Typical Uses of Surrogate Endpoints
Earlier Clinical Endpoints
The evaluation methods described in this book can be used in the straightforward situation where a clinical endpoint is repeatedly measured over time, and the treatment effect is assessed at a pre-specified time such as one year after treatment initiation. A natural question in such situations is whether an earlier measurement of the same endpoint would essentially be equivalent to the measurement at the later time point, thus reducing time to analysis (as well as time to market for new drugs). The age-related macular degeneration (ARMD) example is typical: patients with this disease lose vision over time, and the purpose of treatment is to prevent vision loss or, hopefully, restore normal vision. The regulatory agencies commonly require the change in vision one year after starting therapy as the primary endpoint in clinical trials. The dataset described came from a trial of interferon-a, a drug that has no efficacy in ARMD (Pharmacological Therapy for Macular Degeneration Study Group, 1997). In contrast, anti-angiogenic drugs such as ranibizumab have more recently been shown to have remarkable efficacy in in ARMD, and the clinical effect of these agents is already manifest a couple of months following treatment initiation (Rosenfeld et al., 2006). It makes sense, in this setting, to investigate whether vision change at earlier time points (perhaps as early as 3 months) would be an equally good endpoint as vision change at 12 months. An earlier endpoint would considerably reduce the clinical development time of new treatments for this condition, even if the regulatory agencies would still require evidence of maintenance of benefit and absence of safety issues at 1 year and 2 years after initiation of therapy.