Comprehensive genomic profiling to identify molecular determinants of efficacy of IDH1 inhibition in intrahepatic cholangiocarcinoma.

4147 Background: Ivosidenib (IVO) in IDH1 -mutant iCCA yields modest mPFS (2.7 m) with heterogenous responses. Molecular features driving this variability remain undefined. We sought to identify determinants of therapeutic efficacy via comprehensive analyses of co-occurring genomic alterations and oncogenic pathway activation. Methods: We analyzed 93 IDH1 -mutant iCCA cases (MD Anderson n=59, Moffitt n=23, Yonsei n=11) treated with IVO (n=78), IDH305 (n=15). We stratified patients (pts) into 3 groups based on co-occurring alterations: EC (Epigenetic/Chromatin remodeling: BAP1/ARID1A/PBRM1), PMF (Proliferative/Mitogenic signaling: PI3K/MAPK/FGFR), ACT (Amplified cyclins/CDKN2A-B/TP53), yielding 4 groups: G1 (EC-intact/low-oncogenic: EC+/PMF-/ACT-), G2 (signaling-driven: EC+/PMF+), G3 (cell-cycle-deregulated: EC+/ACT+), and G4 (EC-deficient: EC-). Cox models assessed group-specific treatment effects; TME characterization utilized 29 mRNA gene signatures (Bagaev, Cancer Cell 2021) (n=31) and paired pre/post-progression biopsies (n=4). Results: Survival analysis (n=93) revealed distinct stratification (p<0.0001): G1, the longest mPFS (11.4 m), significantly superior to G4 (3.5 m) and G3 (1.8 m). G1 had superior durability (12-m PFS 46.5%) over G2 and G4 (23.9%, 14.6%); G3 progressed rapidly (6-m 0%). G3 outcomes suggest that ACT co-mutations confer poor prognosis regardless of epigenetic context (HR 6.85, P=0.004). Transcriptomic analysis identified G1 as "inflamed-suppressed" (highest angiogenesis/Treg); G4 was "proliferative-inflammatory" (highest M1-macrophage/Ki67, p=0.04). Paired longitudinal analysis (n=4) identified genomic bypass via acquired kinase drivers (eg, NTRK1 amplification) and stromal adaptation with dense fibrotic remodeling. Radiographic hyperprogression on IVO (Kato criteria) was observed in 2 pts in G3 harboring CDKN2A loss. Conclusions: Co-occurring genomic alterations may serve as prognostic markers in IDH1 -mutant iCCA, suggesting distinct trajectories of response and resistance to IDH1 inhibition. This molecular classification could guide clinical decision-making by identifying pts likely to benefit from monotherapy vs pts requiring combination strategies to overcome intrinsic resistance. G1 G2 G3 G4 EC+ / PMF- / ACT- EC+ / PMF+ EC+ / ACT+ EC- N (%) 28 (30%) 19 (20%) 5 (5%) 41 (44%) mPFS (95% CI) 11.4 mo (6.2-NR) 5.6 mo (3.7-9.2) 1.8 mo (0.9-3.5) 3.5 mo (2.1-5.8) 3-M Rate 80.9% (66.3-95.5) 88.9% (74.8-100) 20.0% (0.0-55.1) 59.5% (44.5-74.5) 6-M Rate 72.4% (55.8-89.0) 47.9% (25.4-70.4) 0.0% (0.0-0.0) 23.4% (10.4-36.4) 12-M Rate 46.5% (28.0-65.0) 23.9% (4.7-43.1) 0.0% (0.0-0.0) 14.6% (3.8-25.4) Hazard Ratio Reference (1.0) 2.15 (p=0.03) 6.85 (p=0.004) 2.94 (p=0.002) TME Phenotype Angiogenesis / Treg-High Fibrotic (CAF-High) Senescent / Dormant Proliferative / M1-High