Mendelian randomization.

5. Mendelian randomization

International Agency for Research on Cancer, Lyon, France

Associations between modifiable exposures and disease seen in observational epidemiology are sometimes confounded and thus misleading, despite our best efforts to improve the design and analysis of studies. Mendelian randomization, the random assortment of genes from parents to offspring that occurs during gamete formation and conception provides one method for assessing the causal nature of some environmental exposures. The asso-ciation between a disease and a polymorphism that mimics the biological link between a proposed exposure and disease is not generally susceptible to the reverse causation or con-founding that may distort interpretations of conventional observational studies. This ap-proach, which uses gene-intermediate phenotype-disease associations as a way of inferring the causal nature of associations between potentially modifiable environmental, dietary or physiological characteristics and disease [1] is an area of growing interest.

Several examples where the phenotypic effects of polymorphisms are well documented and provide encouraging evidence of the explanatory power of Mendelian randomization are described below.

Organophosphates and ill-health in farmers

Agricultural workers who have been exposed to sheep dips containing organophosphates attribute a variety of symptoms of poor health to this exposure, but such claims may be false and may reflect secondary gain from compensation or paid early retirement on health grounds. Thus it is difficult to obtain reliable evidence in this area, and rand-omized controlled trials are not feasible. People who become cases in studies of health-related outcomes of organophosphate exposure generally know that the exposure is hy-pothesized to cause health problems, and it is thus difficult, if not impossible, to conduct unbiased case-control studies.

An enzyme that deactivates a potentially toxic component found in many sheep dips, paraoxonase (PON1), has isoforms with different biological activity. If the component of sheep dip that is detoxified by this enzyme does cause symptoms of ill-health, then among people exposed to sheep dip a higher proportion of those reporting symptoms would be expected to be poor detoxifiers. A study designed along these lines found that the PON1 genetic variant associated with lower detoxification was related to reporting symptoms of poor health among people exposed to sheep dip [2]. Since it is unlikely that genotype is related to potential confounding factors, to the tendency to report symptoms differen-tially, or to a desire for compensation or early retirement, these findings provide evidence that there is a causal effect of the sheep dip exposure on health outcomes.

92

Methyl-tetrahydrofolate reductase polymorphisms and neural tube defects Examining the effects of genotype of parents on the health outcomesof their children introduces the idea of ‘intergenerational’Mendelian randomization. In these circumstanc-es, exposures ofinterest relate to aspects of the intrauterine environment thatare difficult to measure, but are modified by parental genotype.For example, folate deficiency in pregnancy is now known tobe a cause of neural tube defects (NTDs), an effect con-firmedby randomized controlled trial evidence. The methyl‑tetrahydrofolate reductase (MTHFR) gene 677C‑>T polymorphism is associated with increased blood levelsof homo-cysteine (equivalent to the situation resulting fromlower levels of folate intake) and in a meta‑ analysis of case-controlstudies of NTDs, the MTHFR TT genotype carrying mothers had a 2-fold higher risk of havingan infant with a neural tube defect than the CC genotype car-rying mothers [3]. Therelative risk of a neural tube defect associated with the MTHFR TT genotype in the infant was less than that observed with respectto maternal genotype, and there was no effect of paternal genotypeon offspring neural tube defect risk. This suggests that itis the intra-uterine environment, which is influenced by maternalgenotype rather than the genotype of offspring thatincreases the risk of NTD.

The above findings do not mean that only people with a particular genotype will ben-efit from the characterization of an environmental risk factor, since individuals who are not genetically susceptible will still present with the disease and hence the whole population would benefit from findings which lead to modification of a exposure. This method pro-vides strong evidence, more robust than from conventional observational epidemiological studies, of manipulations to environmental and behavioral exposures that could benefit population health.

Possible environment-disease associations, which could be investigated in the future are: Milk intake and prostate cancer

Milk intake is thought to be a risk factor for prostate cancer [4] although associations found in epidemiological studies may be confounded. By investigating whether polymorphisms in the lactase gene, which determine lactose intolerance [5,6] are associated with milk intake and prostate cancer risk, we will have strong evidence on whether the association between milk and prostate cancer is causal.

Cruciferous vegetable consumption and lung cancer

Observational studies have provided consistent evidence for a protective role of vegetable consumption against lung cancer, with the evidence being most apparent for green crucifer-ous vegetables such as broccoli and cabbage [7]. Such vegetables are rich in isothiocyanates (ITC), which have been shown in animal studies to have strong chemo‑preventative prop-erties against the development of lung cancer [8], and are therefore likely agents for ex-plaining any chemo-preventative effect in humans. However, interpreting a causal associa-tion between the observaassocia-tional evidence and a reducassocia-tion in lung cancer risk is problematic, as the potential for bias and confounding are extremely difficult to exclude. Further causal Paul Brennan, Rayjean Hung

93 evidence may however be provided by adopting a Mendelian randomization approach.

ITC are known to be eliminated by glutathione S‑transferase (GST) enzymes (see Chap‑ ter 3), most notably GSTM1 and GSTT1 [9]. Both GSTM1 and GSTT1 genes have null al-leles with homozygous null genotypes resulting in no enzyme being produced. Individuals who are homozygous for the inactive form for either one or both genes are likely to have higher ITC concentrations due their reduced elimination capacity.

Summary and implications

The above examples illustrate the potential for studies of Mendelian randomisation to ask challenging questions regarding the role of environmental, lifestyle and endogenous factors in the development of chronic disease that are beyond the limit of traditional epidemiologi-cal studies, although they also indicate some potential pitfalls. One could think of other possible carcinogens where such an approach would be useful. For example, specific pesti-cides and sunlight exposure have both been linked to the development of non-Hodgkins lymphoma [10], although any traditional study of either of these exposures is prone to both bias and confounding.

Identification of genes that are responsible for metabolising potential pesticide carcino-gens, or modulating any immune suppression effect from sunlight, would prove instrumen-tal in identifying the causal nature of these exposures. Similarly, potential risk factors for breast cancer that remain to be clarified include dietary fat, passive smoking and specific endogenous hormones. Again, identification of functional genes that resulted in signifi-cant between person variations for these potential carcinogens would enable the testing of specific causal hypotheses. However, we are still very much in ignorance regarding the role of specific genes in metabolising most of the above exposures and it is likely that the biggest drawback to the conduct of useful Mendelian randomisation studies will the lim-ited amount of information available on genetic metabolism of potential risk factors. Fur-ther identification of metabolising genes, including estimation of between person-variation could clearly become an area of higher priority.

Regarding other potential problems, pleitropy, or the multiple function of individu-al genes, only becomes a problem if another compound being metabolised by a gene individu-also affect the risk for the disease under question, and the two compounds are associated, resulting in confounding. If this function is known then it may be possible to adjust for the effect of the secondary compound. Possibilities for such confounding are however likely to be much more limited than those encountered with lifestyle or environmental exposures. Confounding may also occur if the gene influences the potential for exposure, such as that with alcohol dehydrogenase 1B gene (ADH1B; see Chapter 2) and alcohol consumption. Such situations are likely to be restricted to individuals with an increased genetic tendency to expe-rience a toxic reaction against a particular substance, or to addictive tendencies that are at least partially genetically regulated, e.g., for nicotine and dopamine receptor genes [11]. Similarly, genetic confounding may occur via linkage disequilibrium with other causal genes in the close vicinity of the gene under study. Again, the potential for this is likely to be limited. Mendelian randomization

94

The final limitation of Mendelian randomisation studies is one shared by traditional randomised controlled trials, i.e., the lack of replicability of studies due to their small sam-ple size. While Mendelian randomisation studies will clearly allow for a focused testing of specific hypothesis, it will be important not to forget the lessons of the first generation of genetic association studies, that results based on small sample sizes are more likely to confuse than enlighten.

References

1. Smith GD, Ebrahim S. ‘Mendelian randomization’: can genetic epidemiology contribute to un-derstanding environmental determinants of disease? Int J Epidemiol 2003;32:1–22.

2. Cherry N, Mackness M, Durrington P, Povey A, Dippnall M, Smith T, et.al. Paraoxonase (PON1) polymorphisms in farmers attributing ill health to sheep dip. Lancet 2002;359:763–4. 3. Botto LD, Yang Q. 5,10‑Methylenetetrahydrofolate reductase gene variants and congenital

anomalies: a HuGE review. Am J Epidemiol 2000;151:862–77.

4. Le Marchand L, Kolonel LN, Wilkens LR, Myers BC, Hirohata T. Animal fat consumption and prostate cancer: a prospective study in Hawaii. Epidemiology 1994;5:276–82.

5. Enattah NS, Sahi T, Savilahti E, Terwilliger JD, Peltonen L, Järvelä I. Identification of a variant associated with adult‑type hypolactasia. Nat Genet 2002;30:233–7.

6. Swallow DM. Genetics of lactase persistence and lactose intolerance. Annu Rev Genet 2003;37:197–219.

7. World Cancer Research Fund (WCRF). Food, Nutrition and the Prevention of Cancer: a Global Perspective. WCRF/AICR 1997.

8. Hecht SS. Chemoprevention of lung cancer by isothiocyanates. Adv Exp Med Biol 1996;401:1–11.

9. Brennan P. Commentary: Mendelian randomization and gene‑environment interaction. Int J Epidemiol 2004;33:17–21.

10. Thomas DC, Conti DV. Commentary: the concept of ‘Mendelian Randomization’. Int J Epide-miol 2004;33:21–5.

11. Spitz MR, Shi H, Yang F, Hudmon KS, Jiang H, Chamberlain RM, et.al. Case‑control study of the D2 dopamine receptor gene and smoking status in lung cancer patients. J Natl Cancer Inst 1998;90:358–63.