Clinical practice and implications of biomarker testing in biliary tract cancer: An observational study

Abstract

Background & aims: Biliary tract cancers (BTC) are aggressive malignancies with limited treatment options. Owing to the high frequency of actionable genomic alterations (GA) and the availability of targeted therapies, molecular testing has become increasingly important; however, its clinical implementation remains inconsistent. This study aimed to evaluate real-world molecular testing practices, characterize the BTC molecular landscape, and assess the prognostic and predictive relevance of selected GA.

Methods: We retrospectively analyzed genomic and clinical data from 1,521 patients treated at 18 centers in Germany and Austria. A side-by-side comparison of clinical grade reports generated on two different sequencing platforms was performed for 90 patients.

Results: Twenty-four different NGS panels were used across 18 centers. A comparative analysis highlighted the significant variability in reports used to inform therapeutic decisions in clinical practice. Although there were substantial differences in the number of GA covered, the broader panels identified a similar number of actionable GA, indicating that key therapeutic targets are sufficiently represented. Integration with clinical data suggested that certain GA, such as HER2 amplifications (3%), BRAF ^V600E mutations (2%), and FGFR2 alterations (14%), may have prognostic significance beyond their predictive value. Patients with actionable alterations (610, 40%) that were treated accordingly (n = 204, 13%) had prolonged overall survival (31.8 months vs. 22.8 months, p <0.01).

Conclusion: Standardized biomarker testing is crucial for effective integration of targeted therapies in the management of BTC. Our findings reinforce the value of targeted treatments and underscore the predictive and prognostic significance of selected GA.

Impact and implications: Genomic profiling is recommended in patients with biliary tract cancers (BTC) but lacks harmonization across platforms and centers. By retrospectively analyzing genomic and clinical information from 1,521 patients with BTC diagnosed and treated at 18 centers in Germany and Austria, we provide real-world insights into the implementation of molecular profiling in BTC, highlighting variability in next generation sequencing-based testing and its impact on the detection of genomic alterations. Standardized molecular testing strategies will be key to enable the integration of more consistent and comparable genomic datasets across studies. Further, by elucidating the prognostic relevance of individual genomic alterations, our insights carry significant implications for interpreting single-arm clinical trials within genomically stratified patient cohorts and underscore the importance of randomized studies to delineate the benefit of targeted therapies.

Keywords: biopsy; molecular targeted therapy; next-generation sequencing; panel sequencing; precision oncology.

Graphical abstract

Fig. 1

Percentage of patients with any alteration in specific genes separated by the applied NGS panel and the participating center. (A) Only included assays applied in at least 25 cases. Blanks indicate a lack of coverage, as per the technical specifications of the assay. (B) Depicts only participating centers that included more than 20 patients in the analysis. Zero indicates that no GA in the respective genes was reported, either because of a lack of coverage or non-detection in the respective cohort. Archer, Archer FusionPlex Core Solid Tumor Panel; FMI, FoundationOne CDx; GA, genomic alterations; Handle, HANDLE Classic NGS Panel; MAPK1, GeneRead DNA seq Custom Panel v2; NGS, next-generation sequencing; OCC, Oncomine Comprehensive Cancer Panel v3; OFA, Oncomine Focus Assay; TSO170, TruSight Oncology 170; TSO500, TruSight Oncology 500.

Fig. 2

Direct comparison of the results of two independently generated reports for 90 patients sequenced on two different platforms (FMI CDx and TSO 500, DNA part). Each column represents an individual patient. Grey boxes indicate that the alteration was detected but was reported as a variant of uncertain significance. Only GA reported by at least one assay as clinically significant are shown. FMI, FoundationOne CDx; GA, genomic alterations; TSO, TruSight Oncology 500.

Fig. 3

Frequency of GA. (A) Frequency of the respective GA in the full cohort and according to anatomic subtype and (B) co-alteration spectrum when stratified according to key actionable GA. Calculations were based only on panels that covered the respective GA. eCCA, extrahepatic cholangiocarcinoma; FGFR2, FGFR2 fusions and activation mutations; GA, genomic alterations; GBCA, gallbladder carcinoma; iCCA, intrahepatic cholangiocarcinoma.

Fig. 4

Actionable GA. (A) 735 actionable GA were identified in 610 patients (503 patients: one GA, 91 patients: two GA, 14 patients: three GA, 2 patients: four GA). Treatment allocation for six key actionable GA (small circles). For patients with sequential targeted therapies, only the first targeted treatment is shown (further specified in Table S1). (B) OS from diagnosis for patients with and without actionable GA (p = 0.207). (C) OS from diagnosis for patients with an actionable GA who received or did not receive targeted therapy, and for patients without actionable GA (+ TTx vs. - TTx: p <0.001; HR 0.62; 95% CI 0.50–0.77). (D) OS from start of palliative treatment (+ TTx vs. - TTx: p <0.001; HR 0.54; 95% CI 0.43–0.68 and (E) OS from start of second-line treatment (+ TTx vs. - TTx: p = 0.005; HR 0.66; 0.50–0.87). Amp, amplification; CTx, chemotherapy; Del, deletion; GA, genomic alteration; HR, hazard ratio, mOS, median overall survival; TTx, targeted therapy. Statistical tests: Kaplan-Meyer estimates, log-rank tests.

Fig. 5

OS for patients with an actionable FGFR2 GA (fusions and activating mutations) or IDH1 mutations who received TTx, and for patients with and without the respective actionable GA. (A) OS for FGFR2 GA from diagnosis (+ TTx vs. - TTx: p = 0.033; HR 0.65; 95% CI 0.45–0.96). (B) OS for FGFR2 GA from the start of palliative treatment (+ TTx vs. - TTx: p = 0.001; HR 0.57; 95% CI 0.38–0.86). (C) OS for FGFR2 GA from the start of second-line treatment (+ TTx vs. - TTx: p = 0.235; HR 0.72; 95% CI 0.44–1.19). (D) OS for patients with IDH1 mutations from diagnosis (+ TTx vs. - TTx: p = 0.279; HR 0.81; 95% CI 0.53–1.24), (E) from start of palliative treatment (+ TTx vs. - TTx: p = 0.073, HR 0.71; 95% CI 0.45–1.11), (F) and from start of second-line treatment (+ TTx vs. - TTx: p = 0.310; HR 0.75; 95% CI 0.43–1.30). HR, hazard ratio; mOS, median overall survival; TTx, targeted therapy. Statistical tests: Kaplan-Meyer estimates, log-rank tests.

Fig. 6

OS for patients with a BRAF^V600E mutation, ERBB2 amplification or BRCA1/2 alteration who received or did not receive TTx, and for patients without the respective GA. (A) OS for BRAF^V600E patients from diagnosis (+ TTx vs. - TTx p = 0.015; HR 0.27; 95% CI 0.11–0.69, (B) from the start of palliative treatment (+ TTx vs. - TTx: p = 0.003; HR 0.19; 95% CI 0.07–0.51), and (C) from the start of second-line treatment (+ TTx vs. – TTx: p <0.001; HR 0.12; 95% CI 0.04–0.38). (D) OS for ERBB2 amplified patients from diagnosis (+ TTx vs. - TTx: p = 0.218; HR 0.64; 95% CI 0.29–1.38), (E) from the start of palliative treatment (+ TTx vs. - TTx: p = 0.012; HR 0.42; 95% CI 0.19–0.95), and (F) from the start of second-line treatment (+ TTx vs. - TTx: p = 0.113; HR 0.62; 95% CI 0.19–2.07). (G) OS for BRCA1/2 GA from diagnosis (+ TTx vs. - TTx: p = 0.340; HR 0.71; 95% CI 0.33–1.55), (H) from start of palliative treatment (+ TTx vs. - TTx: p = 0.179; HR 0.62; 95% CI 0.29–1.36) and (I) from the start of second-line treatment ((+ TTx vs. – TTx: p = 0.402; HR 0.68; 95% CI 0.27–1.72). GA, genomic alterations; HR, hazard ratio; mOS, median overall survival; TTx, targeted therapy. Statistical tests: Kaplan-Meyer estimates, log-rank tests.

Fig. 7

Overall survival from diagnosis for patients with and without specific GA. Overall survival from diagnosis for patients with and without GA in KRAS (A: p = 0.618), TP53 (B: p = 0.003; HR 1.29; 95% CI 1.09–1.52), MTAP (C: p = 0.003; HR 1.76; 95% CI 1.2–2.6), BAP1 (D: p = 0.008; HR 0.69; 95% CI 0.53–0.91), CDKN2A/B (E: p = 0.022; HR 1.36 1.1–1.8]), and MDM2 (F: p = 0.819; HR 0.96; 95% CI 0.66–1.38). GA, genomic alterations; HR, hazard ratio; mOS, median overall survival. Statistical tests: Kaplan-Meyer estimates, log-rank tests.