Development and validation of a prognostic nomogram for overall survival in patients with platinum-resistant ovarian cancer treated with chemotherapy.

ABSTRACT

BACKGROUND: Platinum-resistant ovarian cancer (PROC) is associated with a variable prognosis and unpredictable survival times. We have developed and validated a prognostic nomogram with the objective of improving the prediction of overall survival (OS) in patients treated with chemotherapy. METHODS: The nomogram was developed using data from a training cohort of patients from two trials, including the chemotherapy-only arm in AURELIA and all randomised patients in CARTAXHY. Multivariable proportional hazards models were generated based on pretreatment characteristics to develop a nomogram that classifies patients based on OS. We subsequently assessed the performance of the nomogram in terms of discrimination and calibration in independent validation patient cohorts PENELOPE and the bevacizumab-chemotherapy arm of AURELIA. RESULTS: The nomogram included six significant OS predictors, in order of importance performance status, ascites, size of the largest tumour, CA-125, platinum-free interval and primary platinum resistance (C-statistic 0.69). In the training cohort, the median OS in the good, intermediate and poor prognosis groups was 25.3, 15.2 and 7.4 months, respectively. In the PENELOPE validation cohort (C-statistic 0.59), the median OS in the good, intermediate and poor prognosis groups was 18.5, 10.3 and 5.8 months, respectively. In the AURELIA bevacizumab-chemotherapy validation cohort (C-statistic 0.67), the median OS in good, intermediate and poor prognosis groups was 26.7, 13.8 and 10.0 months, respectively. CONCLUSIONS: This nomogram with six pretreatment characteristics allows prediction of OS in PROC and could be used for stratification of patients in clinical trials as well as for counselling patients about prognosis.