Abstract
PURPOSE: To evaluate the supplemental diagnostic value of T2WI for assessing VPI in NSCLC ≤3 cm with indeterminate CT features (pleural contact/tags). MATERIALS AND METHODS: This prospective single-center study enrolled 138 participants with NSCLC ≤3 cm and CT-suspected pleural involvement, of whom 98 underwent surgical resection and histopathologic VPI confirmation from January 2021 to March 2023. All participants underwent thoracic MRI (3.0T), including T2WI sequences. Two radiologists independently graded tumor-pleural interface signals (Grades 0-3: absent to wedge-shaped hyperintensity). Pathologic VPI served as the reference standard. Logistic regression and ROC analysis were performed to develop a predictive model integrating tumor size and T2WI grades. RESULTS: VPI-positive lesions (35/98, 35.7%) demonstrated larger mean tumor size (22.32 ± 5.58 mm vs. 16.17 ± 4.22 mm; P = 0.024) and higher T2WI hyperintensity frequencies (85.71% vs. 52.38%; P = 0.001). Grade 3 T2WI signals (wedge-shaped) achieved 93.65% specificity and 76.47% PPV for VPI, with a positive likelihood ratio of 5.85. Multivariate analysis identified tumor size (OR = 1.126/mm, P = 0.024), Grade 2 (OR = 8.826, P = 0.003), and Grade 3 signals (OR = 29.890, P < 0.001) as independent VPI predictors. The combined model achieved an AUC of 0.837 (95% CI: 0.758-0.916), demonstrating superior diagnostic performance versus CT-based criteria. CONCLUSION: T2WI-based grading of tumor-pleural interface hyperintensity serves as a radiation-free method for preoperative VPI prediction in small NSCLC, offering potential as a complementary tool to CT diagnosis. While promising, these findings require multicenter validation due to the single-center design.