Abstract A135: Are survival benefits of new drugs in combinations due to modest benefits in most patients or large benefits in few patients?

Abstract

Listen

Translate article

Abstract INTRODUCTION Combination therapy is a cornerstone in cancer treatment, often developed by adding new therapies to a standard-of-care treatment. However, clinical trials of adding a drug to a combination (A versus A+B) face an ambiguity in interpretation, because a higher Kaplan-Meier survival curve can result from modest benefits to overall survival (OS) or progression-free survival (PFS) in most patients, or from larger benefits in a few patients. This ambiguity exists even with complete patient data because no individual patient is simultaneous enrolled in control and test arms. Thus, clinical trials of drug combinations do not reveal the percentage of patients who benefit from an added drug and the extent of that benefit. METHODS We developed a deconvolution algorithm to estimate the distributions of survival benefits of new drugs in combination regimens. Among FDA approvals of combination therapies for advanced cancer between 1995 and 2020, we identified and analyzed 92 trials which compared a ‘standard-of-care’ control arm with ‘standard-of-care plus a new therapy’. We used deconvolution to infer all possible distributions of benefits of the new therapy. These possibilities differ by cross-resistance between drugs, and we therefore leveraged cross-resistance data from large-scale pre-clinical drug response encyclopedias to estimate the most likely distribution of survival benefits of new drugs. Our methods were further validated using patient-derived xenograft (PDX) studies and clinical monotherapy data. RESULTS Clinical trials with seemingly modest PFS or OS benefits from combination therapy were commonly explained by longer survival benefits in a minority of patients. Across many trials, on average 50% of patients experience a PFS or OS benefit of at least 1 month from new drugs, and median duration of benefit in such patients was on average 3 months longer than median benefit reported in the clinical trial. Data from the Novartis PDX encyclopedia (in which single tumors’ responses to both control and test treatments can be measured) confirmed that our deconvolution algorithm was successful at resolving the contributions of individual drugs in combinations. CONCLUSION We developed a broadly applicable method to estimate the proportion of patients that benefit from additional drugs and the durations of those benefits, thus resolving an ambiguity in the interpretation of drug combination trials. Most approved drug combinations had a proportion of patients who experienced much longer survival benefits than was visually apparent in Kaplan-Meier curves. These findings reconcile conflicting interpretations of drug trials in oncology, where health economists often see modest benefits, whereas clinicians observe long-lasting and worthwhile benefits in (some of) their patients. Seemingly modest survival benefits of new drugs in combinations may often be attributed to a lack of patient stratification rather than a lack of drug efficacy, which calls for greater use of biomarkers and precision strategies in the development of combination therapies. Citation Format: Haeun Hwangbo, Adam C Palmer. Are survival benefits of new drugs in combinations due to modest benefits in most patients or large benefits in few patients? [abstract]. In: Proceedings of the AACR-NCI-EORTC Virtual International Conference on Molecular Targets and Cancer Therapeutics; 2023 Oct 11-15; Boston, MA. Philadelphia (PA): AACR; Mol Cancer Ther 2023;22(12 Suppl):Abstract nr A135.