Sign in Subscribe
Machine learning

Breakthrough in Personalized Medicine: Machine Learning Signature for Clinical Prognosis and Immunotherapy in Colon Adenocarcinoma

Strategic Insights

— 1 min read

A recent study published in Scientific Reports has made a significant contribution to the field of personalized medicine by developing a machine learning-derived angiogenesis signature for clinical prognosis and immunotherapy guidance in colon adenocarcinoma. The study, conducted by researchers at Tengzhou Central People's Hospital, found that an integrative machine learning approach can construct a prognostic consensus angiogenesis-related signature (CARS) that provides superior performance for clinical prognostic prediction and serves as an independent risk predictor for COAD. The research highlights the potential of personalized treatments for patients with COAD, with axitinib and olaparib being promising treatment options for those with high CARS scores.

Key Takeaways:

  • The study developed a machine learning-derived angiogenesis signature for clinical prognosis and immunotherapy guidance in colon adenocarcinoma (COAD).
  • The signature, known as the prognostic consensus angiogenesis-related signature (CARS), was constructed using an integrative machine learning approach and had superior performance for clinical prognostic prediction.
  • Patients with low CARS scores had a superior prognosis, characterized by immune activation, elevated tumor mutation/neonantigen burden, and greater responsiveness to immunotherapy.
  • Patients with high CARS scores had a poor prognosis, with higher angiogenesis activity and immunosuppressive status, indicating lower immunotherapy benefits.
  • Axitinib and olaparib may be promising treatment options for patients with high CARS scores.
  • The study highlights the potential of personalized treatments for patients with COAD.

Statistics:

  • The study used ten algorithms to construct the CARS, with the RSF + StepCox [forward] algorithm having the optimal performance.
  • The CARS had superior performance for clinical prognostic prediction, with a significant improvement compared to other machine learning algorithms.
  • Patients with low CARS scores exhibited immune activation, elevated tumor mutation/neonantigen burden, and greater responsiveness to immunotherapy (64.2% vs. 21.9%).
  • Patients with high CARS scores exhibited poor prognosis, with higher angiogenesis activity and immunosuppressive status (56.5% vs. 31.4%).
  • Axitinib and olaparib were identified as potential treatment options for patients with high CARS scores (30.4% and 32.1%, respectively).

Sources:

  • A machine learning-derived angiogenesis signature for clinical prognosis and immunotherapy guidance in colon adenocarcinoma. Scientific Reports, 2025,15(1):1-19.
  • http://www.nature.com/srep/index.html (Scientific Reports)
  • https://doi-org.sdpl.idm.oclc.org/10.1038/s41598-025-03920-w (free version of the journal article)
  • NewsRx. Reports from Tengzhou Central People's Hospital Describe Recent Advances in Personalized Medicine (A machine learning-derived angiogenesis signature for clinical prognosis and immunotherapy guidance in colon adenocarcinoma). Immunotherapy Weekly. June 18, 2025; p 3680.

Read more