Prognostic factors in gestational trophoblastic tumors: a proposed new scoring system based on multivariate analysis.

All 1391 patients treated for gestational trophoblastic tumors (invasive mole and choriocarcinoma) at the John I. Brewer Trophoblastic Disease Center of Northwestern University between 1969 (when use of combination chemotherapy for initial treatment of high-risk disease came into general use) and 1988 were evaluated. Univariate and multivariate analyses were used to determine the relative importance of prognostic factors with respect to survival. The overall cure rate was 93% (363/391): 100% for 223 patients with nonmetastatic disease and 83% for 168 patients with metastatic disease. The only patients who died had a clinicopathologic diagnosis of metastatic choriocarcinoma. In addition to presence of metastasis (83% vs 100%, p less than 0.0001) and diagnosis of choriocarcinoma (67% vs 100%, p less than 0.0001), number of metastases (47% if greater than 8 vs 92% if less than or equal to 8, p less than 0.0001), metastases to sites other than the lung or vagina (52% vs 91%, p = 0.0002), and previous failed chemotherapy (46% vs 84%, p = 0.0014) demonstrated independent significant effects on survival in patients with metastatic disease. A Brewer score, based on our multivariate analysis of survival in patients with metastatic disease, provided predictability of outcome (likelihood ratio chi 2 statistic, chi 2 = 49.8) comparable to that with the World Health Organization score (chi 2 = 45.3), both of which, in turn, were better predictors than either the traditional Hammond clinical classification system (chi 2 = 34.4) or the International Federation of Gynecology and Obstetrics stage (chi 2 = 22.9).