AI Pre-Screening Helps Identify Lung Cancer Cases for Additional Biomarker Testing

The research examines whether routine pathology images can help identify patients who may benefit from additional biomarker testing to guide treatment decisions.

The multicenter study analyzed 60,459 advanced non-small cell lung cancer (NSCLC) cases from 56,969 patients. The AI-enabled panel uses routinely generated hematoxylin and eosin (H&E)-stained whole-slide pathology images to predict potentially clinically relevant genomic and immunohistochemistry (IHC) biomarker status, helping identify patients who may require confirmatory molecular testing.

The panel achieved sensitivity above 80% across all 12 biomarkers evaluated, with sensitivity exceeding 90% for several markers. The findings suggest that AI analysis of existing pathology images could help prioritize additional diagnostic testing and identify patients who may benefit from further molecular characterization of their tumors.

As biomarker-directed therapies for advanced NSCLC continue to expand, identifying clinically actionable alterations can require multiple tests and sufficient tumor tissue. AI-based pre-screening could help laboratories determine which molecular tests are most relevant, potentially reducing diagnostic delays while conserving limited tissue for confirmatory analysis.

The study was conducted through an ongoing collaboration between Tempus and Pfizer focused on addressing barriers to biomarker testing. The research evaluates how AI-assisted pathology analysis could support diagnostic decision-making across the broader NSCLC treatment landscape, regardless of the therapies patients ultimately receive.

The findings build on Tempus’ existing AI-based biomarker prediction technologies, including Paige Predict and Tempus Preview. Paige Predict analyzes H&E whole-slide images to estimate the likelihood of 123 biomarkers across multiple cancer types when tissue is insufficient for sequencing, while Tempus Preview provides early predictions of clinically relevant biomarkers approximately 24 hours after sample receipt.

The H&E pre-screening panel remains in development and is not intended to replace established molecular or IHC testing. Instead, the technology is being evaluated as a tool to guide subsequent diagnostic testing. Its potential clinical applications include improving biomarker testing workflows and helping clinicians identify patients who may require further testing before treatment selection.

“Precision medicine only works when clinicians have the right information at the right time,” said Ezra Cohen, MD, Chief Medical Officer of Oncology at Tempus. He added that the findings demonstrate the potential for AI to extract useful information from routine pathology images and help clinicians determine which additional diagnostic tests may be most important.

Related Links
Tempus AI