MRI-Based Quantification of Intratumoral Heterogeneity for Predicting Progression-Free Survival in Patients with Lung...

Area of Science:

• Radiology and Oncology
• Medical Imaging Analysis
• Cancer Biomarkers

Background:

• Lung cancer brain metastasis (LCBM) presents a significant clinical challenge.
• Predicting progression-free survival (PFS) is crucial for guiding radiotherapy treatment strategies.

Purpose of the Study:

• To investigate the efficacy of MRI-based habitat features in predicting PFS for LCBM patients receiving radiotherapy.
• To compare the predictive performance of habitat features against conventional radiomics and clinical factors.

Main Methods:

• Retrospective analysis of 146 lesions from 68 LCBM patients treated with radiotherapy.
• Extraction of conventional radiomics and MRI-based habitat features from tumor regions.
• Development and evaluation of machine learning risk models (clinical, radiomics, habitat, and combined) using concordance index (C-Index) and Brier scores.

Main Results:

• The habitat risk model demonstrated superior prediction ability in the time-independent test cohort (C-Index: 0.716).
• Habitat and radiomics models outperformed the clinical model in both training and time-independent test cohorts.
• Habitat features showed strong performance in predicting PFS, highlighting intratumoral heterogeneity as a key factor.

Conclusions:

• MRI-based habitat features, reflecting intratumoral heterogeneity, serve as a reliable biomarker for predicting PFS in LCBM patients undergoing radiotherapy.
• Habitat risk modeling offers a promising tool for personalized treatment planning in LCBM.