Quantitative MRI-Based T Category in Pleural Mesothelioma: Analysis of a Large Real-World Cohort

Accurate clinical staging in pleural mesothelioma (PM) is limited by the subjectivity of conventional imaging assessment, which introduces inter- and intraobserver variability. Tumor volume, pleural thickness and diaphragmatic thickness measurements individually prognostic - offer objective, quantifiable surrogates for locoregional tumor burden. We evaluated these Magnetic Resonance Imaging (MRI)-derived metrics to develop a reproducible quantitative clinical T-classification strategy for PM that offers objective, reproducible surrogates for tumor burden with reduced interobserver variability compared to qualitative MRI staging.

Patients referred for surgical evaluation of PM between 2009 and 2014 who underwent MRI using an IRB-approved protocol were included. For survival analyses, only patients with complete data and uniform treatment were analyzed. Quantitative measures included MR-derived tumor volume (VolMR), summed unidimensional pleural thickness (Psum) and Fissural and diaphragmatic thickness (Ptotal= Psum+Fmax+Dtotal) Optimal categorical thresholds were derived by survival-based optimization using the log-rank statistic, implemented via Classification and Regression Trees (CART) with 20-fold cross-validation. Predictive accuracy was assessed using univariable Cox models and C-statistics, and survival curves were generated using Kaplan-Meier estimates. Continuous variable Cox models were also constructed for VolMR and Ptotal. Analyses were stratified by surgical modality to assess robustness.

Of 695 patients evaluated, 377 had eligible MR imaging. Median age was 68 years; 77% were male and 60% had epithelioid tumors. Clinical Tumor Node Metastasis staging was concordant with pathology in 52.5%, understaged in 33%, and overstaged in 14% of patients. Among evaluated metrics, diaphragmatic thickness (C-statistic = 0.5996) and total pleural thickness (0.5868) demonstrated the highest predictive accuracy. Survival curves showed strong stratification using Ptotal and VolMR. Head-to-head comparison with AJCC 8th edition qualitative T staging (C-statistic 0.6022) showed that Ptotal-based categories (C=0.6217) and VolMR-based categories (C=0.6364) provided superior survival stratification.

Quantitative MR-derived Ptotal and VolMR improve upon qualitative T category in PM by offering objective, reproducible surrogates for tumor burden with strong prognostic stratification when appropriate cut points are applied. These metrics could support future Tumor Node Metastasis refinements and individualized treatment planning.