A novel morphology-based risk stratification model for stage I uterine leiomyosarcoma: an analysis of 203 cases

Abstract

Uterine leiomyosarcoma is the most common uterine mesenchymal malignancy. The majority present at stage I, and clinical outcomes vary widely. However, no widely accepted risk stratification system for stage I uterine leiomyosarcoma is currently available. We studied 17 routinely evaluated clinicopathologic parameters in 203 stage I uterine leiomyosarcoma from three institutions to generate a novel risk stratification model for these tumors. Mitoses >25 per 2.4 mm² (10 high-power fields), atypical mitoses, coagulative necrosis, lymphovascular invasion, and serosal abutment were significantly associated with disease-free and disease-specific survival in univariate and multivariate analyses. These prognostic parameters were each scored as binary ("yes" or "no") variables and fitted to a single optimized algebraic risk model:Risk score = (coagulative necrosis)(1) + (mitoses > 25 per 2.4 mm²)(2) + (atypical mitoses)(2) + (lymphovascular invasion)(3) + (serosal abutment)(5)By logistic regression, the risk model was significantly associated with 5-year disease-free (AUC = 0.9270) and 5-year disease-specific survival (AUC = 0.8517). Internal and external validation substantiated the model. The continuous score (range, 0-13) was optimally divided into 3 risk groups with distinct 5-year disease-free and disease-specific survival: low risk (0-2 points), intermediate risk (3-5 points), and high risk (6-13 points) groups. Our novel risk model performed significantly better than alternative uterine leiomyosarcoma risk stratification systems in predicting 5-year disease-free and disease-specific survival in stage I tumors. A simplified risk model, omitting terms for serosal abutment and lymphovascular invasion, can be accurately applied to myomectomy or morcellated specimens. We advocate routine application of this novel risk model in stage I uterine leiomyosarcoma to facilitate patient counseling and proper risk stratification for clinical trials.