall survival (OS) and time to treatment discontinuation (TTD) were analyzed using the Kaplan–Meier and Cox models, adjusting for baseline characteristics. Results Four hundred fifty-two uHCC pts were included. Median age: 68 years; 77% male; 81% white. Most common etiologies were viral hepatitis (38.9%) and metabolic dysfunction-associated steatohepatitis (19.5%). Disease progression was the primary reason for treatment discontinuation, atezo/bev (56%) and durva/treme (42%). Outcomes were not statistically significant (median OS [month, m]: 14.0 vs 14.6 [P = .66]; median TTD [m]: 4.9 vs 3.9 [P = .42] for atezo/bev vs durva/treme). Outcomes were significantly different between Child-Pugh classes (CP: A, B7, B8/9, C) respectively, median OS(m): 19.0, 6.1, 5.1, 2.0 (P &lt; .001); median TTD(m): 6.1, 2.3, 3.0, 1.3 (P &lt; .001). Conclusions In this real-world study of uHCC, no significant difference in clinical outcomes was observed between atezo/bev and durva/treme in the first-line setting. CP scores were a key prognostic variable with both regimens.

Importance The prognosis of advanced pancreaticobiliary tumors is poor. Next-generation sequencing (NGS) of tissue samples is utilized to identify actionable alterations, but there are occasional limitations due to inadequate tissue acquisition. Circulating tumor DNA (ctDNA) is an alternative method, but its correlation with tissue-based NGS remains unexplored. Objectives To assess the mutation concordance (mCR) rates between ctDNA and tissue testing for patients with pancreatic ductal adenocarcinoma (PDAC) and cholangiocarcinoma (CC) and to evaluate how ctDNA performs as a treatment response biomarker. Design, Setting, and Participants This retrospective cohort study was conducted among patients with PDAC and CC treated at academic institutions from January 2014 to January 2025. Patients underwent ctDNA testing using 1 platform and tissue NGS testing using 2 platforms. Main Outcomes and Measures mCR, which measures the shared gene alterations observed by both ctDNA and tissue-based NGS testing, was calculated using the Spearman correlation for PDAC and CC. The performance of ctDNA as a treatment response biomarker was assessed by comparing serial ctDNA data with restaging scans and cancer antigen 19-9 levels in patients with PDAC. Results Our cohort included 790 patients: 570 with advanced PDAC (265 [46.5%] female; median [IQR] age, 64 [33-84] years) and 220 with advanced CC (95 [43.2%] female; median [IQR] age, 66 [42-88] years). Overall, 461 patients with PDAC (80.9%) and 192 patients with CC (87.2%) underwent ctDNA testing, while 239 patients with PDAC (41.9%) and 70 patients with CC (31.8%) had tissue NGS testing. Among patients with PDAC, 85 of 130 patients (65.4%) showed shared specific gene alterations between ctDNA and tissue testing; there was a significant mCR, with a Spearman correlation of 0.47 (95% CI, 0.28-0.62; P &amp;lt; .001). For patients with CC, 32 of 48 (66.7%) had shared alterations between ctDNA and tissue testing; there was a significant mCR, with a Spearman correlation of 0.56 (95% CI, 0.35-0.70; P &amp;lt; .001). A subgroup analysis of patients with PDAC who underwent serial ctDNA testing suggested that new TP53 subclones and increased ctDNA variant allele frequency levels of TP53 and KRAS were associated with higher odds of progressive disease (eg, increased TP53 frequency: odds ratio, 7.28; 95% CI, 2.15-24.66; P = .001). Conclusions and Relevance In this cohort study of patients with pancreaticobiliary tumors, there was a significant mCR between ctDNA and tissue NGS testing. Additionally, results suggest that ctDNA might detect resistant clones and enable assessment of treatment response.

Abstract Background Cabozantinib is indicated for hepatocellular carcinoma (HCC) following first-line (1L) sorafenib, but the 1L standard has shifted to immuno-oncology (IO)-based regimens. This study evaluated real-world outcomes among patients with advanced HCC receiving second-line (2L) cabozantinib following 1L therapies, including newer regimens. Patients and Methods US claims data were used to identify adults with advanced HCC initiating 2L cabozantinib monotherapy (index date). Patients were stratified into three 1L treatment cohorts: IO monotherapy/IO + IO combination therapy; IO + non–IO combination therapy; or tyrosine kinase inhibitor (TKI) monotherapy. Real-world time to treatment discontinuation (rwTTD), real-world time to next treatment or death (rwTNTD), real-world overall survival (rwOS), and cabozantinib dosing were assessed collectively and by cohort. Adverse events were assessed before/after cabozantinib initiation. Results Among 148 patients who received 2L cabozantinib, 28 were in the IO monotherapy/IO + IO combination therapy cohort, 54 in the IO + non–IO combination therapy cohort, and 66 in the TKI monotherapy cohort. Median rwTTD among all patients was 3.2 months; median rwTNTD was 7.6 months; 12-month rwOS rate was 61.6%. There were no significant differences in these outcomes among the three cohorts. Overall, 44.6% of patients initiated 2L cabozantinib at 60 mg/day, of whom 39.4% required a dose reduction; 37.8% initiated at 40 mg/day, of whom 16.1% had a dose reduction. Adverse event rates were similar before/after cabozantinib initiation. Conclusions Cabozantinib shows consistent effectiveness and safety in the 2L HCC setting following prior TKIs or IO-based regimens in real-world clinical practice. These findings may inform 2L treatment decisions.