ABSTRACT
Molecular epidemiology has been defined as the study of the distribution and deter-
minants of disease in human populations using techniques of molecular biology and
epidemiology (1). The field started to grow in the early 1980s (2). Studies in the 1980s
were based primarily on markers of exposure to carcinogens such as DNA and protein
adducts and then expanded into the analysis of somatic genetic alterations in tumor and
precancerous tissues (1). Since then, there has been a great deal of work on genetic markers of
susceptibility, initially focused on candidate genes and more recently on genomewide
association studies. Also, since the 1980s, there has been the development of large-scale
analyses of somatic mutations and a range of phenotypical assays, including gene expression
arrays, genomic instability, and DNA repair. In some instances, molecular epidemiological
investigation has added value to causal inference, for example, the clarification of the dose-
response effects of aflatoxin exposure (1), the role of human papillomavirus infection in the
etiology of cervical cancer (3), and the detection of protein and DNA adducts in workers
exposed to chemicals such as ethylene oxide (1). However, in a worryingly high number of
other instances, promising initial results have not been confirmed by subsequent
investigations that have usually had greater statistical power to detect effects. Indeed, lack
of replication became such an issue in genetic association studies (4-12) that it threatened to
discredit the field. However, as research has matured, a somewhat greater proportion of
associations with candidate gene variants has been replicated (11) and recently consistent
findings have emerged from genomewide association studies (13-35). A similar pattern of
failure to confirm early promising findings has been observed in relation to prognostic
markers (36). Indeed, the opening salvo of a commentary in 2005was, “The number of cancer
prognostic markers that have been validated as clinically useful is pitifully small, despite
decades of effort and money invested in marker research” (37).