MRI-based perfusion-diffusion habitat analysis for characterizing intratumoral heterogeneity in rectal adenocarcinoma

BACKGROUND: This study aimed to evaluate whether MRI-based perfusion-diffusion habitat imaging derived from dynamic contrast-enhanced (DCE) and diffusion-weighted imaging (DWI) can characterize intratumoral heterogeneity and evaluate associations with pathologic stage, synchronous distant metastasis, and exploratory survival outcomes in rectal adenocarcinoma. METHODS: This retrospective study included 73 patients with pathologically confirmed rectal adenocarcinoma who underwent preoperative DCE-MRI and DWI scans. The volume transfer constant (Ktrans) and apparent diffusion coefficient (ADC) maps were generated from DCE and DWI data, respectively. After manual tumor segmentation and image co-registration, voxel-wise Ktrans and ADC values were clustered using a k-means algorithm to identify distinct spatial habitats. Quantitative metrics extracted from each habitat were compared across pathologic subgroups and entered into multivariable logistic regression models for T staging and association with synchronous distant metastasis. RESULTS: Three reproducible habitats were identified: Habitat 1, hyper-vasopermeability; Habitat 2, hypo-vasopermeability with hypo-cellularity; and Habitat 3, hypo-vasopermeability with hyper-cellularity. Compared with T1-2 tumors, T3-4 lesions showed significantly higher mean Ktrans and the extracellular-extravascular volume fraction (Ve) in Habitats 1 and 2 and lower mean ADC in Habitats 2 and 3. Multiple perfusion metrics from Habitat 1 significantly correlated with distant metastasis. The optimized logistic regression models achieved AUCs of 0.81 for T staging and 0.78 for distant metastasis prediction. CONCLUSION: Combined DCE-MRI and DWI habitat imaging enables voxel-level assessment of intratumoral heterogeneity in rectal cancer and may provide a noninvasive imaging biomarker for preoperative T staging and assessment of tumor heterogeneity and systemic disease-associated imaging features.