ADC-Based MRI Achieves Superior Accuracy in Preoperative LVSI Prediction for Endometrial Cancer

Background: Preoperative identification of lymphovascular space invasion (LVSI) in endometrial cancer remains challenging yet critical for surgical planning and adjuvant therapy. This study compared qualitative diffusion-weighted imaging (DWI) and quantitative apparent diffusion coefficient (ADC) analysis for preoperative LVSI detection. Methods: A total of 95 patients with histologically confirmed endometrial cancer underwent preoperative DWI. Two blinded radiologists performed a visual assessment of high b-value images, and manual regions of interest were placed on the ADC maps to derive the mean ADC values. Optimal ADC thresholds were determined and compared with published cut-offs. Diagnostic performance metrics were calculated for both methods in the overall cohort and stratified by tumor size, FIGO stage, and grade. Results: Qualitative DWI yielded 85.4% sensitivity, 21.3% specificity, and 53.7% accuracy. ADC at a threshold of ≤0.690×10⁻³ mm²/s achieved 75.0% sensitivity, 76.6% specificity, and 75.8% accuracy, with an AUC of 0.770. Published ADC cut-offs (0.690-0.820×10⁻³ mm²/s) consistently outperformed visual assessment (accuracy 62.1-71.6%). In subgroup analyses, ADC maintained superior accuracy across all tumor sizes (<2 cm: 90.0%; 2-4 cm: 79.3%; >4 cm: 80.3%), FIGO stages (I: 71.4%; II: 82.1%; III: 92.9%), and grades (1: 72.4%; 2: 78.4%; 3: 84.0%). Combining visual and quantitative assessments did not improve the AUC beyond that of ADC alone. Conclusions: Quantitative ADC analysis significantly outperforms qualitative DWI for preoperative detection of LVSI in endometrial cancer, providing an objective biomarker that enhances risk stratification and informs surgical and adjuvant treatment strategies.