Correspondence

Genetically Elevated C-Reactive Protein and Vascular Disease

N Engl J Med 2009; 360:933-935February 26, 2009

Article

To the Editor:

Zacho and colleagues (Oct. 30 issue)1 provide considerable support for other studies2 that suggest that since polymorphisms in the C-reactive protein (CRP) gene are not associated with ischemic heart disease, CRP itself is not causally associated with ischemic heart disease. We agree that “the meaning of theoretically predicted hazard ratios . . . may be somewhat difficult to understand.” Rather than comparing observed and predicted associations between the genotype and ischemic heart disease for each genotype, we suggest that the causal effect of CRP on ischemic heart disease should be estimated on the basis of the associations between the genotype and CRP and the genotype and ischemic heart disease.3

Figure 1Figure 1The Effect of C-Reactive Protein (CRP) on the Risk of Ischemic Heart Disease. shows an alternative presentation of the results reported by Zacho et al. The slope of the dashed line indicates the estimated causal odds ratio for ischemic heart disease. We cannot calculate this odds ratio precisely from the published data, but the use of a method for approximation4 results in an odds ratio of 0.84 (95% confidence interval [CI], 0.66 to 1.07) associated with a doubling in the CRP level. The upper confidence limit of this causal effect is considerably lower than the observed association and, despite the wider confidence intervals associated with estimates based on mendelian randomization, suggests that elevated CRP levels are unlikely to cause a substantial increase in the risk of ischemic heart disease.

Both the authors' and our methods involve approximations and extra assumptions when used with binary outcomes,5 and further work is required to establish the optimal method.

Roger M. Harbord, M.Sc.
Bristol University, Bristol BS8 2PS, United Kingdom
roger.harbord@bristol.ac.uk

Debbie A. Lawlor, Ph.D.
George Davey Smith, D.Sc.
Medical Research Council Centre for Causal Analyses in Translational Epidemiology, Bristol BS8 2BN, United Kingdom

5 References

1. 1

   Zacho J, Tybjaerg-Hansen A, Jensen JS, Grande P, Sillesen H, Nordestgaard BG. Genetically elevated C-reactive protein and ischemic vascular disease. N Engl J Med 2008;359:1897-1908
   Full Text | Web of Science | Medline

2. 2

   Casas JP, Shah T, Hingorani AD, Danesh J, Pepys MB. C-reactive protein and coronary heart disease: a critical review. J Intern Med 2008;264:295-314
   CrossRef | Web of Science | Medline

3. 3

   Lawlor DA, Harbord RM, Sterne JAC, Timpson NJ, Davey Smith G. Mendelian randomization: using genes as instrumental for making causal inferences in epidemiology. Stat Med 2008;27:1133-1163
   CrossRef | Web of Science | Medline

4. 4

   Greenland S, Longnecker MP. Methods for trend estimation from summarized dose-response data, with applications to meta-analysis. Am J Epidemiol 1992;135:1301-1309
   Web of Science | Medline

5. 5

   Didelez V, Sheehan N. Mendelian randomization as an instrumental variable approach to causal inference. Stat Methods Med Res 2007;16:309-330
   CrossRef | Web of Science | Medline

To the Editor:

Zacho et al. contrast an increased risk of ischemic vascular disease related to elevations in the CRP level with an absence of increased risk attributable to genetic polymorphisms that affect CRP expression. They conclude that CRP levels are a marker, rather than a causal factor. However, this conclusion is predicated on an unstated assumption. The authors assume that the increased risk of ischemic vascular disease will occur in proportion to the elevation in the CRP level; that is, any causal relationship between CRP and this disease will be linear.

In this study, the association between the disease and CRP levels was identified by comparing the prevalence of vascular disease among persons with a low level (<1.0 mg per liter) versus a high level (>3.0 mg per liter) of CRP, but the variation of CRP levels attributable to genetic polymorphisms occurred over a much smaller range (only a 64% increase). If CRP causes ischemic vascular disease only at high levels of CRP or protects against this disease only at low levels of CRP, then the polymorphism data would not have been expected to reveal this relationship. Binding isotherms are highly nonlinear, and nonlinear biologic responses should be the default assumption.1

Michael R. Bubb, M.D.
University of Florida, Gainesville, FL 32610
bubbmr@medicine.ufl.edu

1 References

1. 1

   Minton AP. How can biochemical reactions within cells differ from those in test tubes? J Cell Sci 2006;119:2863-2869
   CrossRef | Web of Science | Medline

To the Editor:

Zacho et al. used the mendelian randomization approach1 to examine the causal relationship between CRP levels and the onset of ischemic vascular disease. In this analysis, the apolipoprotein E genotype was used as a positive control. However, apolipoprotein E variants are well known to have highly pleiotropic effects. Indeed, as shown in their study, the observed risk of ischemic heart disease tended to exceed the risk theoretically predicted by the apolipoprotein E genotype for the highest-risk genotype. Thus, this “positive control” seems inappropriate to justify the method of this analysis. Also, CRP levels were clearly shown to be regulated by the CRP genotype. If CRP is not a cause but simply a marker for the disease, how should we take the CRP genotype into account when CRP levels are evaluated in a clinical setting?

Hiroyuki Morita, M.D., Ph.D.
Ryozo Nagai, M.D., Ph.D.
University of Tokyo, Tokyo 113-8655, Japan
hmrt-tky@umin.net

1 References

1. 1

   Smith GD, Ebrahim S. Mendelian randomization: prospects, potentials, and limitations. Int J Epidemiol 2004;33:30-42
   CrossRef | Web of Science | Medline

Author/Editor Response

Harbord et al. suggest an alternative way of presenting the results in Figure 5 of our article by estimating a causal odds ratio for ischemic heart disease of 0.84 (95% CI, 0.66 to 1.07) for a doubling in genetically elevated levels of CRP. Because this risk estimate differs markedly from the hazard ratio of 1.45 (95% CI, 1.36 to 1.54) observed in the Copenhagen City Heart Study for a doubling in common elevated levels of CRP, this calculation provides further support for the notion that elevated levels of CRP are not the cause of ischemic vascular disease.

As Bubb suggests, the interpretation of our data depends on the assumption that the relationship between levels of CRP and the risk of ischemic vascular disease is linear. Such a linear relationship has been shown previously1 and also appears to be present within the Copenhagen General Population Study, in which there were 34,860 participants (Figure 1Figure 1Risk of Ischemic Heart Disease as a Function of Plasma Levels of High-Sensitivity C-Reactive Protein in the General Population.) (P for trend, <0.001). Furthermore, using a likelihood-ratio test2 we found no evidence of nonlinearity for an increasing risk of ischemic heart disease with increasing levels of CRP up to 9 mg per liter (P=0.28).

Morita and Nagai question the use of the apolipoprotein E genotype as a positive control because it has pleiotropic effects. In the Copenhagen City Heart Study, the apolipoprotein E genotypes that are associated with elevated cholesterol levels are also associated with elevated levels of apolipoprotein B and nonfasting triglycerides and lipoprotein(a), and with decreased levels of high-density lipoprotein cholesterol and apolipoprotein A-I.3 The larger observed rather than theoretically predicted risk of ischemic heart disease as a function of the apolipoprotein E genotype therefore is probably explained by changes in these other lipid traits, which together with cholesterol levels are risk factors for ischemic vascular disease. We therefore agree that Figure 5 of our article shows a simplification of the relationship between the apolipoprotein E genotype and ischemic heart disease. However, we still think that the apolipoprotein E genotype is a valid positive control for our mendelian randomization design.

Finally, on the second issue raised by Morita and Nagai, CRP genotypes only explain about 2% of the variance in plasma levels of CRP, as shown in Figure 2 of our article. Therefore, in a clinical setting, genetics will only have a minor influence on plasma levels and, consequently, a simple plasma measurement will be sufficient to identify persons who are at increased risk for ischemic vascular disease.

Jeppe Zacho, M.D.
Copenhagen University Hospital Herlev, DK-2730 Herlev, Denmark

Anne Tybjærg-Hansen, M.D.
Copenhagen University Hospital Rigshospitalet, DK-2100 Copenhagen, Denmark

Børge G. Nordestgaard, M.D.
Copenhagen University Hospital Herlev, DK-2730 Herlev, Denmark
brno@heh.regionh.dk

3 References

1. 1

   Ridker PM, Cook N. Clinical usefulness of very high and very low levels of C-reactive protein across the full range of Framingham Risk Scores. Circulation 2004;109:1955-1959
   CrossRef | Web of Science | Medline

2. 2

   Freiberg JJ, Tybjaerg-Hansen A, Jensen JS, Nordestgaard BG. Nonfasting triglycerides and risk of ischemic stroke in the general population. JAMA 2008;300:2142-2152
   CrossRef | Web of Science | Medline

3. 3

   Frikke-Schmidt R, Nordestgaard BG, Agerholm-Larsen B, Schnohr P, Tybjaerg-Hansen A. Context-dependent and invariant associations between lipids, lipoproteins, and apolipoproteins and apolipoprotein E genotype. J Lipid Res 2000;41:1812-1822
   Web of Science | Medline

Citing Articles (1)

Citing Articles

1. 1

   Børge G Nordestgaard. (2009) Does elevated C-reactive protein cause human atherothrombosis? Novel insights from genetics, intervention trials, and elsewhere. Current Opinion in Lipidology 20:5, 393-401
   CrossRef