Promises and perils of validating biomarkers for cancer risk.

Abstract

The number of biomarkers potentially relevant to diet and disease has exploded in recent years. Examples include adenomas and intraepithelial neoplasia (tissue); proliferation and apoptosis indices (cell); DNA adducts, strand breaks, genetic polymorphisms (molecular); human papillomavirus (HPV) infection, Helicobacter pylori antibodies (infection); estrogens, PSA (blood analytes); mammographic densities, ovarian ultrasound abnormalities (imaging). Biomarkers can serve three functions in diet and disease research: 1) enhancing the biologic plausibility of diet–disease relations; 2) increasing the strength of diet–disease associations (‘‘relative risk sharpening’’); and 3) serving as surrogate endpoints. With respect to biologic plausibility, an example is the role of blood estrogen levels in explaining the connection between alcohol and breast cancer. Design options for investigating this biologic plausibility include metabolic studies, larger observational epidemiologic studies, and nutrition–gene studies. Nutrition–gene studies can add to biologic plausibility by helping to unravel mixtures (of foods or supplements, for example) or via the Mendelian randomization strategy, which helps rule out bias resulting from confounding and measurement error. Relative risks can be enhanced through the use of specific intake markers (of diet or supplement use—recent studies of serum selenium in relation to colorectal nepolasia, for example) and through investigation of diet–gene interactions (e.g., studies of red meat intake, N-acetyltransferase genotype, and colorectal cancer). Studies with surrogate endpoints can be smaller, faster, and less expensive than those with explicit disease outcomes. This holds for both intervention studies (trials) and observational epidemiologic studies. Surrogate endpoint validity is critical and can be conceived as follows: the study of a dietary factor in relation to the putative surrogate marker gives the correct answer for the relation of that dietary factor to the disease in question. This is a tough requirement. Surrogate validity depends on 3 specific criteria: 1) the marker is associated with disease; 2) the exposure (or intervention) is associated with the marker; and 3) the marker mediates the association between exposure and cancer. Examples of all 3 criteria can be seen in recently published prospective studies of obesity, serum estrogen levels, and breast cancer as well as reports of the interrelations of reproductive factors, HPV infection, and cervical cancer. Standard epidemiologic measures, such as relative risk and attributable proportion, can be used for criteria 1 and 2; regression and stratified analyses can be used to evaluate the third (mediation) criterion. Being on the causal pathway does not in itself guarantee surrogate endpoint validity. A theoretical example of this is given for a well-studied marker, rectal mucosal proliferation assays. Hyperproliferation may not be necessary for colorectal cancer; that is, there may be an alternative pathway bypassing hyperproliferation. The effect of an intervention on the alternative pathway may counterbalance the effect on the hyperproliferation pathway. An agent may reduce proliferation but have no effect on colorectal cancer because it also reduces apoptosis; similarly, an agent may have no effect on proliferation but actually reduce colorectal cancer incidence because it increases apoptosis. The totality of causal connection is key in evaluating surrogate endpoint validity. A high attributable proportion as well as mediation of exposure–disease relations supports the validity of surrogacy. Validity tends to be more assured for surrogates both necessary for and relatively close developmentally to disease (e.g., cervical intraepithelial neoplasia, CIN3, for cervical cancer). A surrogate endpoint may be valid for 1 exposure or intervention in relation to disease but not necessarily valid for another exposure or intervention. That is because, once again, an alternative pathway to the disease may exist. Colorectal adenomas have been studied as surrogates for invasive cancer. The rationale for such use includes the high prevalence of adenomas, the high recurrence rate (2 orders of magnitude greater than that for cancer), the assessment of the endpoint by standard clinical practice (endoscopy), and, reflecting the key biologic rationale, the adenoma–carcinoma 1 Published in a supplement to The Journal of Nutrition. Presented as part of the conference ‘‘The Use and Misuse of Biomarkers as Indicators of Cancer Risk Reduction Following Dietary Manipulation’’ held July 12–13, 2005 in Bethesda, MD. This conference was sponsored by the Center for Food Safety and Applied Nutrition (CFSAN), Food and Drug Administration (FDA), Department of Health and Human Services (DHHS); the Office of Dietary Supplements (ODS), National Institutes of Health, DHHS; and the Division of Cancer Prevention (DCP), National Cancer Institute, National Institutes of Health, DHHS. Guest Editors for the supplement publication were Harold E. Seifried, National Cancer Institute, NIH; and Claudine Kavanaugh, CFSAN, FDA. Guest editor disclosure: H.E. Seifried, no relationships to disclose; C. Kavanaugh, no relationships to disclose. 2 Author disclosure: no relationships to disclose. 3 Abbreviations used: CIN, cervical intraepithelial neoplasia; HPV, human papillomavirus; PSA, prostate-specific antigen. * To whom correspondence should be addressed. E-mail: schatzka@mail.nih. gov.