CA 125 regression after two completed cycles of chemotherapy: lack of prediction for long-term survival in patients with advanced ovarian cancer

The prognostic influence of CA 125 regression between the time point before surgery and after two completed courses of chemotherapy was studied in 210 patients with advanced ovarian cancer, and was compared to other well established prognostic factors. CA 125 blood samples were collected preoperatively (CA 125 pre) and 3 months after surgery (CA 125 3 mo) (at the beginning of the 3rd cycle of chemotherapy). The parameter CA 125 regression defined as log(10) (CA 125 3 mo/CA 125 pre) was used for statistical analysis. In a survival analysis using a Cox proportional hazards model, CA 125 regression (P = 0.0001), residual tumour (P = 0.0001), age (P = 0.0095) and grading (P = 0.044) were independent variables, whereas stage of disease, histology, ascites and type of surgery failed to retain significance. Using log(10) (CA 125 3 mo/CA 125 pre) as simple covariate in a Cox model showed a hazard ratio of 1.70 (95% confidence interval 1.32–2.19, P = 0.0001). However, a detailed analysis of the interaction of time with the prognostic factor CA 125 regression on survival revealed a strong time-dependent effect with a hazard ratio of more than 6 immediately after two courses of chemotherapy, whereas within approximately 1 year the hazard ratio for the surviving patients dropped quickly to the neutral level of 1. In summary, CA 125 regression is an independent prognostic factor for survival of women with advanced ovarian cancer and allows an identification of a high-risk population among patients with advanced ovarian cancer. However, the discriminating power of serial CA 125 for long-term survival seems to be temporary and prediction of individual patients outcome is far less precise. © 1999 Cancer Research Campaign