Tempus AI Shows IPS Test Cuts Hazard Ratio to 0.45, Improves ICI Predictions

TEMTEM

Tempus AI’s Immune Profile Score delivered a hazard ratio of 0.45 across four pan-cancer metastatic cohorts, outperforming TMB, MSI and PD-L1 in predicting immunotherapy benefit. IPS also flagged 13% of microsatellite stable colorectal cancer patients (HR=0.2) and 17% of rare metastatic tumors as IPS-High (HR=0.26).

1. Tempus’ Immune Profile Score Outperforms Conventional Biomarkers in Predicting Immunotherapy Outcomes

In a new multicenter study of pan-cancer metastatic solid tumor patients, Tempus’ Immune Profile Score (IPS) demonstrated superior prognostic power compared to tumor mutational burden, microsatellite instability and PD-L1. Across four independent validation cohorts, IPS achieved a hazard ratio of 0.45 for overall survival in patients treated with immune checkpoint inhibitors (ICIs), nearly doubling the predictive accuracy of existing biomarkers. Notably, IPS identified 13% of microsatellite stable colorectal cancer patients with a hazard ratio of 0.20, indicating they derive significant survival benefit from ICI therapy despite being classified as low-priority by conventional tests. In a cohort of rare metastatic solid tumors, 17% were classified as IPS-High and showed a hazard ratio of 0.26 for median real-world overall survival, underscoring the algorithm’s ability to uncover new candidate populations for immunotherapy beyond current FDA labels. The IPS test, available as an add-on to Tempus’ xT and xR sequencing assays, integrates DNA and RNA markers—both established and novel—to deliver a multimodal assessment that empowers oncologists to tailor ICI treatment decisions with greater confidence and precision.

2. Tempus Expands AI Pathology Footprint with Paige Acquisition and Paige Predict Launch

Tempus AI has broadened its pathology offering through the strategic acquisition of Paige, a leader in AI-driven digital pathology, and the rollout of a new platform, Paige Predict. This combined solution now supports biomarker insights across 16 major cancer types—including breast, lung, colorectal and prostate—by integrating whole-slide image analysis with genomic and transcriptomic data. Paige Predict uses deep-learning models validated on over 200,000 annotated slides and correlates image-based features with clinically relevant genomic alterations, enabling pathologists to identify actionable mutations such as ERBB2 amplifications and TP53 mutations in real time. The integration with Tempus’ existing multimodal database of over two million patient records accelerates discovery of novel histogenomic signatures and promises to reduce diagnostic turnaround by up to 40%. By embedding Paige Predict into its AI Operating System, Tempus aims to support clinical trial enrollment, companion diagnostic development and personalized treatment planning at hundreds of partner hospitals and biopharma programs worldwide.

Sources

ZNB