Documents

Association between C reactive protein and coronary heart disease

Description
RESEARCH Association between C reactive protein and coronary heart disease: mendelian randomisation analysis based on individual participant data C Reactive Protein Coronary Heart Disease Genetics Collaboration (CCGC) Correspondence to: crpgenetics@phpc.cam.ac.uk Cite this as: BMJ 2011;342:d548 doi:10.1136/bmj.d548 ABSTRACT Objective To use genetic variants as unconfounded proxies of C reactive protein concentration to study its causal role in coronary heart disease. Design Mendelian randomisat
Categories
Published
of 8
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
Share
Transcript
  RESEARCH Association between C reactive protein and coronary heartdisease: mendelian randomisation analysis based onindividual participant data C Reactive Protein Coronary Heart Disease Genetics Collaboration (CCGC)  ABSTRACT Objective To use genetic variants as unconfoundedproxies of C reactive protein concentration to study itscausal role in coronary heart disease. Design Mendelian randomisation meta-analysis of individual participant data from 47 epidemiologicalstudies in 15 countries. Participants 194418 participants, including 46557patients with prevalent or incident coronary heartdisease. Information was available on four  CRP  genetagging single nucleotide polymorphisms (rs3093077,rs1205, rs1130864, rs1800947), concentration of Creactive protein, and levels of other risk factors. Main outcome measures Risk ratios for coronary heartdisease associated with genetically raised C reactiveprotein versus risk ratios with equivalent differences in Creactive protein concentration itself, adjusted for conventional risk factors and variability in risk factor levels within individuals. Results CRP  variants were each associated with up to30% per allele difference in concentration of C reactiveprotein (P<10 − 34  ) and were unrelated to other risk factors.Risk ratios for coronary heart disease per additional copyofanalleleassociatedwithraisedCreactiveproteinwere0.93 (95% confidence interval 0.87 to 1.00) for rs3093077; 1.00 (0.98 to 1.02) for rs1205; 0.98 (0.96 to1.00) for rs1130864; and 0.99 (0.94 to 1.03) for rs1800947. In a combined analysis, the risk ratio for coronary heart disease was 1.00 (0.90 to 1.13) per 1 SDhigher genetically raised natural log (ln) concentration of C reactive protein. The genetic findings were discordantwith the risk ratio observed for coronary heart disease of 1.33 (1.23 to 1.43) per 1 SD higher circulating lnconcentrationof C reactive protein in prospectivestudies(P = 0.001 for difference). Conclusion Human genetic data indicate that C reactiveprotein concentration itself is unlikely to be even amodest causal factor in coronary heart disease. INTRODUCTION Persistent inflammation has been implicated in thepathogenesis of coronary heart disease, but causalityhasnotbeenestablishedforanyspecificinflammatorymediator. 1 C reactive protein, an acute phase proteinproduced by the liver, is the most extensively studiedsystemic marker of inflammation. 2 Observational epi-demiological studies have shown that C reactive pro-tein concentration is log linearly related to risk of subsequent coronary heart disease, though this asso-ciationdependsconsiderablyonlevelsofconventionalrisk factors. 3 C reactive protein binds to low densitylipoproteins 4 and is present in atheroscleroticplaques. 5 There is, therefore, interest in whether long termaverage( “ usual ” )concentrationofCreactivepro-tein is itself causally relevant to coronary heart disease. 67 Randomised trials of interventions specificto C reactive protein, however, have not yet beendone in relation to vascular disease outcomes.In the absence of such trials, focused genetic studiescan be used to help judge causality. This approach isknown as “ mendelian randomisation ” because it isbasedonMendel ’ ssecondlaw,whichstatesthatallelesof different genes assort independently of one anotherduring gamete formation. Consequently, mendelianrandomisation analyses are based on Mendel ’ s obser-vation that inheritance of one trait should be indepen-dent of inheritance of other traits. 8 For the causalassessment of C reactive protein, a mendelian rando-misation analysis should reduce confounding, pro-vided the genetic variants used as proxies forconcentration of C reactive protein are unrelated toconventional vascular risk factors and other diseasemarkers. Such studies should also avoid distortionscaused by factors occurring later in life (such as theonset of disease) because genetic variants are fixed at conception. 8 Hence, mendelian randomisation ana-lyses should confer certain design advantages akin tothose in randomised trials. When applied to other riskfactors in coronary heart disease, this approach haspreviously confirmed the causal relevance of low den-sity lipoprotein cholesterol, 9 increased the likelihoodof causality for Lp(a) lipoprotein, 10 and reduced thelikelihood of causality for fibrinogen. 11 Findings from previous human genetic studies havereducedthelikelihoodofamajorcausalroleforCreac-tive protein concentration in coronary heart disease. 12-18 As most known genetic variants related toC reactive protein account for a relatively small por-tion of the variability in its concentration, however,even larger and more detailed analyses are needed to Correspondence to:crpgenetics@phpc.cam.ac.uk Cite this as: BMJ  2011;342:d548 doi:10.1136/bmj.d548BMJ | ONLINE FIRST | bmj.com page 1 of 8  Ln C reactive protein (mg/L)Age at survey (years)BMI (kg/m 2  )Systolic BP (mm Hg)Diastolic BP (mm Hg)Total cholesterol (mmol/L)Non-HDL cholesterol (mmol/L)HDL cholesterol (mmol/L)Ln triglycerides (mmol/L)LDL cholesterol (mmol/L)Apolipoprotein A I (g/L)Apolipoprotein B (g/L)Albumin (g/L)Lp(a) lipoprotein (mg/dL)Ln interleukin 6 (ng/L)Fibrinogen (µmol/L)Ln leucocyte count (×10 9 /L)Glucose (mmol/L)Smoking amount (pack years)Weight (kg)Height (cm)Waist:hip ratio0.207 (0.174 to 0.239)-0.002 (-0.024 to 0.019)0.011 (-0.012 to 0.034)0.024 (0.001 to 0.047)0.009 (-0.015 to 0.032)-0.001 (-0.026 to 0.023)0.004 (-0.019 to 0.028)-0.011 (-0.040 to 0.017)0.01 (-0.014 to 0.033)0.005 (-0.019 to 0.029)0.012 (-0.029 to 0.053)0.013 (-0.020 to 0.046)-0.097 (-0.437 to 0.242)-0.025 (-0.079 to 0.029)0.006 (-0.045 to 0.056)0.014 (-0.013 to 0.041)-0.078 (-0.246 to 0.089)-0.014 (-0.051 to 0.024)-0.151 (-0.350 to 0.048)0.015 (-0.009 to 0.038)0.011 (-0.017 to 0.040)0.001 (0.025 to 0.025)15/70 11718/81 64816/73 66316/74 30916/74 29216/72 93816/70 96916/70 97116/70 47616/68 2478/58 6788/58 8411/24363/16 5776/13 27413/64 1902/293812/60 9612/92614/68 76014/70 3858/62 358 -0.3-0.2-0.100.10.20.3  VariableSD (95% CI) changein biomarker per allele change in SNPSD (95% CI) changein biomarker per allele change in SNPSD (95% CI) changein biomarker per allele change in SNPNo of studies/participants 5.44x10 -35 0.830.340.040.460.910.710.440.420.690.570.450.570.370.830.300.360.480.140.210.440.97 -0.3-0.2-0.100.10.20.3 P valuers3093077 (frequency of risk allele: G = 0.06) 0.169 (0.153 to 0.185)-0.003 (-0.011 to 0.005)-0.003 (-0.013 to 0.008)0 (-0.009 to 0.009)0.006 (-0.004 to 0.017)-0.001 (-0.010 to 0.007)-0.004 (-0.013 to 0.005)0.005 (-0.004 to 0.014)0.002 (-0.010 to 0.013)-0.004 (-0.013 to 0.005)0.007 (-0.004 to 0.018)-0.001 (-0.011 to 0.010)0.007 (-0.015 to 0.029)-0.001 (-0.019 to 0.017)-0.002 (-0.024 to 0.021)-0.007 (-0.019 to 0.005)-0.012 (-0.034 to 0.010)0.008 (-0.004 to 0.021)-0.021 (-0.055 to 0.012)0.006 (-0.007 to 0.020)0.013 (0.004 to 0.022)0.009 (-0.003 to 0.020)30/105 47637/129 71733/116 19131/116 46431/116 43933/111 42233/109 36233/109 40432/103 90632/101 30814/74 52516/7625010/21 48013/26 95313/21 81023/811939/18 33223/83 7079/753429/106 57429/108 93920/91 199 No of studies/participants 1.00x10 -40 0.490.640.980.220.750.400.320.780.390.200.900.510.930.870.220.300.200.210.360.0030.15 P valuers1205 (frequency of risk allele: C = 0.67) Ln C reactive protein (mg/L)Age at survey (years)BMI (kg/m 2  )Systolic BP (mm Hg)Diastolic BP (mm Hg)Total cholesterol (mmol/L)Non-HDL cholesterol (mmol/L)HDL cholesterol (mmol/L)Ln triglycerides (mmol/L)LDL cholesterol (mmol/L)Apolipoprotein A I (g/L)Apolipoprotein B (g/L)Albumin (g/L)Lp(a) lipoprotein (mg/dL)Ln interleukin 6 (ng/L)Fibrinogen (µmol/L)Ln leucocyte count (×10 9 /L)Glucose (mmol/L)Smoking amount (pack years)Weight (kg)Height (cm)Waist:hip ratio0.127 (0.113 to 0.141)0.004 (-0.005 to 0.013)0.007 (-0.004 to 0.017)0 (-0.009 to 0.009)0.005 (-0.004 to 0.014)-0.007 (-0.016 to 0.002)-0.007 (-0.016 to 0.002)0 (-0.009 to 0.009)-0.003 (-0.013 to 0.007)-0.010 (-0.020 to 0)-0.007 (-0.018 to 0.004)-0.008 (-0.019 to 0.003)0 (-0.023 to 0.023)-0.004 (-0.024 to 0.015)-0.007 (-0.029 to 0.016)0 (-0.010 to 0.011)-0.008 (-0.032 to 0.016)0.006 (-0.005 to 0.017)-0.022 (-0.057 to 0.012)0.007 (-0.004 to 0.019)0.005 (-0.006 to 0.017)0.015 (-0.009 to 0.038)30/98 41135/118 91732/109 14931/109 79931/109 78032/104 37531/101 82231/101 84531/96 99530/93 90214/70 70716/72 5488/17 97011/23 48512/17 69424/79 1668/15 73122/80 23610/767528/99 41028/101 72018/86 541 -0.3-0.2-0.100.10.20.3  VariableSD (95% CI) changein biomarker per allele change in SNPSD (95% CI) changein biomarker per allele change in SNPSD (95% CI) changein biomarker per allele change in SNPSD (95% CI) changein biomarker per allele change in SNPSD (95% CI) changein biomarker per allele change in SNPNo of studies/participants 1.00x10 -40 0.370.230.970.300.140.150.990.550.050.220.160.990.660.570.930.530.260.210.210.360.22 -0.3-0.2-0.100.10.20.3 P valuers1130864 (frequency of risk allele: T = 0.30) 0.232 (0.202 to 0.261)-0.019 (-0.045 to 0.008)-0.015 (-0.042 to 0.012)0 (-0.032 to 0.031)0.002 (-0.025 to 0.028)0.013 (-0.023 to 0.049)0 (-0.043 to 0.043)0.024 (-0.005 to 0.053)-0.014 (-0.057 to 0.029)0.009 (-0.029 to 0.047)-0.003 (-0.045 to 0.039)0.010 (-0.034 to 0.054)0.003 (-0.040 to 0.046)-0.050 (-0.109 to 0.010)-0.002 (-0.050 to 0.046)0.001 (-0.042 to 0.044)0.001 (-0.054 to 0.056)0.002 (-0.034 to 0.038)-0.036 (-0.113 to 0.041)-0.018 (-0.052 to 0.016)-0.002 (-0.030 to 0.026)0.003 (-0.034 to 0.039)19/38 57325/58 38522/47 50920/48 51520/48 50222/42 73722/41 50422/41 54522/42 26222/40 8069/19 70311/21 2729/18 0789/10 7028/13 97012/17 8926/10 55015/25 5636/647720/42 65020/44 76011/26 347 No of studies/participants 1.00x10 -40 0.170.280.980.900.490.990.100.520.640.880.650.890.100.930.960.970.920.360.300.880.89 P valuers1800947 (frequency of risk allele: G = 0.94) Fig 1 |  Associations of four principal single nucleotide polymorphisms (SNP) related to C reactive protein with various characteristics in individuals free fromknown coronary heart disease at time of measurement. Estimates presented are based on random effects meta-analysis of study specific associations of eachSNP with panel of risk factors, adjusted, where appropriate, for ethnicity. Per allele model corresponds to association per addition of risk allele for each SNP RESEARCH page 2 of 8 BMJ | ONLINE FIRST | bmj.com  reliably assess the possibility of any moderate causalrole. We report such an analysis based on individualdata from 194418 participants in 47 epidemiologicalstudies.WestudiedthesegeneticvariantsinrelationtoC reactive protein concentration, other risk factors,and risk of coronary heart disease. METHODS Design and rationale The study had five inter-related components. Firstly,we selected, a priori, a panel of four single nucleotidepolymorphismsthatexplain98%ofthevariationinthe CRP   gene in populations of European descent. Thesevariantshavebeenshowntoinfluenceconcentrationof C reactive protein without affecting its proteinsequence. 19 Secondly, we studied whether these poly-morphisms are likely to be exclusively associated withCreactiveproteinconcentrationbyevaluatingtheminrelation to a range of other risk factors. Thirdly, wecalculated risk ratios for coronary heart disease withgenetically raised concentration using information onthese CRP   variants. Fourthly, we calculated risk ratiosfor coronary heart disease with circulating C reactiveprotein concentration after adjustment for conven-tional risk factors and variability in risk factors withinindividuals.Fifthly,wecomparedriskratiosforcoron-ary heart disease with genetically raised concentrationof C reactive protein versus risk ratios seen withequivalent differences in circulating concentrations. Genetic variants We used detailed information about the compositionof the CRP   gene to select a parsimonious set of  “ tag-ging  ” single nucleotide polymorphisms (rs3093077,rs1205, rs1130864, and rs1800947) that fully coversthe common variations of this gene in populations of Europeandescent(thatis,minorallelefrequency ≥ 0.05and an r  2  threshold of  ≥ 0.8). 1920 Data available on 36further single nucleotide polymorphisms enabled useofproxyvariantswhenprincipalpolymorphismswerenot measured. To enhance power, we also studiedcombinations of alleles inherited together, or “ haplo-types ” (see table A in appendix 1 on bmj.com). Contributing studies Detailsof theC ReactiveProtein CoronaryHeart Dis-ease Genetics Collaboration (CCGC) have beendescribed previously. 20 Tables B-E in appendix 1 onbmj.com provides details of contributing studies, andappendix 2 lists study acronyms. Studies were identi-fied through registry approaches and systematicsearches of the literature (see fig A in appendix 3, andappendix 4 on bmj.com). Individual data were sup-plied on 194418 participants, including 46557 withincidentorprevalentcoronaryheartdisease,in 47stu-dies. Studies used different genotyping platforms: 23usedTaqManassays,threeusedKASPARtechnology(KBioscience), three used restriction fragment lengthpolymorphism, 10 used the ITMAT-Broad-CARe50K SNP array, and eight used other multiplex meth-ods. Thirty five studies measured C reactive proteinwith high sensitivity assays, directly or indirectly stan-dardised on the International Reference Standard forC reactive protein immunoassay (WHO 85/506). Theoutcome was defined as fatal coronary heart disease(based on International Classification of Diseasescodes), non-fatal myocardial infarction (using WHOcriteria), or coronary stenosis (>50% narrowing of one of more coronary arteries assessed by angiogra-phy).Allstudyparticipantsprovidedwritteninformedconsent for use of their DNA for genetic testing. Statistical analyses Appendix5onbmj.comprovidesdetailsofthestatisticalmethods. Levels of C reactive protein and other posi-tively skewed variables were natural log transformed.Principal analyses assumed additive effects (per alleleassociations), with subsidiary analyses of other geneticmodels. All analyses of circulating C reactive proteinand other risk factors susceptible to reverse associationbiaseswerelimitedtoparticipantswithoutknowncoron-ary heart disease at time of blood sampling. We calcu-lated study specific associations of baseline (and repeat)values of C reactive protein concentration and othercharacteristics with a linear regression model, adjustedfor sex and ethnicity and combined across studies using  rs3093077rs1205rs1130864rs18009470.21 (0.17 to 0.24)0.18 (0.16 to 0.20)0.13 (0.12 to 0.15)0.26 (0.23 to 0.29)0.060.670.300.94 -0.100.10.20.30.40.850.900.9511.051.10 Single nucleotidepolymorphismPer allele higher mean ln CRP(95% CI), mg/LPer allele higher mean ln CRP(95% CI), mg/LPer allele risk ratio for CHD(95% CI)Per allele risk ratio for CHD(95% CI) 0.93 (0.87 to 1.00)1.00 (0.98 to 1.02)0.98 (0.96 to 1.00)0.99 (0.94 to 1.03)  Allelefrequency* 19/15 133/96 80743/40 527/172 56741/37 145/157 90531/31 636/93 507 No of studies/cases/participants† Fig 2 | Estimates of association of each single nucleotide polymorphism with ln concentrations of C reactive protein and risk of coronary heart disease (CHD). *Frequency of allele for increased concentrations of circulating ln C reactive protein (that is, risk allele). Associations presented per additional copy of risk allele. † For associations between single nucleotide polymorphismand coronary heart disease, studies with <10 cases or <50 participants were excluded from analyses. Study specific estimatesstratified, where appropriate, by sex, ethnicity, and trial arm and combined with random effects models. Maximum availabledata on genetic variants, circulating C reactive protein, and coronary heart disease used for analyses; sensitivity analysesrestricted to participants with data on C reactive protein single nucleotide polymorphisms, circulating C reactive protein, andcoronary heart disease did not differ from current analyses. Fig C in appendix 3 on bmj.com shows study specific associationsbetween single nucleotide polymorphism and C reactive protein and coronary heart disease for each single nucleotidepolymorphism RESEARCH BMJ | ONLINE FIRST | bmj.com page 3 of 8  random effects meta-analysis to obtain summary esti-matesofsinglenucleotidepolymorphismandCreactiveprotein correlates. 21 We calculated risk ratios for coron-aryheartdiseaseperadditionofaprespecifiedriskalleleof each single nucleotide polymorphism using a twostageapproach,stratified,whereappropriate,bysex,eth-nicity,andadjustmentforage.Wecalculatednaturallog (ln) risk ratios separately within each study before pool-ingthemacrossstudiesbyrandomeffectsmeta-analysis. 3 For prospective cohort studies, we used stratified Coxproportional hazard regression models to calculatehazard ratios for incident coronary heart disease risk.For retrospective studies (and for prevalent coronaryheartdiseasecasesinprospectivestudies),weusedeitherconditionalorunconditionallogisticregressionasappro-priate to calculate odds ratios. Hazard ratios and oddsratioswereassumedtoapproximatethesameriskratios.To correct risk ratios for coronary heart disease forpotentialbiascausedbymeasurementerrorandvaria-bility in risk factors within an individual ( “ regressiondilutionbias ” ),wemadeallowanceforregressiondilu-tion bias in both C reactive protein concentration andpotential confounding factors. Regression dilutionratios were obtained by regressing 93998 serial mea-surements (mean interval 2.9 (SD 1.9) years) from35023 participants on baseline levels of the relevant characteristic, adjusted for conventional risk factorsplus baseline ln C reactive protein concentration andduration of follow-up. 22 Risk ratios for coronary heart disease, adjusted for conventional risk factors, werecalculated per 1 SD higher usual ln concentration inparticipants. This difference corresponds to anapproximately threefold difference on the srcinalscaleofCreactiveproteinmeasurement,asthepooledstandard deviation for baseline ln C reactive protein is1.05 (that is, e 1.05 ). To estimate the effect of geneticallypredictedCreactiveproteinoncoronaryheartdisease,we used a hierarchical Bayesian meta-analysisframework. 23 Using information on baseline ln Creactive protein concentration within each genotypeor haplotype group, we obtained a summary estimatefrom meta-analysis of study specific risk ratios for cor-onary heart disease per 1 SD higher genetically pre-dicted C reactive protein concentration. We used I 2 statistic to assess heterogeneity. 24 Diversity at thestudy level was investigated by grouping studies byrelevant characteristics and by meta-regression. Effect modification by several prespecified variables wasinvestigated by formal tests of interaction. Conven-tional analyses were conducted in Stata v 11.0 andBayesian analyses in WinBUGS. RESULTS Mean age at entry was 59 years (SD 10), 89% of parti-cipants were of European descent, and 44% werewomen. Minor allele frequencies ranged from 0.06 to0.33 for the principal single nucleotide poly-morphisms.Foranygivenpolymorphism,minorallelefrequencies weresimilaracrossstudies.Meanbaselineconcentrations of C reactive protein varied across stu-dies, though standard deviations were broadly similar(see fig B in appendix 3 on bmj.com), with an overallmedian of 1.7 (5th, 95th centile 0.3, 12.7) mg/L. Theregression dilution ratio of ln C reactive protein,adjusted for age and sex, was 0.57 (95% confidenceinterval 0.51 to 0.64), similar to those observed herefor systolic blood pressure (0.51, 0.48 to 0.54), andtotal cholesterol (0.55, 0.52 to 0.60). CRP  variants, C reactive protein concentration, and levelsof other risk factors Each of these CRP   variants was associated with base-line C reactive protein, with per allele differences in Creactiveproteinconcentrationof23%(95%confidenceinterval19%to27%)forrs3093077,19%(17%to21%)for rs1205, 14% (12% to 16%) for rs1130864, and 30%(26% to 34%) for rs1800947 (P<0.001; fig 1) (see alsofig C in appendix 3 and table F in appendix 1 onbmj.com). These associations were consistent over (1) T/C/T/G(2) T/C/C/G(3) G/C/C/G(4) T/T/C/G(5) T/T/C/CReference-0.10 (-0.12 to -0.08)0.10 (0.07 to 0.13)-0.21 (-0.24 to 0.17)-0.34 (-0.39 to -0.30)0.300.240.050.130.03 -0.3-0.2-0.4-0.100.10.20.900.951.0511.10 Haplotype(rs3093077/rs1205/rs1130864/rs1800947)Change in mean lnCRP per copy of haplotype versusreference haplotype 1(95% CI), mg/LChange in mean lnCRP per copy of haplotype versusreference haplotype 1(95% CI), mg/L Reference1.01 (0.97 to 1.04)0.98 (0.92 to 1.03)0.99 (0.96 to 1.03)0.98 (0.91 to 1.05) Risk ratio (95% CI)for CHD per copy of haplotype versusreference haplotype 1Risk ratio (95% CI)for CHD per copy of haplotype versusreference haplotype 1Frequency* Fig 3 | Estimates of association of each haplotype with ln concentrations of CRP and risk of coronary heart disease. *Based onseven haplotypes and therefore do not add up to 1. Haplotypes 6 and 7 excluded because they represent individuals for whomit was not possible to assign to primary haplotypes because of missing data in rs3093077 and rs1800947. See table A inappendix 1, fig F in appendix 3, and appendix 5 for details of these haplotypes and explanation of methods. Additivehaplotype model was used to estimate effect of each haplotype relative to two copies of haplotype 1. See appendix 5 for details of this model. Data limited to populations of European descent, for which it was possible to assign haplotypes basedon linkage disequilibrium information between single nucleotide polymorphisms (see appendix 5). Studies with <10 cases or <50 participants excluded. Study specific estimates were stratified, where appropriate, by sex, ethnicity, and trial arm andcombined with multivariate random effects meta-analysis. Data available on up to 25960 cases and up to 139251participants of European descent from 33 studies. Fig G in appendix 3 shows study specific associations between haplotypeand C reactive protein and coronary heart disease for each haplotype RESEARCH page 4 of 8 BMJ | ONLINE FIRST | bmj.com

Cold Calling Pays

Nov 19, 2017
Search
Similar documents
View more...
Tags
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks
SAVE OUR EARTH

We need your sign to support Project to invent "SMART AND CONTROLLABLE REFLECTIVE BALLOONS" to cover the Sun and Save Our Earth.

More details...

Sign Now!

We are very appreciated for your Prompt Action!

x