Abstract
AIM: To establish prediction models for 2-year overall survival of ovarian cancer patients with metastasis. METHODS: In total, 4,929 participants from Surveillance, Epidemiology, and End Results (SEER) database were randomly divided into the training set (n = 3,451) and the testing set (n = 1,478). Univariate and multivariable regression were conducted in the training set to identify predictors for 2-year overall survival of metastatic ovarian cancer patients. The C-index was calculated for assessing the performance of the models. The nomogram for the model was plotted. The prediction value of the model was validated in the testing set. Subgroup analysis were performed concerning surgery and chemotherapy status of patients and the metastatic site of ovarian cancer in the testing set. The calibration curves were plotted and the decision curve analysis (DCA) were conducted. RESULTS: At the end of follow-up, 2,587 patients were survived and 2,342 patients were dead within 2 years. The 2-year survival rate was 52.5%. The prediction models were constructed based on predictors including age, radiation, surgery and chemotherapy, CA125, and bone, liver, and lung metastasis. The prediction model for 2-year overall survival of ovarian cancer patients with metastasis showed good predictive ability with the C-index of the model of 0.719 (95% CI: 0.706-0.731) in the training set and 0.718 (95% CI: 0.698-0.737) in the testing set. In terms of patients with bone metastasis, the C-index was 0.740 (95% CI: 0.652-0.828) for predicting the 2-year overall survival of ovarian cancer patients. The C-index was 0.836 (95% CI: 0.694-0.979) in patients with brain metastasis, 0.755 (95% CI: 0.721-0.788) in patients with liver metastasis and 0.725 (95% CI: 0.686-0.764) in those with lung metastasis for predicting the 2-year overall survival of ovarian cancer patients. CONCLUSION: The models showed good predictive performance for 2-year overall survival of metastatic ovarian cancer patients.