Prognostic nomogram for predicting survival in patients with high grade endometrial stromal sarcoma: a Surveillance Epidemiology, and End Results database analysis.

ABSTRACT

OBJECTIVE: High grade endometrial stromal sarcoma is a rare and highly malignant tumor that lacks a prognostic model. The aim of this study was to develop a prognostic nomogram predicting the overall survival of patients with high grade endometrial stromal sarcoma. METHODS: Clinical data for patients were derived from the Surveillance Epidemiology, and End Results database. Cox analysis and Akaike's information criterion were used to construct the nomogram. The concordance index, time dependent receiver operating characteristic curve, and calibration plot were used to evaluate the discriminative and calibrating capability. The net reclassification index, integrated discrimination improvement, and concordance index change were also compared between the nomogram and the International Federation of Gynecology and Obstetrics (FIGO) stage. Clinical benefit was evaluated using decision curve analysis. The patients were separated into groups with low and high nomogram risk scores. Kaplan-Meier curve analysis and Cox analysis were used to investigate the survival difference between the two groups. RESULTS: The training and validation cohorts had 461 and 195 patients, respectively. A nomogram that incorporated disease stage, age, surgery, lymph node status, radiotherapy, and chemotherapy for predicting overall survival was established and validated. The concordance index of the nomogram was 0.734 (0.708-0.761) in the training cohort and 0.705 (0.659-0.751) in the validation cohort. The calibration plots showed a favorable calibrating ability of the nomogram. The 1 year and 3 year time dependent receiver operating characteristic curves showed the better discriminative ability of the nomogram than the staging system. The concordance index change, net reclassification index, and integrated discrimination improvement also indicated a significantly (p<0.05) better predictive power of the nomogram over disease stage. Furthermore, decision curve analysis suggested that the nomogram was clinically useful and had a larger clinical net benefit than disease stage alone. Patients with a high risk score had distinctly poorer survival than those with low risk scores. CONCLUSIONS: A prognostic nomogram in patients with high grade endometrial stromal sarcoma exhibited favorable prognostic discrimination and survival prediction ability compared with FIGO stage.