First-line GemCis ± immunotherapy vs FGFR inhibition in ctDNA-detected FGFR2 fusion–positive advanced cholangiocarcinoma: A real-world analysis.

4159 Background: Advanced cholangiocarcinoma (aCCA) carries a poor prognosis. Approximately 40% of aCCA harbor targetable biomarkers; however, standard first-line (1L) therapy is gemcitabine-cisplatin (GemCis) ± immune checkpoint inhibitors (ICI), regardless of genomic status. Emerging data suggest that oncogene-driven molecular subsets of aCCA may exhibit reduced sensitivity to ICI, raising questions about the benefit of adding ICI to chemotherapy in genomically defined populations. Therefore, we compared outcomes for aCCA pts with ctDNA-detected FGFR2 fusions ( FGFR2 +) treated with targeted therapy vs. GemCis ± ICI. Methods: Real-world data was sourced from InfinityAI Data Library, a database combining de-identified genomic information from Guardant360 (Guardant Health, Palo Alto, CA) and clinical data from administrative claims. aCCA pts with an FGFR2 fusion detected prior to 1L therapy between Sept. 2018 and June 2025 were analyzed. Outcomes were assessed via real-world time to treatment discontinuation (rwTTD), real-world time to next treatment (rwTTNT) and real-world overall survival (rwOS) in months with 95% confidence intervals. Log-rank tests were used to compare Kaplan-Meier survival curves. Results: 149 FGFR2 + aCCA pts with treatment and outcomes data were identified. 18.1% (27) were treated with GemCis, 45.6% with GemCis + ICI (68), and 19.5% (29) with FGFRi. FGFR2 + pts treated with 1L FGFRi had improved rwTTD, rwTTNT, and rwOS compared to GemCis and GemCis + ICI, with significantly improved rwTTD in those treated with FGFRi vs GemCis + ICI (Table 1). Based on these data, the addition of ICI to GemCis was not associated with improved outcomes in FGFR2+ pts. Conclusions: Real-world outcomes in ctDNA-detected FGFR2+ aCCA did not improve with addition of ICI to 1L GemCis. In contrast, 1L FGFRi was associated with longer treatment duration and delayed need for subsequent therapy. These findings suggest standard chemo-immunotherapy may not be optimal for all biomarker-defined aCCA subgroups and support the use of ctDNA testing to inform biomarker-drive 1L treatment selection abd sequncing strategies, warranting prospective validation. Outcomes in FGFR2+ pts by 1L therapy. rrwTTD rrwTTNT rrwOS FGFRi vs GemCis 9.1 (4.9-11.5) vs 4.2 (2.3-6.2) HR=0.68, p=0.22 20.8 (8.3-20.8) vs 13.8 (5.6-30.3) HR=0.64, p=0.33 NR (16.9-NR) vs 17.5 (10.3-28.4)HR=0.48, p=0.16 FGFRi vs GemCis + ICI 9.1 (4.9-11.5) vs 5.2 (3.9-7.8)HR=0.62, p=0.03 20.8 (8.3-20.8) vs 8.9 (7.9-NR) HR=0.51, p=0.06 NR (16.2-NR) vs 20.9 (10.7-NR)HR=0.43, p=0.11 GemCis vs GemCis + ICI 4.2 (2.3-6.2) vs 5.2 (3.9-7.8) HR=0.91, p=0.91 13.8 (5.6-30.3) vs 8.9 (7.9-NR) HR=0.8, p=0.61 17.5 (10.3-28.4) vs 20.9 (10.7-NR) HR=0.89, p=0.69