Abstract LB116: Real-world implementation of AI in target discovery: CURE AI predicts IL-15 as a therapeutic target in renal cancer

Abstract We developed a machine learning platform that learns and measures the individualized benefit of a therapeutic intervention over standard of care, which we named CURE AI (Clinical trials Uncovering Real Efficacy Artificial Intelligence). To build CURE AI, we trained a foundation model with clinical and multi-omics data from hundreds of thousands of oncology patients using a proprietary deep learning architecture and training schema. CURE AI is designed to understand the interplay between all baseline clinicogenomic information to predict the magnitude of benefit of therapy for each patient enrolled on a clinical trial. The foundation model allows CURE AI to predict a counterfactual outcome for every patient - a predicted outcome for the trial therapy they did not receive - based on finetuning our foundation model with conditional average treatment effect modeling. The result is a continuous benefit prediction ranking of patients (the CURE Index) based on the most important and predictive clinical and genomic factors, which can be utilized to informbiomarker and target discovery. For the phase 3 JAVELIN Renal 101 trial, which compared avelumab/axitinib to sunitinib in advanced renal cell carcinoma patients, CURE AI identified a population of patients who had superior response to axitinib/avelumab. We show that if this CURE AI-selected patient group was utilized for trial enrollment, the JAVELIN Renal 101 trial could have been significant with 25% of the patients who were enrolled on the completed trial (n=886 to n=222). Using our novel Accelerated Approval Modeling method, we show that the trial could have sought FDA approval at least 6 months earlier (while enrolling one-fourth of the patient population) with the use of CURE AI. Target discovery analysis of the CURE AI-selected patient groups revealed that IL-15 was predictive of patient response, and we predict success of IL-15 agonists in patients with renal cell cancer. In summary, we developed and validated a causal clinicogenomic deep learning platform that calculates individualized patient benefit of investigational therapy compared to an existing standard of care. CURE AI is available to implement on completed or actively enrolling clinical trials to identify the patients who receive a true benefit from investigational therapy as well as to perform biomarker and target discovery. Citation Format: Neil Thomas Pfister, Amit Weiss, Vitalay Fomin. Real-world implementation of AI in target discovery: CURE AI predicts IL-15 as a therapeutic target in renal cancer [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 2 (Late-Breaking, Clinical Trial, and Invited Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_2):Abstract nr LB116.