The aim of this retrospective single-center study was to evaluate the relationship between maximum tumor size, tumor volume, tumor volume ratio (TVR) based on preoperative magnetic resonance (MR) volumetry, and negative histological prognostic parameters (deep myometrial invasion [MI], lymphovascular space invasion, tumor histological grade, and subtype) in International Federation of Gynecology and Obstetrics stage I endometrial cancer.Methods/Materials
Preoperative pelvic MR imaging studies of 68 women with surgical-pathologic diagnosis of International Federation of Gynecology and Obstetrics stage I endometrial cancer were reviewed for assessment of MR volumetry and qualitative assessment of MI. Volume of the tumor and uterus was measured with manual tracing of each section on sagittal T2-weighted images. Tumor volume ratio was calculated according to the following formula: TVR = (total tumor volume/total uterine volume) × 100. Receiver operating characteristics curve was performed to investigate a threshold for TVR associated with MI. The Mann-Whitney U test, Kruskal-Wallis test, and linear regression analysis were applied to evaluate possible differences between tumor size, tumor volume, TVR, and negative prognostic parameters.Results
Receiver operating characteristics curve analysis of TVR for prediction of deep MI was statistically significant (P = 0.013). An optimal TVR threshold of 7.3% predicted deep myometrial invasion with 85.7% sensitivity, 46.8% specificity, 41.9% positive predictive value, and 88.0% negative predictive value. Receiver operating characteristics curve analyses of TVR, tumor size, and tumor volume for prediction of tumor histological grade or lymphovascular space invasion were not significant. The concordance between radiologic and pathologic assessment for MI was almost excellent (κ value, 0.799; P < 0.001). Addition of TVR to standard radiologic assessment of deep MI increased the sensitivity from 90.5% to 95.2%.Conclusions
Tumor volume ratio, based on preoperative MR volumetry, seems to predict deep MI independently in stage I endometrial cancer with insufficient sensitivity and specificity. Its value in clinical practice for risk stratification models in endometrial cancer has to be studied in larger cohort of patients.