Abstract
CONVENTIONAL randomized clinical trials of new treatments usually rely on clinical end points that directly show the effect of the new treatment on outcomes such as quality or length of life. In chronic diseases, like cancer, cardiovascular disease, and human immunodeficiency virus (HIV) infection, standard clinical end points may be either distant in time or rare, and thus require a trial of long duration, large sample size, or both to demonstrate the value of a new treatment. A surrogate marker for that clinical end point would allow more rapid completion of clinical investigations of new therapies. The use of surrogate markers in evaluating effective therapies for HIV infection has drawn considerable attention. See also pp 158 and 161. A perfect surrogate marker for use in the study of a new therapeutic agent must be biologically plausible, measurable in all patients with the disease, predictive of disease progression (ie, worsening with