Mendelian randomization

The Mendelian randomization refers to a method of epidemiology and biostatistics for non-experimental studies to determine the impact of varying risk factors ( english modifiable risk factors ) on diseases using the variation of genes of known function. A false inverse causal relationship (event B causes event A, effect creates cause) and confounding factors can be controlled without an epidemiological study .

properties

Mendelian randomization is used to make causal inferences that are important to public health. Mendelian randomization here uses Mendelian genetics to one of Quasirandomisierung to reach and thus a quasi-randomized controlled clinical trial to create. The strength of Mendelian randomization is that the number of cases can quickly and cheaply become very large, which increases the statistical significance . For example, Mendel's randomization can be used to investigate the effect of alcohol consumption on mortality . Mendelian randomization provides good control of reverse causation and confounding factors which normally reduce or even falsify the informative value of epidemiological studies. Gray and Wheatley introduced the term in 1991 using Mendelian randomization to biostatistically compare bone marrow transplantation with chemotherapy .

principle

“Genetics is actually in a fortunate state in that it protects the geneticist from the many difficulties of controlled comparison (what is meant here is the controlled randomized trial). The different possible genotypes have already been randomized in a nice way by the meiotic process. A more perfect control of the conditions than that of different genotypes as they occur in a population is not possible. "

- Ronald A. Fisher

The epidemiologist or biostatistician uses the natural basis of genetics to have the randomization carried out for him. This allows him to draw causal inferences even from non-randomized observation data. Mendelian randomization uses existing genetic polymorphisms , for example those that control alcohol consumption or that have an influence on blood cholesterol. The genotypes are randomly passed on from parents to children during meiosis. If one also acceptance of the panmixia added takes (the choice of partner is not associated with the genotype), then the distribution of the genotypes in the population is independent of the disturbing factors which are typically interfere with other designs observational study in epidemiology. And so Mendelian randomization can be viewed as a “nature-provided” controlled randomized study (the gold standard in epidemiology and all empirical sciences). Because genetic polymorphism is the tool, Mendelian randomization requires genetic association studies that provide good candidate genes for the response to risk exposure. From a statistical point of view, Mendelian randomization is an application of the technique of instrumental variables, with the genotype as a tool for exposure of interest. Potential false conclusions threaten if the instrument has direct effects on the development of the disease, as well as if there are linkage disequilibria with unmeasured direct-causal variants or genetic heterogeneity, pleiotropy or population stratification.

literature

R. Gray, K. Wheatley: How to avoid bias when comparing bone marrow transplantation with chemotherapy . In: Bone Marrow Transplant . 7 Suppl. 3, 1991, p. 9-12 , PMID 1855097 .
George Davey Smith , Shah Ebrahim: Mendelian randomization: can genetic epidemiology contribute to understanding environmental determinants of disease? In: International Journal of Epidemiology . tape 32 , 2003, p. 1-22 , doi : 10.1093 / ije / dyg070 .
George Davey Smith, Shah Ebrahim, Sarah Lewis, Anna L Hansell, Lyle J Palmer, Paul R Burton: Genetic epidemiology and public health: hope, hype, and future prospects . In: The Lancet . tape 366 , no. 9495 , October 2005, p. 1484-1498 , doi : 10.1016 / S0140-6736 (05) 67601-5 .

Individual evidence

^ G. Davey Smith: Mendelian Randomization for Strengthening Causal Inference in Observational Studies: Application to Gene × Environment Interactions . In: Perspectives on Psychological Science . 2010, pp. 527-545. doi : 10.1177 / 1745691610383505 .
↑ Martjin B. Katan: Apolipoprotein E isoforms, serum cholesterol, and cancer . In: Lancet . tape 1 , no. 8479 , March 1986, ISSN 0140-6736 , p. 507-508 , PMID 2869248 .
^ DC Thomas, DV Conti: Commentary: The concept of 'Mendelian Randomization' . In: International Journal of Epidemiology . tape 33 , no. 1 , February 1, 2004, ISSN 1464-3685 , p. 21-25 , doi : 10.1093 / ije / dyh048 .
↑ V. Didelez, N. Sheehan: Mendelian randomization as to instrumental variable approach to causal inference . In: Statistical Methods in Medical Research . tape 16 , no. 4 , August 1, 2007, ISSN 0962-2802 , p. 309-330 , doi : 10.1177 / 0962280206077743 .
^ G. Davey Smith, S. Ebrahim: What can mendelian randomization tell us about modifiable behavioral and environmental exposures? In: BMJ . tape 330 , no. 7499 , May 5, 2005, ISSN 0959-8138 , p. 1076-1079 , doi : 10.1136 / bmj.330.7499.1076 , PMID 15879400 .

[DaveySmith2010-1] G. Davey Smith: Mendelian Randomization for Strengthening Causal Inference in Observational Studies: Application to Gene × Environment Interactions . In: Perspectives on Psychological Science . 2010, pp. 527-545. doi : 10.1177 / 1745691610383505 .

[Katan-2] Martjin B. Katan: Apolipoprotein E isoforms, serum cholesterol, and cancer . In: Lancet . tape 1 , no. 8479 , March 1986, ISSN 0140-6736 , p. 507-508 , PMID 2869248 .

[Thomas-3] DC Thomas, DV Conti: Commentary: The concept of 'Mendelian Randomization' . In: International Journal of Epidemiology . tape 33 , no. 1 , February 1, 2004, ISSN 1464-3685 , p. 21-25 , doi : 10.1093 / ije / dyh048 .

[Didelez-4] V. Didelez, N. Sheehan: Mendelian randomization as to instrumental variable approach to causal inference . In: Statistical Methods in Medical Research . tape 16 , no. 4 , August 1, 2007, ISSN 0962-2802 , p. 309-330 , doi : 10.1177 / 0962280206077743 .

[Smith-5] G. Davey Smith, S. Ebrahim: What can mendelian randomization tell us about modifiable behavioral and environmental exposures? In: BMJ . tape 330 , no. 7499 , May 5, 2005, ISSN 0959-8138 , p. 1076-1079 , doi : 10.1136 / bmj.330.7499.1076 , PMID 15879400 .