Predicting HIF-2α Levels in Clear Cell Kidney Cancer Using Machine Learning

Development of a Machine Learning-Based Nomogram for Predicting HIF-2α Expression Levels in Clear Cell Renal Cell Carcinoma

This project aims to conduct a multicenter retrospective study to collect clinical, CT imaging, and pathological data from patients. A comprehensive data management system will be established, and radiomic features will be extracted to integrate and analyze multicenter data. We will develop a predictive model based on CT radiomic features and perform both internal and external cohort validation. The model will predict HIF-2α expression levels and clinically relevant prognostic factors in ccRCC, enabling precise identification of patient populations responsive to the HIF-2α antagonist Belzutifan, thereby facilitating personalized treatment decisions, minimizing unnecessary therapeutic risks, and ultimately improving patient quality of life and clinical outcomes.

Inclusion Criteria:

• Pathologically confirmed clear cell renal cell carcinoma (ccRCC)
• Availability of comprehensive clinical, pathological, and follow-up information
• Access to preoperative non-contrast and contrast-enhanced CT images through the PACS database
• Adequately preserved pathological slides for subsequent immunohistochemical (IHC) or tissue microarray analysis
• Minimum of one post-treatment follow-up with documented treatment response or efficacy evaluation

Exclusion Criteria:

• Patients considered ineligible for treatment owing to severe comorbid conditions or inability to undergo any therapeutic intervention
• Patients with concurrent malignancies, including prior treatment for other cancers or presence of untreated active malignancies
• Patients with inadequate CT image quality or missing imaging data
• Patients with missing or incomplete clinical, pathological, or follow-up information