209P - A novel method for early evaluation of drug-specific predictive biomarker

Background

A key challenge in precision oncology is to develop predictive markers that identify patients likely to respond to a given treatment, ideally before clinical trials are performed so that they can be effectively used in clinical trial design. Even when markers are developed early, evaluating them is challenging in the absence of interventional data. We introduce the Surrogate Clinical Outcome Test (SCOT), which assesses, using non-interventional data, the predictivity of a drug-specific marker in clinical settings.

Methods

SCOT assumes that patients with low (native) expression of the investigated drug target(s) have a similar survival profile to patients actually receiving the drug. A marker score predictive of response to the drug will therefore have a negative hazard ratio (HR) in patients with low expression of the drug target(s). To exclude markers that are merely prognostic, SCOT requires that the HR is higher when the target expression is high. We tested SCOT on predictive markers previously published or generated using our ENLIGHT platform for 7 drug classes and on two known prognostic biomarkers: KI67 and Proliferation Index (PI).

Results

Of the 7 treatment response markers tested, SCOT classified 5 as predictive to the respective drug. 4 of the 5 were confirmed as predictive on interventional data (response prediction AUC > 0.6). Of the 2 markers SCOT classified as non-predictive, 1 was confirmed as such on interventional data. SCOT also correctly classified KI67 and PI as prognostic, non-predictive markers. Overall, SCOT’s classification was confirmed for 7 out of 9 biomarkers (78% accuracy).

Conclusions

SCOT is a novel method to test predictive markers for virtually any targeted cancer drug using big non-interventional clinical data. A key feature of this method is that it can evaluate predictive biomarkers for drugs, based on human data, even before the drug enters clinical development. Combined with our ENLIGHT platform for generating predictive markers in the absence of direct response data, this creates a powerful pipeline for early biomarker development and evaluation.

Clinical trial identification

Editorial acknowledgement

Legal entity responsible for the study

The authors.

Funding

Has not received any funding.

Disclosure

G. Dinstag, Y. Kinar, R. E. Bell, D. S. Ben-Zvi, E. Elis, E. Maimon, R. Aharonov: Financial Interests, Personal, Full or part-time Employment: Pangea Biomed. T. Beker: Financial Interests, Personal, Full or part-time Employment: Pangea Biomed; Financial Interests, Personal, Leadership Role: Pangea Biomed.