Comparison of the predictive power of different prognostic indices for overall survival in patients with advanced ovarian carcinoma.

Validation of a 5-covariate prognostic index (PI5) (performance status, stage, residual tumor size, histological grade, and ascites) derived from a group of 268 Dutch patients with advanced ovarian carcinoma was performed in a similar study of 301 Danish patients. Analysis of survival suggested an alternative 4-covariate PI (performance status, residual tumor size, age, weight/body surface) for the latter group. As residual tumor size and performance status were common to both indices, the predictive power of this 2-covariate PI (PI2) was also assessed in the Danish study and, subsequently, validated in the Dutch patients. The PI5 defined 10 and 9% of the Dutch patients as low and high risk patients with 3-year survival rates of 80 and 8%, respectively. In the Danish study the PI5 classified 14% to be high risk patients with a 3-year survival rate of 18%. Only a few patients (3%) were classified as low risk, making comparisons irrelevant. The PI2 classified 33 and 26% of the Danish patients as having 3-year survival rates of 67 and 13%, respectively. The corresponding Dutch PI2 values were 41 and 15% of the patients with 3-year survival rates of 60 and 8%, respectively. Although the PI5 possessed a better validity in the Danish study than the PI2 in the Dutch study, the PI2 may be the best prognostic index available for general use.