AI-Powered Cancer Biomarker Detection Shows Promising Results in Treating Metastatic Colorectal and Breast Cancers

Strategic Insights -

Researchers from Caris Life Sciences have developed a novel AI framework that achieves clinical-grade performance in detecting cancer biomarkers, particularly microsatellite instability (MSI)/mismatch repair deficiency (MMRd) and programmed death-ligand 1 (PD-L1), in colorectal and breast cancers. The study, published in Communications Medicine, reveals that patients with biomarker-positive model predictions demonstrate prolonged time-on-treatment (TOT) and overall survival (OS) when treated with pembrolizumab. The AI framework integrates features from both hematoxylin & eosin (H&E) and immunohistochemistry (IHC) stained whole slide images, improving predictive accuracy and prognostic precision compared to current biomarker assessments.

Key Takeaways:

  • Researchers at Caris Life Sciences developed a dual-modality transformer-based model for predicting MSI/MMRd and PD-L1 status using H&E and IHC stained whole slide images.
  • The AI framework achieved clinical-grade performance, with area under the receiver operating curve (AUROC) exceeding 0.97 for MSI/MMRd prediction in colorectal cancer and 0.96 for PD-L1 prediction in breast cancer.
  • Patients with biomarker-positive model predictions demonstrated prolonged TOT and OS when treated with pembrolizumab.
  • The model's predictions were superior to PD-L1 IHC in stratifying patients with improved outcomes on pembrolizumab, suggesting a reevaluation of existing PD-L1 status thresholds.
  • The study promotes the integration of advanced AI tools in clinical pathology to enhance the precision and efficiency of cancer biomarker evaluation.
  • The AI framework offers a customizable framework for varied clinical scenarios, enhancing predictive accuracy and prognostic precision.

Statistics:

  • AUROC for MSI/MMRd prediction in CRC: 0.97
  • AUROC for PD-L1 prediction in breast cancer: 0.96
  • TOT for patients with biomarker-positive model predictions: prolonged
  • OS for patients with biomarker-positive model predictions: prolonged
  • Hazard ratios (HR) for TOT and OS: determined using the Cox proportional hazard model

Sources:

  • Synergistic H&E and IHC image analysis by AI predicts cancer biomarkers and survival outcomes in colorectal and breast cancer. Communications Medicine, 2025,5(1):1-15.
  • https://doi-org.sdpl.idm.oclc.org/10.1038/s43856-025-01045-9 (free journal article available online)
  • Yating Cheng, Caris Life Sciences, [yating.cheng@carislifesciences.com](mailto:yating.cheng@carislifesciences.com)