![[Feedback]](/icons/banner/feedbackACT.gif){kind=icon}
![[For Subscribers]](/icons/banner/subscriberACT.gif){kind=icon}
![[Archive]](/icons/banner/archiveACT.gif){kind=icon}
![[Search]](/icons/banner/searchACT.gif){kind=icon}
![[Contents]](/icons/banner/tocACT.gif){kind=icon}
|
|
|
|
|
||
Program in Molecular and Genetic Epidemiology, Harvard School of Public Health, Boston, MA 02115
* To whom correspondence should be addressed. E-mail: david.hunter{at}channing.harvard.edu.
Introduction
The completion of the Human Genome Project gave us a canonical human genome sequence. Subsequent efforts such as the SNP Consortium and the International HapMap project have defined the spectrum of common inherited variation in DNA sequence. Targeted gene resequencing projects such as the NIEHS Environmental Genome Project have provided comprehensive documentation about polymorphisms in some key candidate genes relevant to nutrient metabolism. We are only just beginning to understand how this new information about human genetic variation can help in the complex task of unraveling which components of diet cause, or protect from, specific cancers (1,2).
Core findings
Information about the interaction between food and nutrient intakes and inherited polymorphisms may be useful in several ways and can serve to 1) define susceptible subpopulations, thus strengthening dietary associations; 2) help establish causality of food and nutrient associations in epidemiology studies; 3) aid in distinguishing causal components of complex dietary mixtures; and 4) eventually provide the basis for gene-based screening tests. This could assist in providing individual nutritionally based prevention strategies to adults.
First, definition of a susceptible or resistant stratum may permit stronger nutrient–disease associations to be observed, compared with the associations observed when susceptible and resistant persons are mixed. To date we have only been able to study a weak surrogate for inherited susceptibility to each disease: family history. Family history can be a good indicator of inherited susceptibility for highly penetrant gene mutations that cause readily evident familial clusters or "cancer pedigrees"; however, these are relatively rare in the population. The molecular mechanisms by which high-penetrance genes cause cancer, for instance mutations in the Mismatch Repair Pathway causing colon and other cancers in Lynch Syndrome, may also be sufficiently powerful that environmental factors such as diet have little influence on the lifetime penetrance among mutation carriers, or only influence at the age at onset rather than lifetime penetrance. Because of the rarity of mutations in these genes (e.g., pathogenic mutations in the Mismatch Repair genes are carried by <1% of most populations), it may also be very difficult to assemble large enough samples of mutation carriers to assess interactions with diet observationally, far less conduct randomized trials of dietary interventions. Thus, the influence of dietary factors in the context of high-penetrance gene mutations will be difficult to establish convincingly.
Much of the inherited predisposition to common cancers, however, is thought to be related to more common polymorphic variants that do not give rise to cancer in most carriers but modestly increase risk, sometimes only after exposure to specific lifestyle and dietary risk factors. The variants may be carried by up to 50% of persons (by definition if >50% of persons carry a "variant," it is the "wild-type" allele, not the variant), so statistical power to study these alleles is less of an issue if they are common. These "low-penetrance" variants do not tend to give rise to strong family histories, as many members of a family who are carriers have not developed the cancer of interest; that is, family history is a very weak surrogate for this form of inherited susceptibility.
If 1 or a small number of these variants predict risk of a specific cancer, then establishing this risk should be feasible, and the question then turns to whether the risk is modified by dietary and/or other "environmental" factors. By genotyping cases and controls in typical case-control or prospective epidemiologic studies, it may be possible to establish which members of these studies are truly susceptible and which are not. If particular dietary factors increase risk only in susceptible individuals, then studying these in a mix of susceptible and nonsusceptible people will attenuate the risk of cancer seen for these factors. The strength of association will be stronger if the analysis can be limited to those who are susceptible, and stronger associations are more likely to reproduce and to satisfy classic causal criteria.
Second, gene–nutrient interaction may provide evidence that the primary nutrient–disease association is causal through the principle of "Mendelian randomization" (3,4). Even reproducible nutrient–disease associations may be results of confounding by risk factors that are correlated both with nutrient intake and the cancer outcome being studied. If the relation between genotype and nutrient intake is null (i.e., removing 1 of the criteria for confounding), then a genotype–disease association or genotype–nutrient interaction implies an unconfounded association with the nutrient of interest. Thus, the consistent finding that the relation between diets low in methyl groups (i.e., diets low in folate, methionine, and high in ethanol intake) and colorectal cancer is stronger among persons who carry the low-activity variant of the methylenetetrahydrofolate reductase (MTHFR) enzyme (5–7), implies that the primary relation between the dietary variable and colorectal cancer risk is not merely a result of confounding by other risk factors. Relationships between biomarkers of nutritional intake and risk of specific cancers may also be confounded by risk factors that alter the biomarker level. The principle of Mendelian randomization may be particularly helpful with biomarkers. The key concept is that if a genetic variant alters levels of a biomarker (e.g., plasma folate levels are lower in carriers of the MTHFR common functional variant), then an association of the variant directly with cancer risk would be strong supportive evidence that the biomarker–cancer risk is unconfounded (because it is unlikely that the genetic variant would be associated with potentially confounding lifestyle variables, thus removing one of the criteria for the occurrence of confounding). Although this concept may be useful when these conditions are met, it may be of limited utility until more genetic variants associated with biomarker levels are found.
A third way in which knowledge of genetic polymorphisms may help untangle complex food–cancer associations is by assisting with the identification of the causal components that underlie dietary associations (8). Diet in general, and many foods, are complex mixtures in which the component causal foods and nutrients are inextricably correlated and cannot be reliably distinguished from each other. Finding that a particular food is associated with increased or reduced risk of cancer opens up the question of which nutrients or nonnutrient compounds are responsible. Finding that a food–metabolic enzyme polymorphism interaction exists implies that the substrates of the enzyme are the relevant causal actors in the food or complex mixture. In this manner, genetic polymorphism can help define the causal nutrients in diet that could not be ascertained observationally though nonexperimental means. For instance, a relatively consistent association has been observed between red meat intake and risk of colorectal cancer. There are competing theories, however, on what components of red meat are causal. Candidates include dietary fat, heme iron (because of its pro-oxidant activities), or the formation of heterocyclic amines and/or other carcinogens during the high-temperature cooking of animal protein. The enzyme N-acetyltransferase 2 is a key step in the metabolism of heterocyclic amines from procarcinogenic compounds to ultimate carcinogens that can cause DNA mutations. Several nonsynonymous polymorphisms in the gene encoding the enzyme distinguish between people who are slow metabolizers (about 55% of whites) and fast metabolizers (the other 45%). The finding that the relation between red meat intake and colorectal cancer is stronger among fast metabolizers suggests that compounds metabolized by this enzyme are causing this cancer. The most logical compounds are the heterocyclic amines formed during cooking. Le Marchand et al. (9) have extended these observations to include variation in the function of the CYP1A2 enzyme, another component of the pathway that metabolizes heterocyclic amines. Thus, the specificity of the relation between red meat and colorectal cancer to a specific functional form of the enzyme implicates the enzyme's substrates as the causal actors in red meat. In this way, knowledge of the way in which metabolic activation or detoxification enzymes interact with food intake may help distinguish the causal compounds in these foods.
Research needs
Projects such as the HapMap (10) have meant that common variation in human genes is increasingly being well documented. Developments in genotyping technology make it possible to ascertain thousands, even hundreds of thousands, of variants in a single DNA sample. Distinguishing nonfunctional from functional variation, and functional variation that is relevant to nutritional intake from functional variation relevant to other exposures, is a major need. The ability to reduce the very large number of genetic variants to those that are of functional relevance would substantially reduce the multiple-comparisons problem associated with examining all variants simultaneously.
Another substantial need is the development of databases to capture the huge amount of information that is being generated daily, and which will increase almost exponentially in the coming years. Only a fraction of this information could ever reasonably be published in conventional research papers. The phenomenon of publication bias, the tendency of authors to submit, reviewers to favor, and editors to accept "positive" findings, means that the published literature becomes increasingly biased over time, and relevant, often larger and superior "null" studies are not made available for literature reviews or meta-analyses. Capturing unpublished data in accessible form will be important to the realistic assessment of diet–gene interactions but is a formidable task, especially in cancer research where so many different types of tumors are involved. Information from different studies is not always in compatible formats. Different variants may be genotyped in the same gene by different groups, inhibiting comparisons of the findings. Formation of networks and consortia may be helpful in this regard (11), as common protocols, data formats, data analyses, and tabular output substantially enhance the ability to archive and interpret these large amounts of data.
In conclusion, new knowledge about between-person differences in response to diet should help untangle the complex relations of diet with cancer risk. It is possible that, in the future, dietary advice may be able to be tailored to a specific inherited susceptibility or resistance to specific cancers. Much data will need to be generated, shared, and analyzed before this promise of "personalized prevention" will be evidence-based. In the meantime, however, there are several ways in which establishing robust diet–gene interactions in cancer risk will bolster the evidence for the association of individual foods, or components of these foods, with cancer risk. These developments should improve our ability to make more confident statements about the role of diet in cancer prevention.
 |
FOOTNOTES |
|---|
2 Author disclosure: no relationships to disclose.
|
LITERATURE CITED |
|---|
 TOP LITERATURE CITED |
|---|
1. Hunter DJ. Gene–environment interactions in human diseases. Nat Rev Genet. 2005;6:287–98.[Medline]
2. Gibney MJ, Gibney ER. Diet, genes and disease: implications for nutrition policy. Proc Nutr Soc. 2004;63:491–500.[Medline]
3. Smith GD, Ebrahim S. Mendelian randomization: prospects, potentials, and limitations. Int J Epidemiol. 2004;33:30–42.
4. Little J, Khoury MJ. Mendelian randomization: a new spin or real progress? Lancet. 2003;362:930–1.[Medline]
5. Sharp L, Little J. Polymorphisms in genes involved in folate metabolism and colorectal neoplasia: a Huge review. Am J Epidemiol. 2004;159:423–43.
6. Le Marchand L, Wilkens LR, Kolonel LN, Henderson BE. The MTHFR C677T polymorphism and colorectal cancer: the multiethnic cohort study. Cancer Epidemiol Biomarkers Prev. 2005;14:1198–203.
7. Eaton AM, Sandle R, Carethers JM, Millikan RC, Galanko J, Keku TO. 5, 10-Methylenetetrahydrofolate reductase 677 and 1298 polymorphisms, folate intake, and microsatellite instability in colon cancer. Cancer Epidemiol Biomark Prev. 2005;14:2023–9.
8. Rothman N, Wacholder S, Caproraso NE, Garcia-Closas M, Buetow K, Fraumeni JF. The use of common genetic polymorphisms to enhance the epidemiologic study of environmental carcinogens. Biochim Biophys Acta. 2001;1471:C1–10.[Medline]
9. Le Marchand L, Hankin JH, Pierce LM, Sinha R, Nerurkar PV, Franke AA, Wilkens LR, Kolonel LN, Donlon T, et al. Well-done red meat, metabolic phenotypes and colorectal cancer in Hawaii. Mutat Res. 2002;506–507:205–14.
10. The International HapMap Consortium. The International HapMap Project. Nature. 2003;426:789–96.[Medline]
11. The National Cancer Institute Breast and Prostate Cancer Cohort Consortium. A candidate gene approach to searching for low-penetrance breast and prostate cancer genes. Nat Rev Cancer. 2005;5:977–85.[Medline]
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
