572 Background: Cholangiocarcinoma (CCA) is a rare, and aggressive malignancy with limited treatment options, where gemcitabine plus cisplatin (GemCis) has long been the first-line standard. Immunotherapy (IO) has shown promise across cancer types, but its benefit remains uncertain. This meta-analysis assesses the efficacy and safety of IO-based regimens compared with GemCis alone in phase III trials. Methods: For this meta-analysis, we searched PubMed, Embase, Scopus, Google Scholar for studies published from 2023. Eligible studies included randomized clinical trials evaluating ICPIs plus GemCis versus GemCis plus placebo as first-line treatment for unresectable or metastatic CCA. The primary outcome was OS, and the secondary outcomes included progression-free survival (PFS) and adverse events. For studies presenting Kaplan-Meier curves, individual patient data (IPD) tool was used to reconstruct the data for outcome calculations. Results: A total of 1,754 patients with Unresectable or metastatic CCA from TOPAZ-1, and KEYNOTE-966 were included: 880 received GemCis + placebo, 533 received GemCis + pembrolizumab, and 341 received GemCis + durvalumab; median age was 63-64 years, 50% male. The meta-analysis showed that median OS was 11.2 months with GemCis alone, compared with 12.9 months for GemCis+ durvalumab and 12.8 months for GemCis + pembrolizumab groups (p<0.001); median PFS was 5.7, 7.2, and 6.6 months, respectively (p<0.011).Grade 3/4 events were highest with GemCis + pembrolizumab (78.8%) vs GemCis + placebo (68.6%) and GemCis + durvalumab (65.1%; p=0.0001), mainly thrombocytopenia, asthenia, and vomiting. Other toxicities showed no significant differences. Conclusions: Adding IO to GemCis in unresectable or metastatic CCA modestly improved OS and PFS versus GemCis alone, particularly GemCis+ durvalumab.

325 Background: Metastatic gastric cancer (mGC) is associated with dismal outcomes, and most patients eventually receive only palliative care. Accurate prognostic models are urgently needed to individualize survival estimates, optimize therapeutic decisions, and inform trial design. Data from Latin American populations are scarce. We aimed to develop and internally validate a prognostic nomogram for 12-month overall survival (OS) in patients with mGC. Methods: We retrospectively analyzed 1,323 patients with histologically confirmed mGC treated at the Instituto Nacional de Cancerología (Mexico) between 2010 and 2022. Demographic, clinical, biochemical, and histologic data were collected. OS was estimated using the Kaplan–Meier method. Multivariate Cox regression was performed in R Studio, and independent prognostic factors were incorporated into a nomogram predicting 12-month OS. Discrimination and calibration were assessed Results: Median age was 54 years (±13.5), with 63.4% younger than 50 years; 52.5% were male. ECOG performance status was 0–1 in 71.2% of patients. Median weight loss at diagnosis was 10 kg. Anemia grade ≥2 occurred in 27.7% and hypoalbuminemia (<3.5 g/dL) in 29.3%. Peritoneal carcinomatosis (58.1%), ascites (28.4%), and hepatic metastases (20.8%) were the most frequent metastatic sites. Median OS was 6.04 months (95% CI, 5.61–6.47). Independent adverse prognostic factors included hepatic metastases, ascites, peritoneal carcinomatosis, retroperitoneal involvement, anemia grade 2–3, and low serum albumin. The nomogram demonstrated strong discrimination and calibration, stratifying patients into clinically meaningful risk groups with differential 12-month OS probabilities. Conclusions: This is the first prognostic nomogram for mGC developed in a Latin American cohort, integrating both clinical and nutritional parameters. The tool provides individualized prediction of 12-month OS, offering oncologists an evidence-based resource to identify patients who may benefit from systemic therapy versus those more likely to transition early to supportive care. External validation is needed, but this nomogram has the potential to shift decision-making in a setting where most patients face limited therapeutic opportunities.

491 Background: Hepatocellular carcinoma (HCC) is the sixth most common cancer worldwide and the third leading cause of cancer related mortality. Outcomes are often shaped not only by the cancer itself but also by the burden of underlying cirrhosis and treatment related complications. Hospitalizations are common and carry substantial costs, driven by both disease progression and liver decompensation. This study examines resource use and the leading causes of 90-day readmissions after HCC admissions. Methods: We conducted a retrospective study using the 2022 Nationwide Readmissions Database (NRD), part of the Healthcare Cost and Utilization Project. We identified patients with Hepatocellular carcinoma who were hospitalized between January and October. The first unplanned readmission within 90 days of discharge was used to define the readmission cohort. Survey-weighted logistic, linear, and Cox regression models were used to assess clinical and hospital-level factors associated with readmission. Results: In multivariable analysis, younger age (aHR 0.99, p<0.001) and thrombocytopenia (aHR 0.92, p=0.035) were associated with lower readmission risk, while longer length of stay during index admission (aHR 1.00, p=0.005), diabetes (aHR 1.08, p=0.007), hyperlipidemia (aHR 1.09, p=0.004), ascites (aHR 1.23, p<0.001), varices (aHR 1.10, p=0.027), renal impairment (aHR 1.07, p=0.023), and treatment under 'other’ insurance payers was associated with lower risk compared with Medicare (aHR 0.79, p=0.030). Alcohol-related liver disease was protective (aHR 0.84, p<0.001). Top 90-day readmission causes were sepsis (2,181), cancer progression (1,822), cirrhosis (1,157), hepatic failure (965), acute kidney failure (504), cholangitis/biliary disorders (305), chronic pain (166), COVID-19 (145), hypertensive heart and kidney disease with heart failure (145), and pneumonia (137). Conclusions: In patients with hepatocellular carcinoma, 90-day readmissions are driven not only by tumor progression but also by sepsis and complications of portal hypertension, including ascites and varices. Metabolic comorbidities such as diabetes and hyperlipidemia further increased risk, Targeting infection prevention and better management of cirrhosis related decompensation may reduce readmissions and improve outcomes in this high-risk population. Comparative resource utilization between index hospitalizations and 90-day readmissions in patients with hepatocellular carcinoma. Category Index Admission 90-Day Readmission Weighted admissions 36,143.38 11,576.02 In-hospital deaths 3267.847 N/A Mean length of stay (days) 7.25 7.29 Mean charges (USD) $113,925 $111,415.6 Mean costs (USD) $27,787.15 $27,179.18 Cumulative hospital days 262,174.8 84,459.19 Cumulative charges (USD) $4.08 billion $1.28 billion Cumulative costs (USD) $996 million $312 million

826 Background: Oxaliplatin (OXA) is a cornerstone therapy for gastrointestinal (GI) cancers, but chemotherapy-induced peripheral neuropathy (CIPN) is its most common dose-limiting toxicity, impairing quality of life and limiting efficacy. While mechanisms of CIPN remain incompletely defined, inflammation has been implicated in its development. Clonal hematopoiesis (CH)—the expansion of mutated hematopoietic clones without overt malignancy—is linked to systemic inflammation and adverse clinical outcomes. We evaluated the association between CH and CIPN in older patients with GI cancers treated with OXA. Methods: Using the Moffitt Cancer Analytics Platform, we identified patients >65 years with GI cancers treated with OXA between 2020–2022 who also had tissue or blood-based next-generation sequencing (NGS). CH was defined by the presence of somatic mutations in canonical myeloid genes in blood or tissue. For variants such as TP53 , CH was suspected if detected only in blood in patients with both blood and tissue NGS, or when variant allele frequency was significantly lower than known tumor mutations in tissue. Results: Among 486 patients with colorectal (n=173, 35.6%), esophageal (n=74, 15.2%), gastric (n=66, 13.6%), pancreatic (n=161, 33.1%), or unknown primary (n=14, 2.9%) cancers, CIPN was reported in 110 (25%). CH was identified in 109 patients (22.4%), including 21 (21.6%) with multiple CH mutations. The most frequent mutations were DNMT3A (47.4%), TET2 (25.8%), and TP53 (24.7%), with less common alterations in ASXL1 (9.3%), SF3B1 (7.2%), and others. Mutation prevalence varied by cancer type, with DNMT3A enriched in gastric, pancreatic, and colorectal cancers, TP53 in unknown primary, and ATM largely restricted to esophageal cancer. No significant association was observed between CH and CIPN (p=0.917). Multigene CH did not differ by CIIP status (2.5% vs 4.9%, p=0.312), and neither TET2 nor DNMT3A , individually or grouped as an epigenetic CH category, was linked to CIPN (all p≥0.99). In the subset of 82 patients with diabetes-independent PN, CH was not associated with increased risk (CH– 17.5% vs CH+ 14.7%; RR 0.84; p=0.583). Conclusions: CH was common among older adults with GI cancers but was not associated with CIPN or diabetic independent PN. While CH reflects a pro-inflammatory state, it does not appear to influence oxaliplatin-related neurotoxicity.

846 Background: There is an alarming rise in early-onset (EO) gastrointestinal (GI) cancers, which may have distinct genomic profiles compared with later-onset (LO) cases. OncoKB is a validated database that provides evidence-based annotation of genomic alterations with therapeutic relevance. Utilizing the American Association for Cancer Research (AACR) Project Genomics Evidence Neoplasia Information Exchange (GENIE) v18.0 and OncoKB annotator, we performed a comprehensive analysis of OncoKB-actionable alterations across all GI tumors. Methods: The AACR GENIE v18.0 database was used to select GI cancers samples by OncoTree code, which were classified by tumor group. We utilized OncoKB to annotate all samples and their somatic mutations, copy number alterations, and structural variants by therapeutic level of evidence (Levels 1-4). We considered a tumor sample to have a potentially actionable alteration if there was at least one Level 1-3B alteration present. We defined EO as age < 50 and LO as age ≥ 50, and excluded samples without age. Chi-square testing compared frequency of actionable alterations between EO and LO tumor groups, and Benjamini-Hochberg procedure controlled for false discovery rate (statistical significance for q<0.05). Results: 43748 samples were analyzed with 19.2% EO and 80.8% LO samples. There were decreased potentially actionable alterations in EO colorectal (43.0% vs 46.3%, q<0.001), gastrointestinal stromal tumor (78.6% vs 93.4%, q<0.001), pancreatic (59.1% vs 67.2%, q<0.001), and stomach/esophagus tumors (50.8% vs 57.2%, q=0.001). No significant differences were observed in other GI tumor groups. The most frequent potentially actionable alterations were in KRAS , PIK3CA , and ERBB2 . Conclusions: To our knowledge, this study represents the largest molecular analysis of actionability of GI tumors. EO GI cancers have decreased frequency of potentially actionable genomic alterations compared to respective LO GI cancers in numerous GI tumor groups. These findings suggest that several EO GI cancers have more limited precision oncology targets, highlighting the need for novel therapeutic strategies. Tumor Group EO samples (n) LO samples (n) EO samples with potentially actionable alterations (n, %) LO samples with potentially actionable alterations (n, %) Most common potentially actionable altered gene q-value Appendix 222 624 103 (46.4%) 282 (45.2%) KRAS 0.85 Colorectal 5274 13989 2270 (43.0%) 6474 (46.3%) PIK3CA <0.001 Gastrointestinal stromal tumor 322 1554 253 (78.6%) 1451 (93.4%) KIT <0.001 Hepatobiliary 589 3995 344 (58.4%) 2274 (56.9%) KRAS 0.64 Neuroendocrine 345 1219 161 (46.7%) 534 (43.8%) ATM 0.52 Other 77 544 37 (48.1%) 295 (54.2%) PTEN 0.52 Pancreatic 614 7874 363 (59.1%) 5290 (67.2%) KRAS <0.001 Small bowel 85 480 62 (72.9%) 351 (73.1%) KRAS 0.97 Stomach/Esophagus 861 5080 437 (50.8%) 2904 (57.2%) ERBB2 0.001

211 Background: Pyroptosis is a lytic and inflammatory form of programmed cell death mediated by gasdermin proteins. It may play an important role in cancer pathogenesis, metastasis, and modulation of the tumor immune microenvironment. However, the roles of pyroptosis and gasdermins in colorectal cancer (CRC) remain incompletely defined. We evaluated the prognostic significance of gasdermin expression in patients with metastatic CRC (mCRC) treated on CALGB (Alliance)/SWOG 80405. Methods: RNA sequencing was performed on tumors from 433 patients with mCRC in CALGB/SWOG (Alliance) 80405 who received first line FOLFOX or FOLFIRI with bevacizumab (n = 226) or cetuximab (n = 207). RNA was extracted from FFPE tumor samples and sequenced on HiSeq 2500 platform (Illumina). Expression of five gasdermin proteins (GSDMA, GSDMB, GSDMC, GSDMD, and GSDME) were stratified into tertiles. Progression-free survival (PFS) and overall survival (OS) were assessed using Kaplan-Meier and Cox regression analyses. Results: High GSDME expression was associated with shorter survival (PFS HR 1.50, 95% CI 1.18-1.90, p=0.003; OS HR 1.64, 95% CI 1.28-2.11, p=0.00012). The effect was most pronounced with cetuximab (median OS 23.1 vs 39.1 mo; HR 2.04, 95% CI 1.40-2.98, p = 0.00065), while treatment with bevacizumab showed a similar but non-significant trend (median OS 25.6 vs 34.1 mo; HR 1.31, 95% CI 0.93-1.85, p = 0.051). Conversely, high GSDMC expression was associated with longer OS in the bevacizumab arm (median 38.6 vs 29.1 mo; HR 0.66, 95% CI 0.46-0.96, p = 0.0028), but not in patients treated with cetuximab. Expression of GSDMA, GSDMB and GSDMD showed no consistent associations with survival. Conclusions: In mCRC, gasdermin expression patterns highlight the dual roles of pyroptosis in tumor biology. GSDME was associated with poor prognosis, particularly in patients receiving cetuximab, suggesting a potential relationship with EGFR signaling or resistance. In contrast, GSDMC expression trended toward improved survival and conferred benefit in the bevacizumab subgroup, pointing to context-dependent effects possibly related to angiogenesis. These findings suggest pyroptosis as a clinically relevant pathway in CRC and support further investigation of gasdermins as prognostic biomarkers and potential therapeutic targets. Trial identifiers: NCT00265850; Support: P30CA014089, U10CA180821, U10CA180882, U10CA180820 and UG1CA233277; U10CA180888; Pfizer, Genentech. https://acknowledgments.alliancefound.org Clinical trial information: NCT00265850 .

588 Background: Biliary tract carcinomas (BTC) affect about 8,000 people annually across the U.S. and present with dismal prognoses. Disparities in mortality trends across the last two decades are not well-characterized. Methods: A retrospective analysis of death certificate data from the CDC WONDER database was performed for adults ≥25 years from 1999 to 2023 (ICD-10 codes C23, C24.0, C24.1, C24.9). Age-adjusted mortality rates (AAMRs) per 100,000 were standardized to the 2000 U.S. population. Data was stratified by age groups, race, gender, and census regions. Trends were evaluated using Joinpoint. Results: From 1999 to 2023, 103,589 deaths resulted from BTC. Overall AAMR declined from 2.18 in 1999 to 1.85 in 2023 (AAPC: -0.858, 95% CI: -1.139 to -0.576), decreasing significantly from 1999 to 2008 (APC-2.507) with a non-significant rise from 2008 onwards. Women consistently exhibited higher AAMRs (AAMR:1.958, AAPC: -0.931) than men (AAMR:1.745, AAPC: -0.763). 65+ years individuals exhibited highest death rates (AAMR: 7.382 , AAPC:-1.176). American Indians had the highest AAMR (2.295) but a significant decline (AAPC: -3.829) compared to Hispanic (AAMR: 2.448) and Black individuals (AAMR: 2.144). Metropolitan areas had higher AAMR (1.84) than non-metropolitan areas (1.81). Conclusions: The steady rise in BTC-related mortality from 2008 onwards with persistent demographic and temporal disparities underscores the need for timely and targeted interventions. Deaths and AAMRs per 100,000 for mortality trends related to BTC in adults in the U.S. between 1999 and 2023. Variable Deaths (n) AAMR (95% CI) Overall 103,589 1.852 (2.18-1.85) Male 41,713 1.745 (1.96-1.74) Female 61,876 1.958 (2.37-1.93) Non-Hispanic American Indian 840 2.295 (4.15-0.9) Non-Hispanic Asian 4,600 2.042 (2.59-1.84) Non-Hispanic Black 11,727 2.144 (2.2-2.19) Non-Hispanic White 86,344 1.811 (2.17-1.79) Hispanic 10,823 2.448 (3.32-2.19)

546 Background: Durvalumab or pembrolizumab combined with GemCis are established first-line treatment for patients with unresectable or metastatic BTC. As patient profiles in daily practice is different from those in prospective trials, we performed the retrospective analysis of patients treated with first-line durvalumab or pembrolizumab plus GemCis. Methods: In this retrospective analysis, 255 patients with histologically or cytologically documented BTC, including intrahepatic and extrahepatic cholangiocarcinoma (CCA) and gallbladder cancer (GBCA), treated with first-line durvalumab (n=210) or pembrolizumab (n=45) plus GemCis for unresectable or metastatic disease between March 6, 2023, and May 16, 2025, were included in this analysis. The primary endpoints were overall survival (OS) and progression-free survival (PFS). Secondary endpoints were objective response rate (ORR) and disease control rate (DCR). Results: The median age was 66 years, and 58.8% were male; 77% of patients had recurrent or metastatic disease while 23% had locally advanced disease. Primary tumor sites were intrahepatic CCA (40.4%), extrahepatic CCA (42.0%), and GBCA (16.5%). In overall patients, median OS was 14.4 months (95% CI, 12.6–16.5), and median PFS was 6.9 months (95% CI, 6.0–8.0). The ORR was 15.3%, and the DCR was 72.5%. In the multivariable analysis, OS was significantly worse in patients with elevated serum CA 19-9 level (HR 1.87 [95% 1.11-3.15], p=0.018), GBCA (vs. extrahepatic CCA; HR 1.99 [95% CI, 1.10-3.59], p=0.023), and recurrent or metastatic disease (vs. locally advanced; HR 2.50 [95% CI, 1.31=4.77], p=0.005). Conclusions: Current analysis demonstrated that the clinical effectiveness of durvalumab or pembrolizumab plus GemCis in real-world practice are consistent in those in pivotal prospective trials.

TPS256 Background: Currently, there are no biomarkers validated prospectively in randomized studies for resected colon cancer (CC) to determine need for adjuvant chemotherapy (AC). However, circulating tumor DNA (ctDNA) represents a highly specific and sensitive approach (especially with serial monitoring) for identifying minimal/molecular residual disease (MRD) post-surgery in CC patients (pts), and may outperform traditional clinical and pathological features in prognosticating risk for recurrence. CC pts who do not have detectable ctDNA (ctDNA-) are at a much lower risk of recurrence and may be spared the toxicities associated with AC. Furthermore, for CC pts with detectable ctDNA (ctDNA+) who are at a very high risk of recurrence, the optimal AC regimen has not been established. We hypothesize that for pts whose CC has been resected, ctDNA status may be used to risk-stratify for making decisions about AC. Methods: In this prospective phase II/III trial, up to 1,912 pts with resected stage IIB, IIC, and III CC will be enrolled. Based on the post-operative ctDNA status using personalized and tumor-informed assay (Signatera, bespoke assay), those who are ctDNA- (Cohort A) will be randomized to immediate AC with fluoropyrimidine (FP)+oxaliplatin (Ox) for 3-6 mos per established guidelines v serial ctDNA monitoring. Patients who are ctDNA+ post-operatively, or with serial monitoring (Cohort B), will be randomized to FP+Ox v more intensive AC with addition of irinotecan (I) for 6 mos. One cycle of chemotherapy is allowed while awaiting ctDNA testing results for cohort assignment. The primary endpoints for Cohort A are time to ctDNA+ status (phase II) and disease-free survival (DFS) (phase III) in the immediate v delayed AC arms. The primary endpoint for Cohort B is DFS in the FP+Ox v FP+Ox+I arms for both phase II and phase III portions of the trial. Secondary endpoints include prevalence of detectable ctDNA post-operatively, time-to-event outcomes (overall survival and time to recurrence) by ctDNA status, and the assessment of compliance to adjuvant therapy. Biospecimens including archival tumor tissue, as well as post-operative plus serial matched/normal blood samples, will be collected for exploratory correlative research. Active enrollment across the NCTN started in June 2022 with CCTG sites joining in August 2023. Current accrual (as of 9-4-2025): 967/1,912. Drs. Lieu and Dasari contributed equally to the development of this trial. Clinical trial information: NCT05174169 .

669 Background: Pancreatic cancer is among the most lethal gastrointestinal malignancies in the United States. Identifying where and in whom the burden is highest can guide smarter prevention, earlier detection, and more equitable healthcare planning. Geographic and demographic analysis is essential to help reduce disparities in outcomes. Methods: Data were obtained using the Institute for Health Metrics and Evaluation (IHME) Global Burden of Disease (GBD) 2021 Results Tool. Age-standardized death and DALY rates were analyzed by state and sex. Age-stratified crude mortality and DALYs identified high-burden subgroups. Risk attribution was assessed using DALY percentages and temporal risk factor rankings from 1990 to 2021. Descriptive statistics were performed using Microsoft Excel. Results: In 2021, Louisiana and Mississippi consistently ranked among the highest states across incidence, mortality, and DALY rates, while Colorado, Utah, and North Dakota showed the lowest burden. Males demonstrated higher incidence and mortality than females, and rates rose sharply after age 65, peaking in those ≥80 years. State-level variation persisted across three decades, with the South and Midwest showing the greatest burden. Smoking and high body mass index accounted for the largest share of attributable DALYs, with variation across states. DALY rates peaked in men aged 65–69 and women aged ≥80. Prevalence was highest in Hawaii, Louisiana, and Michigan, highlighting a growing survivor population with long-term care needs. Conclusions: Pancreatic cancer burden in the United States shows striking geographic variation, with consistently high rates in southern states such as Louisiana and Mississippi. The greatest burden occurs in older adults, particularly men. These findings underscore the importance of targeted prevention and improved resource allocation to reduce disparities in pancreatic cancer outcomes. Pancreatic cancer burden in the United States, 2021 (age-standardized per 100,000). Metric Highest State (Rate) Lowest State (Rate) Notable Pattern Incidence Louisiana (12.93) North Dakota (8.78) South vs. Midwest contrast Prevalence Hawaii (11.46) Montana (8.83) Coastal variation Mortality Louisiana (12.39) North Dakota (8.08) Mirrors incidence DALY Louisiana (281.6) Colorado (177.4) Southern burden highest Male Incidence Louisiana (14.46) North Dakota (10.04) Higher than female Female Incidence Louisiana (11.53) North Dakota (7.55) Lower overall Age Group Peak M 80+ yrs (6,651 cases) <20 yrs (2.3 cases) Sharp rise >65 yrs Age Group Peak F 80+ yrs (9,332 cases) <20 yrs (2.1 cases) Sharp rise >65 yrs