Precision oncology promises tailored treatments based on tumor molecular profiles, but the interpretation of complex genomic data remains a challenge. The diversity and intricacy of molecular alterations, coupled with a rapidly expanding landscape of targeted therapies, create a data interpretation bottleneck in clinical practice. Genomate Health’s Digital Drug Assignment (DDA)technology — commercially known as Genomate — addresses this challenge through a knowledge-based computational reasoning model that scores and ranks treatment options based on full tumor genomic profiles.
Previously, DDA was evaluated using clinical outcome data from the SHIVA01 trial, where 113 patients were treated with molecularly targeted agents (MTAs). This study demonstrated that high-score MTAs are more effective than other treatments, thus supporting the model's clinical utility. However, broader validation in a scalable and transparent model system was needed to confirm its predictive power.
This study, first presented at the 2025 Annual Meeting of the American Association for Cancer Research (AACR), extends the validation of the DDA system using large-scale patient-derived xenograft (PDX) data to assess its predictive power across a broad spectrum of cancer types and MTAs.
PDX models are created by implanting cancerous cells or tissues from patients’ tumors into immunodeficient mice to simulate human tumor biology in vivo. PDX systems have been extensively used in cancer research, and several platforms have validated that PDX models faithfully recapitulate human tumor biology and predict patient drug response. A unique advantage of the PDX model is that multiple treatments can be tested concurrently for a single tumor, thereby enabling direct evaluation of drug ranking.
We analyzed genomic profiles from 178 solid tumors utilizing the DDA model. These profiles were sourced from a PDX Clinical Trial (PCT) study conducted by Gao et al., which involved over 1000 PDX models and mimicked a clinical trial design. The tumors were subjected to 1151 monotherapy treatments using 23 distinct MTAs that target key oncogenic signaling pathways.
DDA was used to assign personalized scores to all available treatments for each tumor. Treatments were then grouped into ranked cohorts based on their DDA scores: highest-ranked drugs formed the top cohort, followed by progressively lower-scoring cohorts. This structure allowed direct comparison of predicted versus actual treatment outcomes using the following clinical endpoints: disease control rate (DCR), overall response rate (ORR), and progression-free survival (PFS).
A consistent, statistically significant pattern emerged: drugs ranked highest by DDA provided the most clinical benefit. The predictive gradient showed a clear, stepwise decline in treatment efficacy with increasing distance from the top-ranked cohort. This correlation held true across all outcome metrics (DCR, ORR, PFS), reinforcing the robustness of the DDA model.
To the best of our knowledge, at the time of this publication, this study represents the only large-scale experimental validation of a computational drug ranking system for precision oncology.
PDX systems have long been recognized for faithfully simulating human tumor biology and predicting patient drug response. This study capitalizes on their power to provide empirical support for the DDA system’s ability to transform genomic data into actionable treatment rankings.
The current paradigm of treatment selection often relies on forming subgroups based on one or two biomarkers within tumor types categorized by location and histology. This reductionist approach stands in contrast to the principles of precision oncology, which seeks to personalize therapy based on the unique molecular architecture of each tumor.
Despite advances in molecular diagnostics, the full potential of these data is often underutilized. Clinicians may overlook valuable genomic information due to the inherent complexity of interpretation and the limitations of manual review. While Molecular Tumor Boards (MTBs) aim to address this gap, they are labor-intensive, subject to inter-observer variability, and not scalable for widespread clinical adoption.
Genomate solves these problems by providing:
By transforming data into precise, ranked treatment options, Genomate empowers oncologists to make confident, individualized therapy decisions, even in complex cases where standard options are unclear or exhausted.
Clinical trial failure is a major concern in oncology, with nearly 90% of drug candidates failing to demonstrate efficacy. A major contributor to this attrition is the inadequacy of current enrollment strategies, which often select participants based on narrow biomarker criteria or disease subtypes, overlooking the full molecular complexity of the tumor.
Genomate can reshape clinical trial design and execution by:
By incorporating Genomate into the clinical trial process, sponsors and investigators gain a fast, scalable, and scientifically rigorous tool to increase trial efficiency, reduce costs, and accelerate time-to-approval for targeted therapies.
Molecularly targeted therapies have unlocked the potential for causal, personalized cancer treatment. Yet, the field continues to grapple with the complexity of tumor heterogeneity and the limitations of single-biomarker paradigms. Genomate Health's DDA (commercially known as Genomate) addresses these challenges by interpreting the full spectrum of tumor molecular data to rank therapies in a clinically meaningful way. Initially validated using patient outcomes from the SHIVA01 clinical trial, DDA has been further substantiated through this large-scale patient-derived xenograft (PDX) treatment data analysis.
This dual-layered validation, across both human clinical trials and preclinical PDX models, confirms DDA as a predictive, objective, reproducible, and scalable solution for precision oncology. By fully leveraging comprehensive tumor molecular profiles, DDA enables clinicians and researchers to move beyond traditional paradigms and toward a future where every patient receives the treatment most likely to benefit them, from day one.
Want to learn more about this breakthrough in computational precision oncology?
📍 Visit Poster 3647 / 9 (Section 45) Validation of a computational reasoning model for precision oncology with large-scale patient-derived xenograft data
🕐 4/28/2025, 2:00 PM - 5:00 PM
👨🔬 Presented by Dr. Istvan Petak, Genomate Health
Session Title: Artificial Intelligence and Machine Learning for Therapeutic Election and Discovery
Come see the full data and speak with our team about how Genomate can power smarter, faster, and more personalized cancer treatment and trial design.
.png)
