Q J Med 2004; 97: 163-166
QJM vol. 97 no. 3 (c) Association of Physicians 2004; all rights reserved.
Commentary |
Fibrinogen, C-reactive protein and coronary heart disease: does Mendelian randomization suggest the associations are non-causal?
From the Department of Social Medicine, University of Bristol, Bristol, UK
 |
Introduction |
|---|
 Top Introduction What is Mendelian randomization?Mendelian randomization,...ConclusionsReferences |
|---|
Two recent reviews in QJM have evaluated the role of fibrinogen1 and C-reactive protein (CRP)2 as risk factors for cardiovascular disease. Both raised fibrinogen and CRP levels clearly predict future risk of cardiovascular events, and could be used to identify those patients who would get the greatest absolute benefit from intervention targeting blood pressure and cholesterol lowering, smoking cessation or exercise promotion. The reviews also raise the issue of a causal relationship between fibrinogen or CRP and cardiovascular disease, and discuss the roles of pharmacological or lifestyle changes as means of modifying fibrinogen or CRP levels.1,2 However, doubt is cast on the causal nature of either factor by evidence produced by studies using the ‘Mendelian randomization’ paradigm.3 Here we briefly outline Mendelian randomization, and review what studies using this approach have to say regarding the potential causal roles of fibrinogen and CRP with respect to cardiovascular disease.
|
What is Mendelian randomization? |
|---|
|
TopIntroductionWhat is Mendelian randomization?Mendelian randomization,...ConclusionsReferences |
|---|
Epidemiology aims to inform public-health strategies by identifying modifiable risk factors for disease. However, in the past decade, several well-publicized examples of the misleading identification of such factors by observational epidemiological studies have been reported: ß-carotene and cancer; vitamin C and coronary heart disease; and hormone replacement therapy and cardiovascular disease among them. In observational epidemiological studies these factors appeared to have clinically important protective effects, but randomized controlled trials demonstrated that they were, if anything, harmful. The probable reason for these misleading findings in the observational studies is that there is considerable confounding between, for example, vitamin C consumption (or taking ß-carotene supplements) and various behavioural and socio-economic factors related to increased risk of disease. Because of measurement problems, observational epidemiology simply cannot deal with this issue (despite valiant attempts—and strident claims—to the contrary).
Genetic epidemiology is often seen as the antipathy of public-health epidemiology, but paradoxically it now offers one of the most attractive solutions to this quandary. Mendelian randomization—the random assortment of genes from parents to offspring that occurs during gamete formation and conception—provides a method of assessing whether certain environmental exposures are causally related to a disease. The association between risk of a disease and a genetic variant that mimics the biological link between a proposed exposure and disease is not generally susceptible to the reverse causation or confounding that may distort interpretations of conventional observational studies. Several examples where the phenotypic effects of polymorphisms are well-documented provide encouraging evidence of the explanatory power of Mendelian randomization.3
The example of homocysteine and CHD illustrates the approach. The association between homocysteine and CHD, which has been repeatedly reported in observational epidemiological studies, is heavily confounded by factors such as smoking and blood pressure, which tend to be related to elevated homocysteine levels. Polymorphisms of the methyltetrahydrofolate reductase gene result in lifelong differences in the levels of serum homocysteine. The TT MTHFR genotype is associated with increased homocysteine, so examination of its relation with CHD risk provides an unconfounded test of the hypothesis that homocysteine increases CHD risk. Indeed, the association between MTHFR genotype and CHD is mathematically close to that predicted by a combination of the strength of association between MTHFR genotype and homocysteine, and the strength of association between homocysteine and CHD.3 Figures 1 and 2 contrast the logic of the associations studied in conventional epidemiology and those investigated through use of the ‘Mendelian randomization’ paradigm. Figure 3 also demonstrates that the principle behind Mendelian randomization holds in the MTHFR 
homocysteine CHD case. There is no confounding of genotype and other CHD risk factors, but considerable confounding of measured blood homocysteine levels by these risk factors is seen.
|
|
|
|
Mendelian randomization, fibrinogen and CRP |
|---|
|
TopIntroductionWhat is Mendelian randomization?Mendelian randomization,...ConclusionsReferences |
|---|
The approach described above has been applied with respect to fibrinogen and CRP. Youngman et al.4 examined a polymorphism in the promoter region of the ß-fibrinogen gene which had a per-allele influence on plasma fibrinogen levels of 0.12 g/l. In their large case-control study, fibrinogen was related to coronary heart disease (CHD) risk, with 0.12 g/l higher fibrinogen being associated with a relative risk of CHD of 1.20 (95%CI 1.13–1.26). Since 0.12 g/l higher fibrinogen is the per allele difference in fibrinogen, it would therefore be predicted that there should be a per allele effect on CHD, with a relative risk of approximately 1.20. However, when genotype was related to CHD risk, essentially no relationship was seen, with a per allele relative risk of 1.03 (0.96–1.10). This effect excludes the estimate obtained by examining the fibrinogen-CHD association.
In this case-control study, therefore, individuals with a genotype that would have subjected them to long-term elevated fibrinogen levels did not experience any increased risk of CHD. This suggests that circulating fibrinogen may not be a causal factor with respect to CHD, despite being associated with CHD risk in observational studies. It seems likely that the fibrinogen-CHD association was over-estimated in this case-control study, since previous pooling of data from prospective studies, in which fibrinogen levels were measured before the occurrence of CHD, found a pooled relative risk of 1.8 (95%CI 1.6–2.0) for a 1 g/l difference in fibrinogen,5 equivalent to a relative risk of 1.07 (95%CI 1.06–1.09) for a 0.12 g/l difference, assuming a linear relationship between fibrinogen level and log-risk. This suggests that reverse causality (existing atheroma increasing the levels of fibrinogen, and therefore fibrinogen automatically being associated with the future risk of CHD) plays a part in explaining the case-control association, and that confounding will also still influence prospective studies.
Reverse causation and confounding may explain why conventional observational epidemiological studies consistently find a positive association between fibrinogen and risk of CHD. Fibrinogen levels are related to many CHD risk factors in a way that would generate a positive fibrinogen-CHD association. For example, fibrinogen levels are higher in smokers, are higher in people who come from deprived backgrounds or have low educational achievement, shorter people (who have higher risk of coronary heart disease) have higher fibrinogen levels, moderate alcohol consumption (which appears to protect against CHD) is associated with lower fibrinogen levels, and fibrinogen is positively correlated with serum cholesterol level.6 In one study,7 plasma fibrinogen was strongly associated with confounding factors such as these, but the genotype associated with higher plasma fibrinogen was not related to these confounding factors. These data illustrate the basic principle of Mendelian randomization, that genotype-disease associations can provide an unconfounded test of the association between a particular phenotype and disease.
In the case of CRP, a common genetic variant has been identified that is associated with raised plasma CRP concentrations. In a nested case-control study, the association of the variant with CRP was replicated, and CRP was related to CHD risk, but genotype had no association with CHD.8
While there are several potential problems with interpreting the findings from studies based on principles of Mendelian randomization,3 it is a potentially powerful technique for determining whether associations found in observational epidemiology are likely to be causal. One major issue is sample size, since very large studies are required to give sufficiently precise results to exclude a meaningful genotype-disease association, as these effects are generally likely to be small. With respect to both fibrinogen and CRP, the studies suggest that, at the very least, the observed association between measured fibrinogen or CRP and CHD is overestimated—the best estimate being close to the null—but there are wide confidence intervals around the estimates, particularly with respect to CRP, where the study was only of moderate size. The relationships between genotype and disease risk and genotype and fibrinogen levels observed in the study by Youngman et al.4 imply a risk ratio of 1.3 (95%CI 0.7–2.2) for a 1 g/l increase in fibrinogen, assuming a linear relationship between fibrinogen and log-risk of CHD (compared with the relative rate of 1.8, 95%CI 1.6–2.0, for 1 g/l difference in fibrinogen reported in the meta-analysis5). A test between these two estimates yields a p value of 0.24. Similarly for CRP, the relationships observed in the study by Zee and Ridker8 imply an odds ratio of 0.96 (95%CI 0.2–2.6) for a 1.4 mg/l increase in CRP (the difference in mean estimated usual values of CRP between top and bottom thirds of the population found by pooling prospective studies,9 in which a relative rate of 2.0 (95%CI 1.6–2.5) was seen. A test between these estimates yields a p value of 0.23.
|
Conclusions |
|---|
|
TopIntroductionWhat is Mendelian randomization?Mendelian randomization,...ConclusionsReferences |
|---|
Combining estimates of genotype–CHD associations with genotype-phenotype associations to predict phenotype–CHD associations produces values that are considerably lower than those reported from the observational studies, suggesting that reverse causation and confounding may generate the findings seen in the observational data: the confidence intervals are wide, however, as the difference in mean phenotype between genotypes is small compared to the overall spread of phenotype levels in the population. Sample sizes required to produce more precise genotype–disease estimates of association are large: in the cases examined here, sample sizes of around 30 000 and 9000 cases (and the same number of controls) for fibrinogen and CRP, respectively, would be required, to have 80% power to exclude phenotype–CHD associations of 1.5 for the difference in phenotype between top and bottom thirds of the population, if there is no true phenotype–CHD association. The inferences that can be drawn from such studies are, however, more secure than those of conventional observational studies, plagued as they are by problems of reverse causation and confounding. The principles of Mendelian randomization can be expanded to many domains,3 and illustrate the exciting possibilities that can arise from a combination of genetic and conventional epidemiology.
|
Footnotes |
|---|
Address correspondence to Professor G. Davey-Smith, Department of Social Medicine, University of Bristol, Canynge Hall, Whiteladies Road, Bristol BS8 2PR. e-mail: zetkin{at}bristol.ac.uk
|
References |
|---|
|
TopIntroductionWhat is Mendelian randomization?Mendelian randomization,...ConclusionsReferences |
|---|
1. Kamath S, Lip GYH. Fibrinogen: biochemistry, epidemiology and determinants. Q J Med 2003; 96:711–29.
2. Hirschfield GM, Pepys MB. C-reactive protein and cardiovascular disease: new insights from an old molecule. Q J Med 2003; 96:793–807.
3. Davey Smith G, Ebrahim S. ‘Mendelian randomization’: can genetic epidemiology contribute to understanding environmental determinants of disease? Int J Epidemiology 2003; 32:1–22.
4. Youngman LD, Keavney BD, Palmer A, et al. Plasma fibrinogen and fibrinogen genotypes in 4685 cases of myocardial infarction and in 6002 controls: test of causality by ‘Mendelian randomization’. Circulation 2000; 102(suppl. II):31–2.
5. Danesh J, Collins R, Appleby P, Peto R. Association of fibrinogen, C-reactive protein, albumin or leucocyte count with coronary heart disease. Meta-analysis of prospective studies. JAMA 1998; 279:1477–82.
6. Brunner E, Davey Smith G, Marmot M, Canner R, Beksinska M, O’Brien J. Childhood social circumstances and psychosocial and behavioural factors as determinants of plasma fibrinogen. Lancet 1996; 347:1008–13.[CrossRef][Web of Science][Medline]
7. Tybjaerg-Hansen A, Agerholm-Larsen B, Humphries SE, Abildgaard S, Schnohr P, Nordestgaard BG. A common mutation (G455 A) in the ß-Fibrinogen Promoter is an Independent Predictor of Plasma Fibrinogen, but not of Ischaemic Heart Disease. J Clin Invest 1997; 99:3034–9.[Web of Science][Medline]
8. Zee RYL, Ridker PM. Polymorphism in the human C-reactive protein (CRP) gene, plasma concentrations of CRP, and the risk of future arterial thrombosis. Atherosclerosis 2001; 162:217–19.
9. Danesh J, Whincup P, Walker M, Lennon L, Thomson A, Appleby P, Gallimore JR, Pepys MB. Low grade inflammation and coronary heart disease: prospective study and updated meta-analyses. Br Med J 2000; 321:199–204.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  M. Kivimaki, D. A. Lawlor, G. D. Smith, C. Eklund, M. Hurme, T. Lehtimaki, J. S. A. Viikari, and O. T. Raitakari Variants in the CRP Gene as a Measure of Lifelong Differences in Average C-Reactive Protein Levels: The Cardiovascular Risk in Young Finns Study, 1980 2001 Am. J. Epidemiol., October 1, 2007; 166(7): 760 - 764. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 K. Heikkila, S. Ebrahim, and D. A Lawlor A systematic review of the association between circulating concentrations of C reactive protein and cancer J Epidemiol Community Health, September 1, 2007; 61(9): 824 - 833. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 L. A. Lange, C. S. Carlson, L. A. Hindorff, E. M. Lange, J. Walston, J. P. Durda, M. Cushman, J. C. Bis, D. Zeng, D. Lin, et al. Association of Polymorphisms in the CRP Gene With Circulating C-Reactive Protein Levels and Cardiovascular Events JAMA, December 13, 2006; 296(22): 2703 - 2711. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M May, D A Lawlor, P Brindle, R Patel, and S Ebrahim Cardiovascular disease risk assessment in older women: can we improve on Framingham? British Women's Heart and Health prospective cohort study Heart, October 1, 2006; 92(10): 1396 - 1401. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. P Casas, T. Shah, J. Cooper, E. Hawe, A. D McMahon, D. Gaffney, C. J Packard, D. S O'Reilly, I. Juhan-Vague, J. S Yudkin, et al. Insight into the nature of the CRP-coronary event association using Mendelian randomization Int. J. Epidemiol., August 1, 2006; 35(4): 922 - 931. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
G. D. Smith Randomised by (your) god: robust inference from an observational study design J Epidemiol Community Health, May 1, 2006; 60(5): 382 - 388. [Full Text] [PDF]
|
|
|
|
 |
|
 S. J Lewis, S. Ebrahim, and G. Davey Smith Meta-analysis of MTHFR 677C->T polymorphism and coronary heart disease: does totality of evidence support causal role for homocysteine and preventive potential of folate? BMJ, November 5, 2005; 331(7524): 1053. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 G. D. Smith, R. Harbord, J. Milton, S. Ebrahim, and J. A.C. Sterne Does Elevated Plasma Fibrinogen Increase the Risk of Coronary Heart Disease?: Evidence from a Meta-Analysis of Genetic Association Studies Arterioscler Thromb Vasc Biol, October 1, 2005; 25(10): 2228 - 2233. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. Kivimaki, D. A. Lawlor, M. Juonala, G. Davey Smith, M. Elovainio, L. Keltikangas-Jarvinen, J. Vahtera, J. S.A. Viikari, and O. T. Raitakari Lifecourse Socioeconomic Position, C-Reactive Protein, and Carotid Intima-Media Thickness in Young Adults: The Cardiovascular Risk in Young Finns Study Arterioscler Thromb Vasc Biol, October 1, 2005; 25(10): 2197 - 2202. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
G. D. Smith, D. A. Lawlor, R. Harbord, N. Timpson, A. Rumley, G. D.O. Lowe, I. N.M. Day, and S. Ebrahim Association of C-Reactive Protein With Blood Pressure and Hypertension: Life Course Confounding and Mendelian Randomization Tests of Causality Arterioscler Thromb Vasc Biol, May 1, 2005; 25(5): 1051 - 1056. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 N. Krieger and G. Davey Smith "Bodies Count," and Body Counts: Social Epidemiology and Embodying Inequality Epidemiol. Rev., July 1, 2004; 26(1): 92 - 103. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||