Search

There is significant need for rapid yet reliable evaluation of promising new interventions for the treatment of patients with cancer or HIV infection. Considerable attention has been given to identifying replacement or 'surrogate' endpoints for the true clinical efficacy endpoints, in order to reduce the cost, size and duration of clinical trials. We discuss issues which affect the validity of surrogate markers. The reliability of the CD4 lymphocyte count marker is carefully considered in clinical trials of anti-retroviral agents in HIV infected individuals. The nature of surrogate markers and their reliability is discussed in cancer prevention, screening and treatment trials. Some suggested uses of marker information are also considered.

The integrity and credibility of randomized clinical trials, designed to provide a definitive evaluation of therapeutic interventions, can be substantially enhanced if periodic review of results on the relative efficacy of treatments is conducted by a data monitoring committee (DMC). The role of the DMC can include some important activities which promote obtaining relevant information of high quality. In the final stage of study development, these include review of protocol-specified study objectives and design, and review of procedures planned for capturing high-quality data and for reporting to the DMC. At analyses during the trial, these include careful review of data completeness and accuracy to assist in early detection and resolution of evolving problems.

Use of inappropriate stopping rules for clinical trials results in an excess of false positive conclusions when no true survival differences exist. Overviews of such trials, however, consist mainly of trials which were not stopped early, plus a few of reduced sample size which were. Simulations confirm that the level of such an overview is minimally elevated. Additional follow-up for survival further corrects the level. In fact, for individual trials conducted according to inappropriate rules, late tests have nearly correct level. On the other hand, publication bias (differential reporting of positive results) can substantially increase the level of an overview if only published studies are included.