Real-world performance of the digital drug-assignment system for precision oncology in lung cancer.

Abstract

e13590 Background: Precision medicine has profoundly transformed lung cancer treatment, an array of molecularly targeted agents (MTAs) are approved and used in clinical practice. Comprehensive molecular testing is key in treatment decisions, nevertheless, interpretation of complex molecular profiles can be challenging and subjective. Previously we demonstrated that digital drug assignment (DDA), an algorithmic computational reasoning model that ranks associated targeted therapies based on the totality of individual tumor genomic data, rather than matching one drug to one biomarker, was predictive of relative benefit of the agents as used in the SHIVA01 trial (1). Here, we collected real-world clinical outcome data from lung cancer patients who received decision support where DDA was integrated to aid a molecular tumor board (MTB) and investigated the effectiveness of administered therapies. Methods: Between 2018 and 2022, 111 lung cancer patients were involved in our precision oncology program. Following molecular diagnostic tests, analysis of the individual molecular profiles and scoring of associated drugs (DDA score) was performed by the DDA system as previously described. The output was assessed by the MTB that provided a strategy to the clinicians who made the final therapy decisions. Treatment courses and outcomes were collected from hospital health records and analyzed. Treatment lines were categorized as the following: high-score MTA (1000≤DDA score), low-score MTA (DDA score<1000), and standard of care (SOC). Results: Overall response rate (ORR) of high-score MTA lines administered following DDA-based decision support (n=13) was 69%, whereas of other treatment lines (low-score MTAs and SOC; n=55) it was 16% (p<0.001); and 26% in low-score MTA lines (n=34; p=0.017). Median progression-free survival (mPFS) of high-score MTA lines administered following DDA-based decision support (n=13) was 63 months (95% CI: 12.0 – 66.0), whereas it was 9 months both for any other treatment lines (low-score MTAs and SOC; n=44; 95% CI: 5.0 – 66.0; p<0.001) and for low-score MTA lines only (n=27; 95% CI: 5.0 – 66.0; p<0.001). Median overall survival (mOS) of patients who received at least one line of high-score MTA after decision support (n=13) was significantly longer than of those who only received non-high score treatments after decision support (patients only treated with low-score MTAs and/or SOC, n=34): not reached vs. 32 months (95% CI 24.0 – not reached), respectively (p=0.012). Conclusions: This study revealed that the DDA system is predictive of relative benefit of the various agents as used in lung cancer care, in line with our previous clinical validation. Validated algorithmic AI-guided drug assignment methods represent a new class of tools that can potentially address challenges in utilizing the results of complex molecular profiles in routine clinical setting. 1. Petak et al., 2021, npj Precis. Onc. 5, 59.