Clinical Chemistry Siemens Point of Care - Urinalysis
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

Clinical Chemistry 53: 8-16, 2007. First published November 27, 2006; 10.1373/clinchem.2006.074591
This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow 074591.Supplemental Data
Right arrow All Versions of this Article:
clinchem.2006.074591v1
53/1/8    most recent
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via ISI Web of Science (15)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Humphries, S. E.
Right arrow Articles by Miller, G. J.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Humphries, S. E.
Right arrow Articles by Miller, G. J.
Related Collections
Right arrow Molecular Diagnostics and Genetics
Right arrow Lipids, Lipoproteins, and Cardiovascular Risk Factors
(Clinical Chemistry. 2007;53:8-16.)
© 2007 American Association for Clinical Chemistry, Inc.

Molecular Diagnostics and Genetics

Candidate Gene Genotypes, Along with Conventional Risk Factor Assessment, Improve Estimation of Coronary Heart Disease Risk in Healthy UK Men

Steve E. Humphries1,a, Jackie A. Cooper1, Philippa J. Talmud1 and George J. Miller2

1 Centre for Cardiovascular Genetics, Department of Medicine, British Heart Foundation Laboratories, Royal Free and University College Medical School, London, United Kingdom.
2 Medical Research Council Cardiovascular Group, Department of Environmental and Preventive Medicine, Wolfson Institute of Preventive Medicine, London, United Kingdom.

aAddress correspondence to this author at: Centre for Cardiovascular Genetics, Department of Medicine, British Heart Foundation Laboratories, Rayne Bldg., Royal Free and University College Medical School, 5 University St., London WC1E 6JF, United Kingdom. Fax 0107-679-6212; e-mail rmhaseh{at}ucl.ac.uk.


   Abstract
Top
 Abstract
Introduction
 Materials and Methods
 Results
 Discussion
 References
 
Background: One of the aims of cardiovascular genetics is to test the efficacy of the use of genetic information to predict cardiovascular risk. We therefore investigated whether inclusion of a set of common variants in candidate genes along with conventional risk factor (CRF) assessment enhanced coronary heart disease (CHD)-risk algorithms.

Methods: We followed middle-aged men in the prospective Northwick Park Heart Study II (NPHSII) for 10.8 years and analyzed complete trait and genotype information available on 2057 men (183 CHD events).

Results: Of the 12 genes previously associated with CHD risk, in stepwise multivariate risk analysis, uncoupling protein 2 (UCP2; P = 0.0001), apolipoprotein E (APOE; P = 0.0003), lipoprotein lipase (LPL; P = 0.007), and apolipoprotein AIV (APOA4; P = 0.04) remained in the model. Their combined area under the ROC curve (AROC) was 0.62 (0.58–0.66) [12.6% detection rate for a 5% false positive rate (DR5)]. The AROC for the CRFs age, triglyceride, cholesterol, systolic blood pressure, and smoking was 0.66 (0.61–0.70) (DR5 = 14.2%). Combining CRFs and genotypes significantly improved discrimination (P = 0.001). Inclusion of previously demonstrated interactions of smoking with LPL, interleukin-6 (IL6), and platelet/endothelial cell adhesion molecule (PECAM1) genotypes increased the AROC to 0.72 (0.68–0.76) for a DR5 of 19.1% (P = 0.01 vs CRF combined with genotypes).

Conclusions: For a modest panel of selected genotypes, CHD-risk estimates incorporating CRFs and genotype–risk factor interactions were more effective than risk estimates that used CRFs alone.


   Introduction
Top
 Abstract
 Introduction
Materials and Methods
 Results
 Discussion
 References
 
One goal of genetic epidemiology is the development of DNA-based tests for assessment of genetic predisposition to coronary heart disease (CHD).1 To be of clinical value, such tests must provide information over and above that given by conventional risk factors (CRFs). Risk algorithms for determining the Framingham (1) or Prospective Cardiovascular Munster (PROCAM) (2) risk scores use CRFs such as age, sex, blood lipid concentrations, blood pressure (BP), glucose tolerance, adiposity, and cigarette smoking. Using data from the prospective Second Northwick Park Heart Study (NPHSII), we recently compared the predictive powers of these risk algorithms for 3052 British middle-aged men with 250 CHD events during a 10-year period (3)(4). All algorithms had a false-negative rate [for a 5% false-positive rate (DR5)] >85% (5), suggesting that they failed to take account of other important contributors to risk. Adding other classical CHD risk variables to the scores improved predictive power, but this improvement was limited and not statistically significant, in part because they are correlated with other factors already included or are at low prevalence and so of minimal impact.

Inclusion of genotypes from selected candidate gene loci may improve the predictive accuracy of risk algorithms for CHD (6). Although many candidate genes for CHD have been tested (7), the optimal set of risk genotypes has yet to be identified. Only a relatively modest risk can be expected in association with any single genotype; published estimates are in the range 1.2–1.4 (6)(8)(9). Furthermore, given the multifactorial nature of CHD, lifestyle will set the operational context for genetic variants. Thus, a genotype may be associated with a high CHD risk only with exposure to a certain environment, for example, the reported interaction between smoking and apolipoprotein E genotype (APOE)2 (10). Thus, multiple-locus batteries of genotypes (11) may be significantly associated with CHD, but their usefulness over and above CRFs in risk prediction has not been demonstrated. A study using 5 genes involved in cancer risk suggested that only ~20 genes are usually needed to explain 50% of the burden of a disease in the population if the predisposing genotypes are common (>25%), even if the individual risk ratios are relatively small (relative risk, 1.2–1.5) (12). To explore the use of genes in risk prediction for CHD, we used previously published NPHSII data on CRFs (4) and 14 common single-nucleotide polymorphisms (SNPs) in 12 candidate genes involved in determining plasma lipids, hemostasis, and vascular cell biology (13)(14)(15)(16)(17)(18)(19). We evaluated and compared ROC curves based on combinations of genotypes, CRFs, and previously demonstrated genotype–CRF interactions (17)(18)(20)(21)(22).


   Materials and Methods
Top
 Abstract
 Introduction
 Materials and Methods
Results
 Discussion
 References
 
study participants
NPHSII is a prospective study of healthy middle-aged men (50–64 years) recruited from 9 UK general medical practices (3). Full details have been reported elsewhere (3)(5) and are presented in the Methods of the Data Supplement that accompanies the online version of this article at http://www.clinchem.org/content/vol53/issue1. In the 2057 Caucasian men with complete trait data (and with full genotype data), by April 2005 there had been 183 CHD events comprising 128 acute CHD events (41 fatal), 40 coronary artery revascularization procedures, and 15 silent myocardial infarctions.

genotype analysis
Genotypes were determined for 12 genes: uncoupling protein 2 (UCP2), 866G>A (15); apolipoprotein (apo) E (APOE) {epsilon}2, {epsilon}3, {epsilon}4 (20); lipoprotein lipase (LPL) D9N (21), apoAIV (APOA4) T347S (13); interleukin-6 (IL6), 174G>C (22); platelet endothelial cell adhesion molecule-1 (PECAM1) G650R (17); angiotensin type 1 receptor (AGTR1) 1166A>C (18); angiotensin type 2 receptor (AGTR2) 1675G>A (18); bradykinin type 1 receptor (BDKRB1) –699G>C (16); bradykinin type 2 receptor (BDKRB2) +9/–9 (16); peroxisome proliferator-activated receptor-{alpha} (PPARA); intron 7 and L162V (19); and factor VII (F7) A670C (14) (see Table 1A and 1B in the online Data Supplement for references for genotyping methods and Table 1C in the online Data Supplement for genotype distributions and allele frequencies).

statistical analysis
We log-transformed serum triglyceride (TG) data before statistical analysis. Associations with CHD risk were assessed with Cox proportional hazards models, and hazard ratios (HR; 95% confidence interval) were derived. Analyses were stratified by recruitment center. Coefficients and HRs for systolic BP, TG, and cholesterol were adjusted for regression to the mean by correction factors obtained by the Rosner intraclass correlation coefficient method from repeat measurements made within 6 weeks on a subset of 228 men in NPHSII (5). For the conventional model, a score was derived based on age, TG, cholesterol, smoking, and systolic BP (5). For the genotype-only and genotype-plus-CRF models, scores were obtained by weighting according to the ß coefficients presented in Table 2 . Thus, for the genotype score, those negative for all risk genotypes were scored as 0, and for those positive for any genotype, the coefficient associated with each positive genotype was added to the score. This method gives more weight to the more strongly associated genotypes and thus should yield better predictions than those obtained by basing estimates simply on the number of risk genotypes. Only genotypes and interactions previously shown to be statistically significant in NPHSII were considered for entry into the models. Genotypes were selected for inclusion in the model by a stepwise procedure. Departure from a multiplicative model was then assessed by adding interaction terms into the model. Because of low power to pick up interaction effects, a significance level of P = 0.1 was used. In addition to models including the individual genotypes, models were also fitted according to the number of risk genotypes carried by each study participant, and HRs were obtained for participants carrying any 1, 2, 3, or more risk genotypes. The ability of the several algorithms to predict CHD was assessed using the area under the ROC curve (AROC). The ROC curve was produced by plotting the sensitivity (true positive rate) against 1 – specificity (false-positive rate) for all possible scores. Differences in the areas under the curves were tested with an algorithm suggested by Cooper et al. (5). We also calculated the Harrell C index, which extends the ROC analysis to the case of right-censored survival data. Confidence intervals were calculated as described (23). Detection rates (or sensitivities) for a 5% false-positive rate were calculated by interpolation (DR5). It was not appropriate to use a random 50% of the sample to derive the best-fit model and the 2nd 50% as a test cohort, because the number of cases available limits the power to detect the expected modest effects associated with the SNPs.


View this table:
[in this window]
[in a new window]

  		Table 2. Genotype only and genotype plus conventional models based on 183 men with CHD and 1874 without CHD.


   Results
Top
 Abstract
 Introduction
 Materials and Methods
 Results
Discussion
 References
 
CRFS and risk
We used the whole data set for initial analyses (n = 3012; 254 CHD events), irrespective of whether genotype data were available for all 14 SNPs. After we obtained the best model of 7 SNPs (6 genes), we repeated the analysis and included only study participants with complete genotype data. The characteristics of the whole group and those with complete genotype data did not differ significantly (see Table 2 in the online Data Supplement), and the results obtained for the whole data set and for those with complete genotype data were essentially the same (not shown). For the 2057 Caucasian men with complete trait and genotype data, the baseline CRFs age, TG, cholesterol, smoking rate, and systolic BP were higher in those who developed CHD (Table 1 ). The HR associated with each CRF was statistically significant. The AROC given by this set of CRFs was 0.66 (0.61–0.70), with a DR5 of 14.2%.


View this table:
[in this window]
[in a new window]

  		Table 1. Baseline characteristics by CHD event status and HRs for risk traits for men with complete trait (CRF) and genotype data.

genotype and risk
To reduce the possibility of type I errors due to multiple comparisons, we did not examine the complete list of genes and previously determined genotypes examined for CHD risk associations in NPHSI (see Table 1 in the online Data Supplement). We examined only the 12 candidate genes (see Materials and Methods) that showed statistically significant CHD-associated risk. Only UCP2, APOE, LPL, and APO4 were independently associated with risk in the stepwise regression procedure (Table 2Up ), with no evidence for greater-than-additive effects for any pairwise combination (see Table 3 in the online Data Supplement). The AROC for this 4-genotype model was 0.62 (0.58–0.66), with a DR5 of 12.6%, not statistically significantly different from that obtained using CRFs alone (P = 0.20) (Fig. 1A ). Compared with the risk associated with homozygosity for the common alleles, the risk associated with the less frequent alleles was modest, with HRs ranging from 1.38 to 2.41 and 0.35 for the protective e2 allele. These HRs were similar and remained statistically significant after adjustment for CRFs.


Figure 1
View larger version (26K):
[in this window]
[in a new window]

  		Figure 1. ROC curves using the models in Tables 2Up and 3Up .

(A), Figure 1, conventional risk factors only. Figure 1, genotypes only. Performance of 5 SNPs at the UCP2, APOE, LPL, and APOA4 loci and CRFs in discriminating between men with and without coronary events in NPHSII. (B), Figure 1, conventional risk factors and genotypes. Figure 1, conventional risk factors, genotypes, and interactions. Performance of 7 SNPs at the UCP2, APOE, and APOA4 loci and LPL, PECAM1, and IL6 loci with interaction with smoking, vs a model with no interactions in discriminating between men with and without coronary events.

conventional plus genotype risk factors
When both genotype data and CRFs were included in the score, the AROC increased to 0.70 (0.66–0.74) with a DR5 of 18.0%, a significant improvement over the score obtained with conventional factors alone (P <0.001). The predictive accuracy of these 4-gene genotypes was not significantly different in study participants in the lowest, middle, or top tertile of CRF score [0.65 (0.55–0.75), 0.63 (0.55–0.72), and 0.60 (0.53–0.67), respectively; P = 0.66]. Adding APOE genotype alone significantly improved the AROC [0.68 (0.64–0.72); P <0.01 vs CRFs only], but none of the other genotypes singly or in pairwise combinations did so (see Table 4 in the online Data Supplement). Study participants did not differ significantly in age, TG, systolic BP, or smoking habit according to the number of genetic risk factors. Mean (SD) serum cholesterol, however, was significantly lower in those with no genetic risk factors than in all other groups [5.30 (1.04) vs 5.77 (1.00) mmol/L, respectively; P <0.001], due in part to the higher prevalence of the cholesterol-lowering {epsilon}2{epsilon}3 genotype in this group (see Table 5 in the online Data Supplement). With CHD event rates in men with none of the risk genotypes (9.0% of the sample) as reference, the CRF-adjusted HRs for those carrying any 1 (53.3%), any 2 (33.4%), or any 3 (4.3%) genotypes were 3.17 (1.16–8.60), 4.58 (1.67–12.56), and 12.05 (4.12–35.24), respectively (no study participant was found to have all 4 risk genotypes).

risk score with gene–environment interactions
Several of the gene variants have shown evidence for interaction with smoking [LPL (21), APOE (20), IL6 (22), PECAM1 (17)] or a baseline systolic BP of ≥160 mmHg [AGTR2 (18), BDKRB2 (16)], and we included these interactions in a 3rd model. The reported interactions with increased systolic BP did not significantly improve the model (P >0.2 for both). However, as shown in Table 3 , in a model including the CRFs and the reported smoking interactions, UCP2, APOE, and APO4 were retained in the model, as was the interaction term between smoking and LPL; 2 additional genotypes also added to the model, PECAM1 and IL6. The AROC for this model was 0.72 (0.68–0.76) with a DR5 of 19.1% (Fig. 1BUp ), which was a significant improvement (P = 0.01) on the model based on CRFs and genotypes alone. These estimates were not materially altered in any way when we accounted for those having CHD events by use of the Harrell C index, which takes into account the censored data (see Table 6 in the online Data Supplement). For LPL, IL6, and PECAM1, significant risk was found only in smokers, with the highest HR of 8.02 observed in LPL N9 smokers, compared with LPL D9 nonsmokers, although the prevalence of this group was low (0.9%). For IL6, men carrying 1 or more copy of the –174C allele, nonsmokers had no significant evidence for higher CHD risk than GG men, but smokers had a 1.81-fold higher risk. For PECAM1, nonsmokers homozygous for the common G allele had the lowest risk (and were set as the reference group) but smokers had the highest risk [3.1 (1.79, 5.36)], whereas men carrying the R allele had a much smaller increase in risk, only 30%, if they smoked.


View this table:
[in this window]
[in a new window]

  		Table 3. Model including genotypes and gene-environment interactions in 1874 non-CHD and 183 CHD men.

A Kaplan–Meier survival plot of the combined effects of these variants and their interactions on survival is presented in Fig. 2 . No significant differences in age, serum cholesterol, or TG were observed between the 5 groups of men with different numbers of genetic risk factors (i.e., either a risk genotype alone or a genotype–smoking risk factor; Table 4 ).


Figure 2
View larger version (16K):
[in this window]
[in a new window]

  		Figure 2. Kaplan–Meier survival curve for CHD in study participants with different numbers of genetic risk factors defined as UCP2–866G>A = AA; APOE, {epsilon}3/{epsilon}3/{epsilon}4; APOA4 T347S = TS/SS; LPL D9N = N9 smokers; PECAM G650R = GG smokers; and IL6 G174C = GC/CC smokers.


View this table:
[in this window]
[in a new window]

  		Table 4. Mean (SD) levels, or prevalence, of CRFs, and adjusted HR by number of any genetic risk factors (i.e., genotype or genotype-CRF risk factors; model with interactions).1


   Discussion
Top
 Abstract
 Introduction
 Materials and Methods
 Results
 Discussion
References
 
For any novel CHD risk factor to be useful in a clinical setting, it must significantly enhance event prediction based on easily measurable CRFs, such as age, cholesterol, BP, or body mass index. Thus, any such factor(s) must have a major impact on risk (24). In NPHSII, the relative odds for individual CRFs were 2.5–3.3, whereas the final CRF risk score model had a relative odds of 4.3. For categorical risk factors, lack of variation in exposure within the population may partly explain the lack of discrimination. In this study, diabetes was the categorical factor most strongly associated with CHD, but because almost all of the study participants were nondiabetic, the majority of CHD cases were among nondiabetics.

The primary aim of this study was to examine whether common SNPs in any of 12 candidate genes, individually or in combination, could meet the criterion of enhancing event prediction. Our a priori hypothesis was that risk prediction would not be significantly improved, because of the low reported risk associated with any particular SNP, the low prevalence of risk genotypes, and the fact that many of the genes will be influencing CHD through factors already in the CRF algorithm. Thus, although metaanalysis demonstrates a significant impact of CETP genotype on risk, this effect was considerably attenuated (and no longer statistically significant) by adjustment for HDL concentrations (9), demonstrating that CHD-risk genotypes will not necessarily improve prediction over CRFs. APOE did so singly, however, and the combination of SNPs in the genes for UCP2, APOA4, and LPL with APOE improved prediction further. The inclusion of reported interactions between certain genotypes (LPL, IL6, and PECAM1) and smoking improved the model still further, with a final 19.1% specificity of prediction, for a false-positive rate of 5%. These results suggest that it may be possible to develop a relatively small, clinically useful set of SNPs that can enhance CHD risk prediction based on CRFs alone. The set of SNPs (and their interactions with smoking) that gave the best prediction in NPHSII are unlikely to be universally applicable, and without validation in larger data sets cannot be considered for clinical practice. Although others have reported multilocus genotype assays for CHD risk (11), and at least 1 report suggests that inclusion of a limited set of candidate SNPs may improve prediction of stroke risk (25), to our knowledge our study is the first demonstration of the predictive value of a combined SNP genotype for CHD risk. For cancer risk, such an approach has yielded variable results (26)(27).

We have suggested that data from large metaanalyses be used to select genotypes for inclusion in risk algorithms (6). For the genes examined here, only metaanalysis confirmation for APOE (28) and for LPL (29) have been reported. Such confirmation is particularly important in regard to gene–environment interactions but has not been reported for any of the genes studied here. In this analysis, the previously reported interaction of APOE and smoking in determining risk of CHD (20), although not seen in all studies (30), appears relatively robust (31)(32). In stepwise regression analysis, however, it was not statistically significant (P = 0.12) and was therefore not retained in the final model. Thus, considerable caution must be exercised and many additional data are required before such SNPs and their interactions can be accepted for risk prediction.

Our data suggest that although the risk associated with any 1 genotype included here is modest, in combination, they may be associated with a clinically significant risk. Because the allele frequency for many of these variants is high, many men carry several alleles associated with CHD risk. Compared with the 9.0% of men carrying none of the risk alleles, the observed 10-year HR for CHD in the 33% of men with 2 risk alleles was >4-fold higher, and in the 4.3% with 3 risk alleles, it was 12-fold higher (see Table 5 in the online Data Supplement). This risk is comparable to that seen in untreated familial hypercholesterolemia (33). For the 29% of men in NPHSII who were smokers, the proportion with a similarly increased HR attributable to the combination of smoking and their genotype is even larger (11.7%; Table 4Up ). To put the magnitude of the genetic contributions to risk prediction in context, the contributions of cholesterol and systolic BP to the AROC were 2.5% and 0.7%, respectively, and the genotype and genotype–smoking contributions were 4.1% and 6.4%.

mechanism of chd risk effects
All SNPs included in the final model were functional in that the base change creating the SNP alters the amino-acid sequence and function of the cognate protein [LPL (34), APOE (35), APO4(36), PECAM1 (17)] or the control of transcription and gene expression [UCP2 (37), IL6(38)]. In contrast, nonfunctional SNPs are likely to have little or no utility unless they show complete or strong linkage disequilibrium with a functional SNP. The CHD-risk mechanism of action of the functional SNPs is thus attributable to modulation of the anti- or proatherogenic actions of the cognate proteins, with the APOE effect probably related to the associated atherogenicity of plasma lipoproteins (28)(35), the influence of UCP2 on oxidative stress (39), the reverse cholesterol metabolism and lipid oxidation effects of APOA4 (40), the influence of LPL (34) and PECAM1 (41) on monocyte adhesion and entry into the developing plaque, and the modulation by IL6 of IL-6 plasma and tissue concentrations and thus damage to endothelial or other vascular wall cells (reviewed in Ref. (42).

limitations to the study
One limitation to our findings is that evaluating the effects on CHD risk of the combined genotypes in the same sample in which they were originally tested likely leads to overestimation of their predictive power. Both the Framingham (1) and PROCAM (2) risk algorithms, however, were developed in the same sample in which the constituent CRFs were originally tested, with PROCAM based on 325 events in 5389 men followed for 10 years, and Framingham based on 383 events in 2489 men followed for 12 years (i.e., of a magnitude similar to NPHSII). Therefore, although the actual risk estimates seen here are unlikely to be extrapolated to other studies, we believe that this limitation did not affected the exploration of whether a modest panel of SNPs can add significantly to the CHD-predictive power of classical risk factors. We and others (43) have argued that correction for multiple comparisons is too conservative in hypothesis-deriving analyses such as these, but for some of the genotypes, the statistical significance is high and likely to be a true effect.

The number of CHD events was a major limitation in our study, restricting the power of detection to a risk of >1.7-fold for a variant with a carrier frequency of 10% ({alpha} = 0.05; ß = 0.2). To decrease the potential problem of multiple testing, we included in the modeling only those genes showing a statistically significant effect on risk in NPHSII, excluding 13 other genes (26 SNPs; see Table 1C in the online Data Supplement). Thus, several genotypes with modest effects on CHD risk confirmed by metaanalysis [e.g., ENOS (8) and CETP (9)] were not statistically significant in NPHSII (see Table 1B in the online Data Supplement) and were therefore not included in the current analysis. We also acknowledge the limitations set by the limited number of SNPs examined. Other genes reported to influence CHD risk (7), such as those for lymphotoxin {alpha} (44) or the Toll-like receptor 4 (45), may also improve predictive power, but studies larger than NPHSII will be needed to test this possibility. Simulations in NPHSII show that the AROC for conventional factors would improve by ~3% (P = 0.03) after addition of a genotype of 10% frequency associated with a 3-fold risk, a level unlikely for common genetic variants for which estimates are usually <2-fold (6)(28)(29).

In summary, we explored the potential scope of genetic approaches to risk assessment in CHD and examined the potential predictive utility of genotype interactions with CRFs, such as smoking. For mathematical reasons, adding independent risk factors to risk prediction is associated with diminishing returns (24), and with risk factors of modest screening performance (DR5 ≤15%), all independent of each other, ~15 such factors would be required to increase the DR5 to 80%. Our genetic risk factor results for 5 classical factors and 10 genes indicate that this goal is feasible. For the genotypes included here, if the true HR values are lower, their individual contribution to the ROC will be smaller, and for the combined genetic contribution to improve the predictive power of the CRFs, more genotypes of modest effect would have to be included. Thus, to have a similar improvement for AROC, 12 SNPS with a 10% genotype frequency and a 1.5-fold risk would be required, whereas for a 30% frequency, 3 SNPs with a 1.5-fold risk would be required. Nevertheless, these findings confirm the reported utility of a relatively small set of common SNPs of modest effect on risk to explain a significant proportion of the burden of disease (12), suggesting that a relatively small number of selected SNPs can be used with CRFs to identify high-risk individuals who would derive the most benefit from correction of modifiable risk factors through lifestyle interventions or medications. Thus, these data demonstrate that the discovery of independent gene SNPs or haplotypes of modest predictive power to add to the classical risk score algorithm may have clinical utility in CHD risk prediction and management.


   Acknowledgments
 
We dedicate this paper to Prof. George Miller, who died on August 14, 2006, after a long illness. The NPHSII was supported by the British Medical Research Council, NIH Grant NHLBI 33014, and Du Pont Pharma (Wilmington, DE). J.A.C. and S.E.H. are supported by British Heart Foundation Grant PG2000/015.


   Footnotes
 
1 Nonstandard abbreviations: CHD, coronary heart disease; CRF, conventional risk factor; PROCAM, Prospective Cardiovascular Munster; BP, blood pressure; NPHSII, Northwick Park Heart Study II; DR5, detection rate for a 5% false-positive rate; SNP, single-nucleotide polymorphism; TG, triglyceride; HR, hazard ratio; AROC, area under the ROC curve.

2 Human genes: APOE, apolipoprotein E; UCP2, uncoupling protein 2 (mitochondrial, proton carrier); LPL, lipoprotein lipase; APOA4, apolipoprotein A-IV; IL6, interleukin 6 (interferon ß2); PECAM1 platelet/endothelial cell adhesion molecule (CD31 antigen); AGTR1, angiotensin II receptor, type 1; AGTR2, angiotensin II receptor, type 2; BDKRB1, bradykinin receptor B1; BDKRB2, bradykinin receptor B2; PPARA, peroxisome proliferator-activated receptor-{alpha}; F7, coagulation factor VII (serum prothrombin conversion accelerator); CETP, cholesteryl ester transfer protein.


   References
Top
 Abstract
 Introduction
 Materials and Methods
 Results
 Discussion
 References
 

  1. Wilson PW, D’Agostino RB, Levy D, Belanger AM, Silbershatz H, Kannel WB. Prediction of coronary heart disease using risk factor categories. Circulation 1998;97:1837-1847.
  2. Assmann G, Cullen P, Schulte H. Simple scoring scheme for calculating the risk of acute coronary events based on the 10-year follow-up of the prospective cardiovascular Munster (PROCAM) study. Circulation 2002;105:310-315.
  3. Miller GJ, Bauer KA, Barzegar S, Cooper JA, Rosenberg RD. Increased activation of the haemostatic system in men at high risk of fatal coronary heart disease. Thromb Haemost 1996;75:767-771.[Web of Science][Medline] [Order article via Infotrieve]
  4. Cooper JA, Miller GJ, Bauer KA, Morrissey JH, Meade TW, Howarth DJ, et al. Comparison of novel hemostatic factors and conventional risk factors for prediction of coronary heart disease. Circulation 2000;102:2816-2822.
  5. Cooper JA, Miller GJ, Humphries SE. A comparison of the PROCAM and Framingham point scoring systems for estimation of individual risk of coronary heart disease in the Second Northwick Park Heart Study. Atherosclerosis 2004;181:93-100.[CrossRef][Web of Science]
  6. Humphries SE, Ridker PM, Talmud PJ. Genetic testing for cardiovascular disease susceptibility: a useful clinical management tool or possible misinformation?. Arterioscler Thromb Vasc Biol 2004;24:628-636.[Abstract/Free Full Text]
  7. Yamada Y, Izawa H, Ichihara S, Takatsu F, Ishihara H, Hirayama H, et al. Prediction of the risk of myocardial infarction from polymorphisms in candidate genes. N Engl J Med 2002;347:1916-1923.[Abstract/Free Full Text]
  8. Casas JP, Bautista LE, Humphries SE, Hingorani AD. Endothelial nitric oxide synthase genotype and ischemic heart disease: meta-analysis of 26 studies involving 23028 subjects. Circulation 2004;109:1359-1365.
  9. Boekholdt SM, Sacks FM, Jukema JW, Shepherd J, Freeman DJ, Mcmahon AD, et al. Cholesteryl ester transfer protein TaqIB variant, high-density lipoprotein cholesterol levels, cardiovascular risk, and efficacy of pravastatin treatment: individual patient meta-analysis of 13,677 subjects. Circulation 2005;111:278-287.
  10. Stengard JH, Kardia SL, Tervahauta M, Ehnholm C, Nissinen A, Sing CF. Utility of the predictors of coronary heart disease mortality in a longitudinal study of elderly Finnish men aged 65–84 years is dependent on context defined by Apo E genotype and area of residence. Clin Genet 1999;56:367-377.[CrossRef][Web of Science][Medline] [Order article via Infotrieve]
  11. Cheng S, Grow MA, Pallaud C, Klitz W, Erlich HA, Visvikis S, et al. A multilocus genotyping assay for candidate markers of cardiovascular disease risk. Genome Res 1999;9:936-949.[Abstract/Free Full Text]
  12. Yang Q, Khoury MJ, Friedman J, Little J, Flanders WD. How many genes underlie the occurrence of common complex diseases in the population?. Int J Epidemiol 2005;34:1129-1137.[Abstract/Free Full Text]
  13. Wong WR, Hawe E, Li LK, Miller GJ, Nicaud V, Pennacchio LA, et al. Apolipoprotein AIV gene variant S347 is associated with increased risk of coronary heart disease and lower plasma apolipoprotein AIV levels. Circ Res 2003;92:969-975.[Abstract/Free Full Text]
  14. Carew JA, Basso F, Miller GJ, Hawe E, Jackson AA, Humphries SE, et al. A functional haplotype in the 5' flanking region of the factor VII gene is associated with an increased risk of coronary heart disease. J Thromb Haemost 2003;1:2179-2185.[CrossRef][Web of Science][Medline] [Order article via Infotrieve]
  15. Dhamrait SS, Stephens JW, Cooper JA, Acharya J, Mani AR, Moore K, et al. Cardiovascular risk in healthy men and markers of oxidative stress in diabetic men are associated with common variation in the gene for uncoupling protein 2. Eur Heart J 2004;25:468-475.[Abstract/Free Full Text]
  16. Dhamrait SS, Payne JR, Li P, Jones A, Toor IS, Cooper JA, et al. Variation in bradykinin receptor genes increases the cardiovascular risk associated with hypertension. Eur Heart J 2003;24:1672-1680.[Abstract/Free Full Text]
  17. Elrayess MA, Webb KE, Bellingham GJ, Whitall RA, Kabir J, Hawe E, et al. R643G polymorphism in PECAM-1influences transendothelial migration of monocytes and is associated with the progression of CHD and CHD events. Atherosclerosis 2004;177:127-135.[CrossRef][Web of Science][Medline] [Order article via Infotrieve]
  18. Jones A, Dhamrait SS, Payne JR, Hawe E, Li P, Toor IS, et al. Genetic variants of angiotensin II receptor in cardiovascular risk in hypertension. Hypertension 2003;42:500-506.[Abstract/Free Full Text]
  19. Flavell DM, Jamshidi Y, Hawe E, Pineda Torra I, Taskinen M-R, Frick MH, et al. Peroxisome proliferator-activated receptor gene variants influence progression of coronary atherosclerosis and risk of coronary artery disease. Circulation 2002;105:1440-1445.
  20. Humphries SE, Talmud PJ, Hawe E, Bolla M, Day IN, Miller GJ. Apolipoprotein E4 and coronary heart disease in middle-aged men who smoke: a prospective study. Lancet 2001;358:115-119.[CrossRef][Web of Science][Medline] [Order article via Infotrieve]
  21. Talmud PJ, Bujac S, Hall S, Humphries SE, Miller GJ. Substitution of asparagine for aspartic acid at residue 9 (D9N) of lipoprotein lipase markedly augments risk of coronary heart disease in male smokers. Atherosclerosis 2000;149:75-81.[CrossRef][Web of Science][Medline] [Order article via Infotrieve]
  22. Humphries SE, Luong LA, Ogg MS, Hawe E, Miller GJ. The interleukin-6–174 G/C promoter polymorphism is associated with risk of coronary heart disease and systolic blood pressure in healthy men. Eur Heart J 2001;22:2243-2252.[Abstract/Free Full Text]
  23. Pencina MJ, D’Agostino RB. Overall C as a measure of discrimination in survival analysis: model specific population value and confidence interval estimation. Stat Med 2004;23:2109-2123.[CrossRef][Web of Science][Medline] [Order article via Infotrieve]
  24. Wald NJ, Morris JK, Rish S. The efficacy of combining several risk factors as a screening test. J Med Screen 2005;12:197-201.[Medline] [Order article via Infotrieve]
  25. Zee RY, Cook NR, Cheng S, Reynolds R, Erlich HA, Lindpaintner K, et al. Polymorphism in the P-selectin and interleukin-4 genes as determinants of stroke: a population-based, prospective genetic analysis. Hum Mol Genet 2004;13:389-396.[Abstract/Free Full Text]
  26. Dwyer T, Stankovich JM, Blizzard L, FitzGerald LM, Dickinson JL, Reilly A, et al. Does the addition of information on genotype improve prediction of the risk of melanoma and nonmelanoma skin cancer beyond that obtained from skin phenotype?. Am J Epidemiol 2004;159:826-833.[Abstract/Free Full Text]
  27. Comings DE, Gade-Andavolu R, Cone LA, Muhleman D, MacMurray JP. A multigene test for the risk of sporadic breast carcinoma. Cancer 2003;97:2160-2170.[CrossRef][Web of Science][Medline] [Order article via Infotrieve]
  28. Song Y, Stampfer MJ, Liu S. Meta-analysis: apolipoprotein E genotypes and risk for coronary heart disease. Ann Intern Med 2004;141:137-147.[Abstract/Free Full Text]
  29. Wittrup HH, Tybjaerg Hansen A, Nordestgaard BG. Lipoprotein lipase mutations, plasma lipids and lipoproteins, and risk of ischemic heart disease. A meta-analysis. Circulation 1999;99:2901-2907.
  30. Keavney B, Parish S, Palmer A, Clark S, Youngman L, Danesh J, et al. Large-scale evidence that the cardiotoxicity of smoking is not significantly modified by the apolipoprotein E epsilon 2/epsilon 3/epsilon 4 genotype. Lancet 2003;361:396-398.[CrossRef][Web of Science][Medline] [Order article via Infotrieve]
  31. Pezzini A, Grassi M, Del Zotto E, Bazzoli E, Archetti S, Assanelli D, et al. Synergistic effect of apolipoprotein E polymorphisms and cigarette smoking on risk of ischemic stroke in young adults. Stroke 2004;35:438-442.[Abstract/Free Full Text]
  32. Talmud PJ, Stephens JW, Hawe E, Demissie S, Cupples LA, Hurel SJ, et al. The significant increase in cardiovascular disease risk in APOEepsilon4 carriers is evident only in men who smoke: potential relationship between reduced antioxidant status and ApoE4. Ann Hum Genet 2005;69:613-622.[CrossRef][Web of Science][Medline] [Order article via Infotrieve]
  33. Risk of fatal coronary heart disease in familial hypercholesterolaemia. Scientific Steering Committee on behalf of the Simon Broome Register Group. BMJ 1991;303:893-896.[Abstract/Free Full Text]
  34. Fisher RM, Benhizia F, Schreiber R, Makoveichuk E, Putt W, al Haideri M, et al. Enhanced bridging function and augmented monocyte adhesion by lipoprotein lipase N9: insights into increased risk of coronary artery disease in N9 carriers. Atherosclerosis 2003;166:243-251.[CrossRef][Web of Science][Medline] [Order article via Infotrieve]
  35. Davignon J, Cohn JS, Mabile L, Bernier L. Apolipoprotein E and atherosclerosis: insight from animal and human studies. Clin Chim Acta 1999;286:115-143.[CrossRef][Web of Science][Medline] [Order article via Infotrieve]
  36. Wong WR, Stephens JW, Acharya J, Hurel SJ, Humphries SE, Talmud PJ. The APOA4 T347S variant is associated with reduced plasma TAOS in subjects with diabetes mellitus and cardiovascular disease. J Lipid Res 2004;45:1565-1571.[Abstract/Free Full Text]
  37. Krempler F, Esterbauer H, Weitgasser R, Ebenbichler C, Patsch JR, Miller K, et al. A functional polymorphism in the promoter of UCP2 enhances obesity risk but reduces type 2 diabetes risk in obese middle-aged humans. Diabetes 2002;51:3331-3335.[Abstract/Free Full Text]
  38. Fishman D, Faulds G, Jeffery R, Mohamed-Ali V, Yudkin JS, Humphries S, et al. The effect of novel polymorphisms in the interleukin-6 (IL-6) gene on IL-6 transcription and plasma IL-6 levels, and an association with systemic-onset juvenile chronic arthritis. J Clin Invest 1998;102:1369-1376.[Web of Science][Medline] [Order article via Infotrieve]
  39. Fleury C, Mignotte B, Vayssiere JL. Mitochondrial reactive oxygen species in cell death signaling. Biochimie 2002;84:131-141.[Medline] [Order article via Infotrieve]
  40. Ostos MA, Conconi M, Vergnes L, Baroukh N, Ribalta J, Girona J, et al. Antioxidative and antiatherosclerotic effects of human apolipoprotein A-IV in apolipoprotein E-deficient mice. Arterioscler Thromb Vasc Biol 2001;21:1023-1028.[Abstract/Free Full Text]
  41. Thompson RD, Noble KE, Larbi KY, Dewar A, Duncan GS, Mak TW, et al. Platelet-endothelial cell adhesion molecule-1 (PECAM-1)-deficient mice demonstrate a transient and cytokine-specific role for PECAM-1 in leukocyte migration through the perivascular basement membrane. Blood 2001;97:1854-1860.[Abstract/Free Full Text]
  42. Yudkin JS, Kumari M, Humphries SE, Mohamed-Ali V. Inflammation, obesity, stress and coronary heart disease: is interleukin-6 the link?. Atherosclerosis 2000;148:209-214.[CrossRef][Web of Science][Medline] [Order article via Infotrieve]
  43. Rothman KJ. No adjustments are needed for multiple comparisons. Epidemiology 1990;1:43-46.[Medline] [Order article via Infotrieve]
  44. Ozaki K, Ohnishi Y, Iida A, Sekine A, Yamada R, Tsunoda T, et al. Functional SNPs in the lymphotoxin-{alpha} gene that are associated with susceptibility to myocardial infarction. Nat Genet 2002;32:650-654.[CrossRef][Web of Science][Medline] [Order article via Infotrieve]
  45. Ameziane N, Beillat T, Verpillat P, Chollet-Martin S, Aumont MC, Seknadji P, et al. Association of the Toll-like receptor 4 gene Asp299Gly polymorphism with acute coronary events. Arterioscler Thromb Vasc Biol 2003;23:e61-e64.[Abstract/Free Full Text]



The following articles in journals at HighWire Press have cited this article:


Home page
 Circ Cardiovasc GenetHome page
L. E. Viiri, K. M. Viiri, E. Ilveskoski, H. Huhtala, M. Maki, P. J. Tienari, M. Perola, T. Lehtimaki, and P. J. Karhunen
Interactions of Functional Apolipoprotein E Gene Promoter Polymorphisms With Smoking on Aortic Atherosclerosis
Circ Cardiovasc Genet, December 1, 2008; 1(2): 107 - 116.
[Abstract] [Full Text] [PDF]

Home page
 Hum Mol GenetHome page
A. C. J.W. Janssens and C. M. van Duijn
Genome-based prediction of common diseases: advances and prospects
Hum. Mol. Genet., October 15, 2008; 17(R2): R166 - R173.
[Abstract] [Full Text] [PDF]

Home page
 Eur J Public HealthHome page
R. Leidl
Promoting economic value in public health
Eur J Public Health, June 1, 2008; 18(3): 216 - 216.
[Full Text] [PDF]

Home page
 NEJMHome page
S. Kathiresan, O. Melander, D. Anevski, C. Guiducci, N. P. Burtt, C. Roos, J. N. Hirschhorn, G. Berglund, B. Hedblad, L. Groop, et al.
Polymorphisms Associated with Cholesterol and Risk of Cardiovascular Events
N. Engl. J. Med., March 20, 2008; 358(12): 1240 - 1249.
[Abstract] [Full Text] [PDF]

Home page
 Clin. Chem.Home page
P. J. Talmud, J. A. Cooper, J. Palmen, R. Lovering, F. Drenos, A. D. Hingorani, and S. E. Humphries
Chromosome 9p21.3 Coronary Heart Disease Locus Genotype and Prospective Risk of CHD in Healthy Middle-Aged Men
Clin. Chem., March 1, 2008; 54(3): 467 - 474.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow 074591.Supplemental Data
Right arrow All Versions of this Article:
clinchem.2006.074591v1
53/1/8    most recent
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via ISI Web of Science (15)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Humphries, S. E.
Right arrow Articles by Miller, G. J.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Humphries, S. E.
Right arrow Articles by Miller, G. J.
Related Collections
Right arrow Molecular Diagnostics and Genetics
Right arrow Lipids, Lipoproteins, and Cardiovascular Risk Factors

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS