Main

Atherosclerosis is a partly heritable disorder,1 although the genes involved and the associated risks are still unclear.2 Ischemic heart disease (IHD) is a group of diseases mainly caused by coronary atherosclerosis. It is widely accepted that inflammation and infection play a key role in atherosclerosis and IHD.3–6 Immune cells are thought to be involved in many aspects related to IHD (e.g., lipoprotein retention and activation of oxidized low-density lipoprotein, plaque development and rupture) and dominate the atherosclerotic lesions.6 Infection has also been reported to be involved in inflammation and plaque activation, thus infection affects the progressions of atherosclerosis and IHD and elicits clinical symptoms.6 The CD14 gene has been proposed as a susceptibility gene for IHD.7 It is located on chromosome 5q31 and encodes CD14, a protein which is a component of lipopolysaccharide receptor complex mainly expressed by monocytes and macrophages.8 By binding to lipopolysaccharide, CD14 mediates the activation of inflammatory cells and is thus involved in inflammatory reactions and contributes to the production of inflammatory mediators and cytokines.9

GENE VARIANT

Baldini et al.10 first reported the existence of a single nucleotide polymorphism in the 5′ genomic region of CD14 at position -260 (allele C and T) with respect to the translation start site (-159 when counting from the transcription start site) and found that the frequencies of the two alleles were similar in the white population from Tucson. The -260T allele was soon reported to be associated with myocardial infarction (MI).7 The substitution of C → T leads to an increased transcriptional activity, which is paralleled by a decreased affinity of DNA/protein interactions between the Sp1, 2, 3 proteins and the GC box in the CD14 promoter. This may be important for the pathogenesis of inflammatory diseases.11 A higher expression of CD14 on the surface of monocytic lineage cells has been observed in TT homozygotes carriers.7,10 All of these lead to the hypothesis that the increased activity of the CD14 promoter results in a higher expression of CD14, hence triggers production of inflammatory cytokines and increases IHD risk.

OBJECTIVES

Many studies have been carried out to investigate the association between the CD14 C-260T polymorphism and IHD risk, but the results were inconsistent. This may be partly because of small sample size, different IHD endpoints (e.g., MI and angiographic coronary artery stenosis [CAS]), and different populations (e.g., European, East Asian, and others). Demonstration of an association may require a much larger number of subjects, which may be beyond the resources of one single study site. IHD continues to be the leading cause of morbidity and mortality worldwide, affecting about 9.4% US white men, 6.0% US white women, 3.8% Asians, and 2.5% American Indians.12 We thus conducted meta-analyses of all available data in accordance with the guidelines of the Human Genome Epidemiology Network (HuGENet™)13,14 to clarify the role of the CD14 C-260T polymorphism in IHD. To help elucidate the gene effect in different ethnicities, we performed the analyses based on ethnicities.15

METHOD

Selection criteria

For inclusion, studies had to be case-controlled or cohort in design, use proper IHD diagnosis criteria (e.g., angiographically confirmed; elevations of cardiac enzymes, changes of electrocardiographic and clinical symptoms according to the World Health Organizations criteria; a documented history of coronary artery bypass graft, percutaneous transluminal coronary angioplasty, or percutaneous coronary intervention), involve unrelated participants, and examine the association between IHD and the presence of CD14 C-260T polymorphism. IHD was defined as acute coronary syndrome (ACS), prior MI, stable angina pectoris (AP), and angiographic CAS.16–18 If essential information of a study was not presented, authors were contacted for details. The study would be excluded if the information could not be obtained.

Search strategy

All studies reporting the association between the CD14 C-260T polymorphism and IHD risk published before December 2008 were identified by comprehensive computer-based searches of Medline, EMBase, BIOSIS, Global Health, LILACS (http://bases.bireme.br), CBMDisc (http://cbmwww.imicams.ac.cn), and HuGENet.19 Terms used for the searches were “CD14,” “ischemic heart disease,” “coronary heart disease,” “coronary artery disease,” “acute coronary syndrome,” “myocardial infarction,” and “angina pectoris,” combined with “gene,” “genetic,” “variant,” “mutation,” or “polymorphism.” Hand searches for related articles were also performed.

Data extraction

The first author, published year, country, populations, mean age of participants, study design, sample size, outcome, diagnostic criteria, genotyping method, characteristics of the controls, allele frequencies, genotype distributions, and cardiovascular risk factors were extracted independently by two authors (H.-F.Z. and B.-L.Z.). Results were then compared, and disagreements were resolved by discussion.

Quality score assessment

The quality of included studies was assessed independently by the same two authors using modified quality assessment scores reported previously.20 Dissensions were resolved by discussion. Scores ranged from 0 (lowest) to 13 (highest).

Statistical analysis

First, deviance from Hardy-Weinberg equilibrium (HWE) was assessed for the controls of each study using Fisher's exact test. Second, genotype distributions of controls were used to estimate the frequency of the putative risk allele (-260T) in various ethnic groups (if the ethnicity of participants was not indicated in the study, ethnicity was assumed according to geographical location from which the participants were recruited) using the inverse variance method.20 Third, estimation of the gene effect on IHD was performed by a logistic regression approach described previously.21 Briefly, a Cochrane Q-test22 for heterogeneity with a significance level of P < 0.1 rather than 0.05 was used separately for three odds ratios (ORs): TT versus CC (OR1), CT versus CC (OR2), and TT versus CT (OR3). If there was heterogeneity on at least one of the three ORs, the sources of heterogeneity were explored by fitting a covariable (e.g., ethnicity, sample size, published language, outcome, or quality score) in the meta-regression model.23–25 If there was no heterogeneity, logistic regression with the fixed-effects model was used to evaluate the overall gene effect; or else, the random-effects model was used. A likelihood ratio (LR) test, but not the three ORs, was used to estimate whether the overall gene effect was significant. If a significant overall gene effect was observed, further comparisons of OR1, OR2, and OR3 were made and the indicated genetic models were selected as follows:

  1. 1

    Dominant model if OR1 = OR2 ≠ 1 and OR3 = 1.

  2. 2

    Recessive model if OR1 = OR3 ≠ 1 and OR2 = 1.

  3. 3

    Overdominant model if OR2 = 1/OR3 ≠ 1 and OR1 = 1.

  4. 4

    Codominant model if OR1 > OR2 > 1 and OR1 > OR3 > 1 (or OR1 < OR2 < 1 and OR1 < OR3 < 1).

Finally, once the appropriate genetic model was identified, results were pooled again under this genetic model.

Publication bias was assessed using Egger's test.26 In addition, subgroup analysis based on whether the IHD was related to vulnerable plaque or not was performed.27–29 ACS and prior MI were defined as vulnerable plaque IHD, whereas AP and angiographic CAS were indicated as other IHD.27–29 Sensitivity analysis was carried out by including studies that deviated from HWE.

All analyses were performed using STATA software, version 9.2 (StataCorp. 2005. Stata Statistical Software: Release 9. College Station, TX: StataCorp LP).

RESULTS

Study inclusion and characteristics

The combined searches yielded 603 records. Of the 603, 531 were excluded by reading titles and abstracts. Finally, 19 studies were included.7,27,30–46 One Chinese study47 met the inclusion criteria but was excluded, because the author obviously misunderstood the locus and no additional information could be obtained by contacting authors. One European study7 was used only for sensitivity analysis because both the cases and the controls in this study might overlap with the latter study by Lorenzova et al.,45 which had a larger sample size.7,45,48

All the included studies used either case-control or nested case-control design. ACS, prior MI, AP, and angiographic CAS cases were included. Appropriate diagnosis criteria and proper genotyping methods were used in most of the studies. Of the 19 studies, 11 involved European populations,7,30–34,36,38,39,41,45 one involved an Indian population,43 and seven involved East Asian populations,27,35,37,40,42,44,46 of which five were published in Chinese journals.37,40,42,44,46 Fourteen studies involved vulnerable plaque IHD patients. Of the 14, 10 included European populations,7,30–32,34,36,38,39,41,45 one included an Indian population,43 and three included East Asian populations,27,35,40 of which one was published in a Chinese journal.40 Twelve studies involved other IHD cases, including six involving European populations30,32–34,38,39 and six involving East Asian populations,27,37,40,42,44,46 of which five were published in Chinese journals.37,40,42,44,46 Genotype distributions deviated from HWE in one study.42 The characteristics of included studies are listed in Table 1, and the sources of participants, as well as the geographic location of the studies, are listed in Supplemental Table, Supplemental Digital Content 1, http://links.lww.com/A842.

Table 1 Characteristics of eligible studies in the meta-analysis
Full size table

Pooled prevalence of CD14-260T in the controls

The -260T allele frequency could be obtained in all included studies. Two European studies reported two38 and four subgroups41 containing different control subjects, respectively.

There was no heterogeneity among the 10 European population studies (χ132 = 11.01, P = 0.61). The pooled frequency using the fixed-effects model was 48.03% (95% confidence interval [CI]: 46.54–49.51%); a sensitivity analysis including the study by Hubacek et al.7 showed a similar result (47.60% [95% CI: 46.14–49.07]). The pooled -260T frequencies were 46.88% (95% CI: 41.41–52.36%), 51.39% (95% CI: 47.43–55.36%), and 45.70% (95% CI: 36.48–52.62%) for overall East Asian studies (random-effects model; χ52 = 17.21, P < 0.01), East Asian studies published in English language journals27,35 (fixed-effects model; χ12 = 0.26, P = 0.61) and Chinese studies37,40,44,46 (random-effects model; χ32 = 12.80, P < 0.01), respectively. An inclusion of the study by Hu et al.42 did not change the result significantly. The -260T allele frequency was 56.47% (95% CI: 50.09–62.85%) in Indian population.43

Association between CD14 C-260T polymorphism and IHD risk

Meta-analysis of European studies

All included studies reported the association between genotype distributions and IHD risk. The summary OR1/OR2/OR3 and Q-test results are listed in Table 2. No significant heterogeneity for OR1, OR2, or OR3 was detected among the 10 European descent studies. The logistic regression with the fixed-effects model comprised 14,263 subjects yielded a nonsignificant overall gene effect (LR2 = 0.85, P = 0.66), indicating no association between the polymorphism and IHD risk in this population. Egger's test showed that publication bias was significant for OR2 but not for OR1 or OR3 (P = 0.75 for OR1; P = 0.04 for OR2 and P = 0.16 for OR3). A sensitivity analysis including the study by Hubacek et al.7 involving 14,576 subjects did not change the results significantly (data not shown).

Table 2 Estimated ORs for CD14 polymorphism and ischemic heart disease risk
Full size table

An initial analysis of a vulnerable plaque IHD subgroup including 10 European studies involving 8738 subjects yielded homogenous results (Table 2). The logistic regression with the fixed-effects model showed a nonsignificant overall gene effect (LR2 = 0.33, P = 0.85), indicating no association between the polymorphism and vulnerable plaque IHD. An inclusion of the study7 showed very similar results (data not shown). A meta-analysis of other IHD subgroup including six studies involving 7157 subjects showed a similar outcome (LR2 = 0.17, P = 0.92).

Meta-analysis of East Asian studies

A Q-test of the six East Asian studies in HWE27,35,37,40,44,46 including 2546 subjects, showed a significant heterogeneity and published language was found to be the source of it (Q-test for OR1, P = 0.27; Q-test for OR2, P = 0.05, meta-regression P < 0.01; Q-test for OR3, P = 0.67). We further performed subgroup analyses of English or Chinese studies. A meta-analysis of the two English studies27,35 including 1240 subjects showed a significant overall gene effect (LR2 = 11.46, P < 0.01). The estimated OR1, OR2, and OR3 suggested a recessive model (Table 2). The pooled OR under this genetic model indicated that East Asians who had the TT genotype were about 53% more likely to have IHD (Table 2). A meta-analysis of the four Chinese studies involving 1306 subjects showed a significant overall gene effect (LR2 = 36.54, P < 0.01). The estimated ORs (Table 2) implied a codominant model. The pooled OR under this genetic model indicated that both TT and CT genotypes conferred increased susceptibility to the disease (Table 2). A sensitivity analysis including the study41 that deviated from HWE did not change the results significantly (data not shown).

A Q-test of the two East Asian studies published in English language journals on vulnerable plaque IHD27,35 involving 1193 subjects showed a significant heterogeneity (Table 2) and the logistic regression with the random-effects model showed a nonsignificant overall gene effect, but the Chinese study39 on vulnerable plaque IHD reported a significant difference. The four Chinese studies37,40,44,46 on other IHD involving 1210 subjects showed a homogenous result (Table 2) and the logistic regression with the fixed-effects model showed a significant overall gene effect (LR2 = 27.97, P < 0.01). The pooled ORs (Table 2) indicated a codominant model. The estimated OR under this genetic model indicated that both TT and CT genotypes were associated with IHD risk (Table 2). An inclusion of the study42 that deviated from HWE did not change the result significantly (data not shown). However, the English study on other IHD showed a nonsignificant result.27

DISCUSSION

A large number of studies have been carried out to test the hypothesis that the CD14 C-260T polymorphism might be associated with the risk of IHD, but data have yielded conflicting results. There is concern that a positive association might be spurious; on the other hand that a negative result might be due to a small sample size. To produce a more precise result, we explored sources of heterogeneity by meta-regression and performed meta-analyses in various populations to evaluate the association between the polymorphism and IHD risk.

Population stratification because of ethnicities may lead to inconsistency, especially when both allele frequencies and incidence rates of the diseases vary across ethnic groups.49 In the present study, results from populations with different genetic backgrounds were not the same. The combinations of the European studies showed nonsignificant results. We further found that the genotype distributions between the cases and the controls were almost the same (proportions of CC, CT, and TT were 26.88%, 49.60%, 23.18% and 27.24%, 49.02%, 23.35%, respectively, for cases and controls) in this population. This indicates that the CD14 C(-260)T polymorphism might be of little importance for clinical practice and public health. The use of this polymorphism as a predictor for the risk of IHD may not be efficient and the screening utility of this genetic variant in asymptomatic individuals may not be warranted in European population. Results from East Asian studies were distinct and the pooled CD14-260T allele frequency of the controls showed a modest difference across ethnicities (European studies: 48.03%; East Asian studies published in English language journals: 51.39%; Chinese studies: 45.70%; Indian study: 56.47%). These may be explained by the different genetic backgrounds across ethnicities or a much smaller sample size in the East Asian and Indian studies.

Evidence of heterogeneity was found among East Asian studies and genetic backgrounds may not account for it.15 Published language was found to be the main source and a larger genetic effect was observed in most Chinese studies. Besides the small sample size, the plausible explanations may be that there is a publication bias in favor of positive results, selection bias in pursuing a significant finding for which the Chinese language may be a marker50; on the other hand, there may be differences in methods for selecting controls used in these studies.35,37,40,42,44,46 Subgroup analysis of English or Chinese studies indicated that the -260T allele was associated with overall IHD risk. However, there is chance, because of a relatively small sample size and publication bias50 this might lead to spurious results. The indicated genetic models21 were also different between English and Chinese studies involving East Asian subjects, for which the publication bias50 and relative small sample size may account. Interestingly, heterogeneity was not found in the analysis of overall IHD but was observed in the subgroup analysis of vulnerable plaque IHD of the two East Asian studies published in English language journals. This might be explained by the much smaller sample size of the study.27 Results from East Asian studies published in English language journals on overall IHD and vulnerable plaque IHD were also different, which was probably because the random-effects model produced a more conservative result. Therefore, given that a combination of heterogeneous studies may lead to a less clear result,22 the potential risk of publication bias may lead to a deviation from the true effect size50–54 and the relative small sample, results drawn from the East Asian studies should be interpreted with caution.

Evidence of publication bias was significant for OR2 but not for OR1 or OR3 among European studies. Many have argued that Egger's test26 has been likely to yield a false-positive result because of high type I error rate, especially when used for a binary outcome with a larger OR.55 Moreover, we further explored publication bias for all genetic models using Egger's test, and found that it was significant for the dominant model with a larger OR but not for the other genetic models with smaller ORs (data not shown). Therefore, the evidence of publication bias by Egger's test was probably false-positive.

Some limitations of the meta-analyses should be considered. First, the phenotypes that could be affected by both the low-penetrance susceptibility gene56 and other risk factors of IHD (e.g., environment factors) might not be the same even though the genotypes were the same and this might lead to the misclassification bias.56,57 Second, the relatively small sample size, representativeness of controls, potential heterogeneity, and publication bias might affect results drawn from East Asian studies; in addition, there was only one study with a relatively small sample size involving an Indian population. Third, there might be eligible studies that were not published, not indexed by electronic databases, or published in the journals we did not cover. Lastly, a lack of individual participants' data has restricted further adjustments by other covariables, such as specific outcome, sex, smoking, diabetes, dyslipidemia, etc.

Despite limitations, the present study has suggested a possible association between the CD14 C-260T polymorphism and IHD risk in an East Asian population but not in European and Indian populations. In addition, there is a need for larger and more rigorous studies than is now customary, and a need to support the publication of negative results. There is also a greater need for updated genetic epidemiology quantitative systematic reviews with proper methodology,13 to help minimize random error, explore heterogeneity, address publication bias and enhance statistical power; thereby helping to better understand the association between the CD14 C-260T polymorphism and IHD risk.