Association of candidate surrogate endpoints with overall survival in advanced biliary tract cancer

Background & Aims

The use of surrogate endpoints in trials including biliary tract cancers (BTC) is growing. While this may expedite drug approval and decrease costs, it may not always correlate with an overall survival (OS) advantage. We aimed to explore the association of progression-free survival (PFS), objective response rate (ORR) and disease-control rate (DCR) with OS at the trial- and patient-level.

Methods

For the trial-level analysis, we performed a systematic review of Pubmed/MEDLINE, Embase, Cochrane, clinicaltrials.gov and conference proceedings for phase II-III trials in advanced BTC. We used a weighted linear regression to measure the correlation of OS with PFS, ORR and DCR. For the patient-level analysis, we analyzed patients included in five randomized trials and three real-world datasets. The protocol is registered with PROSPERO, CRD42023398279.

Results

For the trial-level analysis, we included 41 studies, involving 88 treatment arms and 7817 patients. The coefficient of determination (R²) of the model was 0.71 (95% CI 0.56-0.86) for PFS, 0.01 (0-0.08) for ORR and 0.39 (0.14-0.64) for DCR. Predefined subgroup analysis showed consistent results. For the patient-level analysis, we included a total of 2506 patients, 783 in randomized trials (first-line 512, second-line 271) and 1723 in routine clinical care (first-line chemotherapy 773, first-line chemotherapy-durvalumab 628, second-line chemotherapy 322). Across the distinct datasets, the correlation coefficient ranged from 0.73 to 0.86 for PFS. A responder analysis found no association between response and survival.

Conclusion

PFS shows a moderate correlation with OS both at the trial- and patient-level, while ORR and DCR show a low correlation. Whilst PFS is currently the best candidate surrogate marker for OS, our results highlight the need for novel endpoints in this field.

IMPACT AND IMPLICATIONS

The use of validated surrogate endpoints in biliary tract cancer trials may decrease costs and improve study feasibility, particularly with agents that only target small subsets of patients or in trials that incorporate a cross-over design. A formal statistical validation of surrogacy requires patient-level and trial-level data. This is the first comprehensive analysis to incorporate novel agents (including immunotherapies and targeted agents), include patient-level data and rigorously and homogeneously extract appropriate measures of treatment effect for endpoint correlation. These results show a moderate correlation for progression-free survival both at the trial- and patient-level and a low correlation for disease-control rate and response rate. This information will aid clinicians in appropriately interpreting contemporary clinical trials and guide clinical researchers and trial sponsors involved in clinical trial design. Furthermore, it has important implications for the regulatory approval process and may aid agencies in appropriately evaluating novel drugs.