Predictive Value of the ATP Chemosensitivity Assay in Epithelial Ovarian Cancer

Abstract

Objective. The aims of this study were to define the specific parameters of the ATP chemosensitivity assay which most accurately predict a patient's clinical response to chemotherapeutic agents in epithelial ovarian cancer and to assess the clinical utility of the ATP assay.Methods. In our laboratory from 1992 to 1994, fresh tumor specimens from patients with epithelial ovarian carcinomas were assayed with the ATP chemosensitivity assay (ATP-CSA) for their in vitro responses to several chemotherapeutic agents including cisplatin, paclitaxel, and cyclophosphamide. Clinical data on 161 of those patients including all follow-up assessments were then collected, and an investigator blinded to the in vitro assay results determined the patients' responses to chemotherapy. In order to determine which parameter of the assay was the best predictor of clinical response for each drug, receiver-operator characteristic (ROC) curves were constructed for several parameters, including the amount of cell kill at particular dosage levels of drug, the slope of the dose–response curve, and the IC50, or the average concentration of drug at which 50% of the cells were nonviable.Results. The specific parameter of the ATP-CSA which was most predictive of clinical response differed for each drug tested. The resulting positive predictive values for cisplatin, cyclophosphamide, and paclitaxel ranged from 70.0 to 78.3% and negative predictive values from 46.2 to 60.9%, with overall ATP-CSA positive and negative predictive values of 83.0 and 56.5%. Overall, patients whose tumors tested sensitive to an agent in vitro were almost twice as likely (83% versus 43%) to show a clinical response (RR 1.91, 95% CI 1.34–2.71).Conclusion. Analysis of the ROC curves in this study shows that different parameters of the ATP-CSA need to be utilized for each drug tested in order to give the best prediction of clinical chemosensitivity. Although the ATP-CSA shows predictive ability, routine use of the ATP-CSA for clinical selection of drug therapy in patients with epithelial ovarian cancer would not be warranted without a prospective study comparing chemotherapy treatment based on assay results versus clinician selection of drug.