Valar Labs has published peer-reviewed evidence in the Journal of Clinical Oncology showing its AI can predict which chemotherapy will work better for individual pancreatic cancer patients - and the survival differences are significant.
The study validated Vitara Pancreas ChemoPredict on 173 patients who had already received one of two standard chemotherapy regimens: FOLFIRINOX or gemcitabine/nab-paclitaxel. By analyzing their tumor pathology slides, the AI correctly identified which patients would have responded better to each treatment.
Patients who received the chemotherapy the AI predicted would work best for them lived 2.7 to 2.9 months longer than those who received the opposite treatment. In a disease where median survival is measured in months, that difference matters.
The Problem This Solves
When oncologists treat advanced pancreatic cancer, they typically choose between two main chemotherapy regimens. Both work for some patients. Neither works for everyone. Without a reliable way to predict response, treatment selection often comes down to clinical judgment and patient factors like overall health.
The study found that more than 40% of patients in the cohort likely would have fared better if they had received the opposite treatment from what they got. That is a lot of patients going through toxic chemotherapy that was never their best option.
How It Works
Valar Labs analyzes standard H&E-stained pathology slides - the same slides pathologists already create for diagnosis. The AI extracts visual patterns from the tissue that correlate with treatment response but are not visible to human eyes.
“There is a wealth of predictive information hidden within standard pathology images that the human eye cannot quantify, but AI can,” said Viswesh Krishna, co-founder and CTO of Valar Labs.
The test requires no additional tissue samples, no genetic sequencing, and no new procedures. It works with slides that already exist as part of routine diagnosis.
Clinical Availability
Following the JCO publication, Valar Labs is making Vitara Pancreas ChemoPredict available through early access. Oncologists can order the test to help guide first-line chemotherapy selection for advanced pancreatic cancer patients.
The company has not disclosed pricing or insurance coverage details.
The Bigger Picture
Valar Labs is one of several companies racing to bring AI-powered cancer diagnostics to market. The appeal is obvious: these tools analyze existing data to provide actionable treatment recommendations. If they work as advertised, they could meaningfully improve outcomes while adding minimal cost or complexity to clinical workflows.
“I think it’s a really promising direction for getting the right treatment to the right person,” said Danielle Bitterman, an oncologist and AI researcher at Dana-Farber Cancer Institute.
But integration remains a challenge. Oncologists must decide whether to trust AI recommendations that sometimes contradict their clinical intuition, and liability questions remain unresolved when AI-guided treatment decisions lead to poor outcomes.
What We Do Not Know Yet
The 173-patient study provides validation but not definitive proof. Prospective trials - where the AI actually guides treatment decisions and outcomes are tracked - would provide stronger evidence. The company has not announced such trials.
The survival improvements are clinically meaningful but modest. Whether they hold up in broader, more diverse patient populations remains to be seen. And the test currently only addresses the choice between two specific regimens; it does not predict response to newer therapies or combination approaches.
Still, for a cancer where treatment options are limited and prognosis is poor, even incremental improvements in treatment matching could help.