|
 
 |
| REVIEW ARTICLE |
|
| Year : 2016 | Volume
: 2
| Issue : 3 | Page : 151-156 |
|
Mendelian randomization: A biologist's perspective
Subhoshree Ghose, Akash Kumar Bhaskar, Anju Sharma, Shantanu Sengupta
Proteomics Lab, Institute of Genomics and Integrative Biology, Mathura Road, New Delhi, India
| Date of Web Publication | 2-Mar-2017 |
Correspondence Address:
Subhoshree Ghose
Proteomics Lab, IGIB, Mathura Road, New Delhi
India
 Source of Support: None, Conflict of Interest: None  | Check |
DOI: 10.4103/jpcs.jpcs_62_16
Mendelian randomization (MR) is a statistical technique used by genetic epidemiologists to determine causal effects, i.e. whether a biomarker actually influences disease risk, is it just a statistical association. It uses common genetic polymorphisms with known biological effects (propensity to drink alcohol) or effects that mimic modifiable exposures (raised blood cholesterol). For MR, it is necessary that the genotype only affects the disease status indirectly through its effect on the exposure of interest. Keywords: Genetic polymorphism, Mendelian randomization, statistics
How to cite this article:
Ghose S, Bhaskar AK, Sharma A, Sengupta S. Mendelian randomization: A biologist's perspective. J Pract Cardiovasc Sci 2016;2:151-6 |
How to cite this URL:
Ghose S, Bhaskar AK, Sharma A, Sengupta S. Mendelian randomization: A biologist's perspective. J Pract Cardiovasc Sci [serial online] 2016 [cited 2023 Jun 7];2:151-6. Available from: https://www.j-pcs.org/text.asp?2016/2/3/151/201386 |
| Introduction | |  |
The basic principle of Mendelian randomization (MR) is that genetic variants mirror the biological effects of biological risk factors which themselves alter disease risk. Therefore, genetic polymorphisms which have a clear biological function can be used to study the effect of a suspected environmental exposure on disease risk. Let us explain this with a simple example: To study the link between alcohol consumption and coronary artery disease (CAD). It would be difficult to conduct a randomized trial to study the effects of alcohol on CAD. Alcohol is oxidized to aldehyde which is oxidized by aldehyde dehydrogenases (ALDHs) to acetate. Half of the Japanese are heterozygous or homozygous for a null variant of ALDH2 and alcohol levels after a challenge goes very high in homozygotes or heterozygotes of the null variant leading to facial flushing, palpitations, and drowsiness. Therefore, there was a significant difference in drinking patterns according to genotype. This would imply that possession of the null gene, associated with lower alcohol consumption, would lead to higher myocardial infarction risk, and this has been borne out by Japanese studies. Going one step further, studies show that alcohol increases high-density lipoprotein (HDL), and it has also been shown that there is a strong link between ALDH2 and HDL with lower levels of HDL with the null gene.
| The History | | |
MR was conceptualized by a mere observation made by Martin Katan in the mid-1980s while much debate was happening over the association of low serum cholesterol levels and cancer. Katan reasonably argued that it was important to genotype the cancer patients and controls for APOE alleles, an important determinant of serum cholesterol levels to claim any causal association between serum cholesterol levels and cancer. The rationale behind this hypothesis was that if the alleles were randomly allocated to gametes according to “Mendel's second law of independent assortment,” they would not be exposed to the effects of confounders and reverse causation. Any association between APOE alleles and cancer would give indirect evidence of association with serum cholesterol levels and cancer. The situation was further complicated when the patients with a rare genetic disorder abetalipoproteinemia (rare autosomal recessive disorder hampering absorption of dietary fats) did not show any tendency to develop cancer. Although the term “MR” was not coined by Katan, the idea was attributed to him and further followed by a series of authors like Davey Smith and Ebrahim and Thomas and Conti. A report in early 2001 advocated the use of beta-carotene supplements in reducing lung cancer mortality based on an observational study;[1] in addition, beta-carotene,[2] Vitamin E,[3] Vitamin C supplements,[4] and hormone replacement therapy [5] were claimed to be protective based on large-scale trials. Repeated failures of claims made from observational epidemiology have given the impetus for thinking about new methods which take into account the residual confounding factors, selection bias, and reverse causation, in which the disease itself influences the apparent exposure generating strong and replicable associations.[6] The concept of MR can also be explained using the example of “inflammatory hypothesis” in cardiovascular disease. C-reactive proteins (CRPs), acute phase proteins which aggravate during inflammation, are known to be a risk factor for coronary heart disease (CHD) from observational studies, but after robust MR studies, the role of fibrinogen has been elucidated as a confounder. After adjusting for fibrinogen levels, the association between CRP levels and risk of CHD was found to be near null.[7] A majority of the MR studies reported in observational epidemiology are based upon cancer and cardiovascular disease epidemiology because of their multifactorial nature. The concept of MR was perceived to overcome the shortcomings of classical epidemiology which majorly suffers from the discrimination of “causation” and “correlation” in the context of multifactorial disease etiology. The major focus of this approach does not rely on identification of a causal gene or its function rather its utility is even more in ruling out the spurious claims of observational epidemiology. MR approaches are also named as “natural randomized controlled trial (RCT)” and “Mendelian deconfounders.”[8] Although RCT is the gold standard for determining the causal status of a particular risk factor, it is not applicable for cases of prolonged follow-up or rare outcomes and also ethical reasons at times. In the approaches used by MR, genetic variants are considered as an instrumental variable (IV) which is solely associated with the exposure of interest but do not have any association with any of the confounders. To simplify, MR is a strategy to reassess observational estimates using genetic variation related to the risk factor of interest. Several case–control studies have mentioned several dietary factors to be responsible for causation of noncommunicable diseases, but many of these have not been consistently replicated in cohort studies. In this light, MR strategy has used genomic information in combination with epidemiological observations to further strengthen the etiologic role of nutritional factors. Thus, observational studies produce associations that possibly are not true indicators of disease risk, and this eventually forms the basis of MR framework. However, it should be emphasized here that MR does not discover de novo genetic variation that is a causative factor for any disease. It also does not explain the intermediate mechanistic pathways through which exposure influences the outcome of the disease.[9] With all the definitions considered, this review illustrates on the approaches used by MR studies and its applications in finding some causal association between a risk factor and disease occurrence.
| Approaches Used by Mendelian Randomization | | |
There are majorly three specific methods used in MR analysis:[10]
- Using genetic variables as a proxy for modifiable environmental exposures since they are not widely associated with other behavioral, psychological, and social factors
- Performing an IV analysis: A genotype is considered as an IV only if it meets the criteria that the genotype should be associated with the exposure, a genotype should be associated with the outcome through the studied exposure only, and it should also be independent of other factors that affect the outcome
- Comparing the observed with the expected genotype-outcome association.
Broadly, MR has been performed in a single study population or data using multiple study populations. For studies using data from single study population, it uses triangulation approach where it compares the observed association between the genotype and outcome with that of the expected outcome if the associations were causal. Another statistical method termed as formal IV analysis has also been used to obtain a causal inference about the effect of modifiable exposure on the outcome.[11] The latter approach employs data from multiple study populations and performs a formal IV analysis on all individual populations followed by a meta-analysis. It also does all the above-mentioned analyses after pooling the data from different subpopulations. Although all of these approaches are valid, its pertinence depends on the aim of the experiment, i.e. whether it is just a test for causality or it is a quantification of the causal effect of the exposure on the outcome.
The merit of the IV was first proposed in the field of econometrics literature following which it has been used in molecular epidemiology to draw causal inferences use this.[12] In a simplistic manner, it can be explained that in the triangulation approach, causal effect of intermediate phenotype (IP) on G (genotype) can be estimated by the ratio of the coefficients for the regression of D (disease) on G and IP on G.
Several underlying complications in this approach give rise to false-positive inferences when the genotype (G) has a direct effect on the disease (D) or when the gene has a pleiotropic effect.
| A Few Case Studies of Multifactorial Disease | | |
Alcohol intake and colorectal cancer
Epidemiological evidence have already reported an association between alcohol consumption and risk of colorectal cancer, but these observations are subjected to potential confounding effects such as smoking, diet, and other socioeconomic factors [13] and reverse causation effect. This apparent dilemma has been resolved using MR. Aldehyde dehydrogenase 2 (ALDH2) variant has been used as a proxy for alcohol exposure to establish a link between alcohol consumption and other relevant diseases. The advantage of using ALDH2 genotype as an instrument variable is that it is not affected by reporting bias or other effects, and it can categorize the population into different groups of varying alcohol intake based on their genotype information [Figure 1]. | Figure 1: Concept of Mendelian randomization to explain the link of alcohol consumption and coronary artery disease.
Click here to view |
Adiposity and cardiometabolic traits
MR-based meta-analysis has also been done to establish a causal association between adiposity and cardiometabolic traits using adiposity-associated variant rs9939609 at the FTO (obesity-associated gene) locus as an IV.[14] These reports have confirmed the novel association of body mass index (BMI)-related traits to incident heart failure and liver enzymes such as alanine aminotransferase and gamma-glutamyltransferase.
Obesity and multiple sclerosis
A recent meta-analysis has reported BMI to be a causal risk factor for multiple sclerosis (MS), an autoimmune disease using data available in genetic investigation of anthropometric test consortium, one of the largest genome-wide association studies (GWASs) for BMI and MS.[15] Since genetic variants for BMI are quite stable, these studies provide evidence for better measures of MS, but they do not explain the exact molecular pathophysiology.
| Mendelian Randomization in One-Carbon Metabolism | | |
Application of MR studies will be discussed here to address practical problems in etiological epidemiological studies in relevance to CHD. In this section, we majorly stress on the triangulation of genotype, phenotype, and disease risk with respect to one-carbon metabolism and CHD. Metabolites such as folate, homocysteine, S-adenosyl methionine, S-adenosyl homocysteine, and some gene variants of one-carbon metabolism have been found to be associated with CHD. Out of all, the association of homocysteine with CHD has generated considerable interest. Data from the observational studies have regularly confirmed that the higher plasma homocysteine levels are one of the major risk factors for CHD,[16] and RCTs have shown that an adequate increase in folate intake can lower down the level of homocysteine in plasma.[17] Consequently, the intervention studies of folate intake could lead to the decline of CHD risk if the association between the homocysteine and CHD is causative. Conversely, in these studies, the homocysteine–CHD association might be confounded by a variety of factors such as smoking, socioeconomic background, and also existing atherosclerosis itself.[18] The strength of the cohort and case–control studies derived homocysteine–CHD association can be explained by the phenomena of reverse causality.[19] Regardless of favorable results from RCTs of folate supplementation, existing evidence are mainly based on the epidemiological observational studies which suffer from the limitation of confounding.
Hence, the question remains in the absence of RCTs can we say that the folate is a good candidate for intervention studies and will the relation between folate, homocysteine, and CHD be causal. Here, MR could elucidate the problem by integrating genetic epidemiology with functional genomics.
One-carbon metabolism pathway which involves homocysteine and folate is well understood and conserved throughout species. The functional gene variant, methylenetetrahydrofolate reductase (MTHFR) C677T of enzyme MTHFR, encodes for a thermolabile enzyme where C is substituted at position 677 in place of T that results in reduced enzyme activity. MTHFR converts 5, 10-methylenetetrahydrofolate to 5-methyltetrahydrofolate, which in turn results in the remethylation of homocysteine to methionine [Figure 1]. Hence, in a population, people having this polymorphism mimic a low dietary folate condition and should have higher plasma homocysteine levels. TT homozygous individuals on an average have 2.6 µmol/L homocysteine levels higher than individuals with common CC homozygous allele.[20] If homocysteine level is a causal factor, then the individuals with homozygous TT genotype should be at higher risk of CHD than the individual with CC genotype. Here, the possibility of confounding is almost negligible because the observation is totally based on the genotype, where the genotype of the person is decided by the random assortment of the allele (Mendel's second law) at the time of gamete formation. The total plasma homocysteine level of TT individual is high and is no more likely to be influenced by the confounding factors than the individual with CC genotype.
A meta-analysis of observational studies involving 6186 CHD events had suggested that the elevated total plasma homocysteine level is an independent predictor of CHD risk in healthy population.[21] However, from here, we cannot say whether elevated homocysteine level is causally related to CHD because health behaviors, socioeconomic factors, and reverse causation can influence homocysteine levels, and we cannot dilute out the effect of confounding factors. Some more robust system based on genetic variant (MTHFR-C677T) which has an impact on homocysteine level and disease risk can give a reliable inference about the causality. Another meta-analysis of observational studies conducted on individuals with the MTHFR 677TT genotype involving 11162 CHD cases and 12758 controls showed that the TT individuals have 16% higher odds of CHD compared to CC individuals.[22] MTHFR C677T genotype mimics the low folate condition and cannot be determined by the confounding factors (health behaviors, socioeconomic factors, and reverse causation). These results support the hypothesis that perturbed folate metabolism, resulting in high homocysteine levels, has a causal impact to CHD.
Thus, the findings from the genetic association give the actual picture of the epidemiological studies and are neither confounded by the health behaviors, socioeconomic factors, nor diluted by the statistical errors. MR helps in understanding the contribution of environmental determinants of disease in genetic epidemiological studies. The triangulation of the association between genotype (MTHFR C677T), phenotype (elevated homocysteine level), and disease (CHD) risk provides a worthy evidence of causal effect of homocysteine and protective outcome of folate on CHD [Figure 2]. | Figure 2: (a) Simplistic representation of one-carbon metabolism, (b) triangulation approach in understanding genotype–phenotype association with disease.
Click here to view |
| Biological Applications of Mendelian Randomization | | |
Drug development
MR could be used at various stages in the drug development pipeline such as predicting efficacy, side effects, discovery, and repositioning. MR studies could be applied specifically to design small molecule libraries against the desired drug target rather than focusing on biomarkers, exposures, and other intermediates downstream of the intended target. For example, if variants within a gene encoding a drug target have a marginal effect on disease risk, then other proteins should be prioritized to design novel targets. Thus, it could save substantial time and cost in developing new drugs and also failures of many drugs due to lack of efficacy at the final stages of clinical trials.[23]
One example of an MR study predicting efficacy is the development of antibodies against the proprotein convertase subtilisin/kexin type 9 serine protease (PCSK9), where genetic variants of PCSK9 have found to be linked with decreased low-density lipoprotein-C and cardiovascular heart disease.[24] Another study showed loss-of-function mutations within the APOC3 gene that are linked with low levels of triglycerides and also related to decreased risk of ischemic cardiovascular disease.[25],[26] In another example, studies utilizing CRP variants targeting inflammation in CHD have proved to be futile owing to the possibility of CRP being an off target in CHD.[27],[28]
MR can also be beneficial in examining on- and off-target drug effects, potential side effects. A few recent MR studies have revealed that statins known to reduce the risk of CHD also increase the risk of developing Type II diabetes because of two variants in the gene HMG-CoA reductase. Thus, it indicates this side effect to be a possible on-target effect of the drug.[23],[29] On the strength of MR studies, it has been possible to discover the correlation between different biomarkers and genetic variants of a drug target, which makes it crucial to monitor these biomarkers in clinical trials, for example, P446 L variant of glucokinase regulator (GCKR) gene has been linked to both, lower plasma glucose levels and increased triglyceride levels, thus suggesting additional monitoring of triglyceride levels in future clinical trials of GCKRs.[23],[30]
MR could also be employed to search for new causal associations, which has both potential to result in new drug targets and also repurposing or repositioning of licensed drugs for other conditions. Repositioning of the drug has an advantage since it has already passed all safety and toxicity tests of preclinical research and Phase I clinical trials and hence will be approved rapidly.[31] A MR-pheWAS methodology has been recently proposed which is a “hypothesis-free” approach where a given exposure is examined for causality of distinctive outcomes using linked genetic variants.[32] Many existing therapies have already been repositioned effectively through clinical trials, and other potential candidates are currently undergoing trials for being repurposed. The example includes tocilizumab (anti-interleukin 6 [IL-6] antibodies) which targets variant of IL6R gene has already been approved for treating rheumatoid arthritis. The nonsynonymous variants of IL6R gene have also been found to be related to decreased risk of CHD, suggesting that the monoclonal antibody (tocilizumab) designed to target IL-6 receptor could also be repositioned and therefore might play a crucial role in the prevention of CHD.[33],[34]
| Limitations of Mendelian Randomization | | |
Despite wide applications of MR studies with a series of quantitative and nonquantitative examples, there are certain disadvantages with MR studies which should be considered while assessing.
Low statistical power
MR studies often have low statistical power and thus effective assessments are inaccurate. Genetic variants represent a small fraction of the population variance in a phenotype. Hence, any difference in the outcome because of the variant is possibly small and thus requires a large sample size to get statistically significant results.[35]
Pleiotropy
It is a genetic variant having multiple functions. MR studies have a major disadvantage that assumes the use of genetic variants as IVs which only directly affect the exposure variable. Here, the core assumption of MR approach gets violated if genetic variants affect multiple phenotypes.[36] This means that the genetic variant is linked with a risk factor for the outcome and does not depend on the causal path of exposure, or genetic variant affects outcome independent of exposure.[37] Proteins are the complex molecules that work in a very organized network as few proteins play a role in large complexes and their variants can affect several phenotypes.[38]
| Canalization or Developmental Compensation | | |
Canalization or developmental compensation refers to the buffering of the effects of genetic variants or environmental forces during development of an organism that can offset the effect of a genetic variant to produce an invariant phenotype. There are various molecular mechanisms through which buffering can be achieved like genetic redundancy (more than one gene having similar function) or through substitute metabolic routes, where the complex metabolic pathways allow recruitment of distinctive pathways to have similar phenotypic endpoint, feedback regulation, and cooperative biochemical interactions.[39],[40] These buffering mechanisms invalidate the MR assumptions as they alter the effect of a genotype on the outcome at maturity, without affecting the relation between genotype and the modifiable exposure of interest. One example is myoglobin inhibition in mice interrupts the function of the myocardium that indicates similar effects of deleterious genetic variants. Whereas myoglobin knockout mice evidently showed normal myocardium function, this implies the presence of a compensatory mechanism during development.[41]
| Population Stratification | | |
Population stratification occurs where sample consists of different subpopulations that encounter both difference in the frequency of alleles of interest and mean of their exposure and have different disease rates (or different distributions of traits). This might produce a spacious link between genotype, exposure, and disease in the whole study population.[23] For example, in studies like GWAS, the genetic variants used as IVs vary in frequency between populations, this could result in biased examination and lead to false positives.[42] This gives rise to concerns that population stratification may cause false-positive inferences in population-based genetic association studies and generate biased findings in MR studies. This encourages the use of family-based studies in genetic epidemiology.[43]
| Linkage Disequilibrium | | |
Linkage disequilibrium is an exemption to Mendel's second law of independent assortment. Genetic variants which are placed in adjacent loci are expected to be inherited together as they have lower frequency of recombination between them. This results in another trait-influencing genetic associations that have the aptitude to confound MR studies, possibly as associated single nucleotide polymorphisms (SNPs) often cluster together.
Lack of suitable polymorphisms
Another caveat of MR studies is that it only explores functional polymorphisms (or markers linked to such functional polymorphisms) that are significant to the modifiable exposure of interest. In genetic association studies, it has been indicated that in several cases, even if a locus is implicated in a disease-related metabolic process, then there may be no functional polymorphism or suitable marker to allow this study.[44] One of the examples is an association of Vitamin C, and CHD is linked to an SLC23A1 gene encoding for the Vitamin C transporter SVCT1, responsible for Vitamin C transport by intestinal cells which could be a candidate for MR studies. However, there was no common SNP found in the variant that could be used.[45] Hence, the possibility will further depend on empirical evidence within the human genome, regarding the density of markers and functional polymorphisms.
| Conclusions | | |
Thus, MR has its own advantages and disadvantages just as RCTs do. HoweverHowever, on a broader scale, it has the ability to go beyond the limits of observational epidemiology to have meaningful contribution in the field of pharmacological development. In this review, we have tried to exemplify the role of MR in the etiological investigation of complex diseases. In spite of the limitations described, its emergence in the field of genetic epidemiology has been due to its robustness in finding potential causal associations for necessary health interventions.
Acknowledgment
We would like to thank the Centre for Cardiovascular and Metabolic Disease Research CARDIOMED, BSC 0122.
Financial support and sponsorship
Nil.
Conflicts of interest
There are no conflicts of interest.
| References | | |
| 1. | Willett WC. Vitamin A and lung cancer. Nutr Rev 1990;48:201-11.  |
| 2. | Manson JE, Colditz GA, Stampfer MJ, Willett WC, Rosner B, Monson RR, et al. A prospective study of obesity and risk of coronary heart disease in women. N Engl J Med 1990;322:882-9. |
| 3. | Rimm EB, Stampfer MJ, Ascherio A, Giovannucci E, Colditz GA, Willett WC. Vitamin E consumption and the risk of coronary heart disease in men. N Engl J Med 1993;328:1450-6. |
| 4. | Osganian SK, Stampfer MJ, Rimm E, Spiegelman D, Hu FB, Manson JE, et al. Vitamin C and risk of coronary heart disease in women. J Am Coll Cardiol 2003;42:246-52. |
| 5. | Stampfer MJ, Colditz GA. Estrogen replacement therapy and coronary heart disease: A quantitative assessment of the epidemiologic evidence. Prev Med 1991;20:47-63. |
| 6. | Smith GD, Ebrahim S. Data dredging, bias, or confounding. BMJ 2002;325:1437-8. |
| 7. | Emerging Risk Factors Collaboration, Kaptoge S, Di Angelantonio E, Lowe G, Pepys MB, Thompson SG, et al. C-reactive protein concentration and risk of coronary heart disease, stroke, and mortality: An individual participant meta-analysis. Lancet 2010;375:132-40. |
| 8. | Tobin MD, Minelli C, Burton PR, Thompson JR. Commentary: Development of Mendelian randomization: From hypothesis test to 'Mendelian deconfounding'. Int J Epidemiol 2004;33:26-9. |
| 9. | Schatzkin A, Abnet CC, Cross AJ, Gunter M, Pfeiffer R, Gail M, et al. Mendelian randomization: How it can – and cannot – help confirm causal relations between nutrition and cancer. Cancer Prev Res (Phila) 2009;2:104-13. |
| 10. | Boef AG, Dekkers OM, le Cessie S. Mendelian randomization studies: A review of the approaches used and the quality of reporting. Int J Epidemiol 2015;44:496-511. |
| 11. | Didelez V, Sheehan N. Mendelian randomization as an instrumental variable approach to causal inference. Stat Methods Med Res 2007;16:309-30. |
| 12. | Thomas DC, Conti DV. Commentary: The concept of 'Mendelian Randomization'. Int J Epidemiol 2004;33:21-5. |
| 13. | Wang J, Wang H, Chen Y, Hao P, Zhang Y. Alcohol ingestion and colorectal neoplasia: A meta-analysis based on a Mendelian randomization approach. Colorectal Dis 2011;13:e71-8. |
| 14. | Fall T, Hägg S, Mägi R, Ploner A, Fischer K, Horikoshi M, et al. The role of adiposity in cardiometabolic traits: A Mendelian randomization analysis. PLoS Med 2013;10:e1001474. |
| 15. | Mokry LE, Ross S, Timpson NJ, Sawcer S, Davey Smith G, Richards JB. Obesity and multiple sclerosis: A Mendelian randomization study. PLoS Med 2016;13:e1002053. |
| 16. | Ford ES, Smith SJ, Stroup DF, Steinberg KK, Mueller PW, Thacker SB. Homocyst (e) ine and cardiovascular disease: A systematic review of the evidence with special emphasis on case-control studies and nested case-control studies. Int J Epidemiol 2002;31:59-70. |
| 17. | Lowering blood homocysteine with folic acid based supplements: Meta-analysis of randomised trials. Homocysteine Lowering Trialists' Collaboration. BMJ 1998;316:894-8. |
| 18. | Hallberg L, Hulthén L. Prediction of dietary iron absorption: An algorithm for calculating absorption and bioavailability of dietary iron. Am J Clin Nutr 2000;71:1147-60. |
| 19. | Clarke R, Bennett DA, Parish S, Verhoef P, Dötsch-Klerk M, Lathrop M, et al. Homocysteine and coronary heart disease: Meta-analysis of MTHFR case-control studies, avoiding publication bias. PLoS Med 2012;9:e1001177. |
| 20. | Brattström L, Wilcken DE, Ohrvik J, Brudin L. Common methylenetetrahydrofolate reductase gene mutation leads to hyperhomocysteinemia but not to vascular disease: The result of a meta-analysis. Circulation 1998;98:2520-6. |
| 21. | Homocysteine Studies Collaboration. Homocysteine and risk of ischemic heart disease and stroke: A meta-analysis. JAMA 2002;288:2015-22. |
| 22. | Klerk M, Verhoef P, Clarke R, Blom HJ, Kok FJ, Schouten EG; MTHFR Studies Collaboration Group. MTHFR 677C-->T polymorphism and risk of coronary heart disease: A meta-analysis. JAMA 2002;288:2023-31. |
| 23. | Evans DM, Davey Smith G. Mendelian randomization: New applications in the coming age of hypothesis-free causality. Annu Rev Genomics Hum Genet 2015;16:327-50. |
| 24. | Cohen JC, Boerwinkle E, Mosley TH Jr., Hobbs HH. Sequence variations in PCSK9, low LDL, and protection against coronary heart disease. N Engl J Med 2006;354:1264-72. |
| 25. | Jørgensen AB, Frikke-Schmidt R, Nordestgaard BG, Tybjærg-Hansen A. Loss-of-function mutations in APOC3 and risk of ischemic vascular disease. N Engl J Med 2014;371:32-41. |
| 26. | TG and HDL Working Group of the Exome Sequencing Project, National Heart, Lung, and Blood Institute, Crosby J, Peloso GM, et al. Loss-of-function mutations in APOC3, triglycerides, and coronary disease. N Engl J Med 2014;371:22-31. |
| 27. | C Reactive Protein Coronary Heart Disease Genetics Collaboration (CCGC), Wensley F, Gao P, Burgess S, Kaptoge S, Di Angelantonio E, et al. Association between C reactive protein and coronary heart disease: Mendelian randomisation analysis based on individual participant data. BMJ 2011;342:d548. |
| 28. | Timpson NJ, Lawlor DA, Harbord RM, Gaunt TR, Day IN, Palmer LJ, et al. C-reactive protein and its role in metabolic syndrome: Mendelian randomisation study. Lancet 2005;366:1954-9. |
| 29. | Swerdlow DI, Preiss D, Kuchenbaecker KB, Holmes MV, Engmann JE, Shah T, et al. HMG-coenzyme A reductase inhibition, type 2 diabetes, and bodyweight: Evidence from genetic analysis and randomised trials. Lancet 2015;385:351-61. |
| 30. | Beer NL, Tribble ND, McCulloch LJ, Roos C, Johnson PR, Orho-Melander M, et al. The P446L variant in GCKR associated with fasting plasma glucose and triglyceride levels exerts its effect through increased glucokinase activity in liver. Hum Mol Genet 2009;18:4081-8. |
| 31. | Ashburn TT, Thor KB. Drug repositioning: Identifying and developing new uses for existing drugs. Nat Rev Drug Discov 2004;3:673-83. |
| 32. | Millard LA, Davies NM, Timpson NJ, Tilling K, Flach PA, Davey Smith G. MR-PheWAS: Hypothesis prioritization among potential causal effects of body mass index on many outcomes, using Mendelian randomization. Sci Rep 2015;5:16645. |
| 33. | Interleukin- Receptor Mendelian Randomisation Analysis (ILR MR) Consortium, Swerdlow DI, Holmes MV, Kuchenbaecker KB, Engmann JE, Shah T, et al. The interleukin-6 receptor as a target for prevention of coronary heart disease: A mendelian randomisation analysis. Lancet 2012;379:1214-24. |
| 34. | Genovese MC, McKay JD, Nasonov EL, Mysler EF, da Silva NA, Alecock E, et al. Interleukin-6 receptor inhibition with tocilizumab reduces disease activity in rheumatoid arthritis with inadequate response to disease-modifying antirheumatic drugs: The tocilizumab in combination with traditional disease-modifying antirheumatic drug therapy study. Arthritis Rheum 2008;58:2968-80. |
| 35. | Pierce BL, Burgess S. Efficient design for Mendelian randomization studies: Subsample and 2-sample instrumental variable estimators. Am J Epidemiol 2013;178:1177-84. |
| 36. | Davey Smith G, Hemani G. Mendelian randomization: Genetic anchors for causal inference in epidemiological studies. Hum Mol Genet 2014;23:R89-98. |
| 37. | Wagner GP, Zhang J. The pleiotropic structure of the genotype-phenotype map: The evolvability of complex organisms. Nat Rev Genet 2011;12:204-13. |
| 38. | Solovieff N, Cotsapas C, Lee PH, Purcell SM, Smoller JW. Pleiotropy in complex traits: Challenges and strategies. Nat Rev Genet 2013;14:483-95. |
| 39. | Smith GD, Ebrahim S. 'Mendelian randomization': Can genetic epidemiology contribute to understanding environmental determinants of disease? Int J Epidemiol 2003;32:1-22. |
| 40. | Ziegler A, Pahlke F, König IR. Comments on 'Mendelian randomization: Using genes as instruments for making causal inferences in epidemiology' by Debbie A. Lawlor, R. M. Harbord, J. A. Sterne, N. Timpson and G. Davey Smith, Statistics in Medicine, DOI: 10.1002/sim. 3034. Stat Med 2008;27:2974-6. |
| 41. | Garry DJ, Ordway GA, Lorenz JN, Radford NB, Chin ER, Grange RW, et al. Mice without myoglobin. Nature 1998;395:905-8. |
| 42. | Burgess S, Butterworth A, Malarstig A, Thompson SG. Use of Mendelian randomisation to assess potential benefit of clinical intervention. BMJ 2012;345:e7325. |
| 43. | Gauderman WJ, Witte JS, Thomas DC. Family-based association studies. J Natl Cancer Inst Monogr 1999;26:31-7. |
| 44. | Weiss KM, Terwilliger JD. How many diseases does it take to map a gene with SNPs? Nat Genet 2000;26:151-7. |
| 45. | Erichsen HC, Eck P, Levine M, Chanock S. Characterization of the genomic structure of the human Vitamin C transporter SVCT1 (SLC23A2). J Nutr 2001;131:2623-7. |
[Figure 1], [Figure 2]
|
Search Pubmed for