Predictive chemosensitivity testing in malignant melanoma: reliable methodology--ineffective drugs.

A retrospective and a prospective trial were carried out in patients with malignant melanomas to investigate the predictive value of an in vitro chemosensitivity assay based on the Courtenay and Mills soft agar cultivation method. Evaluable in vitro chemosensitivity data for the three agents DTIC, CCNU, and vinblastine were obtained in 153 cases. In the retrospective study in which the patients received chemotherapy without prior knowledge of the test results, 50 in vitro/in vivo correlations (40 patients) were made, and in the prospective study, where patients received the single agent most active in vitro, 55 correlations (45 patients) were performed. In both studies the sensitivity of the test (the ability to identify patients who will respond to chemotherapy) was approximately 100% and the specificity (the ability to identify patients who will not respond) was 87-98%. Depending on whether 'no change' and 'mixed response' were classified as sensitivity or resistance, the predictive value of a negative test was approximately 100% and that of a positive test 37.5-87.5%. The response rate was low in both series, and although it was somewhat higher in the prospective than in the retrospective trial, the difference was not significant. The median survival time was not significantly different in the two treatment series. We conclude that the chemosensitivity assay here used is reliable and has predictive value, but that the chemotherapeutic agents currently available for treatment of melanoma are too ineffective to warrant routine use of the assay in this disease.