Can a marker be a surrogate for development of cancer, and would we know it if it exists?

Carcinogenesis proceeds through a very long preclinical period. Our collective hope is that multiple opportunities exist for chemoprevention to arrest or reverse progression towards malignancy. In the hope of faster progress with fewer subjects and lower total cost, much effort is being expended on the search for reliable biomarkers to predict the likelihood of developing cancer and/or to signal the effectiveness of chemopreventive therapy. Considerable attention is paid to identifying those markers that can act as surrogate markers for cancer development, since favorable modulation of the surrogate end-point biomarker (SEBM) may demonstrate effectiveness of a putative preventive treatment. However, the complexity of the biology challenges our ability to measure the effectiveness of attempts to arrest or reverse carcinogenesis, other than through costly and time-consuming prospective trials with disease state as the endpoint. Despite much work, to date no prehistologic biological or molecular intermediate marker has been validated for sporadic cancers. Several factors accounting for the difficulties encountered in SEBM development are reviewed. Discussion is focused on the common thread of the complexity of the underlying biological changes in carcinogenesis limiting the effectiveness of any single biomarker. Additionally, the incidence of sporadic cancers is also low, further limiting the positive predictive value of any putative prognostic marker. Recent successes in development of chemopreventive agents show the concept is valid and worth pursuing, but the current strategies to develop biochemical and genetic markers to identify surrogate biomarkers is flawed, and need to be reassessed in light of the difficulties faced over the last 20 years.