Abstract LB129: Clinical efficacy of computational reasoning for personalized treatment planning in a pan-cancer cohort discussed by a French multidisciplinary tumor board: A real-world experience-based analysis

Abstract Digital Drug Assignment (DDA) is a computational reasoning model that recommends cancer therapies for the complete individual molecular tumor profiles and ranks them by their DDA scores. Prior analysis of the SHIVA01 cohort linked higher DDA scores to improved outcomes (Petak et al., 2021). Here, we evaluated predefined DDA tiers in a broad, real-world cohort from Institut Curie’s Molecular Tumor Board (MTB). We retrospectively analyzed 394 MTB cases (2018-2022, adults w solid tumors) with NGS/WES/WGS data, treatment records, and outcomes. Patients receiving molecularly targeted agents (MTAs; n=134) or chemotherapy (n=177) were included. Administered MTAs (including ICIs) were assigned DDA scores and stratified into low (<0), intermediate, and high (≥1000) tiers. PFS, OS, ORR, DCR, and survival rates were compared across tiers and in relation to chemotherapy. Clinical outcomes improved consistently with higher DDA tiers (see table). Median PFS increased from 3.4 (low) to 6.8 months (high), and median OS from 7.8 to 16.6 months. Intermediate-tier MTAs performed similarly to chemotherapy (mPFS 4.5 vs 4.9 months; mOS 9.0 vs 9.8 months). ORR, DCR, 6-month PFS and 24-month OS all showed positive trends across tiers. DDA-high therapies provided the largest benefit, while DDA-low MTAs underperformed chemotherapy. Cases with no molecular-drug link (n=5) had the poorest outcomes (mPFS and mOS: 2.6 and 6.5 months). In this large, real-world pan-cancer cohort, DDA robustly differentiated therapies by clinical efficacy using each patient’s full molecular profile, independently validating the consistency of treatment-outcome associations across pre-established DDA tiers. These results urge the integration of DDA’s computational reasoning into MTB workflows to ensure consistent, high clinical performance and safety in the implementation of precision oncology. Citation Format: Barbara Vodicska, Eniko Kispeter, Dora Lakatos, Gabor G. Kalmar, Robert Doczi, Dora Gorog-Tihanyi, Anna Dirner, William T. Beck, Ivan Bieche, Edith Borcoman, Nicolas Servant, Kenza Nedara, Sarah Watson, Celia Dupain, Istvan Petak, Christophe Le Tourneau. Clinical efficacy of computational reasoning for personalized treatment planning in a pan-cancer cohort discussed by a French multidisciplinary tumor board: A real-world experience-based analysis [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2026; Part 2 (Late-Breaking, Clinical Trial, and Invited Abstracts); 2026 Apr 17-22; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86(8_Suppl):Abstract nr LB129.