Irving, Texas, United States, Aug 27, 2026
Caris Life Sciences has announced the publication of a study in npj Precision Oncology describing the development and validation of an AI-driven molecular signature designed to guide first-line treatment selection in pancreatic cancer. The study evaluates Caris AI Insights™ and its ability to predict treatment benefit using complex molecular patterns derived through machine learning. In the testing cohort, patients classified as having standard molecular risk who were predicted to benefit from FOLFIRINOX achieved a median overall survival of 16.0 months, compared with 9.9 months for patients receiving gemcitabine plus nab-paclitaxel (gem/nab-p). The findings highlight the potential role of AI-guided precision medicine in helping clinicians incorporate tumor biology into treatment decisions for pancreatic ductal adenocarcinoma (PDAC), one of the most challenging cancers to treat. Caris said approximately half of the patients included in the study received a first-line treatment that differed from the therapy recommended by the model, suggesting an opportunity to further refine treatment selection using comprehensive molecular information.
AI Model Targets Pancreatic Cancer Treatment Selection
The Caris AI Insights pancreatic cancer signature uses machine learning to identify complex molecular patterns associated with real-world treatment benefit, rather than relying on a single biomarker. The validation analysis leveraged Caris’ large-scale clinico-genomic datasets, linking comprehensive molecular profiles with treatment outcomes from thousands of patients. The approach provides clinicians with two key outputs: molecular risk stratification and treatment guidance between FOLFIRINOX and gem/nab-p. Patients are categorized into standard- or high-molecular-risk groups, while the treatment-selection component is intended to identify patients who may derive greater benefit from one regimen over another. According to the published study, the model identified a meaningful subset of patients who may achieve similar or greater benefit from gem/nab-p while also identifying patients more likely to benefit from the greater treatment intensity associated with FOLFIRINOX. The company said this approach could help physicians move beyond treatment selection based primarily on clinical judgment and toward decisions informed by the molecular characteristics of individual tumors.
Study Highlights Potential Survival Benefit
For patients with advanced pancreatic ductal adenocarcinoma, commonly used first-line treatment options include FOLFIRINOX, gem/nab-p and NALIRIFOX. The study focused on the first two regimens, both of which can extend survival but differ in treatment intensity and potential toxicity. Caris noted that treatment decisions are currently not widely guided by an established biomarker capable of determining which patients are most likely to benefit from these therapies. In the study’s testing cohort, patients with standard molecular risk who were predicted by the AI model to benefit from FOLFIRINOX experienced median overall survival of 16.0 months, compared with 9.9 months among those receiving gem/nab-p. The findings suggest that tumor biology may provide clinically relevant information for selecting between treatment strategies. However, the study demonstrates the predictive performance of the investigational AI approach and does not establish that AI-guided treatment selection itself causes longer survival. Further clinical validation will be important to determine how the model performs across broader patient populations and how it may ultimately influence routine treatment decisions.
Advertisement
Caris Expands Precision Oncology Platform
The pancreatic cancer study forms part of Caris Life Sciences’ broader strategy to combine comprehensive molecular profiling, artificial intelligence and machine learning to support precision medicine. The company’s Caris AI Insights findings are incorporated into its Molecular Tumor Board Report and use data generated through whole exome sequencing (WES) and whole transcriptome sequencing (WTS). Caris said its disease-specific algorithms are being developed to support treatment decision-making across several tumor types, including colon, breast, ovarian, pancreatic and lung cancers. The company also stated that its MI Cancer Seek tissue-based assay received FDA approval in November 2024 and is the first and only simultaneous WES- and WTS-based assay with FDA-approved companion diagnostic indications for molecular profiling of solid tumors. By integrating molecular data with treatment outcomes and computational analysis, Caris aims to develop tools capable of identifying patterns that may not be apparent through smaller molecular panels. The latest pancreatic cancer findings add to the growing evidence base around AI-enabled therapy selection and illustrate how multimodal clinical and genomic datasets could increasingly support personalized oncology decisions.
Source: Caris Life Sciences press relese
