Decoding FGFR inhibitor sensitivity in cholangiocarcinoma with interpretable machine learning and cross-platform
Yading Xie1, Honglei Li1, Wujie Zhang1
1Department 2 of Hepatobiliary Surgery, Handan First Hospital, Handan, Hebei, China.
Background:
Fibroblast growth factor receptor (FGFR) inhibitors (FGFRis) benefit selected cholangiocarcinoma (CHOL) patients, yet responses remain heterogeneous and are not fully explained by canonical FGFR alterations. Clinically useful biomarkers are hindered by variable single-compound drug sensitivity readouts with limited concordance across FGFR-targeting agents and by optimistic bias arising from information leakage in pharmacogenomic modeling.
Methods:
We integrated large-scale pharmacogenomic screening with baseline transcriptomes to derive a pathway-level FGFRi sensitivity phenotype. Drug responses were z-normalized per compound and aggregated across an FGFR-targeting panel to generate a composite FGFRi score, with reliability assessed by drug-drug concordance and split-half reproducibility. Transcriptome-based predictors were trained using strict group-aware cross-validation with out-of-fold (OOF) evaluation; performance was quantified by Spearman correlation and tested by permutation. An interpretable transcriptional program was extracted from linear model coefficients and projected into CHOL cohorts. Portability was assessed via cross-platform concordance in an orthogonal resource (GDSC) using matched cell lines and via biological coherence across multiple CHOL patient datasets, including rank-based scoring and a reduced feature panel.
Results:
The composite FGFRi score was more stable than single-drug readouts and showed strong split-half reliability (median ρ = 0.63) without confounding by drug coverage. Baseline transcriptomes predicted FGFRi sensitivity under leakage-safe evaluation and yielded a compact, interpretable signature. In TCGA-CHOL, the signature mapped to structured tumor states and associated with FGFR-axis components, showing inverse correlations with FGFR1/2/3 and a positive correlation with KLB. In matched cell lines, PRISM-derived scores/signatures aligned with GDSC sensitivity for representative FGFR inhibitors. In external CHOL cohorts, rank-based scoring and leakage-controlled proxy-label models showed consistent performance across datasets, and a reduced 15-feature panel preserved concordance with the full signature.
Conclusion:
A multi-compound, pathway-level FGFRi phenotype coupled with leakage-safe transcriptomic modeling identifies a transferable, interpretable FGFRi-associated program. This framework improves reliability relative to single-drug biomarker discovery and supports practical, portable scoring for CHOL stratification. Prospective validation in FGFRi-treated CHOL cohorts is warranted.
Insights
This study developed a reliable method to predict cholangiocarcinoma (CHOL) patient response to fibroblast growth factor receptor inhibitors (FGFRis) using transcriptomic data. The new approach improves biomarker discovery for personalized cancer treatment.
Area of Science:
- Genomics and Precision Medicine
- Cancer Biology and Therapeutics
Background:
- Fibroblast growth factor receptor inhibitors (FGFRis) show promise in treating cholangiocarcinoma (CHOL), but patient responses are inconsistent and not fully explained by known FGFR alterations.
- Current biomarker discovery methods for FGFRis are limited by variable drug sensitivity readouts and potential bias in pharmacogenomic modeling.
Purpose of the Study:
- To develop a robust and reliable method for predicting patient response to FGFR inhibitors (FGFRis) in cholangiocarcinoma (CHOL).
- To identify a transferable and interpretable transcriptomic signature associated with FGFRi sensitivity.
- To improve biomarker discovery for stratifying CHOL patients for targeted therapy.
Main Methods:
- Integrated large-scale pharmacogenomic screening with baseline transcriptomes to define a pathway-level FGFRi sensitivity phenotype.
- Developed a composite FGFRi score using normalized drug responses across an FGFR-targeting panel, assessing reliability via drug-drug concordance and split-half reproducibility.
- Trained transcriptome-based predictors using strict cross-validation with out-of-fold evaluation, quantifying performance by Spearman correlation and testing by permutation.
Main Results:
- A composite FGFRi score demonstrated higher stability and reliability (median ρ = 0.63) compared to single-drug readouts.
- Baseline transcriptomes accurately predicted FGFRi sensitivity, yielding a compact and interpretable signature.
- The derived signature mapped to tumor states in TCGA-CHOL, associated with FGFR-axis components, and showed consistent performance across external CHOL cohorts.
Conclusions:
- A multi-compound, pathway-level FGFRi phenotype combined with leakage-safe transcriptomic modeling identifies a transferable and interpretable FGFRi-associated program.
- This framework enhances reliability in biomarker discovery and supports practical, portable scoring for CHOL patient stratification.
- Prospective validation in FGFRi-treated CHOL cohorts is recommended.
