Utility and costs of surrogate endpoints (SEs) and biomarkers in phase I oncology trials

Abstract

6012 Background: The advent of targeted therapies, functional imaging and translational research has driven the use of a new class of endpoints in phase I oncology trials, known as surrogate endpoints (SEs) or biomarkers. Investigators hope that SEs will improve the efficiency of drug-development. However, whether SEs can substitute for traditional endpoints (e.g. toxicity) is unknown. The role of SEs in drug selection, target validation, dosing and schedule optimization remains to be defined. Methods: We analyzed all phase I single-agent studies in PASCO from 1992 to 2002 that included at least one SE. Subsequent publications of these studies were analyzed with respect to the primary SE. Drugs were classified as cytotoxic, biologic (recombinant), or targeted (small molecules) and SEs classified by their technology (imaging, blood biomarkers or advanced histology). We designed 4 yes-no questions to evaluate the role of SEs, beginning with: “Did the SE help…” 1) “…determine the dose for phase II?”; 2) “…in finding the schedule?”; 3) “… with the author's conclusions?”; and 4) “… validate that the target was affected?”. The utility of SEs across these questions and across drug classes was compared by McNemar's and chi-square tests, respectively. We also related the budgets of 18 single-institutional trials to their number of endpoints. Results: Of 65 trials, 55% tested biologics, 24% small molecules, 15% cytotoxic and 6% other. The percentages of primary SE's by technology were: 68% blood studies, 25% histologic analysis, and 6% imaging. The frequency of “yes” for the 4 questions was 14%, 15%, 34% and 62%, respectively (p<0.001 between questions), with no statistical differences between drug classes. The marginal budgeted cost of adding an additional secondary endpoint was US$6,675 per patient. Conclusions: SEs differ in methods and goals. SEs only modestly aided in defining dose and schedule for future studies and in the overall drug-development process. They helped in validating that the therapy affected the intended target. Better preclinical evaluation of SEs may enhance their utility. Further research should help define how best to incorporate SE's into trial design. No significant financial relationships to disclose.