C-reactive protein and all-cause mortality the Copenhagen City Heart Study

Transcription

1 European Heart Journal (2010) 31, doi: /eurheartj/ehq103 CLINICAL RESEARCH Prevention C-reactive protein and all-cause mortality the Copenhagen City Heart Study Jeppe Zacho 1,4, Anne Tybjærg-Hansen 2,3,4, and Børge G. Nordestgaard 1,3,4 * 1 Department of Clinical Biochemistry, Herlev Hospital, Copenhagen University Hospital, Herlev Ringvej 75, DK-2730, Herlev, Denmark; 2 Department of Clinical Biochemistry, Rigshospitalet, Copenhagen University Hospital, Copenhagen, Denmark; 3 The Copenhagen City Heart Study, Bispebjerg Hospital, Copenhagen University Hospital, Copenhagen, Denmark; and 4 Faculty of Health Sciences, University of Copenhagen, Copenhagen, Denmark Received 29 August 2009; revised 6 January 2010; accepted 2 February 2010; online publish-ahead-of-print 27 April 2010 Aims We tested whether elevated levels of C-reactive protein is robustly and causally associated with all-cause mortality.... Methods We studied white persons from the general population. During 16 years 3124 persons died. We measured and results baseline high-sensitivity C-reactive protein and fibrinogen levels and genotyped for four C-reactive protein polymorphisms and two apolipoprotein E polymorphisms. Levels of C-reactive protein.3 mg/l vs.,1 mg/l associated with a multi-factorially adjusted two-fold increased risk of all-cause mortality. Stratifying C-reactive protein into tertiles, quintiles, or octiles resulted in step-by-step increased risk of all-cause mortality, even after fibrinogen adjustment. Finally, genetically elevated C-reactive protein levels associated with a causal hazard ratio of 0.94 (95% CI: ) for all-cause mortality per doubling of C-reactive protein levels on instrumental variable analysis, for which the corresponding hazard ratio on Cox regression for a doubling in measured plasma C-reactive protein levels was 1.25 ( ). As a positive control, a doubling in genetically elevated cholesterol levels via apolipoprotein E associated with a hazard ratio of 6.3 (1.8 22) for all-cause mortality.... Conclusion A single C-reactive protein measurement robustly associates with increased risk of all-cause mortality; however, this does not appear to be a causal association. Therefore, elevated C-reactive protein levels more likely are a marker of hidden, potentially fatal inflammatory disease Keywords C-reactive protein Polymorphisms Mortality Introduction Elevated levels of C-reactive protein associate with increased risk of all-cause mortality; 1 3 however, whether this is a causal association or whether elevated C-reactive protein levels merely are a marker of hidden, potentially fatal inflammatory disease is presently unclear. This question is of clinical importance as several drugs that specifically lower C-reactive protein levels are being developed with the ultimate aim of preventing cardiovascular disease and early death. 4 To prove causality between elevated C-reactive protein levels and early death would be difficult. A randomized placebocontrolled trial of a C-reactive protein reducing drug with total mortality as the primary endpoint could address this question; however, such a trial would be extremely costly and could potentially cause several side effects, as elevated C-reactive protein levels generally are beneficial and help fight infection. 5 Although the JUPITER trial through aggressive statin treatment reduced C-reactive protein levels by 37%, cardiovascular disease and death by 47% and total mortality by 20%, the statin treatment also reduced LDL cholesterol levels by 50%, 6 which likely explains the beneficial effects. 7 Evidence from epidemiological and genetic epidemiological studies may also suggest causality. First, a robust and step-by-step increased risk of all-cause mortality with step-by-step increased C-reactive protein levels would favour causality, particularly after exclusion of deaths within the first 2 years after C-reactive protein measurement to reduce the risk of reverse causation. Second, if adjustment for an independent marker of inflammation like fibrinogen fails to attenuate the C-reactive protein all-cause mortality association, claims of causality would be strengthened. Finally, if lifelong genetically elevated C-reactive protein levels * Corresponding author. Tel: , Fax: , brno@heh.regionh.dk Published on behalf of the European Society of Cardiology. All rights reserved. & The Author For permissions please journals.permissions@oxfordjournals.org.

2 C-reactive protein and mortality 1625 also associate with increased risk of all-cause mortality, a causality claim would be yet further strengthened. We tested whether the association between elevated C-reactive protein levels and increased risk of all-cause mortality is robust and potentially a causal association. To do so, we examined if step-by-step increases in C-reactive protein associate with step-by-step increases in risk of total mortality, whether exclusion of deaths shortly after C-reactive protein measurement attenuates this association, and whether adjustment for fibrinogen attenuates this association. Finally, using a Mendelian randomization design, 8,9 we tested whether lifelong genetically elevated C-reactive protein levels associate with increased risk of all-cause mortality. As a positive control we sought to evaluate apolipoprotein E genotypes with cholesterol levels and risk of all-cause mortality in our study. For these purposes, we used a cohort of white persons from the general population followed for up to 16 years, during which time 3124 died. Methods The study was approved by Herlev Hospital and a Danish Ethics Committee and was conducted according to the Declaration of Helsinki. Written informed consent was obtained from participants. Study cohort and endpoints The Copenhagen City Heart Study 10,11 is a prospective study of individuals randomly selected from the population of Copenhagen, followed from blood sampling in through Individuals were invited based on their Central Person Registration number, the participation rate was 55% and follow-up was 100% complete, that is, we did not loose track of even a single individual. Data on all-cause mortality were from the national Danish Civil Registration System, whereas information on cause-specific mortality was from the national Danish Causes of Death Registry. For cardiovascular death, we used ICD-8 codes and ICD-10 codes I00 I99, and for cancer deaths ICD-8 codes and ICD-10 codes C00 C97. Genotyping and biochemical analyses The ABI PRISM w 7900HT Sequence Detection System (Applied Biosystems, Inc., Foster City, CA, USA) was used to genotype for four single nucleotide polymorphisms in the C-reactive protein gene (rs , rs , rs1205, and rs ). 9 From Ensemble the sequence OTTHUMG was used and relative to the ATG-start codon the promoter variant rs is termed 2390 C.T.A; 3 rs is termed *223 C.T; 3 rs1205 is termed *1081 G.A; and 3 rs is termed *3678 T.G. Rs modulates transcription factor binding 12,13 and is the single most important C-reactive protein polymorphism influencing C-reactive protein levels. 13,14 Rs , rs1205, and rs can be used as tag single nucleotide polymorphisms together describing most of the variation in plasma C-reactive protein levels. 15 The two polymorphisms in the apolipoprotein E gene were genotyped by PCR followed by restriction enzyme digestion. 16 These two polymorphisms in the APOE gene, rs (Cys112Arg, 14) and rs7412 (Arg158Cys; 12), are known to associate with plasma cholesterol levels and IHD; 16,17 individuals with the 122 genotype were excluded from data analysis, because this genotype is well known to lead to remnant hyperlipidaemia in some individuals. High-sensitivity C-reactive protein was measured using turbidimetry (Dako, Glostrup, Denmark). C-reactive protein levels were classified as low (,1.0 mg/l), average ( mg/l), or high (.3.0 mg/l). Total cholesterol, low-density lipoprotein cholesterol, high-density lipoprotein cholesterol, and triglycerides were measured using standard hospital assays (Boehringer Mannheim, Mannheim, Germany or Konelab, Helsinki, Finland). Low-density lipoprotein cholesterol was calculated using the Friedewald equation if triglycerides were,4 mmol/l (352 mg/dl), but measured directly at higher triglyceride levels. Other covariates Body mass index was expressed as weight in kilograms divided by height in meters squared. Hypertension was diastolic blood pressure 90 mmhg and/or systolic blood pressure 140 mmhg and/or antihypertensive medication. Diabetes mellitus was non-fasting blood glucose levels 11.0 mmol/l and/or insulin treatment or any other anti-diabetic treatment. Smoking was classified into ever and never smokers. Lipid lowering treatment was statin treatment. Physical inactivity was coded as inactive at both work time and leisure time or increasingly active at work time or leisure time. Statistical analysis From the four C-reactive protein polymorphisms, we generated all possible genotype combinations and ranked the nine most common combinations according to increasing plasma C-reactive protein levels. Apolipoprotein E genotypes were ranked in a similar fashion according to increasing plasma cholesterol levels. We used Stata 10.1 for all analyses. For trend tests by C-reactive protein levels, genotype or genotype combination groups were coded 1, 2, 3, etc., ranked according to increasing C-reactive protein levels. As we only asked two pre-specified questions, that is, whether the association between elevated C-reactive protein levels and increased risk of allcause mortality is robust and potentially a causal association, we Table 1 Baseline characteristics of participants in the Copenhagen City Heart Study by death status at the end of follow-up No event Death... No. of individuals Women (%) Age (years) 59 (45 69) 72 (65 77) Total cholesterol (mmol/l) 5.6 ( ) 6.1 ( ) LDL cholesterol (mmol/l) 3.5 ( ) 3.8 ( ) HDL cholesterol (mmol/l) 1.4 ( ) 1.4 ( ) Triglycerides (mmol/l) 1.4 ( ) 1.6 ( ) Body mass index (kg/m 2 ) 25 (23 28) 26 (23 29) Hypertension (%) Diabetes mellitus (%) 5 9 Smokers (%) Alcohol intake 48 g/week (%) Lipid lowering therapy (%) 4 3 Postmenopausal (%) a Hormone replacement therapy (%) a Physical inactivity (%) 2 2 Values are median (inter-quartile range) or percent. LDL, low-density lipoprotein; HDL, high-density lipoprotein. To convert cholesterol in mmol/l to mg/dl multiply by 38.6 and to convert triglycerides in mmol/l to mg/dl multiply by 88. a Women only.

3 1626 J. Zacho et al. did not correct for multiple comparisons in the main analyses. The only exception to this is the comparisons of characteristics of participants by C-reactive protein genotype combination, where we corrected P-values for 16 multiple comparisons using the Bonferroni method. We first analysed the relationship between plasma C-reactive protein and risk of all-cause mortality with the use of Cox regression with left truncation (or delayed entry) (from the examination). 18 To demonstrate robustness in the C- reactive protein all-cause mortality association, individuals were divided by plasma C-reactive protein levels in those,1, 1 3, or.3 mg/l, as well as in tertiles, quintiles, and octiles. Cox regression models with age as the time scale and adjusted for sex and use of statins were used to estimate hazard ratios; when age is used as time scale this implies that age is automatically adjusted for, but for simplicity we refer to this as adjusted for age. Additional regression models that included adjustment for age, sex, and C- reactive protein genotype, multi-factorial adjustment that included a full set of available risk factors, and models that included adjustment for age, sex, and fibrinogen levels were also developed. Data from the serial examinations of the Copenhagen City Heart Study were used as time-dependent covariates for multi-factorial and fibrinogen adjustments. Information on covariates used for adjustments was.99% complete; the few missing covariates were treated as such in the statistical analyses and participants were not excluded due to a few missing covariates. Also, we estimated the risk of all-cause and cause-specific mortality in a Cox regression model with log 2 (C-reactive protein) as the independent predictor, giving the risk increase for every doubling in C-reactive protein levels. For the association between C-reactive protein levels as a continuous predictor variable and risk of all-cause and cause-specific mortality, we assumed that there was a linear relationship between log 2 (C-reactive protein) and the log-hazards of mortality. We tested this assumption by computing a likelihood ratio test comparing a model that contained a second-order and firstorder term of log 2 (C-reactive protein) (a quadratic model) to one containing only the first-order term (linear model). We detected no signs of non-linearity (P ¼ 0.09). To further evaluate the question of causality, we analysed the relationship between C-reactive protein polymorphisms and risk of allcause mortality with the use of Cox regression as described above. On the basis of the associations between C-reactive protein polymorphisms and plasma C-reactive protein levels, and between C-reactive protein polymorphisms and risk of all-cause mortality, we used instrumental variable analysis to estimate the association between genetically elevated levels of C-reactive protein and risk of all-cause mortality. For this purpose, we used generalized least squares regression. 19 Similar calculations were performed for apolipoprotein E genotype Table 2 Study Characteristics of participants by C-reactive protein genotype combinations in The Copenhagen City Heart C-reactive protein genotype combinations #1 #2 #3 #4 #5 #6 #7 #8 #9 P-values... rs1205, *1081 G.A AA GA GG GA GG GA GG GG GG rs , *223 C.T CC CC CC CT CT CC TT CC CT rs , 2390 C.T.A CC CC CC CT CT CA TT CA AT rs , *3678 T.G TT TT TT TT TT GT TT GT GT... No. of individuals Women (%) Age (years) Total cholesterol (mmol/l) LDL cholesterol (mmol/l) HDL cholesterol (mmol/l) Triglycerides (mmol/l) Body mass index (kg/m 2 ) Hypertension (%) Diabetes mellitus (%) Smokers (%) Alcohol intake 48 g/week (%) Lipid lowering therapy (%) Postmenopausal (%) a Hormone replacement therapy (%) a Physical inactivity (%) C-reactive protein (mg/l) b Values represent the last value available for each participant. Values are median or percent. LDL, low-density lipoprotein. HDL, high-density lipoprotein. To convert cholesterol in mmol/l to mg/dl multiply by 38.6 and to convert triglycerides in mmol/l to mg/dl multiply by 88. P-values are for x 2 test or Kruskal Wallis test. Covariates (except C-reactive protein) did not differ between the nine different C-reactive protein genotype combinations after correction for multiple comparisons [Bonferroni corrected P-values,0.003 (0.05/16)]. a Women only. b The values are from the Copenhagen General Population Study, Zacho et al. 9

4 C-reactive protein and mortality 1627 Figure 1 Risk of incident all-cause mortality as a function of plasma C-reactive protein (P-C-reactive protein),1, 1 3, and.3 mg/l in the general population. High-sensitivity C-reactive protein was measured in participants in the examination of The Copenhagen City Heart Study and subsequently followed up to 16 years. Multi-factorial adjustment was for all covariates listed in Table 1. combination, 132, 142, 133, 143, 144, and the risk of all-cause mortality, in order to use the apolipoprotein E genotypes as a positive control 9 for the association between genotypes and risk of all-cause mortality. For Cox regression analyses, we tested the assumption of proportional hazards graphically by plotting of log (cumulative hazard) as a function of age. Suspicion of non-parallel lines was tested using Schoenfeld residuals. We detected no major violations of the proportional hazards assumption. Results Selected clinical characteristics of participants in the study cohort are shown in Table 1. As would be anticipated, those individuals who died were older and more often male. They also tended to have higher levels of total and LDL cholesterol and triglycerides and higher rates of hypertension, diabetes, and smoking. And naturally among the women more were postmenopausal due to the higher age. The same clinical characteristics are shown in Table 2 for the C-reactive protein genotype combinations of the four polymorphisms and none differed. Neither did they differ for any of the four polymorphisms individually (data not shown). Plasma C-reactive protein and risk of all-cause mortality C-reactive protein levels.3 mg/l, compared with levels,1 mg/l, were associated with a 2.1-fold [95% confidence interval (CI): ] increased risk of all-cause mortality after adjustment for age, sex, and use of statins (Figure 1). After adjustment for age, sex, statins, and C-reactive protein genotype, the equivalent hazard ratio was 2.1 ( ). After multi-factorial adjustment, the equivalent hazard ratio was 2.0 ( ). The P-values for trend were all, Dividing participants into tertiles, those in the highest vs. the lowest tertile associated with a 1.6-fold ( ) increased risk after multi-factorial adjustment (trend P, ) (Figure 2). Using quintiles, the ones in the highest vs. the lowest quintile associated with a 1.9-fold ( ) increased risk after multifactorial adjustment (P, ). Finally, for octiles those in the highest vs. the lowest octile associated with a 2.1-fold ( ) increased risk after multi-factorial adjustment (P, ). Excluding those who died within the first 2 years after plasma C-reactive protein measurement resulted in similar but slightly attenuated risk increases (all trend P, ) (Figure 2).

5 1628 J. Zacho et al. Figure 2 Risk of incident all-cause mortality as a function of plasma C-reactive protein (P-C-reactive protein) in tertiles, quintiles, or octiles in the general population. High-sensitivity C-reactive protein was measured in participants in the examination of The Copenhagen City Heart Study and subsequently followed up to 16 years. Deaths within the first 2 years after baseline measurement of P-C-reactive protein were not included in the analysis on the right-side panel. Multi-factorial adjustment was for all covariates listed in Table 1. Further adjustment for fibrinogen levels attenuated the risk estimates for all-cause mortality; however, the associations remained highly significant (all trend P, ) (Figure 3). For all-cause mortality, a doubling in plasma C-reactive protein levels was associated with a 1.25-fold ( ) increased risk after age and sex adjustment (Table 3); results were similar after multi-factorial or fibrinogen adjustment. Corresponding age and sex adjusted risks per doubling in plasma C-reactive protein levels were 1.43-fold ( ) for cardiovascular mortality, 1.38-fold ( ) for cancer mortality, and 1.17-fold ( ) for mortality from all other causes. Genetically elevated C-reactive protein and risk of all-cause mortality For the four C-reactive protein polymorphisms, we estimated the nine C-reactive protein genotype combinations with the highest effect on plasma C-reactive protein levels (Table 2). The hazard ratios for the associations between the nine C-reactive protein genotype combinations and risk of all-cause mortality are given in Figure 4. On the basis of these hazard ratios and using instrumental variable analyses, we estimated a regression to illustrate the connection between the genotype combinations, plasma C- reactive protein level change, and risk of all-cause mortality. As is evident visually and also shown with the trend test (P ¼ 0.77), there was no increased risk of all-cause mortality for genetically increasing plasma C-reactive protein levels, a doubling in genetically elevated C-reactive protein levels associated with a hazard ratio for all-cause mortality of 0.94 ( ) (Figure 4). In contrast, a doubling in plasma C-reactive protein levels associated with a hazard ratio for all-cause mortality of 1.25 ( ) (Figure 4 and Table 3). As a positive control, we examined the influence of apolipoprotein E genotype on the risk of all-cause mortality through elevated plasma cholesterol levels in our study sample. Calculated like in Figure 4, we used instrumental variable analyses to estimate a regression illustrating the connection between the apolipoprotein E genotype combinations, plasma cholesterol level change, and risk of all-cause mortality. We found a 6.3-fold (95% CI ) increased risk of all-cause mortality for a doubling in genetically elevated cholesterol levels. Discussion We found that a single plasma C-reactive protein measurement robustly associated with increased risk of all-cause mortality; however, this does not appear to be a causal association for C- reactive protein per se but more likely reflects association of

6 C-reactive protein and mortality 1629 Figure 3 Risk of incident all-cause mortality as a function of P-C-reactive protein in tertiles, quintiles, or octiles, age and sex adjusted (left panel) and further adjusted for fibrinogen (right panel). High-sensitivity C-reactive protein and fibrinogen were measured in participants in the examination of The Copenhagen City Heart Study and subsequently followed up to 16 years. Table 3 Risk of any death or death by cause for a doubling in plasma C-reactive protein levels in the Copenhagen City Heart Study Deaths/participants Hazard ratio (95% confidence interval)... Age and sex adjusted Multi-factorial adjusted Age, sex, and fibrinogen adjusted... Any death 3079/ ( ) 1.22 ( ) 1.17 ( ) Cardiovascular death 1014/ ( ) 1.38 ( ) 1.27 ( ) Cancer death 654/ ( ) 1.37 ( ) 1.25 ( ) Other causes death 1411/ ( ) 1.16 ( ) 1.12 ( ) High-sensitivity C-reactive protein and fibrinogen were measured in participants in the examination of the Copenhagen City Heart Study and subsequently followed up to 16 years. hidden, potentially fatal inflammatory disease with increased allcause mortality. Robustness of the association was documented by step-by-step increases in risk of all-cause mortality for step-by-step increases in C-reactive protein levels, whether adjusting only for age, sex, and use of statins or also adjusting for genotype or adjusting multi-factorially, whether stratifying C-reactive protein levels on,1 mg/l vs..3 mg/l or in tertiles, quintiles, or octiles, whether leaving out those dying within the first 2 years after blood sampling, and whether adjusting for the other acute phase reactant fibrinogen; however, fibrinogen adjustment did attenuate the C-reactive protein all-cause mortality association. The claim for causality for the C-reactive protein molecule itself was further weakened because genetically elevated C-reactive protein levels did not associate with increased risk of all-cause mortality. In support of our findings, elevated plasma C-reactive protein levels are consistently associated with increased risk of any death. 1 3,20 28 In a study using carotid ultrasound, this was only found for the participants with atherosclerosis, 2 suggesting that C-reactive protein is mainly linked to atherosclerosis and cardiovascular mortality. However, increased C-reactive protein concentrations were also able to predict increased mortality in patients with chronic obstructive pulmonary disease 29 and cancer. 30 In accordance with this, in the present study, we found that increased

7 1630 J. Zacho et al. Figure 4 Hazard ratios for the nine C-reactive protein genotype combinations with 95% confidence intervals with respect to the risk of all-cause mortality as a function of their change in plasma C-reactive protein levels. A regression based on generalized least squares for trend estimation is plotted and P-value for trend given. Also, a regression for a doubling in plasma C-reactive protein levels on risk of all-cause mortality is shown for a comparison with a doubling in genetically elevated C-reactive protein levels. C-reactive protein levels associated with increased cardiovascular death, cancer death, as well as death from any other cause. Therefore, and because elevated C-reactive protein levels per se do not seem to lead to early death, elevated C-reactive protein levels more likely is a marker of hidden, potentially fatal inflammatory disease. For studies of C-reactive protein polymorphisms, one study 31 showed an association with increased risk of any death, whereas another did not. 32 It has, therefore, been unclear whether C- reactive protein was merely a marker of underlying inflammation or in itself a causal factor for early death. In a large study, 47 polymorphisms in 23 other inflammation-related genes were not associated with overall or cause-specific death. 31 However, it may be that associations with polymorphisms were not observed, because that study only looked individually at polymorphisms, and perhaps multiple functional polymorphisms in a single gene or multiple genes are needed in order to modify inflammation levels to the point of increasing mortality. 31 Nevertheless, these negative studies fully concur with our findings that genetically elevated C- reactive protein levels as a function of C-reactive protein genotype combination of four C-reactive protein polymorphisms did not associate with increased risk of all-cause mortality in contrast to common elevated plasma C-reactive protein levels [hazard ratio for a doubling of 0.94 (95% CI ) vs ( )]. The validity of the negative finding for genetically elevated C- reactive protein levels was further strengthened by the strong positive association between genetically elevated cholesterol levels via apolipoprotein E genotype and increased all-cause mortality [hazard ratio for a doubling in cholesterol of 6.3 (1.8 22)]. Overall, these associations between plasma C-reactive protein levels and all-cause mortality may arise because C-reactive protein levels reflect underlying inflammatory disease or, perhaps, as a result of confounding due to association of C-reactive protein with causal factors. 27 Alternatively, it is possible that biomarkers of inflammation reflect a final common biochemical pathway of poor health status, (possibly triggered by cytokines) resulting in increased levels of C-reactive protein, which predispose to cardiovascular and non-cardiovascular mortality in old age. 33 Also, we need to consider the evidence that elevated C- reactive protein levels may indeed be related causally with all-cause mortality and specifically to cardiovascular disease and mortality. In this respect, it is important to consider carefully the results of the recent JUPITER trial: although aggressive statin treatment reduced C-reactive protein levels by 37%, cardiovascular disease and death by 47%, and total mortality by 20%, then the statin treatment also reduced LDL cholesterol levels by 50%, 6 Yet, rosuvastatin lowered LDL cholesterol and C-reactive protein largely independently, reduction in both covariates associated independently with reduced risk of cardiovascular disease and death, and the largest risk reduction was achieved in those with large reductions in both LDL cholesterol and C-reactive protein. 34 If this is due to a direct anti-inflammatory effect of statins beside the LDL lowering effect, this at first hand could seem to contradict the results from the genetic C-reactive protein studies, if these latter studies are thought to imply that less inflammation is not associated with improved outcome. However, in our opinion the interpretation of the data that genetically elevated C-reactive protein levels do not lead to increased risk of cardiovascular disease or allcause mortality rather should be that elevated C-reactive protein per se, i.e. the C-reactive protein molecule itself, is not causing this increased risks. Importantly, if elevated C-reactive protein per se does not lead to increased risks, then a more likely interpretation of previous 7 and present data is that inflammation per se may contribute towards increased risk of cardiovascular disease and all-cause mortality. Limitations to our study should be considered in evaluating our results. As we studied whites only our results may not necessarily apply to other ethnic groups. As we only asked two pre-specified questions, i.e whether the association between elevated C-reactive protein levels and increased risk of all-cause mortality is robust and potentially a causal association, we did not correct for multiple comparisons in the main analyses. For the genetic association, a potential limitation of our study is statistical power. 35,36 Also, findings in the total population may not automatically relate to the individual cases. Thus, in ischaemia reperfusion in a patient with myocardial infarction, a causal role of C-reactive protein may be more likely than for cardiovascular disease in general or for death from any cause. Finally, one should not overestimate the potential of Mendelian randomization studies in refuting a causal association. While lack of such an association certainly does not support a causal link, the final answer to the causality question will only be obtained after evaluating the totality of evidence, including human epidemiology, mechanistic studies, intervention trials in animal models, human genetic studies, and most importantly human intervention trials aimed at reducing C-reactive protein selectively. 7 In conclusion, a single plasma C-reactive protein measurement robustly associated with increased risk of all-cause mortality, whereas genetically elevated C-reactive protein levels did not.

8 C-reactive protein and mortality 1631 Therefore, elevated C-reactive protein levels do not appear to cause early death but more likely are a marker of hidden inflammatory disease possibly leading to early death. Author contribution J.Z. had substantial contributions to conception, design, and acquisition of data, analysis and interpretation of data. He also drafted the first version of the article and had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. A.T.-H. had substantial contributions to acquisition of data and revised the article critically for important intellectual content. B.G.N. had substantial contributions to conception, design, acquisition of data, interpretation of data, revised the article critically for important intellectual content, had full access to all the data in the study, takes responsibility for the integrity of the data, the accuracy of the data analysis, and he had final approval of the version to be published. Acknowledgements We are indebted to the staff and participants of The Copenhagen City Heart Study for their important contributions. Funding This work was supported by The Danish Heart Foundation, a nonprofit organization with no rights to approve or disapprove of the manuscript. Conflict of interest: none declared. References 1. Marsik C, Kazemi-Shirazi L, Schickbauer T, Winkler S, Joukhadar C, Wagner OF, Endler G. C-reactive protein and all-cause mortality in a large hospital-based cohort. Clin Chem 2008;54: Cao JJ, Arnold AM, Manolio TA, Polak JF, Psaty BM, Hirsch CH, Kuller LH, Cushman M. Association of carotid artery intima-media thickness, plaques, and C-reactive protein with future cardiovascular disease and all-cause mortality: the Cardiovascular Health Study. Circulation 2007;116: Currie CJ, Poole CD, Conway P. Evaluation of the association between the first observation and the longitudinal change in C-reactive protein, and all-cause mortality. Heart 2008;94: Lowe GD, Pepys MB. C-reactive protein and cardiovascular disease: weighing the evidence. Curr Atheroscler Rep 2006;8: Hirschfield GM, Pepys MB. C-reactive protein and cardiovascular disease: new insights from an old molecule. QJM 2003;96: Ridker PM, Danielson E, Fonseca FA, Genest J, Gotto AM Jr, Kastelein JJ, Koenig W, Libby P, Lorenzatti AJ, MacFadyen JG, Nordestgaard BG, Shepherd J, Willerson JT, Glynn RJ. Rosuvastatin to prevent vascular events in men and women with elevated C-reactive protein. N Engl J Med 2008;359: Nordestgaard BG. Does elevated C-reactive protein cause human atherothrombosis? Novel insights from genetics, intervention trials, and elsewhere. Curr Opin Lipidol 2009;20: Smith GD, Ebrahim S. Mendelian randomization: prospects, potentials, and limitations. Int J Epidemiol 2004;33: Zacho J, Tybjærg-Hansen A, Jensen JS, Grande P, Sillesen H, Nordestgaard BG. Genetically elevated C-reactive protein and ischemic heart and cerebrovascular disease. N Engl J Med 2008;359: Nordestgaard BG, Benn M, Schnohr P, Tybjaerg-Hansen A. Nonfasting triglycerides and risk of myocardial infarction, ischemic heart disease, and death in men and women. JAMA 2007;298: Schnohr P, Jensen JS, Scharling H, Nordestgaard BG. Coronary heart disease risk factors ranked by importance for the individual and community. A 21 year follow-up of men and women from The Copenhagen City Heart Study. Eur Heart J 2002;23: Szalai AJ, Wu J, Lange EM, McCrory MA, Langefeld CD, Williams A, Zakharkin SO, George V, Allison DB, Cooper GS, Xie F, Fan Z, Edberg JC, Kimberly RP. Single-nucleotide polymorphisms in the C-reactive protein (CRP) gene promoter that affect transcription factor binding, alter transcriptional activity, and associate with differences in baseline serum CRP level. J Mol Med 2005;83: Carlson CS, Aldred SF, Lee PK, Tracy RP, Schwartz SM, Rieder M, Liu K, Williams OD, Iribarren C, Lewis EC, Fornage M, Boerwinkle E, Gross M, Jaquish C, Nickerson DA, Myers RM, Siscovick DS, Reiner AP. Polymorphisms within the C-reactive protein (CRP) promoter region are associated with plasma CRP levels. Am J Hum Genet 2005;77: Miller DT, Zee RY, Suk DJ, Kozlowski P, Chasman DI, Lazarus R, Cook NR, Ridker PM, Kwiatkowski DJ. Association of common CRP gene variants with CRP levels and cardiovascular events. Ann Hum Genet 2005;69: Wang Q, Hunt SC, Xu Q, Chen YE, Province MA, Eckfeldt JH, Pankow JS, Song Q. Association study of CRP gene polymorphisms with serum CRP level and cardiovascular risk in the NHLBI Family Heart Study. Am J Physiol Heart Circ Physiol 2006;291:H2752 H Frikke-Schmidt R, Tybjaerg-Hansen A, Steffensen R, Jensen G, Nordestgaard BG. Apolipoprotein E genotype: epsilon32 women are protected while epsilon43 and epsilon44 men are susceptible to ischemic heart disease: the Copenhagen City Heart Study. J Am Coll Cardiol 2000;35: Bennet AM, Di Angelantonio E, Ye Z, Wensley F, Dahlin A, Ahlbom A, Keavney B, Collins R, Wiman B, de Faire U, Danesh J. Association of apolipoprotein E genotypes with lipid levels and coronary risk. JAMA 2007;298: Klein J, Moeschberger ML. Refinements of the semiparametric proportinal hazards model. Survival Analysis Techniques for Censored and Truncated Data. 2nd ed. New York, USA: Springer -Verlag, Inc; p Greenland S, Longnecker MP. Methods for trend estimation from summarized dose response data, with applications to meta-analysis. Am J Epidemiol 1992; 135: Poole CD, Conway P, Currie CJ. An evaluation of the association between C- reactive protein, the change in C-reactive protein over one year, and all-cause mortality in chronic immune-mediated inflammatory disease managed in UK general practice. Rheumatology (Oxford) 2009;48: Koenig W, Khuseyinova N, Baumert J, Meisinger C. Prospective study of highsensitivity C-reactive protein as a determinant of mortality: results from the MONICA/KORA Augsburg Cohort Study, Clin Chem 2008;54: Harris TB, Ferrucci L, Tracy RP, Corti MC, Wacholder S, Ettinger WH Jr, Heimovitz H, Cohen HJ, Wallace R. Associations of elevated interleukin-6 and C-reactive protein levels with mortality in the elderly. Am J Med 1999;106: Laaksonen DE, Niskanen L, Nyyssonen K, Punnonen K, Tuomainen TP, Salonen JT. C-reactive protein in the prediction of cardiovascular and overall mortality in middle-aged men: a population-based cohort study. Eur Heart J 2005;26: Makita S, Nakamura M, Satoh K, Tanaka F, Onoda T, Kawamura K, Ohsawa M, Tanno K, Itai K, Sakata K, Okayama A, Terayama Y, Yoshida Y, Ogawa A. Serum C-reactive protein levels can be used to predict future ischemic stroke and mortality in Japanese men from the general population. Atherosclerosis 2009; 204: Tuomisto K, Jousilahti P, Sundvall J, Pajunen P, Salomaa V. C-reactive protein, interleukin-6 and tumor necrosis factor alpha as predictors of incident coronary and cardiovascular events and total mortality. A population-based, prospective study. Thromb Haemost 2006;95: Jenny NS, Yanez ND, Psaty BM, Kuller LH, Hirsch CH, Tracy RP. Inflammation biomarkers and near-term death in older men. Am J Epidemiol 2007;165: Clarke R, Emberson JR, Breeze E, Casas JP, Parish S, Hingorani AD, Fletcher A, Collins R, Smeeth L. Biomarkers of inflammation predict both vascular and nonvascular mortality in older men. Eur Heart J 2008;29: Iso H, Cui R, Date C, Kikuchi S, Tamakoshi A. C-reactive protein levels and risk of mortality from cardiovascular disease in Japanese: The JACC Study. Atherosclerosis 2009;207: Dahl M, Vestbo J, Lange P, Bojesen SE, Tybjaerg-Hansen A, Nordestgaard BG. C-reactive protein as a predictor of prognosis in chronic obstructive pulmonary disease. Am J Respir Crit Care Med 2007;175: Allin KH, Bojesen SE, Nordestgaard BG. Baseline C-reactive protein is associated with incident cancer and survival in patients with cancer. J Clin Oncol 2009;27: Gallicchio L, Chang H, Christo DK, Thuita L, Huang HY, Strickland P, Ruczinski I, Hoffman SC, Helzlsouer KJ. Single nucleotide polymorphisms in inflammationrelated genes and mortality in a community-based cohort in Washington County, Maryland. Am J Epidemiol 2008;167:

9 1632 J. Zacho et al. 32. Hindorff LA, Rice KM, Lange LA, Diehr P, Halder I, Walston J, Kwok P, Ziv E, Nievergelt C, Cummings SR, Newman AB, Tracy RP, Psaty BM, Reiner AP. Common variants in the CRP gene in relation to longevity and cause-specific mortality in older adults: the Cardiovascular Health Study. Atherosclerosis 2008;197: Hazzard WR. Depressed albumin and high-density lipoprotein cholesterol: signposts along the final common pathway of frailty. J Am Geriatr Soc 2001;49: Ridker PM, Danielson E, Fonseca FA, Genest J, Gotto AM Jr, Kastelein JJ, Koenig W, Libby P, Lorenzatti AJ, MacFadyen JG, Nordestgaard BG, Shepherd J, Willerson JT, Glynn RJ. Reduction in C-reactive protein and LDL cholesterol and cardiovascular event rates after initiation of rosuvastatin: a prospective study of the JUPITER trial. Lancet 2009;373: Smith GD, Ebrahim S. Mendelian randomization : can genetic epidemiology contribute to understanding environmental determinants of disease? Int J Epidemiol 2003;32: Nitsch D, Molokhia M, Smeeth L, DeStavola BL, Whittaker JC, Leon DA. Limits to causal inference based on Mendelian randomization: a comparison with randomized controlled trials. Am J Epidemiol 2006;163: CARDIOVASCULAR FLASHLIGHT doi: /eurheartj/ehq048 Online publish-ahead-of-print 19 February An unusual cause of cardiogenic shock David R. Altmann, Peter Buser, and Christian Sticherling* Division of Cardiology, University Hospital of Basel, Petersgraben 4, 4031 Basel, Switzerland * Corresponding author. Tel: , Fax: , csticherling@uhbs.ch A 76-year-old man with a history of coronary artery bypass grafting presented with acute onset of chest pain and cardiogenic shock. The ECG showed atrial fibrillation (AF) and severe ST-segment depression in the anterolateral leads. Cardiac angiography revealed an occluded native right coronary artery (RCA) and left anterior descending artery (LAD) and stenosis of the proximal circumflex artery (Panel A). The left subclavian artery was occluded by a large thrombus (Panel B, asterix; note the stenosis of the vertebral artery, arrow). Thrombus aspiration failed and balloon dilatation was not performed in order to avoid dislocation of thrombotic material into the left internal mammary artery (LIMA) graft. We therefore started local intra-arterial thrombolysis (alteplase 10 mg bolus) followed by systemic thrombolysis (alteplase 90 mg over 60 min) resulting in rapid improvement of clinical symptoms and normalization of the ECG. A control angiography showed a patent subclavian and LIMA (Panel C), which supplied the LAD. The right internal mammary artery was used as a free graft from the LIMA to the marginal branches and the RCA (Panel D). Mechanistically, the thrombus may have formed locally over a ruptured subclavian atherosclerotic plaque (Panel C, arrow) or may have originated from the left atrium since the patient was in AF and off anticoagulant therapy. In summary, this case demonstrates the successful treatment of cardiogenic shock secondary to thrombotic occlusion of the subclavian artery proximal to a single LIMA bypass supplying six coronary branches by local and systemic thrombolysis. Published on behalf of the European Society of Cardiology. All rights reserved. & The Author For permissions please journals.permissions@oxfordjournals.org.