The human tumor colony-forming chemosensitivity assay: a biological and clinical review.

Over forty papers describing correlations between in vitro human tumor sensitivity to a variety of chemotherapeutic agents and the in vivo response of patients to those agents have been published since the publication in 1978 by Salmon and Hamburger of their results of a human tumor colony-forming chemosensitivity assay (CFCA). The true positive rate in over 1600 correlations is 71% and the true negative rate is 94%. The biological elements of the assay, its developmental history, its place in the spectrum of in vitro chemosensitivity assays, and its theoretical and practical limitations are discussed. The scope, design, and limitations of key clinical trials are presented and an analysis of the potential errors of statistical interpretation of the trials as well as the results of the trials is given.