A prognostic nomogram to predict overall survival in patients with platinum-sensitive recurrent ovarian cancer

    loading  Checking for direct PDF access through Ovid

Abstract

Background

Patients with platinum-sensitive recurrent ovarian cancer have variable prognosis and survival. We extend previous work on prediction of progression-free survival by developing a nomogram to predict overall survival (OS) in these patients treated with platinum-based chemotherapy.

Patients and methods

The nomogram was developed using data from the CAELYX in Platinum-Sensitive Ovarian Patients (CALYPSO) trial. Multivariate proportional hazards models were generated based on pre-treatment characteristics to develop a nomogram that classifies patient prognosis based on OS outcome. We also developed two simpler models with fewer variables and conducted model validations in independent datasets from AGO-OVAR Study 2.5 and ICON 4. We compare the performance of the nomogram with the simpler models by examining the differences in the C-statistics and net reclassification index (NRI).

Results

The nomogram included six significant predictors: interval from last platinum chemotherapy, performance status, size of the largest tumour, CA-125, haemoglobin and the number of organ sites of metastasis (C-statistic 0.67; 95% confidence interval 0.65–0.69). Among the CALPYSO patients, the median OS for good, intermediate and poor prognosis groups was 56.2, 31.0 and 20.8 months, respectively. When CA-125 was not included in the model, the C-statistics were 0.65 (CALYPSO) and 0.64 (AGO-OVAR 2.5). A simpler model (interval from last platinum chemotherapy, performance status and CA-125) produced a significant decrease of the C-statistic (0.63) and NRI (26.4%, P < 0.0001).

Conclusions

This nomogram with six pre-treatment characteristics improves OS prediction in patients with platinum-sensitive ovarian cancer and is superior to models with fewer prognostic factors or platinum chemotherapy free interval alone. With independent validation, this nomogram could potentially be useful for improved stratification of patients in clinical trials and also for counselling patients.

Related Topics

    loading  Loading Related Articles