Heterogeneity of the Phenotypic Definition of Coronary
Artery Disease and Its Impact on Genetic
Association Studies
Georgios D. Kitsios, MD, PhD; Issa J. Dahabreh, MD; Thomas A. Trikalinos, MD, PhD;
Christopher H. Schmid, PhD; Gordon S. Huggins, MD; David M. Kent, MD
Downloaded from http://circgenetics.ahajournals.org/ by guest on June 15, 2017
Background—Variability in phenotypic characterization of coronary artery disease (CAD) may contribute to the
heterogeneity of genetic association studies, and more consistency in phenotype definitions might improve replication
of genetic associations. We assessed the extent of phenotypic heterogeneity and quantified its impact in a large literature
sample of association studies.
Methods and Results—We searched for large (ⱖ15 studies) meta-analyses of genetic associations and reviewed all studies
included therein. From each primary study, we extracted phenotypic definitions, demographics, study design
characteristics, and genotypic data. For each association, we assessed the magnitude and heterogeneity of genetic effects
within and across CAD phenotypes, using meta-analytic methodologies. A total of 965 individual studies investigating
32 distinct variants in 22 genes were included, from which we grouped CAD phenotypes into 3 categories: acute
coronary syndromes (ACS) (426 [44%] studies); angiographically documented disease (323 [34%] studies); and broad,
not otherwise specified CAD (216 [22%] studies). These clinical phenotypes were overlapping. Subgroup meta-analyses
by phenotype showed discordant results, but phenotypic classification generally explained small proportions of
between-study heterogeneity. Differences between phenotypic groups were minimized for associations with robust
statistical support. No CAD phenotype was consistently associated with larger or more homogeneous genetic effects in
meta-analyses.
Conclusions—Substantial phenotypic heterogeneity exists in CAD genetic associations, but differences in phenotype
definition make a small contribution to between-study heterogeneity. We did not find a consistent effect in terms of the
magnitude or homogeneity of summary effects for a specific phenotype to support its preferential use in genetic studies
or meta-analyses for CAD. (Circ Cardiovasc Genet. 2011;4:58-67.)
Key Words: coronary artery disease 䡲 myocardial infarction 䡲 meta-analysis 䡲 genetic association studies
䡲 phenotype 䡲 population characteristics
G
enetic association studies have improved our understanding of the underlying etiology of coronary artery
disease (CAD), although key challenges remain in this
research field.1–3 Most gene-CAD associations proposed by
candidate-gene studies have not been replicated by subsequent, well-powered studies.4 Findings from agnostic
genome-wide association studies (GWAS) have robust statistical support and a strong replication record; nevertheless, the
identified variants confer small increments in risk and explain
only a small proportion of the phenotypic variance of measured coronary atherosclerosis (stenotic lesions and extraluminal calcification) or occurrence of acute cardiac events.5,6
Several biological, methodological, and analytic factors have
been proposed to explain both the failures of the candidate-
gene approach and the missing heritability in GWAS.1,3,6
Because CAD is a clinically heterogeneous entity, the issue of
phenotypic heterogeneity may be of particular relevance to
the study of CAD genetics.
Editorial see p 7
Clinical Perspective on p 67
Problems with phenotypic characterization in CAD are
well recognized.3,7 CAD can manifest with acute events, such
as myocardial infarction (MI), or with chronic, stable symptoms of ischemia; thus, CAD case definitions can vary by
using different clinical or imaging (angiographic) diagnostic
criteria. Furthermore, identifying control subjects is difficult
because CAD can exist during a prolonged asymptomatic
Received May 25, 2010; accepted November 29, 2010.
From the Institute for Clinical Research and Health Policy Studies (G.D.K., I.J.D., T.A.T., C.H.S., D.M.K.), Tufts Medical Center and Sackler School
of Graduate Biomedical Sciences (G.D.K., I.J.D., C.H.S., G.S.H., D.M.K.), Tufts University; and MCRI Center for Translational Genomics (G.S.H.),
Molecular Cardiology Research Institute, Tufts Medical Center, Boston, MA.
The online-only Data Supplement is available at http://circgenetics.ahajournals.org/cgi/content/full/CIRCGENETICS.110.957738/DC1.
Correspondence to Dr. Georgios Kitsios, MD, PhD, Institute for Clinical Research and Health Policy Studies, Tufts Medical Center, 800 Washington
St, #63, Boston, MA 02111. E-mail [email protected]
© 2011 American Heart Association, Inc.
Circ Cardiovasc Genet is available at http://circgenetics.ahajournals.org
58
DOI: 10.1161/CIRCGENETICS.110.957738
Kitsios et al
Downloaded from http://circgenetics.ahajournals.org/ by guest on June 15, 2017
phase and significant atherosclerosis can be missed by coronary angiography and stress testing. In response to the
heterogeneous manner in which CAD can manifest, standardized phenotypic definitions (eg, MI and angiographic measures of disease burden) have been recommended for all
genetic analyses.2,6 Although use of these “purer” phenotypes
can be justified theoretically, there currently is no empirical
evidence for improved sensitivity in detecting genetic effects
compared with more heterogeneous CAD phenotypic definitions. Notably, use of more restrictive definitions is not
without costs in terms of both the burdens of data collection
and the compromised sample sizes.
The complexity of the CAD phenotype poses additional
challenges for meta-analysis, which is a natural framework
for evaluating replication of genetic associations and exploring reasons for inconsistencies among synthesized studies.8
Genetic meta-analyses for CAD typically have included
studies with varying phenotypic definitions, and statistical
heterogeneity is commonly present.9 However, to our knowledge, there has not been a systematic evaluation of the
quantitative consequences of phenotypic heterogeneity, that
is, whether between-study statistical heterogeneity can be
attributed to phenotypic differences among synthesized studies and whether meta-analysis of studies with phenotypically
homogeneous characteristics would result in stronger or more
homogeneous effects.
We addressed these questions in a large-scale empirical
evaluation of genetic association studies for the most extensively studied associations in CAD. Our study aims were to
first describe the degree of heterogeneity of phenotypes in
individual genetic association studies and then to investigate
the impact of different phenotypic definitions on the magnitude and heterogeneity of genetic associations for CAD.
Methods
Study Sample and Data Extraction
All genetic association studies that had been included in large
meta-analyses (quantitative syntheses of at least 15 primary studies)
investigating the association between genetic variants and susceptibility to CAD were eligible for inclusion. The meta-analyses were
identified by systematic searches in the Phenopedia database of the
HuGE (Human Genome Epidemiology) Navigator10 up to June 2009
(more details provided in online-only Data Supplement Tables 1 and
2). We subsequently retrieved and reviewed all primary genetic
association studies included in the eligible meta-analyses. From each
publication, we extracted information on study design, population
demographics, sampling strategy, phenotypic characteristics of cases
and controls, genotypic distributions, and risk estimates for primary
and secondary analyses. When necessary, we supplemented information
from the corresponding meta-analysis publication. Data extraction was
performed by 2 investigators (G.D.K. and I.J.D.), and all extracted data
were cross-validated against the published information in the meta-analyses. Discrepancies were resolved through consensus.
Definition of Disease Phenotypes
We classified each study’s phenotypic definition into 3 prespecified,
mutually exclusive operational categories, which represent 3 distinct
sampling strategies during subject enrollment:
1. Acute coronary syndromes (ACS phenotype), including MI
(fatal and nonfatal) and unstable angina according to standard
criteria (World Health Organization or universal definition
criteria).11
Phenotypic Heterogeneity in Coronary Disease
59
2. Angiographically documented disease (angiographic phenotype) defined as presence of stenosis above a certain threshold
on a major epicardial artery on angiography.
3. Broadly defined CAD (broad phenotype) on the basis of
varying clinical criteria and history of disease, or not otherwise
specified, thus forming a catch-all category.
We also evaluated each study for additional case and control
phenotypic characteristics and study design aspects7,12–14:
a. Premature disease: study populations or subgroups that used
cut-off inclusion criteria below age 55 years for men and 65
years for women.15,16
b. Age matching of controls to cases: used as a measure to avoid
misclassification of subjects with latent, age-related disease.7
c. Angiographic controls: angiographic documentation of absence of disease in controls; performed to avoid misclassification of subjects with developed but asymptomatic disease.7
d. Retrospective or prospective design: if a genetic variant not
only is associated with susceptibility to CAD, but also adversely influences survival, then the variant allele may be
underrepresented in the survivors recruited in retrospective
studies (case control or cross-sectional). Prospective studies
(cohort or nested case control) would be less prone to such
survival bias.
Statistical Analysis
We described the variability of phenotypes in individual genetic
association studies by summarizing their characteristics overall as
well as by phenotypic subgroup. We compared subgroups with
parametric and nonparametric tests, as indicated. Further, we calculated the proportion of studies that satisfied the previously reported
recommendations for phenotypic characterization of cases and
controls.7
We then used meta-analytic techniques to evaluate whether
different phenotypic definitions of the disease can result in summary
associations with differential magnitude and heterogeneity of genetic
effects. The hypothesis tested here is that restrictive definitions (ie,
ACS and angiographic) will have summary genetic effects of
different magnitude (eg, larger odds ratios [ORs]) and smaller extent
of statistical heterogeneity compared with the broad definition.
Data from all genetic association studies available for each genetic
variant were resynthesized (all-inclusive meta-analyses), and the
statistical significance (P⬍0.05) of the summary ORs was recorded.
Our analyses focused on those associations with statistically significant summary ORs because these associations are presumably valid
and can be examined for the impact of phenotypic definitions. We
also separately examined the results from associations meeting more
stringent criteria for statistical significance (P⬍0.01 for the summary
OR). All associations are expressed as ORs with their corresponding
95% CIs. Summary ORs were estimated using the DerSimonian and
Laird random effects model.15 Between-study heterogeneity was
tested with Cochran Q (considered statistically significant at
P⬍0.10) and quantified with the I2 statistic (an estimate of how
much of the heterogeneity is unlikely to be due to chance).15
Subgroup Meta-Analyses for
Phenotypic Definitions
Each of the all-inclusive meta-analyses was stratified in 3 subgroup
meta-analyses according to the phenotypic definition used by
component studies (ACS, angiographic, or broad). Summary
effect estimates and heterogeneity statistics were calculated for
each subgroup.
Subgroup Comparisons of Magnitude of Effects
We examined whether the phenotypic subgroup-specific effects
differed beyond what is expected by chance by calculating the
relative OR (ROR) and its 95% CI for the summary effects between
the subgroups.16 The ROR expresses how much larger the summary
genetic effect is in 1 phenotypic subgroup compared to another. For
60
Circ Cardiovasc Genet
February 2011
Downloaded from http://circgenetics.ahajournals.org/ by guest on June 15, 2017
Figure 1. Flow chart of selected meta-analyses and genetic association studies and studies excluded, with specification of reasons.
example, when comparing the genetic effects for ACS versus broad
phenotypes, ROR ⬎1 means that a genetic variant has a stronger
genetic effect in studies where the disease is defined as ACS
compared with studies in which broad phenotypes are used.
Subgroup Comparisons of Heterogeneity
I2 values obtained from each subgroup were compared with the
Mann-Whitney U test. We also examined how much of the
observed statistical heterogeneity could be explained by consid-
Kitsios et al
Phenotypic Heterogeneity in Coronary Disease
61
Table 1. Descriptive Characteristics of Genetic Association Studies Included in the 32 Meta-Analyses,
Overall and Stratified by Phenotypic Subgroup
Phenotypes
All Studies
Studies
965
ACS
Angiographic
Broad
426 (44)
323 (33)
216 (22)
811 (84)
392 (92)
280 (87)
139 (64)
87 (9)
3 (1)
43 (13)
21 (1)
P*
⬍0.01
Study design
Case-control
Cross-sectional
Cohort
46 (5)
10 (2)
0
36 (17)
Nested case-control
40 (4)
20 (5)
0
20 (9)
Demographics
Sample size of cases
200 (106–427)
200 (103–432)
218 (119–454)
162 (83–352)
⬍0.01
Age of cases, y
55.9⫾7.4
54.8⫾7.9
56.7⫾6.0
56.9⫾8.0
⬍0.01
Male cases, %
77.1⫾21.1
77.9⫾22.6
79.8⫾16.7
71.6⫾22.3
⬍0.01
⬍0.01
Ethnicity
Downloaded from http://circgenetics.ahajournals.org/ by guest on June 15, 2017
White
645 (67)
309 (73)
204 (63)
132 (61)
East Asian
159 (16)
53 (12)
68 (21)
38 (18)
Other
161 (17)
64 (15)
51 (16)
Sample size of controls
Age of controls, y
46 (21)
⬍0.01
211 (121–500)
240 (130–523)
197 (120–338)
253 (117–692)
52.6⫾8.9
51.9⫾9.2
52.4⫾7.9
53.8⫾9.8
89 (21)
44 (14)
30 (14)
0.01
53.3⫾19.6
54.9⫾20.6
0.34
0.48
Phenotypic characteristics
Premature disease assessed
Patients with ACS, %
163 (17)
54.1⫾12.9
100
Studies with angiographic controls
118 (12)
9 (2)
103 (32)
6 (3)
⬍0.01
Age-matched controls
322 (33)
212 (50)
70 (22)
40 (19)
⬍0.01
Blinding
225 (23)
122 (29)
66 (20)
37 (18)
⬍0.01
HWE†
721 (75)
322 (77)
250 (77)
149 (69)
0.07
Study quality characteristics
Data are presented as no. (%), median (IQR), or mean⫾SD. HWE indicates Hardy-Weinberg equilibrium.
*P values for comparisons across the 3 phenotypic definitions by using ␹2, Mann-Whitney, and Kruskal-Wallis tests, as appropriate.
†Studies in which full genotypic distribution was available for the control group, and HWE was met (P⬎0.05 based on a ␹2 test).
ering the phenotypic subgroups. This analysis was done with the
Q statistic partitioning technique as follows: The overall heterogeneity of the all-inclusive meta-analysis (QALL) was partitioned
into that which could be explained by differences between
subgroups (Qphen) and that which remained unexplained within
the subgroups (Qbroad, Qangiographic, QACS). Then, the heterogeneity explained by the differences between the subgroups was
calculated as Qphen⫽QALL⫺(Qbroad⫹Qangiographic⫹QACS) and was
compared with critical values of the ␹2 distribution with 2 degrees of
freedom.17
Secondary Analyses
The additional study characteristics of interest were examined for
their potential impact on genetic effects in additional subgroup
analyses. Complementary to all aforementioned subgroup analyses,
we performed meta-regression analyses that provided similar results
(shown in online-only Data Supplement). Finally, we conducted
exploratory statistical power calculations for the comparisons of
magnitude of effects under a range of representative scenarios.18
For all comparisons, except those for heterogeneity, statistical
significance was defined as P⬍0.05. All tests were 2 sided. Statistical analyses were performed with Stata (StataCorp; College Station, TX), SAS (SAS Institute Inc; Cary, NC), and Meta-Analyst
(Tufts Medical Center; Boston, MA) software.19
Results
Synopsis of Primary Study Characteristics
The HuGE Navigator search identified 71 titles that met the
search criteria (Figure 1). Of those, 19 articles (online-only
Data Supplement Table 1) describing 32 meta-analyses for 22
distinct genes were considered eligible for analysis, resulting
in 965 genetic association studies (meta-analytic strata) that
were finally included in our analyses (Figure 1 and onlineonly Data Supplement Table 2). Summary characteristics of
these genetic association studies are presented in Table 1.
Most (93%) studies were retrospective in design, had small
samples, and examined populations that consisted mainly of
non-Hispanic whites and male participants. When comparing
summary study characteristics against the previously reported
recommendations for phenotypic characterizations,7 we
found that the majority of studies in the field did not follow
most of the recommendations (Table 2).
On the basis of the case sampling strategy used in the
primary analysis of each study, the 965 genetic association
studies were classified into 3 phenotypic subgroups: 426
(44%) studies were classified into the ACS subgroup, 323
(33%) into the angiographic subgroup, and 216 (22%) into
the broad subgroup. The ACS subgroup comprised almost
exclusively MI populations because ⬍10% of these studies
included patients with unstable angina and no history of MI.
The angiographic subgroup included cases in which a coronary stenosis was detected on conventional angiography in at
least 1 epicardial vessel. The 50% stenosis cut-off was used
62
Circ Cardiovasc Genet
February 2011
Table 2. Evaluation of Compliance to the Recommendations
for Phenotypic Characterizations for All Genetic Association
Studies Reviewed
Studies Meeting
Recommendations for
Phenotype Categorizations
Definition of cases with CAD
Angiographic stenosis of ⱖ70% in a major
epicardial artery
Family history of CAD
Lack of risk factors
Definition of cases with MI
Angiographic stenosis of ⱖ70% in a major
epicardial artery
Family history of CAD
Lack of risk factors
Downloaded from http://circgenetics.ahajournals.org/ by guest on June 15, 2017
MI documented of by ECG, enzymes
Definition of controls
Normal coronary arteries by selective coronary
angiography or multidetector CT (⬍10%
narrowing)
No family history of CAD
No history of cerebrovascular or peripheral
artery disease
Age much greater than cases by 10–20 years
539 (100)*
43 (8)
6 (1)
17 (3)†
426 (100)
0 (0)‡
5 (1)
14 (3)†
426 (100)
965 (100)
118 (12)§
69 (7)
779 (81)储
0 (0)¶
Data are presented as no. (%).
*Studies that used the broad or angiographic definitions.
†The definitions of low-risk subgroups (eg, lack of diabetes or normal lipid
levels) were varying across these studies.
‡Angiographic evaluation was not used as an inclusion criterion in ACS
studies.
§All 118 studies used coronary angiography. (Multidetector CT was not used
by any study.)
储One hundred eighteen (19%) studies included patients free from overt
cardiovascular disease but with cardiovascular risk factors (eg, diabetes,
hypertension, obesity, dialysis) or undergoing selective coronary angiography
for evaluation of atypical chest pain or other indications.
¶In 30 (3%) studies, the mean age of control subjects exceeded the mean
age of cases by 5 years. No studies with a 10-year difference were identified.
by the majority of studies, with only 43 (13%) limiting
enrollment to patients with more severe stenosis (ⱖ70%).
Angiographic verification of control status was performed in
32% of the angiographic subgroup studies. The broad sub-
group included variable combinations of patient populations
with a history of MI, revascularization (angioplasty or bypass
surgery), positive exercise test findings or baseline ischemic
electrocardiographic findings, symptomatic angina, fatal
CAD with autopsy findings, or history of hospitalization for
CAD-related diagnoses. Genotypic information for the components of the broad definition was commonly unavailable.
Study design characteristics and demographics of participants were found to be significantly different across the 3
phenotypic subgroups (Table 1). Although these phenotypic
subgroups comprised different study samples and, thus, did
not overlap, more than half of the patients included in studies
with angiographic and broad definitions suffered from ACS
as well (Figure 2, Table 1). In 120 studies with angiographic
or broad definitions, detailed genotypic information was
available for patient subgroups stratified by history of ACS.
We found that the genotypic distribution between these
clinically nonoverlapping patient subgroups was significantly
different (␹2 P⬍0.05) in a small number of studies (n⫽19;
16%).
Analysis of Phenotypic Effects With
Meta-Analytic Techniques
All-Inclusive Meta-Analyses
We resynthesized results from all eligible genetic association
studies with a random-effects model and found statistically
significant effects in 19 meta-analyses, whereas no significant
signal was found in the remaining 13 meta-analyses (onlineonly Data Supplement Tables 3 to 6). The 19 associations
with statistically significant summary ORs involved 15
genes: 5 (APOB, LPL, PON1, CETP, APOE) belong to the
lipid metabolism pathway; 4 are involved in thrombotichemostatic functions (SERPINE1, F13A1, F5, F2); 4 relate to
endothelial dysfunction (NOS3, ESR1, MTHFR, ACE); 1
relates to inflammation (CD14); and 1 genetic variant (9p21
locus) has unknown functions.1,20
Subgroup Meta-Analyses
Results from subgroup analyses are shown in Figure 3 and
online-only Data Supplement Table 3. Statistically significant
ORs are present across phenotypic subgroups, with only 2
(11%) of the 19 associations (9p21-rs1333049 and ACErs4340) being significant in all examined subgroups. The
Figure 2. Venn diagram showing the volume of
studies belonging to each one of the phenotypic sampling strategies (ACS, 426 studies;
angiographic definition, 323 studies; broad definition, 216 studies) and the extent of conceptual overlap (in terms of patients with ACS)
between them. The extent of overlap between
the angiographic and broad definitions (ie,
patients with qualifying angiographic lesions in
the broad definition) is not known.
Kitsios et al
Phenotypic Heterogeneity in Coronary Disease
63
Downloaded from http://circgenetics.ahajournals.org/ by guest on June 15, 2017
Figure 3. Forest plots indicating meta-analyses summary results (ORs and 95% CIs) for all studies included and for phenotypic definition subgroups. The symbols of the point estimates indicate the major pathway where each gene belongs as follows: e indicates lipid
metabolism; f, thrombosis-hemostasis; E, endothelial dysfunction; F, inflammation; ⫻, unknown mechanism.
angiographic subgroup had the highest proportion of significant ORs (10/19; 53%); all significant ORs in this subgroup
were observed for lipid metabolism and endothelial dysfunction genes. For the ACS subgroup, significant ORs were
found in 8 (42%) associations. The ACS phenotype has
significant ORS in 3 of the 4 associations in the thrombotichemostatic pathway genes. The broad phenotype had fewer
statistically significant results (31%), but this could result
from diminished statistical power because smaller numbers of
component studies investigated this phenotype.
Subgroup Comparisons of Magnitude of Effects
We observed statistically significant RORs in 10 (18%) of the
57 comparisons performed (online-only Data Supplement
Table 4), which corresponded to 5 associations (LPL-rs320,
F13A1-rs5985, NOS3-rs2070744, ESR1-rs2234693, and
CD14-rs2569190). In 3 of these associations (LPL-rs320,
F13A1-rs5985, NOS3-rs2070744), the significant RORs were
accounted for by a different directionality of the OR in the
broad subgroup compared to the other subgroups (Figure 4).
The expected genetic effects difference of the restrictive
phenotypes (ACS or angiographic) versus the broad phenotype was not supported by our results; almost half of the
significant RORs for these comparisons were ⬍1. Apart from
LPL-rs320, none of the other lipid metabolism genes had a
significant ROR when the angiographic phenotype was compared to the broad phenotype.
Subgroup Comparisons of Heterogeneity
Pair-wise comparisons of the I2 statistics obtained from the 3
subgroup analyses were not statistically significant (all
Mann-Whitney P⬎0.05), showing that no phenotypic subgroup was found to have a consistently smaller extent of
statistical heterogeneity. The Q test partitioning analysis for
the 3 phenotypic subgroups showed statistically significant
between-subgroup heterogeneity in 9 (47%) of the 19 metaanalyses; however, this between-subgroup heterogeneity explained small proportions of the overall between-study variance (Figure 5, online-only Data Supplement Table 4).
In the sensitivity analysis of 12 associations meeting more
stringent criteria of statistical significance (P⬍0.01) (onlineonly Data Supplement Table 7), we found that phenotypespecific effects were less common. Significant RORs were
found only for NOS3-rs2070744 and the Q partitioning
analysis was statistically significant in 5 (42%) associations.
Secondary Analyses
Detailed results from secondary analyses are presented in
online-only Data Supplement tables 7 to 15. None of the
examined variables was found to be a strong driver of
statistical heterogeneity across all examined associations. The
presumed superiority of premature disease to detect stronger
genetic effects was confirmed for only for 2 of the 11
examined associations (MTHFR-rs1801133 and 9p21rs1333049) (online-only Data Supplement Table 10). Most of
the analyses for control group characteristics were nonsignificant (online-only Data Supplement Tables 12 and 13). The
analysis of study designs showed, with the exception of
9p21-rs1333049, that synthesis of prospective studies resulted in nonsignificant summary ORs. The comparisons of
effect sizes between prospective and retrospective studies
64
Circ Cardiovasc Genet
February 2011
Downloaded from http://circgenetics.ahajournals.org/ by guest on June 15, 2017
Figure 4. Forest plots indicating the
RORs and corresponding 95% CIs for
the between-subgroup comparisons. The
symbols of the point estimates indicate
the major pathway where each gene
belongs as follows: e indicates lipid metabolism; f, thrombosis-hemostasis; E,
endothelial dysfunction; F, inflammation;
⫻, unknown mechanism.
were consistent with a weaker genetic effect in the prospective ones (online-only Data Supplement Table 14). Statistical
power was generally limited for the detection of small RORs
(online-only Data Supplement Table 15).
characteristics (premature disease, control definitions, prospective design), but none of these factors was found to be
a strong driver of statistical heterogeneity; nevertheless,
we found evidence that prospective studies are associated
with attenuated genetic effects for certain associations.
Discussion
Novel Findings
Our systematic evaluation of phenotypic heterogeneity in the
field of CAD identified extensive variability in the way that
the disease is defined at the individual study level. We
observed a broad spectrum of inclusion criteria in terms of
type of disease (acute or stable, recent onset or chronically
present, fatal or not), severity of disease (eg, various thresholds of angiographic stenosis), age at onset (from the second
to the eighth decade of life), health status of controls (ranging
from neonates and healthy blood donors to patients with
cardiovascular risk factors undergoing invasive angiography),
and other design aspects.
Our work provides a large-scale empirical evaluation of
the consequences of this phenotypic heterogeneity. We
found evidence that the phenotypic sampling strategy used
by individual studies accounts for a small proportion of the
observed statistical heterogeneity in certain meta-analyses.
The effect size estimates of subgroup analyses according
to phenotype often were discordant between subgroups in
terms of direction of effects and statistical significance.
However, we failed to identify a specific sampling strategy
that has consistently larger genetic effects or produces less
heterogeneous results. We examined additional study-level
Limitations of Clinical Phenotypes
Although CAD is common with advanced age, only a fraction
of vulnerable patients will ever develop an acute event.21,22
Dissecting the phenotype of MI/ACS from underlying CAD
has been recommended on the basis of clinical differences
between the 2 entities, pathophysiological dissimilarities, and
higher heritability estimates for the MI phenotype.7,21,23
Phenotypic homogeneity of MI is elusive because of the
diverse manner of presentation, such as ST-elevation and
non-ST-segment elevation MIs that may be attributed to
distinct pathophysiological mechanisms,24 and survival bias.7
On the other hand, the angiographic criteria for CAD7 are
problematic because the coronary angiography cannot
quantify the accumulation of atherosclerosis within arterial
walls.25
We found considerable conceptual overlap among the
populations from 3 different sampling strategies (Figure 2),
which may be explained by the natural history of atherosclerosis across the continuum from fatty streaks to vulnerable
plaques or fixed blockages.22 Early stages of atherosclerosis
(ie, lipid oxidation, macrophage infiltration, or platelet reactivity) may be more directly influenced by genetic determinants. Such quantitative intermediate phenotypes for CAD
Kitsios et al
Phenotypic Heterogeneity in Coronary Disease
65
Downloaded from http://circgenetics.ahajournals.org/ by guest on June 15, 2017
Figure 5. Bar diagram showing the
results of the Q partitioning analyses for
the major phenotypic subgroups. The
length of each bar corresponds to the Q
statistic observed in the all-inclusive
meta-analysis. Each bar is partitioned to
the within-subgroup Q statistics (QACS,
Qbroad, Qangiographic) and the remaining
between subgroup heterogeneity (Qphen).
*Statistically significant results for the
Qphen.
(eg, cholesterol levels, blood pressure, coronary calcium
quantity) could serve as proximate and more sensitive physiological markers of genetic effects.26,27 Pathophysiologically
distinct clinical definitions of CAD (eg, acute versus chronic)
also may be useful for identifying the genomic culprits of
different disease processes (eg, CAD onset and progression or
ACS precipitation).28,29 Despite the theoretical advantage of
such possible definitions, we identified a paucity of relevant
analyzable data in the available literature.
Implications for Future Research
Regarding the design of future meta-analyses, we demonstrated that the totality of evidence for CAD phenotypes
should be considered. Subgroup phenotypic analyses have
reduced statistical power, the majority of phenotype-specific
genetic effects do not typically differ beyond chance, and
patient groups show extensive overlap. Phenotypic subgrouping may be informative as an exploratory secondary analysis
because phenotype-specific effects, when present, are
unpredictable.
Most of the associations analyzed herein have emerged
from candidate-gene studies, a body of evidence with
well-recognized limitations.3 Nevertheless, the recommendations to favor particular CAD definitions in discovery
data sets or replication studies are not supported by the
available evidence. This is further reinforced by the fact
that we did not find evidence of phenotype-specific effects
for the 2 variants (9p21-rs1333049 and ACE-rs4340) with
the strongest statistical support (P⬍10⫺8). The 9p21 locus
variant that was discovered by GWAS did show stronger
effects for premature disease (ROR, 1.13; 95% CI, 1.05 to
1.21), in agreement with a recent analysis.30 When conducted on such definitive association findings, phenotypic
heterogeneity analyses are expected to be very informative and
will be worth pursuing once adequate evidence from GWAS
accumulates.
66
Circ Cardiovasc Genet
February 2011
Study Limitations
Downloaded from http://circgenetics.ahajournals.org/ by guest on June 15, 2017
This project analyzed study-level rather than individual
patient-level data. Although significant findings can generate
hypotheses to be tested in well-designed primary studies, we
cannot rule out competing explanations for significant findings by our study design. Selective reporting can hamper our
analyses if investigators perform analyses by using different
phenotypic definitions of outcomes and report only the most
significant results.31 Additionally, differential publication
bias by phenotype definition may exist if reviewers or journal
editors have a positive predisposition to studies with clear
phenotypes and significant results. Because the prevalence of
such practices is unknown, all published evidence was taken
at face value. We have examined a large number of studylevel variables across many genetic associations, and thus,
some positive findings may be due to chance. The racial
diversity of included populations also may have influenced
the examined genetic associations, particularly when the
investigated variant is tagging the causal one in a racespecific haplotypic structure. However, the biological impact
of common variants has been shown to be usually consistent
across racial groups, and the populations analyzed were
mostly white.32
The lack of statistically significant findings for the majority of the phenotypic and the secondary subgroup comparisons may be due to a lack of statistical power for certain
comparisons, as illustrated by our exploratory power analyses
and the relatively broad CIs for many RORs. The identified
limited contribution of phenotypes to explain between-study
heterogeneity does not necessarily mean that there are no true
differences in the influences of genetic variants at the level of
the regulated pathophysiology. Such differences may exist,
but we were not able to detect them because of overlaps in
clinical definitions, small numbers of studies per metaanalytic stratum, and lack of patient-level data. Nevertheless, by compiling the largest compendium of studies for
CAD to date, we were able to detect significant results for
certain genetic associations, and we highlighted the limitations of the available literature.
Conclusions
Our analysis identified extensive phenotypic heterogeneity
among cohorts of patients with CAD studied to date, which
may contribute to the heterogeneity of genetic association
study results. However, we did not find a consistent effect in
terms of the magnitude or homogeneity of summary effects
for a specific phenotype to support its preferential use as a
phenotype for primary analyses. Phenotype testing can be
informative as a secondary analysis after an association
with the disease has been established.28 This project
provides an analytic framework for testing phenotypespecific effects in meta-analysis. Further empirical evidence
on emerging replicated GWAS findings and additional analyses
on primary association data in well-characterized cohorts have
the potential to highlight critical information for the impact of
phenotypes.
Sources of Funding
The project described was partially supported by award number
UL1RR025752 from the National Center for Research Resources.
The content is solely the responsibility of the authors and does not
necessarily represent the official views of the National Center for
Research Resources or the National Institutes of Health.
Disclosures
Dr Kitsios was a Pfizer-Tufts Medical Center fellow in clinical and
translational science. Dr Dahabreh is the recipient of a research
scholarship from the Maria P. Lemos Foundation.
References
1. Damani SB, Topol EJ. Future use of genomics in coronary artery disease.
J Am Coll Cardiol. 2007;50:1933–1940.
2. Dandona S, Roberts R. Genomic view of factors leading to plaque
instability. Curr Cardiol Rep. 2009;11:282–287.
3. Ginsburg GS, Shah SH, McCarthy JJ. Taking cardiovascular genetic
association studies to the next level. J Am Coll Cardiol. 2007;50:
930 –932.
4. Morgan TM, Krumholz HM, Lifton RP, Spertus JA. Nonvalidation of
reported genetic risk factors for acute coronary syndrome in a large-scale
replication study. JAMA. 2007;297:1551–1561.
5. Mayer B, Erdmann J, Schunkert H. Genetics and heritability of coronary
artery disease and myocardial infarction. Clin Res Cardiol. 2007;96:1–7.
6. Manolio TA, Collins FS, Cox NJ, Goldstein DB, Hindorff LA, Hunter DJ,
McCarthy MI, Ramos EM, Cardon LR, Chakravarti A, Cho JH, Guttmacher AE, Kong A, Kruglyak L, Mardis E, Rotimi CN, Slatkin M, Valle
D, Whittemore AS, Boehnke M, Clark AG, Eichler EE, Gibson G, Haines
JL, Mackay TF, McCarroll SA, Visscher PM. Finding the missing heritability of complex diseases. Nature. 2009;461:747–753.
7. Luo AK, Jefferson BK, Garcia MJ, Ginsburg GS, Topol EJ. Challenges
in the phenotypic characterisation of patients in genetic studies of
coronary artery disease. J Med Genet. 2007;44:161–165.
8. Schunkert H, Gotz A, Braund P, McGinnis R, Tregouet DA, Mangino M,
Linsel-Nitschke P, Cambien F, Hengstenberg C, Stark K, Blankenberg S,
Tiret L, Ducimetiere P, Keniry A, Ghori MJ, Schreiber S, El Mokhtari
NE, Hall AS, Dixon RJ, Goodall AH, Liptau H, Pollard H, Schwarz DF,
Hothorn LA, Wichmann HE, Konig IR, Fischer M, Meisinger C,
Ouwehand W, Deloukas P, Thompson JR, Erdmann J, Ziegler A, Samani NJ.
Repeated replication and a prospective meta-analysis of the association
between chromosome 9p21.3 and coronary artery disease. Circulation. 2008;
117:1675–1684.
9. Ntzani EE, Rizos EC, Ioannidis JP. Genetic effects versus bias
for candidate polymorphisms in myocardial infarction: case study
and overview of large-scale evidence. Am J Epidemiol. 2007;165:
973–984.
10. Yu W, Clyne M, Khoury MJ, Gwinn M. Phenopedia and Genopedia:
disease-centered and gene-centered views of the evolving knowledge of
human genetic associations. Bioinformatics. 2010;26:145–146.
11. Thygesen K, Alpert JS, White HD. Universal definition of myocardial
infarction. J Am Coll Cardiol. 2007;50:2173–2195.
12. Zintzaras E, Kitsios G. Identification of chromosomal regions linked to
premature myocardial infarction: a meta-analysis of whole-genome
searches. J Hum Genet. 2006;51:1015–1021.
13. Roberts R. Genetics of premature myocardial infarction. Curr Atheroscler
Rep. 2008;10:186 –193.
14. Zintzaras E, Raman G, Kitsios G, Lau J. Angiotensin-converting
enzyme insertion/deletion gene polymorphic variant as a marker of
coronary artery disease: a meta-analysis. Arch Intern Med. 2008;168:
1077–1089.
15. Trikalinos TA, Salanti G, Zintzaras E, Ioannidis JP. Meta-analysis
methods. Adv Genet. 2008;60:311–334.
16. Altman DG, Bland JM. Interaction revisited: the difference between two
estimates. BMJ. 2003;326:219.
17. Egger M, Smith GD, Altman DG. Systematic Reviews in Health Care:
Meta-Analysis in Context. London: BMJ Books; 2007.
18. Cafri G, Kromrey JD, Brannick MT. A SAS macro for statistical power
calculations in meta-analysis. Behav Res Methods. 2009;41:35– 46.
19. Wallace BC, Schmid CH, Lau J, Trikalinos TA. Meta-Analyst: software
for meta-analysis of binary, continuous and diagnostic data. BMC Med
Res Methodol. 2009;9:80.
20. Debette S, Seshadri S. Genetics of atherothrombotic and lacunar stroke.
Circ Cardiovasc Genet. 2009;2:191–198.
21. Topol EJ. Simon Dack Lecture. The genomic basis of myocardial
infarction. J Am Coll Cardiol. 2005;46:1456 –1465.
Kitsios et al
22. Asheikh-Ali AA, Kitsios GD, Balk EM, Lau J, Ip S. The vulnerable
atherosclerotic plaque: scope of the literature. Ann Intern Med. 2010;153:
387–395.
23. Topol EJ, Smith J, Plow EF, Wang QK. Genetic susceptibility to myocardial infarction and coronary artery disease. Hum Mol Genet. 2006;15:
R117–R123.
24. Di SR, Di B, V, Barsotti MC, Grigoratos C, Armani C, Dell’Omodarme
M, Carpi A, Balbarini A. Inflammatory markers and cardiac function in
acute coronary syndrome: difference in ST-segment elevation myocardial
infarction (STEMI) and in non-STEMI models. Biomed Pharmacother.
2009;63:773–780.
25. Topol EJ, Nissen SE. Our preoccupation with coronary luminology. The
dissociation between clinical and angiographic findings in ischemic heart
disease. Circulation. 1995;92:2333–2342.
26. Pan WH, Lynn KS, Chen CH, Wu YL, Lin CY, Chang HY. Using
endophenotypes for pathway clusters to map complex disease genes.
Genet Epidemiol. 2006;30:143–154.
Phenotypic Heterogeneity in Coronary Disease
67
27. Huang GH, Hsieh CC, Chen CH, Chen WJ. Statistical validation of
endophenotypes using a surrogate endpoint analytic analogue. Genet
Epidemiol. 2009;33:549 –558.
28. Horne BD, Carlquist JF, Muhlestein JB, Bair TL, Anderson JL. Association of variation in the chromosome 9p21 locus with myocardial
infarction versus chronic coronary artery disease. Circ Cardiovasc Genet.
2008;1:85–92.
29. Samani NJ, Schunkert H. Chromosome 9p21 and cardiovascular disease:
the story unfolds. Circ Cardiovasc Genet. 2008;1:81– 84.
30. Palomaki GE, Melillo S, Bradley LA. Association between 9p21
genomic markers and heart disease: a meta-analysis. JAMA. 2010;303:
648 – 656.
31. Williamson PR, Gamble C, Altman DG, Hutton JL. Outcome selection
bias in meta-analysis. Stat Methods Med Res. 2005;14:515–524.
32. Ioannidis JP, Ntzani EE, Trikalinos TA. ‘Racial’ differences in genetic
effects for complex diseases. Nat Genet. 2004;36:1312–1318.
CLINICAL PERSPECTIVE
Downloaded from http://circgenetics.ahajournals.org/ by guest on June 15, 2017
The clinically heterogeneous nature of coronary artery disease (CAD) has caused well-recognized limitations in the
phenotypic characterization of cases and controls for genetic studies of CAD. Experts in the field have recommended the
use of standardized phenotypic definitions (eg, myocardial infarction and angiographic measures of disease burden) for all
genetic analyses, although there currently is no empirical evidence to support the use of these “purer” phenotypes. In this
project, we described the degree of heterogeneity of phenotypic definitions in individual genetic studies and investigated,
with meta-analytic techniques, the impact of phenotypic definitions on the consistency and magnitude of genetic effects
for CAD. We analyzed 965 individual studies for 32 genetic associations and found that the CAD phenotypes could be
classified into 3 categories: acute coronary syndromes (44%); angiographically documented disease (34%); and broad, not
otherwise specified CAD (22%). However, these clinical phenotypes were overlapping. Subgroup meta-analyses by
phenotype showed discordant results, but phenotypic classification generally explained small proportions of between-study
heterogeneity. No CAD phenotype was consistently associated with larger or more homogeneous genetic effects in
meta-analyses to support its preferential use in genetic studies or meta-analyses for CAD. Our findings reinforce the need
of considering the totality of evidence for CAD phenotypes in future meta-analyses of genetic studies. Phenotypic-specific
effects at the clinical phenotype level may exist, but these should be explored in secondary analyses after an association
between a genetic marker and CAD has been established by an all-inclusive meta-analysis.
Heterogeneity of the Phenotypic Definition of Coronary Artery Disease and Its Impact on
Genetic Association Studies
Georgios D. Kitsios, Issa J. Dahabreh, Thomas A. Trikalinos, Christopher H. Schmid, Gordon S.
Huggins and David M. Kent
Downloaded from http://circgenetics.ahajournals.org/ by guest on June 15, 2017
Circ Cardiovasc Genet. 2011;4:58-67; originally published online December 13, 2010;
doi: 10.1161/CIRCGENETICS.110.957738
Circulation: Cardiovascular Genetics is published by the American Heart Association, 7272 Greenville Avenue,
Dallas, TX 75231
Copyright © 2010 American Heart Association, Inc. All rights reserved.
Print ISSN: 1942-325X. Online ISSN: 1942-3268
The online version of this article, along with updated information and services, is located on the
World Wide Web at:
http://circgenetics.ahajournals.org/content/4/1/58
Data Supplement (unedited) at:
http://circgenetics.ahajournals.org/content/suppl/2010/12/13/CIRCGENETICS.110.957738.DC1
Permissions: Requests for permissions to reproduce figures, tables, or portions of articles originally published
in Circulation: Cardiovascular Genetics can be obtained via RightsLink, a service of the Copyright Clearance
Center, not the Editorial Office. Once the online version of the published article for which permission is being
requested is located, click Request Permissions in the middle column of the Web page under Services. Further
information about this process is available in the Permissions and Rights Question and Answer document.
Reprints: Information about reprints can be found online at:
http://www.lww.com/reprints
Subscriptions: Information about subscribing to Circulation: Cardiovascular Genetics is online at:
http://circgenetics.ahajournals.org//subscriptions/
SUPPLEMENTAL MATERIAL
Heterogeneity of the Phenotypic Definition of Coronary Artery Disease and Its
Impact on Genetic Association Studies
Supplementary Table 1. References of included meta-analyses.
1. Bennet AM, Di AE, Ye Z, Wensley F, Dahlin A, Ahlbom A, Keavney B, Collins R, Wiman B,
de FU, Danesh J. Association of apolipoprotein E genotypes with lipid levels and coronary risk.
JAMA. 2007;298:1300-11.
2. Casas JP, Cavalleri GL, Bautista LE, Smeeth L, Humphries SE, Hingorani AD. Endothelial
nitric oxide synthase gene polymorphisms and cardiovascular disease: a HuGE review. Am J
Epidemiol. 2006;164:921-35.
3. Chiodini BD, Barlera S, Franzosi MG, Beceiro VL, Introna M, Tognoni G. APO B gene
polymorphisms and coronary artery disease: a meta-analysis. Atherosclerosis. 2003;167:355-66.
4. Keavney B, Danesh J, Parish S, Palmer A, Clark S, Youngman L, Delepine M, Lathrop M,
Peto R, Collins R. Fibrinogen and coronary heart disease: test of causality by 'Mendelian
randomization'. Int J Epidemiol. 2006;35:935-43.
5. Lewis SJ, Ebrahim S, Davey SG. Meta-analysis of MTHFR 677C->T polymorphism and
coronary heart disease: does totality of evidence support causal role for homocysteine and
preventive potential of folate? BMJ. 2005;331:1053.
6. Li W, Xu J, Wang X, Chen J, Zhang C, Sun K, Hui R. Lack of association between
lymphotoxin-alpha, galectin-2 polymorphisms and coronary artery disease: A meta-analysis.
Atherosclerosis. 2009;208:433-6.
7. Lluis-Ganella C, Lucas G, Subirana I, Escurriol V, Tomas M, Senti M, Sala J, Marrugat J,
Elosua R. Qualitative assessment of previous evidence and an updated meta-analysis confirms
lack of association between the ESR1 rs2234693 (PvuII) variant and coronary heart disease in
men and women. Atherosclerosis. 2009;207:480-6.
8. Sagoo GS, Tatt I, Salanti G, Butterworth AS, Sarwar N, van MM, Jukema JW, Wiman B,
Kastelein JJ, Bennet AM, de FU, Danesh J, Higgins JP. Seven lipoprotein lipase gene
polymorphisms, lipid fractions, and coronary disease: a HuGE association review and metaanalysis. Am J Epidemiol. 2008;168:1233-46.
9. Schunkert H, Gotz A, Braund P, McGinnis R, Tregouet DA, Mangino M, Linsel-Nitschke P,
Cambien F, Hengstenberg C, Stark K, Blankenberg S, Tiret L, Ducimetiere P, Keniry A, Ghori
MJ, Schreiber S, El Mokhtari NE, Hall AS, Dixon RJ, Goodall AH, Liptau H, Pollard H,
Schwarz DF, Hothorn LA, Wichmann HE, Konig IR, Fischer M, Meisinger C, Ouwehand W,
Deloukas P, Thompson JR, Erdmann J, Ziegler A, Samani NJ. Repeated replication and a
prospective meta-analysis of the association between chromosome 9p21.3 and coronary artery
disease. Circulation. 2008;117:1675-84.
10. Thompson A, Di AE, Sarwar N, Erqou S, Saleheen D, Dullaart RP, Keavney B, Ye Z,
Danesh J. Association of cholesteryl ester transfer protein genotypes with CETP mass and
activity, lipid levels, and coronary risk. JAMA. 2008;299:2777-88.
11. Tsantes AE, Nikolopoulos GK, Bagos PG, Vaiopoulos G, Travlou A. Lack of association
between the platelet glycoprotein Ia C807T gene polymorphism and coronary artery disease: a
meta-analysis. Int J Cardiol. 2007;118:189-96.
12. van der AD, Peeters PH, Grobbee DE, Roest M, Marx JJ, Voorbij HM, van der Schouw YT.
HFE mutations and risk of coronary heart disease in middle-aged women. Eur J Clin Invest.
2006;36:682-90.
13. Voko Z, Bereczky Z, Katona E, Adany R, Muszbek L. Factor XIII Val34Leu variant protects
against coronary artery disease. A meta-analysis. Thromb Haemost. 2007;97:458-63.
14. Wheeler JG, Keavney BD, Watkins H, Collins R, Danesh J. Four paraoxonase gene
polymorphisms in 11212 cases of coronary heart disease and 12786 controls: meta-analysis of 43
studies. Lancet. 2004;363:689-95.
15. Xu MQ, Ye Z, Hu FB, He L. Quantitative assessment of the effect of angiotensinogen gene
polymorphisms on the risk of coronary heart disease. Circulation. 2007;116:1356-66.
16. Ye Z, Liu EH, Higgins JP, Keavney BD, Lowe GD, Collins R, Danesh J. Seven haemostatic
gene polymorphisms in coronary disease: meta-analysis of 66,155 cases and 91,307 controls.
Lancet. 2006;367:651-8.
17. Zafarmand MH, van der Schouw YT, Grobbee DE, de Leeuw PW, Bots ML. The M235T
polymorphism in the AGT gene and CHD risk: evidence of a Hardy-Weinberg equilibrium
violation and publication bias in a meta-analysis. PLoS One. 2008;3:e2533.
18. Zhang HF, Zhong BL, Zhu WL, Xie SL, Qiu LX, Zhu LG, Wang Y, Lei L. CD14 C-260T
gene polymorphism and ischemic heart disease susceptibility: a HuGE review and meta-analysis.
Genet Med. 2009;11:403-8.
19. Zintzaras E, Raman G, Kitsios G, Lau J. Angiotensin-converting enzyme insertion/deletion
gene polymorphic variant as a marker of coronary artery disease: a meta-analysis. Arch Intern
Med. 2008;168:1077-89.
Supplementary Table 2. Summary characteristics of meta-analyses for genetic associations of CAD with genetic variants that were
selected from the literature review.
Selection of meta-analyses: we selected associations studied by large meta-analyses (≥15 included primary studies), since the number
of studies synthesized therein would enable meaningful stratified analyses and given that there is less uncertainty regarding the
validity of examined associations when a large volume of evidence is synthesized. The meta-analyses were identified by systematic
searches in the Phenopedia database of Human Genome Epidemiology (HuGE) Navigator up to June 2009 for disease terms related to
CAD and MI (Angina Pectoris; Angina, Unstable; Coronary Arteriosclerosis; Coronary Disease; Coronary Restenosis; Coronary
Stenosis; Coronary Thrombosis; Coronary Vasospasm; Microvascular Angina; Myocardial Infarction; Myocardial Ischemia;
Atherosclerosis) and for studies indexed as meta-analyses (Meta-analysis; HuGE Review; Pooled analysis). When multiple metaanalyses examining the same association were reported, the largest one was selected for our analyses. Meta-analyses that synthesized
only studies with a specific phenotype of the disease (e.g., MI only) and excluded studies with different phenotypic definitions (e.g.,
angina) were not eligible, because they are not informative for our study question; additionally, meta-analyses for outcomes other than
susceptibility to CAD (e.g., response to treatment, disease progression, prognosis) were also excluded. The full-texts of the retrieved
meta-analysis articles were reviewed to evaluate their eligibility for inclusion. Reference lists of retrieved articles were also handsearched for additional relevant meta-analyses not indexed by HuGE Navigator. From each meta-analysis, we extracted information
regarding the investigated gene/variant, the number of studies included, the summary estimate of association and heterogeneity
statistics, the statistical and inheritance models used for primary analysis, the inclusion of unpublished data and their availability in
published tables or supplementary material and the overall study conclusions. Whenever an article reported on multiple meta-analyses
using varying models of inheritance, we selected the one labeled as primary or main; if no main analysis was specified, we selected
the one with the maximal number of synthesized studies.
Statistically significant summary odds ratios (ORs) are shown in bold.
Meta-analysis First
author, Year
Q-test p-value
(Chi-square)
statistical
model
used
Inheritance
model for
main analysis
Unpublished
data
included
N of
studies
included
Gene
Variant (rs-number)
Summary OR
(95%CI)
Bennet,2007
APOE
e2 (rs7412 )
0.80 (0.70-0.90)
<0.001
72
RE
dominant
yes
Chiodini,2003
APOB
EcoRI (rs1042031 )
1.73 (1.19-2.50)
0.99 (4.4)
NA
FE*
recessive
no
15
Chiodini,2003
APOB
XbaI (rs693 )
1.19 (1.01-1.39)
0.08 (26.7)
NA
FE *
recessive
no
20
Chiodini,2003
APOB
Sp I/D (rs11279109 )
1.19 (1.05-1.35)
0.12 (28.7)
NA
FE *
recessive
no
22
Sagoo,2007
LPL
D9N (rs1801177)
1.33 (1.14-1.56)
ns
17
RE
dominant
yes
20
Sagoo,2007
LPL
HindIII (rs320)
0.89 (0.81-0.98)
ns
16
RE
dominant
yes
23
Sagoo,2007
LPL
S447X (rs328)
0.84 (0.75-0.94)
<0.01
50
RE
dominant
yes
26
Sagoo,2007
LPL
N291S (rs268)
1.07 (0.96-1.20)
ns
5
RE
dominant
yes
21
Sagoo,2007
LPL
PvuII (rs285)
0.96 (0.89-1.04)
ns
3
RE
dominant
yes
18
Thompson ,2008
CETP
TaqIB (rs708272 )
0.95 (0.92-0.99)
no
18
RE†
additive
yes
38
Wheeler,2004
PON1
Q192R (rs662)
1.12 (1.07-1.16)
<0.0001 (103.1)
NA
FE *
additive
no‡
35
Wheeler,2004
PON1
L55M (rs854560)
1.00 (0.95-1.06)
0.03 (25.5)
NA
FE*
additive
no‡
21
Voko,2006
F13A1
Val34Leu (rs5985 )
0.81 (0.70-0.92)
0.001
NA
RE
dominant
no
16
Ye,2006
F2
G20210A (rs1799963)
1.31 (1.12-1.52)
ns
27
FE *
additive
yes
40
Ye,2006
SERPINE1
5G/4G (rs1799889 )
1.06 (1.02-1.10)
<0.0001
57
FE *
additive
yes
37
Ye,2006
F5
G1691A (rs6025)
1.17 (1.08-1.28)
ns
17
FE *
additive
yes
60
Ye,2006
F7
G10976A (rs6046)
0.97 (0.91-1.04)
ns
0
FE *
additive
yes
24
I
2
Lipid Metabolism
17
Thrombosis-Hemostasis
Ye,2006
ITGB3
C1565T (rs5918 )
1.03 (0.98-1.07)
<0.001
38
FE *
additive
yes
43
Tsantes,2007
ITGA2
0.95 (0.84-1.08)
0.003 (37.7)
50
FE and RE
recessive
no
19
Keavney,2006
FGB
C807T (rs1126643)
C-148T/G-455A
(rs1800787/rs1800790 )
1.00 (0.95–1.04)
0.77 (15.2)
0
FE
additive
no
20
NOS3
Glu298Asp (rs1799983 )
1.17 (1.07-1.28)
0.00001 (124.7)
68
FE and RE
additive
no‡
42
‡
31
Endothelial Dysfunction
Casas,2006
Casas,2006
NOS3
4a/b (na)
1.12 (1.01-1.24)
0.00001 (67.2)
55
FE and RE
additive
no
Casas,2006
NOS3
T786C (rs2070744 )
1.17 (1.07-1.28)
0.0001 (56.4)
63
FE and RE
no‡
22
yes
80
Lewis ,2005
MTHFR
C677T (rs1801133 )
1.14 (1.05-1.24)
<0.001
38
RE
additive
extreme
homozygotes
Zafarmand,2008
AGT
M235T (rs699 )
1.08 (1.01-1.15)
<0.001
56
FE *
additive
no
38
Xu,2007
AGT
T174M (rs4762)
1.07 (0.93-1.22)
0.03
44
FE and RE
no‡
16
Zintzaras,2008
ACE
I/D (rs4340)
1.25 (1.16-1.35)
<0.01
NA
RE
no
108
Lluis-Ganella ,2009
ESR1
T-397C (rs2234693)
1.17 (1.00-1.32)
0.00003 (29.0)
NA
RE
additive
extreme
homozygotes
extreme
homozygotes
no
16
van der AD,2006
HFE
C282Y (rs1800562 )
1.03 (0.96-1.11)
0.11 (30.3)
NA
FE
dominant
no
19
Zhang,2009
CD14
C-260T (rs2569190 )
1.53 (1.20–1.96);
2.53 (1.84–3.50)§
0.37
NA
FE *
recessive
no
19
Li ,2009
LTA
A252G (rs909253)
1.00 (0.94-1.07)
0.007 (37.5)
49
FE
recessive
no
20
9p21.3
CG (rs1333049)
1.24 (1.20-1.29)
NA
NA
RE†
additive
no
17
Inflammation
Unknown
Schunkert,2008
Abbreviations: OR: odds ratio, CI: confidence interval, FE: fixed effects, RE: random effects, NA: non-available, ns: non-significant.
* When the Q-statistic was significant, the RE model was used.
†
FE was examined in sensitivity analysis.
‡
The meta-analysis did not include unpublished data but authors of individual studies were contacted in order to provide genotypic
distributions not extractable from the original publications.
§
The summary results were provided for studies in whites and East-Asians separately.
Supplementary Table 3. Meta-analyses summary results and heterogeneity metrics for all studies included and for phenotypic
definition subgroups. All meta-analyses had statistically significant results overall or for a particular phenotypic subgroup. Statistically
significant associations are shown in bold.
All phenotypes
Broad
ACS
PQall (I2)
N of studies
(N of
patients)
ORbroad (95%CI)
PQbroad
(I2)
1.74 (1.20, 2.52)
0.99 (0)
9 (1931)
1.64 (0.99, 2.71)
1.26 (1.09, 1.47)
0.33 (10)
2 (2199)
1.32 (0.97, 1.80)
22 (6115)
0.90 (0.81, 0.99)
0.17 (22)
1 (241)
1.73 (1.18, 2.28)
26 (8012)
0.84 (0.75, 0.93)
0.02 (41)
5 (2619)
0.78 (0.53, 1.15)
PON1 (rs662)
36 (20774)
1.15 (1.05, 1.25)
7 (2739)
1.20 (1.01, 1.44)
CETP (rs708272)
25 (44471)
0.94 (0.90, 0.98)
<0.01
(68)
0.08 (30)
6 (12489)
0.91 (0.74, 1.11)
APOE (rs7412)
20 (46907)
0.82 (0.72, 0.93)
<0.01
(68)
6 (25548)
SERPINE1 (rs1799889)
35 (23302)
1.08 (1.02, 1.15)
F13A1 (rs5985)
19 (15027)
0.87 (0.76, 0.99)
F2 (rs1799963)
39 (24720)
1.37 (1.12, 1.67)
<0.01
(55)
<0.01
(68)
0.11 (22)
4 (1581)
1.47 (0.81, 2.64)
F5 (rs6025)
62 (64840)
1.17 (1.06, 1.28)
0.10 (19)
4 (17697)
1.05 (0.83, 1.31)
NOS3 (rs1799983)
41 (25799)
1.17 (1.07, 1.28)
5 (3205)
1.40 (0.92, 2.14)
NOS3 (rs2070744)
21 (23418)
1.17 (1.07, 1.28)
2 (3145)
NOS3 (4a/b)
31 (19152)
1.14 (1.04, 1.26)
ESR1 (rs2234693)
16 (16367)
1.17 (1.01, 1.38)
MTHFR (rs1801133)
75 (23763)
1.11 (1.02, 1.20)
ACE (rs4340)
111
1.24 (1.15, 1.34)
<0.01
(68)
<0.01
(65)
<0.01
(54)
<0.01
(68)
<0.01
(37)
<0.01
Gene (rsnumber)
N of
studies (N
of patients)
ORall (95%CI)
APOB (rs1042031)
15 (3731)
LPL (rs1801177)
19 (10750)
LPL (rs320)
LPL (rs328)
Angiographic
N of studies
(N of
patients)
ORACS (95%CI)
PQACS
(I2)
N of studies (N
of patients)
ORangio (95%CI)
PQangio
(I2)
0.99 (0)
3 (435)
1.72 (0.59, 5.05)
0.46 (0)
3 (1365)
1.92 (1.01, 3.62)
0.57 (0)
0.96 (0)
9 (4787)
1.11 (0.95, 1.29)
0.44 (0)
8 (3764)
1.59 (1.15, 2.19)
0.39 (6)
NA
8 (3867)
0.89 (0.81, 0.99)
0.64 (0)
13 (2007)
0.87 (0.74, 1.01)
0.18 (26)
0.01 (72)
8 (2431)
0.88 (0.75, 1.03)
0.09 (44)
13 (2962)
0.80 (0.68, 0.95)
0.28 (17)
0.22 (28)
11 (6390)
1.06 (0.98, 1.15)
0.66 (0)
18 (11645)
1.19 (1.03, 1.37)
13 (20263)
0.95 (0.90, 1.01)
0.32 (13)
6 (11719)
0.93 (0.88, 0.98)
0.94 (0.76, 1.17)
<0.01
(72)
0.01 (70)
<0.01
(81)
0.89 (0)
8 (15912)
0.74 (0.65, 0.86)
0.13 (37)
6 (5447)
0.78 (0.56, 1.10)
<0.01
(77)
5 (3532)
1.23 (1.01, 1.50)
0.04 (59)
24 (16975)
1.07 (0.99, 1.15)
6 (2795)
1.05 (0.93, 1.18)
0.36 (9)
3 (3380)
1.14 (0.99, 1.31)
0.61 (0)
13 (8818)
0.84 (0.70, 0.99)
3 (2829)
0.77 (0.57, 1.04)
0.05 (67)
0.58 (0)
27 (18108)
1.39 (1.06, 1,81)
<0.01
(60)
<0.01
(70)
0.04 (34)
8 (5031)
1.33 (0.92, 1,92)
0.41 (2)
0.95 (0)
51 (41955)
1.23 (1.09, 1.39)
0.03 (29)
7 (5188)
1.03 (0.87, 1.21)
0.74 (0)
0.01 (72)
18 (10168)
1.11 (0.98, 1.25)
18 (12426)
1.20 (1.05, 1.37)
0.85 (0.71, 1.03)
0.96 (0)
10 (14417)
1.21 (1.06, 1.39)
9 (5856)
1.21 (1.06, 1.39)
<0.01
(76)
0.01 (61)
7 (3937)
1.25 (0.98, 1.60)
0.15 (36)
10 (5025)
1.18 (0.98, 1.41)
<0.01
(57)
<0.01
(66)
0.01 (60)
14 (10190)
1.08 (0.95, 1.24)
0.01 (53)
6 (11043)
1.11 (0.87, 1.41)
6 (4743)
0.97 (0.85, 1.10)
0.58 (0)
4 (581)
2.43 (1.59, 3.73)
0.33 (13)
23 (4639)
1.09 (0.93, 1.28)
<0.01
(77)
0.01 (47)
23 (7184)
1.14 (0.98, 1.32)
0.09 (29)
29 (11940)
1.10 (0.95, 1.26)
0.02 (38)
30 (31063)
1.20 (1.03, 1.41)
<0.01
34 (23374)
1.18 (1.06, 1.31)
0.01 (38)
47 (12141)
1.33 (1.16, 1.52)
<0.01
Lipid Metabolism
Thrombosis-Hemostasis
Endothelial Dysfunction
(66578)
(56)
(62)
(58)
Inflammation
CD14 (rs2569190)
22 (16157)
1.20 (1.04, 1.39)
<0.01
(65)
6 (2245)
1.68 (1.29, 2.20)
0.08 (48)
9 (4428)
1.06 (0.85, 1.33)
0.02 (56)
7 (9484)
1.06 (0.88, 1.28)
0.04 (55)
17 (41158)
1.26 (1.22, 1.31)
0.59 (0)
9 (28581)
1.25 (1.19, 1.32)
0.26 (21)
6 (9583)
1.28 (1.21, 1.36)
0.89 (0)
2 (2994)
1.34 (1.21, 1.49)
0.99 (0)
Unknown
9p21 (rs1333049)
Supplementary Table 4. Results from subgroup analyses: comparison of magnitude of effects
with relative odds ratios (ROR) and 95% confidence intervals (95% CI) and explanation of
heterogeneity with the Q partitioning technique for associations with statistically significant
summary ORs. All statistically significant results are shown in bold.
Gene (rsnumber)
Comparisons of magnitude of effects for phenotypic
subgroups
ROR (95%CI)
ROR (95%CI)
ROR (95%CI)
ACS vs Broad
Angiographic vs
ACS vs
Broad
Angiographic
Q partitioning for phenotypic subgroups
Heterogeneity
of the allinclusive
metaanalysis: Qall
Heterogeneity
within
subgroups
(Qbroad + Qacs +
Qangio)
Heterogeneity
between
subgroups:
Qall - (Qbroad +
Qacs + Qangio)
Heterogeneity
between
subgroupsp-value
(1.15 + 1.53 +
1.12)
(0.00 + 7.95 +
7.41)
(0.00 + 5.13 +
16.1)
(14.0 + 12.4 +
14.3)
(8.29 + 7.72 +
89.7)
(18.0 + 13.7 +
1.68)
(16.5 + 11.0 +
21.5)
0.1
0.93
4.6
0.09
5.6
0.06
1.2
0.55
2.0
0.36
0.9
0.65
9.8
<0.01
2.4
0.29
9.2
0.01
1.2
0.55
1.5
0.48
0.5
0.76
9.6
<0.01
5.2
0.07
18.6
<0.01
0.6
0.73
11.0
<0.01
Lipid Metabolism
APOB (rs1042031)
1.05 (0.32, 3.44)
1.16 (0.51, 2.62)
1.11 (0.31, 3.88)
3.9
LPL (rs1801177)
0.83 (0.59, 1.17)
1.19 (0.76, 1.86)
1.43 (1.00, 2.05)
19.9
LPL (rs320)
0.51 (0.29-0.89)
0.49 (0.28-0.88)
0.96 (0.80, 1.16)
26.9
LPL (rs328)
1.12 (0.74, 1.70)
1.02 (0.67, 1.56)
0.91 (0.72, 1.14)
42.0
PON1 (rs662)
0.88 (0.72, 1.07)
0.98 (0.78, 1.23)
1.11 (0.94, 1.31)
107.0
CETP (rs708272)
1.05 (0.85, 1.29)
1.02 (0.82, 1.26)
0.97 (0.89, 1.05)
34.3
APOE (rs7412)
0.78 (0.60, 1.01)
0.83 (0.55, 1.23)
1.05 (0.73, 1.51)
58.9
SERPINE1 (rs1799889)
0.86 (0.70, 1.06)
0.85 (0.67, 1.07)
0.98 (0.85, 1.13)
75.7
F13A1 (rs5985)
0.73 (0.58, 0.92)
0.67 (0.48, 0.94)
0.91 (0.64, 1.30)
55.6
F2 (rs1799963)
0.94 (0.49, 1.80)
0.90 (0.45, 1.81)
0.95 (0.60, 1.50)
93.6
F5 (rs6025)
1.17 (0.90, 1.51)
0.98 (0.74, 1.30)
0.83 (0.68, 1.02)
75.4
NOS3 (rs1799983)
0.79 (0.50, 1.23)
0.85 (0.54, 1.33)
1.07 (0.89, 1.29)
124.0
NOS3 (rs2070744)
1.41 (1.12, 1.79)
1.42 (1.12, 1.79)
1.00 (0.82, 1.21)
56.6
NOS3 (4a/b)
0.93 (0.69, 1.26)
0.86 (0.65, 1.13)
0.92 (0.73, 1.15)
64.7
ESR1 (rs2234693)
0.87 (0.66, 1.14)
2.19 (1.34, 3.58)
2.50 (1.60, 3.92)
47.2
MTHFR (rs1801133)
1.03 (0.83, 1.29)
1.00 (0.81, 1.23)
0.96 (0.78, 1.18)
117.9
ACE (rs4340)
0.98 (0.81, 1.18)
1.10 (0.89, 1.35)
1.12 (0.94, 1.33)
250.0
0.63 (0.44, 0.89)
0.62 (0.45, 0.87)
0.99 (0.74, 1.33)
60.2
(9.68 + 18.2 +
13.2)
19.0
<0.01
1.02 (0.94, 1.10)
1.06 (0.95, 1.19)
1.04 (0.92, 1.17)
14.1
(10.1 + 1.70 +
0.00)
2.3
0.32
Thrombosis-Hemostasis
(9.76 + 58.0 +
5.48)
(0.97 + 39.4 +
6.07)
(5.01 + 77.2 +
8.92)
(0.34 + 70.1 +
3.49)
Endothelial Dysfunction
(14.3 + 39.9 +
69.8)
(0.00 + 26.4 +
20.6)
(9.41 + 22.4 +
27.6)
(21.4 + 3.76 +
3.43)
(41.3 + 31.1 +
44.9)
(77.0 + 53.0 +
109.)
Inflammation
CD14 (rs2569190)
Unknown
9p21 (rs1333049)
Supplementary Table 5. Genetic associations with statistically non-significant summary odds ratios: meta-analyses summary results
and heterogeneity metrics for all studies included and for phenotypic definition subgroups.
All phenotypes
Gene (rsnumber)
Broad
ACS
Angiographic
N of
studies
(N of
patients)
ORall (95%CI)
PQall (I2)
N of
studies
(N of
patients)
ORbroad (95%CI)
PQbraod
(I2)
N of
studies (N
of
patients)
ORACS (95%CI)
PQACS (I2)
N of
studies
(N of
patients)
ORangio (95%CI)
PQangio
(I2)
21
(9324)
20
(6986)
21
(22316)
18
(11851)
20
(12416)
1.17 (0.95, 1.45)
0.02 (44)
9 (2400)
1.40 (0.96, 2.05)
0.29 (18)
5 (1917)
1.28 (0.95, 1.74)
0.39 (3)
7 (5007)
1.00 (0.70, 1.44)
1.20 (0.99, 1.46)
0.03 (40)
1.17 (0.78, 1.75)
1.15 (0.92, 1.45)
0.58 (0)
5 (2204)
1.15 (0.87, 1.53)
0.41 (4)
<0.01
(67)
0.10 (53)
5 (2152)
1.08 (0.96, 1.21)
9 (9082)
1.03 (0.90, 1.18)
0.94 (0)
8 (3072)
0.95 (0.73, 1.22)
0.50 (0)
0.95 (0.88, 1.02)
0.49 (0)
10
(2630)
4
(10162)
0 (0)
<0.01
(68)
0.83 (0)
5 (9292)
0.95 (0.83, 1.09)
0.29 (19)
0.96 (0.84, 1.10)
0.49 (0)
0.99 (0.93, 1.05)
0.54 (0)
5 (1999)
0.83 (0.64, 1.07)
5 (2612)
0.99 (0.85, 1.16)
0.32 (14)
13
(2559)
10
(7805)
1.01 (0.94, 1.09)
0.96 (0)
0.98 (0.87, 1.09)
0.43 (2)
0 (0)
9 (6650)
1.00 (0.82, 1.22)
0.19 (29)
6 (7341)
0.96 (0.83, 1.11)
0.70 (0)
0.98 (0.93, 1.04)
0.20 (21)
4 (8779)
1.02 (0.80, 1.30)
0.06 (59)
1.00 (0.95, 1.05)
0.83 (0)
4 (2131)
0.91 (0.67, 1.23)
0.03 (68)
1.01 (0.90, 1.12)
<0.01
(74)
0.55 (0)
7 (3111)
0.69 (0.45, 1.08)
1.07 (0.95, 1.20)
1.10 (0.92, 1.31)
0.01 (58)
1.03 (0.90, 1.18)
0.94 (0.84, 1.05)
<0.01
(60)
0.29 (14)
9 (7580)
6 (6540)
<0.01
(87)
0.94 (0)
12
(19980)
26
(21509)
15 (8708)
3 (2386)
0.92 (0.68, 1.24)
0.23 (32)
<0.01
(55)
<0.01
(81)
0.14 (25)
4 (9918)
0.97 (0.88, 1.06)
1.00 (0)
2 (1677)
1.28 (0.77, 2.11)
0.05 (75)
1.14 (0.92, 1.42)
9
(13461)
8
(44076)
1.13 (0.99, 1.28)
0.03 (54)
14 (7732)
1.01 (0.85, 1.21)
1.04 (0.95, 1.13)
0.35 (10)
11 (5156)
1.12 (0.93, 1.36)
<0.01
(85)
0.30 (15)
10
(8111)
17
(11473)
4 (1747)
0.67 (0.43, 1.06)
<0.01
(67)
<0.01
(83)
0.31 (16)
<0.01
(55)
6 (5082)
1.10 (0.80, 1.53)
0.09 (47)
10
(24331)
0.98 (0.85, 1.13)
0.02 (55)
4 (997)
1.51 (0.91, 2.49)
0.18 (39)
Lipid Metabolism
APOB (rs11279109)*
APOB (rs693)*
LPL (rs268)
LPL (rs285)
PON1 (rs854560)
1.66 (0.94, 2.94)
0.13 (43)
Thrombosis-Hemostasis
ITGA2 (rs1126643)
FGB
(rs1800787/rs1800790)
ITGB3 (rs5918)
F7 (rs6046)
15
(13991)
20
(30890)
42
(32200)
24
(17634)
0.98 (0.91, 1.05)
Endothelial Dysfunction
AGT (rs4762)
AGT (rs699)
HFE (rs1800562)
16
(19706)
40
(32666)
23
(50979)
1.09 (0.95, 1.24)
1.04 (0.95, 1.15)
1.03 (0.94, 1.13)
1.03 (0.87, 1.21)
Inflammation
LTA (rs909253)
20
(30410)
1.05 (0.92, 1.20)
* Our re-analysis differed from the previously published meta-analysis in that two associations (APOB-rs639 and APOB-rs11279109)
were no longer significant, which could be due to our use of statistical models that take between study variability into account
(random-effects) on associations that were of borderline statistical significance under fixed effects assumptions or by the fact that we
finally synthesized a slightly different sample of studies compared to the original meta-analyses.
Supplementary Table 6. Results from subgroup analyses for associations with statistically non-significant summary odds ratios (ORs):
comparison of magnitude of effects with relative ORs (ROR) and 95% confidence intervals (95% CI) and explanation of heterogeneity
with the Q partitioning technique. All statistically significant results are shown in bold.
Gene (rsnumber)
Comparisons of magnitude of effects for phenotypic
subgroups
ROR (95%CI)
ROR (95%CI)
ROR (95%CI)
ACS vs Broad
Angiographic vs
ACS vs
Broad
Angiographic
Q partitioning for phenotypic subgroups
Heterogeneity of
the all-inclusive
meta-analysis:
Qall
Heterogeneity
between
subgroups: Qall (Qbroad + Qacs +
Qangio)
Heterogeneity
between
subgroupsp-value
3.13
0.21
5.12
<0.01
0.04
0.98
0.17
0.92
2.97
0.23
(NA + 11.1 +
3.00)
(7.39 + 6.61 +
9.26)
(45.0 + 61.7 +
19.1)
(1.28 + 16.3 +
2.95)
0.11
0.95
31.2
<0.01
3.13
0.21
0.89
0.64
(0.03 + 3.99 +
27.0)
(17.2 + 85.5 +
92.8)
(7.79 + 11.7 +
3.57)
2.36
0.31
6.21
0.04
7.78
0.02
Heterogeneity
within
subgroups
(Qbroad + Qacs +
Qangio)
Lipid Metabolism
APOB (rs11279109)
0.91 (0.56, 1.49)
0.71 (0.42, 1.20)
0.78 (0.48, 1.25)
35.6
APOB (rs693)
0.98 (0.62, 1.56)
0.98 (0.60, 1.61)
1.00 (0.69, 1.44)
20.7
LPL (rs268)
0.61 (0.34, 1.11)
0.56 (0.30, 1.06)
0.91 (0.68, 1.22)
16.4
1.01 (0.83, 1.23)
31.7
1.01 (0.85, 1.21)
17.7
0.95 (0.74, 1.23)
14.3
LPL (rs285)
PON1 (rs854560)
1.19 (0.88, 1.61)
1.21 (0.93, 1.58)
(9.70 + 4.13 +
18.6)
(27.2 + 2.88 +
1.48)
(6.31 + 2.96 +
6.33)
(NA + 4.93 +
11.4)
(7.03 + 4.67 +
3.05)
Thrombosis-Hemostasis
ITGA2 (rs1126643)
FGB
(rs1800787/rs1800790)
ITGB3 (rs5918)
1.53 (0.97, 2.42)
1.58 (0.98, 2.54)
1.02 (0.83, 1.27)
157.0
0.91 (0.76, 1.09)
0.88 (0.63, 1.23)
0.97 (0.70, 1.33)
35.6
F7 (rs6046)
0.91 (0.76, 1.09)
0.88 (0.63, 1.23)
0.97 (0.70, 1.33)
21.4
AGT (rs4762)
1.31 (0.78, 2.19)
1.17 (0.92, 1.49)
0.89 (0.51, 1.55)
33.4
AGT (rs699)
0.89 (0.71, 1.10)
0.90 (0.74, 1.11)
1.01 (0.80, 1.29)
29.3
HFE (rs1800562)
1.08 (0.88, 1.33)
0.65 (0.40, 1.03)
0.59 (0.36, 0.98)
203.0
Endothelial Dysfunction
Inflammation
LTA (rs909253)
0.88 (0.61, 1.26)
1.36 (0.75, 2.48)
1.54 (0.91, 2.60)
42.4
(9.44 + 19.9 +
4.91)
8.10
0.02
Supplementary Table 7: Sensitivity analyses for associations with a stricter threshold of statistical significance for the summary odds
ratio (summary odds ratio of the all-inclusive meta-analysis with a p<0.01).
Gene (rsnumber)
P –value of the
summary effect
size
Comparisons of magnitude of effects for phenotypic
subgroups
Q partitioning for phenotypic subgroups
ROR (95%CI)
ACS vs Broad
ROR (95%CI)
Angiographic vs
Broad
ROR (95%CI)
ACS vs
Angiographic
Heterogeneity
of the allinclusive
metaanalysis: Qall
Heterogeneity
between
subgroups:
Qall - (Qbroad +
Qacs + Qangio)
Heterogeneity
between
subgroupsp-value
1.05 (0.32, 3.44)
1.16 (0.51, 2.62)
1.11 (0.31, 3.88)
3.9
0.1
0.93
Lipid Metabolism
APOB (rs1042031)
0.003
LPL (rs1801177)
0.002
0.83 (0.59, 1.17)
1.19 (0.76, 1.86)
1.43 (1.00, 2.05)
19.9
4.6
0.09
LPL (rs328)
0.001
1.12 (0.74, 1.70)
1.02 (0.67, 1.56)
0.91 (0.72, 1.14)
42.0
1.2
0.55
PON1 (rs662)
0.001
0.88 (0.72, 1.07)
0.98 (0.78, 1.23)
1.11 (0.94, 1.31)
107.0
2.0
0.36
CETP (rs708272)
0.007
1.05 (0.85, 1.29)
1.02 (0.82, 1.26)
0.97 (0.89, 1.05)
34.3
0.9
0.65
APOE (rs7412)
0.002
0.78 (0.60, 1.01)
0.83 (0.55, 1.23)
1.05 (0.73, 1.51)
58.9
9.8
<0.01
0.001
1.17 (0.90, 1.51)
0.98 (0.74, 1.30)
0.83 (0.68, 1.02)
75.4
1.5
0.48
NOS3 (rs1799983)
0.001
0.79 (0.50, 1.23)
0.85 (0.54, 1.33)
1.07 (0.89, 1.29)
124.0
0.5
0.76
NOS3 (rs2070744)
0.001
1.41 (1.12, 1.79)
1.42 (1.12, 1.79)
1.00 (0.82, 1.21)
56.6
9.6
<0.01
NOS3 (4a/b)
0.007
0.93 (0.69, 1.26)
0.86 (0.65, 1.13)
0.92 (0.73, 1.15)
64.7
5.2
0.07
-9
0.98 (0.81, 1.18)
1.10 (0.89, 1.35)
1.12 (0.94, 1.33)
47.2
18.6
<0.01
5.98 x 10-10
1.02 (0.94, 1.10)
1.06 (0.95, 1.19)
1.04 (0.92, 1.17)
14.1
2.3
0.32
Thrombosis-Hemostasis
F5 (rs6025)
Endothelial Dysfunction
ACE (rs4340)
7.89 x 10
Unknown
9p21 (rs1333049)
Supplementary Table 8: Meta-regression analysis results. The observed genetic effect
(logOR) is regressed against the continuous variable “% ACS patients in case group”.
The exponentiated coefficients of the meta-regressions (ROR) and the corresponding
95% CI for the explanatory variable “% ACS patients in case group” for each 5%
increase are presented for the meta-regression models a. for associations with statistically
significant summary ORs in the all-inclusive meta-analyses, and b. for associations with
statistically non-significant summary ORs in the all-inclusive meta-analyses. Significant
results are shown in bold.
a.
Gene (rsnumber)
N of studies
(Subjects included) ROR (95%CI)
Lipid Metabolism
APOB (rs1042031)
11 (3296)
1.14 (0.87, 1.48)
LPL (rs1801177)
10 (2256)
1.91 (0.90, 4.07)
LPL (rs320)
14 (2248)
0.91 (0.59, 1.39)
LPL (rs328)
18 (5581)
1.31 (0.53, 3.28)
PON1 (rs662)
25 (14384)
0.94 (0.81, 1.10)
CETP (rs708272)
10 (15395)
1.03 (0.96, 1.11)
APOE (rs7412)
12 (30995)
0.98 (0.88, 1.08)
Thrombosis - Hemostasis
SERPINE1 (rs1799889)
11 (6327)
0.99 (0.84, 1.16)
F13A1 (rs5985)
6 (6209)
3.55 (1.80, 7.00)
F5 (rs6025)
11 (22885)
0.76 (0.49, 1.18)
F2 (rs1799963)
12 (6612)
1.14 (0.99, 1.31)
NOS3 (rs1799983)
23 (15631)
0.99 (0.90, 1.08)
MTHFR (rs1801133)
52 (16579)
1.04 (0.95, 1.14)
NOS3 (rs2070744)
11 (9001)
1.08 (0.74, 1.59)
ESR1 (rs2234693)
10 (11624)
1.15 (0.62, 2.11)
ACE (rs4340)
77 (43204)
0.98 (0.91, 1.06)
NOS3 (4a/b)
21 (14127)
0.91 (0.83, 0.99)
13 (11729)
1.08 (0.95, 1.24)
11 (31575)
1.01 (0.98, 1.04)
Endothelial dysfunction
Inflammation
CD14 (rs2569190)
Unknown
9p21 (rs1333049)
b.
Gene (rsnumber)
Lipid Metabolism
N of studies
(Subjects
included)
ROR (95%CI)
APOB (rs11279109)
16 (7407)
1.15 (0.89, 1.48)
LPL (rs268)
12 (13234)
0.81 (0.30, 2.16)
LPL (rs285)
13 (2559)
0.79 (0.58, 1.08)
APOB (rs693)
16 (5693)
0.94 (0.80, 1.11)
PON1 (rs854560)
15 (9804)
1.03 (0.92, 1.15)
ITGA2 (rs1126643)
6 (7341)
0.99 (0.94, 1.04)
FGB (rs1800787/rs1800790)
8 (10910)
0.81 (0.44, 1.50)
ITGB3 (rs5918)
16 (10691)
1.00 (0.67, 1.51)
F7 (rs6046)
9 (8926)
1.01 (0.95, 1.07)
12 (45823)
1.02 (0.95, 1.09)
Thrombosis - Hemostasis
Endothelial dysfunction
HFE (rs1800562)
AGT (rs4762)
14 (18029)
1.00 (0.46, 2.15)
AGT (rs699)
26 (24934)
1.09 (0.98, 1.22)
10 (6079)
1.02 (0.82, 1.27)
Inflammation
LTA (rs909253)
Supplementary Table 9: Meta-regression analysis results. The observed genetic effects
(logOR) are regressed against indicator variables representing the different phenotypic
subgroups. The coefficients of the meta-regressions (logROR) and the estimated residual
heterogeneity (τ2) are assessed. The total between study variance (τ2total) from the
intercept model only and the unexplained between-study variance (τ2unexplained) from the
model with the phenotypic indicator variables and the RORs with the corresponding 95%
confidence intervals are shown: a. for associations with statistically significant summary
ORs in the all-inclusive meta-analyses, and b. for associations with statistically nonsignificant summary ORs in the all-inclusive meta-analyses. Significant results are shown
in bold.
Gene (rsnumber)
τ2total
τ2unexplained <
τ2total
τ2unexplained
ROR ACS vs Broad
ROR Angio vs Broad
Lipid Metabolism
APOB (rs1042031)
0
0
0.02953
0.027608
LPL (rs320)
0
0
LPL (rs328)
0.020767
0.032405
LPL (rs1801177)
CETP (rs708272)
APOE (rs7412)
PON1 (rs662)
0.003707
0.005322
0.06301
0.057636
0.051782
0.055141
0.01024
0.013334
0.069678
0.069965
Yes
No
No
Yes
No
1.05 (0.28, 3.98)
1.17 (0.47, 2.90)
0.87 (0.54, 1.42)
1.30 (0.75, 2.27)
1.09 (0.75, 1.58)
1.02 (0.70, 1.48)
1.02 (0.88, 1.19)
0.98 (0.84, 1.16)
0.77 (0.54, 1.09)
0.85 (0.59, 1.24)
0.89 (0.64, 1.23)
0.98 (0.73, 1.32)
0.88 (0.70, 1.11)
0.88 (0.66, 1.18)
0.77 (0.49, 1.22)
0.70 (0.40, 1.24)
1.15 (0.84, 1.57)
0.99 (0.68, 1.47)
0.94 (0.42, 2.08)
0.96 (0.39, 2.33)
0.83 (0.58, 1.20)
0.89 (0.62, 1.28)
1.03 (0.82, 1.30)
0.99 (0.80, 1.23)
1.41 (0.97, 2.06)
1.41 (0.97, 2.07)
0.91 (0.63, 1.30)
2.38 (1.32, 4.30)
1.00 (0.78, 1.28)
1.11 (0.87, 1.41)
Thrombosis - hemostasis
SERPINE1 (rs1799889)
F13A1 (rs5985)
F5 (rs6025)
F2 (rs1799963)
0.014312
0.019171
0.12482
0.1202
0.053428
0.057224
No
No
No
No
Endothelial dysfunction
NOS3 (rs1799983)
MTHFR (rs1801133)
NOS3 (rs2070744)
ESR1 (rs2234693)
ACE (rs4340)
0.039096
0.0298
0.09935
0.095155
0.044622
0.023174
0.043774
0.099656
No
No
Yes
Yes
No
No
NOS3 (4a/b)
0.036657
0.037886
0.083772
0.048138
0.001002
0.000643
0.94 (0.68, 1.29)
0.86 (0.64, 1.16)
0.63 (0.43, 0.92)
0.63 (0.43, 0.93)
1.03 (0.94, 1.12)
1.07 (0.94, 1.22)
Inflammation
CD14 (rs2569190)
Yes
Unknown
9p21 (rs1333049)
Yes
a.
b.
Gene (rsnumber)
τ2total
τ2unexplained
τ2unexplained
2
< τ total
ROR ACS vs
Broad
ROR Angio vs Broad
Lipid metabolism
APOB (rs11279109)
LPL (rs268)
LPL (rs285)
0.112149
0.122332
No
0.89 (0.46, 1.72)
0.71 (0.40, 1.27)
0.00626
0.010207
No
0.73 (0.51, 1.07)
0.66 (0.43, 1.02)
0
0.007689
No
NA
NA
0.040321
0.084149
No
0.99 (0.57, 1.75)
0.96 (0.52, 1.80)
0
0
-
1.17 (0.91, 1.50)
1.18 (0.95, 1.46)
ITGA2 (rs1126643)
0
0
-
NA
NA
FGB (rs1800787/rs1800790)
0
0
-
1.02 (0.86, 1.22)
0.95 (0.73, 1.24)
0.135051
0.107838
Yes
1.58 (1.11, 2.26)
1.59 (1.06, 2.41)
0
0.00642
No
0.92 (0.77, 1.10)
0.94 (0.73, 1.22)
HFE (rs1800562)
0.008873
0.006531
Yes
1.08 (0.87, 1.34)
0.63 (0.41, 0.99)
AGT (rs4762)
0.046103
0.067696
No
1.31 (0.68, 2.51)
1.19 (0.74, 1.91)
AGT (rs699)
0.073103
0.075366
No
0.88 (0.67, 1.16)
0.90 (0.69, 1.18)
0.052121
0.038132
Yes
0.89 (0.61, 1.30)
1.38 (0.75, 2.54)
APOB (rs693)
PON1 (rs854560)
Thrombosis - Hemostasis
ITGB3 (rs5918)
F7 (rs6046)
Endothelial dysfunction
Inflammation
LTA (rs909253)
Supplementary Table 10. Results from subgroup and meta-regression analyses for premature disease.
Subgroup meta-analyses are stratified into analyses for premature (prem) and non-premature (non-prem) disease. Summary
estimates of the association (OR and 95%CI) and heterogeneity statistics (PQ and I2) are shown for each genetic variant separately.
P-values for the statistical significance of the between-subgroup heterogeneity (from the Q-partitioning technique) are also
provided.
Meta-regression analyses results are derived by regressing the observed genetic effects against an indicator variable representing
the premature and non-premature disease subgroups. The total between study variance (τ2total) from the intercept model only and
the unexplained between-study variance (τ2unexplained) from the model with the “premature” indicator variables are compared. The
relative magnitude of effects of premature vs non-premature disease is assessed with the RORs and 95%CI.
All results are displayed separately: a. for associations with statistically significant summary ORs in the all-inclusive meta-analyses,
and b. for associations with statistically non-significant summary ORs in the all-inclusive meta-analyses. Significant results are shown
in bold.
a.
Gene (rsnumber)
Subgroup analysis
N of
studies (N
of
patients)
Lipid metabolism
ORnon-prem (95%CI)
PQnon-prem
(I2)
N of
studies (N
of
patients)
ORprem (95%CI)
Meta-regression analysis
PQprem (I2)
Heterogeneit
y between
subgroupsp-value
τ2unexplained
2
< τ total
ROR (95% CI)
Premature vs nonpremature
APOB (rs1042031)
9 (2159)
1.73 (1.02, 2.92)
0.99 (0)
5 (1572)
1.75 (1.04, 2.94)
0.64 (0)
0.97
-
1.01 (0.45, 2.30)
25
(17187)
12 (8531)
1.11 (1.02, 1.20)
<0.01 (60)
10 (5598)
1.12 (0.97, 1.29)
0.01 (59)
0.77
No
1.00 (0.82, 1.22)
0.83 (0.71, 0.98)
0.01 (58)
7 (4775)
0.78 (0.55, 1.09)
<0.01 (80)
0.76
No
0.98 (0.66, 1.48)
42
(54905)
23
(13104)
1.18 (1.05, 1.33)
0.02 (33)
20 (9370)
1.20 (1.02, 1.41)
0.63 (0)
0.33
No
1.04 (0.84, 1.32)
1.31 (1.00, 1.73)
0.25 (15)
16
(10029)
1.46 (1.07, 1.98)
0.08 (35)
0.75
No
1.09 (0.70, 1.67)
1.14 (1.04, 1.24)
<0.01 (63)
4 (2332)
1.32 (0.86, 2.02)
<0.01 (88)
0.43
No
1.12 (0.81, 1.54)
1.06 (0.98, 1.15)
0.03 (27)
17 (3869)
1.46 (1.06, 2.01)
<0.01 (64)
<0.01
Yes
1.32 (1.02, 1.71)
1.17 (1.05, 1.30)
<0.01 (68)
4 (2436)
1.20 (1.01, 1.41)
0.10 (52)
0.69
No
1.03 (0.78, 1.38)
1.25 (1.16, 1.36)
<0.01 (57)
23 (9949)
1.19 (0.99, 1.44)
<0.01 (50)
0.35
No
0.95 (0.75, 1.20)
1.19 (1.07, 1.33)
<0.01 (53)
4 (2320)
0.91 (0.77, 1.06)
0.98 (0)
<0.01
Yes
0.75 (0.57, 0.99)
7 (24732)
1.21 (1.15, 1.27)
0.46 (0.)
10
(15843)
1.36 (1.30, 1.43)
0.76 (0)
<0.01
Yes
1.13 (1.05, 1.21)
N of
studies (N
of
patients)
ORnon-prem (95%CI)
PQnon-prem
(I2)
Thrombosis-Hemostasis
SERPINE1 (rs1799889)
F13A1 (rs5985)
F5 (rs6025)
F2 (rs1799963)
Endothelial dysfunction
NOS3 (rs1799983)
MTHFR (rs1801133)
NOS3 (rs2070744)
ACE (rs4340)
NOS3 (4a/b)
37
(23338)
58
(19594)
17
(20832)
88
(55814)
27
(16832)
Unknown
9p21 (rs1333049)
b.
Subgroup analysis
Gene (rsnumber)
N of
studies (N
of
patients)
Meta-regression analysis
ORprem (95%CI)
PQprem (I2)
Heterogeneity
between
subgroupsp-value
τ2unexplained <
τ2total
ROR (95% CI)
Premature vs nonpremature
Lipid metabolism
APOB (rs11279109)
15 (6517)
1.15 (0.88, 1.50)
0.08 (36)
6 (2807)
1.23 (0.83, 1.82)
0.02 (63)
0.69
No
1.07 (0.64, 1.78)
LPL (rs268)
19
(21034)
12 (4037)
1.12 (0.96, 1.29)
0.31 (12)
2 (1282)
0.98 (0.71, 1.34)
0.90 (0)
0.25
No
0.87 (1.55, 1.36)
1.33 (0.94, 1.88)
<0.01 (61)
8 (2949)
1.09 (0.89, 1.33)
0.95 (0)
0.55
No
0.82 (0.52, 1.29)
APOB (rs693)
Thrombosis-Hemostasis
ITGA2 (rs1126643)
11 (8876)
0.97 (0.82, 1.16)
0.19 (27)
4 (3344)
2.97 (0.61, 14.56)
<0.01 (82)
0.18
-
NA
FGB
(rs1800787/rs1800790)
18
(28260)
0.98 (0.92, 1.04)
0.12 (29)
2 (2630)
1.00 (0.88, 1.14)
0.96 (0)
0.88
Yes
1.41 (1.01, 1.95)
F7 (rs6046)
ITGB3 (rs5918)
20
(14246)
29
(14891)
0.97 (0.88, 1.06)
0.35 (9)
4 (3388)
0.98 (0.86, 1.13)
0.87 (0)
<0.01
-
1.00 (0.85, 1.19)
0.92 (0.79, 1.08)
<0.01 (76)
13 (9595)
1.26 (1.03, 1.55)
<0.01 (71)
0.93
No
1.01 (0.85, 1.20)
1.05 (0.96, 1.14)
0.22 (18)
2 (857)
0.63 (0.39, 1.03)
0.34 (0)
0.05
yes
0.61 (0.35, 1.05)
1.04 (0.94, 1.15)
<0.01 (83)
5 (1587)
1.08 (0.88, 1.33)
0.09 (50)
0.38
No
1.05 (0.77, 1.44)
Endothelial Dysfunction
HFE (rs1800562)
AGT (rs699)
21
(50122)
35
(31079)
Supplementary Table 11: Meta-regression analysis results: The Relative OR and the
corresponding 95% CI for the explanatory variable “age of patients” (for a 5-year
increase) are presented for each meta-regression model and a. a. for associations with
statistically significant summary ORs in the all-inclusive meta-analyses, and b. for
associations with statistically non-significant summary ORs in the all-inclusive metaanalyses.
a.
Gene (rsnumber)
N of studies
(N of
patients)
ROR (95% CI)
Lipid metabolism
APOB (rs1042031)
14 (3731)
1.07 (0.83, 1.38)
LPL (rs1801177)
19 (4270)
0.90 (0.78, 1.04)
LPL (rs320)
22 (6115)
1.01 (0.92, 1.11)
LPL (rs328)
26 (8012)
0.98 (0.89, 1.07)
PON1 (rs662)
36 (20774)
1.03 (0.93, 1.13)
CETP (rs708272)
17 (28708)
0.97 (0.93, 1.01)
APOE (rs7412)
20 (46907)
1.07 (0.94, 1.21)
SERPINE1 (rs1799889)
35 (23302)
1.02 (0.98, 1.07)
F13A1 (rs5985)
19 (15027)
0.99 (0.91, 1.09)
F5 (rs6025)
62 (64840)
0.96 (0.90, 1.02)
F2 (rs1799963)
39 (24720)
0.99 (0.97, 1.01)
Thrombosis-Hemostasis
Endothelial dysfunction
NOS3 (rs1799983)
41 (25799)
0.97 (0.89, 1.06)
MTHFR (rs1801133)
75 (23763)
0.95 (0.89, 1.01)
NOS3 (rs2070744)
21 (23418)
1.02 (0.94, 1.12)
ESR1 (rs2234693)
16 (16367)
1.08 (0.86, 1.36)
ACE (rs4340)
111 (66578)
1.01 (0.94, 1.08)
NOS3 (4a/b)
31 (19152)
1.00 (0.92, 1.09)
22 (16157)
0.99 (0.80, 1.22)
17 (41158)
0.96 (0.80, 1.14)
N of studies
(N of
patients)
ROR (95% CI)
Inflammation
CD14 (rs2569190)
Unknown
9p21 (rs1333049)
b.
Gene (rsnumber)
Lipid metabolism
APOB (rs11279109)
21 (9324)
0.98 (0.81, 1.19)
LPL (rs268)
21 (22316)
1.02 (0.94, 1.11)
LPL (rs285)
18 (11851)
0.98 (0.91, 1.06)
APOB (rs693)
20 (6986)
0.95 (0.80, 1.14)
PON1 (rs854560)
20 (12416)
1.02 (0.95, 1.09)
15 (13991)
0.98 (0.92, 1.03)
Thrombosis-Hemostasis
ITGA2 (rs1126643)
FGB
(rs1800787/rs1800790)
20 (30890)
0.99 (0.96, 1.03)
ITGB3 (rs5918)
42 (32200)
0.98 (0.89, 1.08)
F7 (rs6046)
24 (17634)
1.01 (0.97, 1.04)
AGT (rs4762)
16 (19706)
0.96 (0.82, 1.13)
HFE (rs1800562)
23 (50979)
1.02 (0.94, 1.11)
AGT (rs699)
40 (32666)
0.96 (0.89, 1.03)
20 (30410)
0.97 (0.79, 1.18)
Endothelial dysfunction
Inflammation
LTA (rs909253)
Supplementary Table 12. Results from subgroup and meta-regression analyses for age-matched controls.
Subgroup meta-analyses are stratified into analyses for studies with age-matched controls and studies with non-age-matched
controls. Summary estimates of the association (OR and 95%CI) and heterogeneity statistics (PQ and I2) are shown for each
genetic variant separately. P-values for the statistical significance of the between-subgroup heterogeneity (from the Q-partitioning
technique) are also provided.
Meta-regression analyses results are derived by regressing the observed genetic effects against an indicator variable representing
the age-matched controls and non-age-matched controls subgroups. The total between study variance (τ2total) from the intercept
model only and the unexplained between-study variance (τ2unexplained) from the model with the “age-matched” indicator variables
are compared. The relative magnitude of effects of “age-matched” vs “non-age-matched controls” is assessed with the RORs and
95%CI.
All results are displayed separately: a. for associations with statistically significant summary ORs in the all-inclusive meta-analyses,
and b. for associations with statistically non-significant summary ORs in the all-inclusive meta-analyses. Significant results are shown
in bold.
a.
Subgroup analyses
N of
studies
(N of
ORnot-agematched(95%CI)
PQnot-age2
matched (I )
N of
studies
(N of
ORage-matched(95%CI)
Meta-regression analyses
PQage-matched
(I2)
Heterogeneity
between
subgroups-
τ2unexplained
2
< τ total
ROR (95%CI)
Age-matched vs nonage-matched
Gene (rsnumber)
patients)
patients)
p-value
Lipid metabolism
APOB (rs1042031)
LPL (rs1801177)
LPL (rs320)
LPL (rs328)
PON1 (rs662)
CETP (rs708272)
APOE (rs7412)
ThrombosisHemostasis
SERPINE1
(rs1799889)
F13A1 (rs5985)
F5 (rs6025)
F2 (rs1799963)
Endothelial
dysfunction
NOS3 (rs1799983)
MTHFR
(rs1801133)
NOS3 (rs2070744)
ESR1 (rs2234693)
ACE (rs4340)
NOS3 (4a/b)
10
(3072)
14
(2806)
18
(4356)
20
(5169)
21
(11090)
16
(35431)
14
(43237)
1.65 (1.10, 2.47)
1.00 (0)
4 (659)
2.26 (0.93, 5.51)
0.51 (0)
0.53
-
1.37 (0.51, 3.65)
1.43 (1.20, 1.70)
0.72 (0)
5 (1464)
1.01 (0.84, 1.21)
0.55 (0)
<0.01
Yes
0.71 (0.54, 0.93)
0.89 (0.78, 1.01)
0.09 (33)
4 (1759)
0.89 (0.77, 1.03)
0.67 (0)
0.92
-
1.00 (0.82, 1.22)
0.85 (0.74, 0.97)
0.03 (42)
6 (2843)
0.80 (0.66, 0.96)
0.16 (37)
0.21
No
0.93 (0.73, 1.17)
1.22 (1.07, 1.39)
<0.01
(73)
0.02 (48)
15
(9684)
9 (9040)
1.06 (0.96, 1.17)
0.02 (49)
<0.01
Yes
0.87 (0.74, 1.02)
0.96 (0.90, 1.02)
0.74 (0)
0.58
No
1.04 (0.94, 1.14)
0.85 (0.73, 0.98)
<0.01
(74)
6 (3670)
0.68 (0.56, 0.82)
0.88 (0)
0.01
Yes
0.82 (0.56, 1.18)
23
(13115)
4 (3000)
1.12 (1.02, 1.22)
0.07 (41)
0.64
No
0.92 (0.82, 1.04)
0.87 (0.74, 1.01)
<0.01 (72)
0.78
No
0.99 (0.72, 1.36)
39
(53025)
24
(13381)
1.20 (1.06, 1.36)
0.04 (31)
1.09 (0.96, 1.24)
0.58 (0)
0.74
No
0.90 (0.75, 1.08)
1.40 (1.11, 1.77)
0.52 (0)
12
(10187)
15
(12027)
23
(11815)
15
(11339)
1.04 (0.96, 1.13)
0.87 (0.66, 1.15)
<0.01
(61)
0.11 (49)
1.28 (0.90, 1.81)
0.02 (46)
0.34
No
0.84 (0.54, 1.30)
33
(19871)
54
(16916)
18
(20877)
12
(11173)
68
(53892)
23
(14128)
1.17 (1.07, 1.29)
<0.01
(65)
<0.01
(39)
<0.01
(59)
<0.01
(71)
<0.01
(61)
<0.01
(57)
8 (5928)
1.15 (0.92, 1.44)
<0.01 (77)
0.06
No
0.97 (0.76, 1.24)
21
(6847)
3 (2541)
1.05 (0.89, 1.24)
0.06 (35)
0.47
No
0.93 (0.76, 1.13)
1.51 (0.93, 2.45)
<0.01 (86)
0.42
No
1.32 (0.80, 2.15)
4 (5194)
0.95 (0.78, 1.16)
0.16 (43)
0.05
Yes
0.77 (0.46, 1.03)
43
(12686)
8 (5024)
1.27 (1.13, 1.43)
<0.01 (41)
0.02
No
1.03 (0.88, 1.20)
1.15 (0.96, 1.37)
0.07 (47)
0.37
No
1.00 (0.80, 1.24)
9 (5361)
0.93 (0.80, 1.07)
0.41 (3)
<0.01
Yes
0.65 (0.48, 0.85)
0.92 (0.86, 0.99)
1.13 (1.02, 1.24)
1.14 (1.04, 1.25)
1.30 (1.04, 1.62)
1.23 (1.12, 1.35)
1.15 (1.02, 1.29)
Inflammation
CD14 (rs2569190)
Unknown
13
(10796)
1.43 (1.18, 1.73)
<0.01
(69)
9p21 (rs1333049)
14
(36181)
1.26 (1.22, 1.31)
0.45 (0)
3 (4977)
1.28 (1.18, 1.39)
0.59 (0)
0.78
No
1.01 (0.92, 1.10)
b.
Gene (rsnumber)
Subgroup analyses
Meta-regression analyses
N of
studies
(N of
patients)
ORnot-agematched(95%CI)
PQnot-age2
matched (I )
N of
studies
(N of
patients)
ORagematched(95%CI)
PQage-matched
(I2)
Heterogeneity
between
subgroupsp-value
τ2unexplained
2
< τ total
ROR (95%CI)
Age-matched vs
non-age-matched
16
(7982)
15
(20678)
15
(10378)
13
(7566)
15
(6138)
1.09 (0.87, 1.37)
0.05 (40)
5 (1342)
1.50 (0.91, 2.49)
0.11 (47)
0.076
Yes
1.37 (0.79, 2.38)
1.15 (0.96, 1.36)
0.29 (15)
6 (1638)
0.98 (0.82, 1.17)
0.73 (0)
0.22
Yes
0.85 (0.66, 1.09)
0.97 (0.89, 1.06)
0.43 (2)
3 (1473)
0.82 (0.66, 1.02)
0.93 (0)
0.16
-
0.84 (0.67, 1.06)
0.97 (0.89, 1.06)
0.53 (0)
7 (4850)
1.00 (0.90, 1.11)
0.38 (7)
0.86
No
1.03 (0.90, 1.18)
1.21 (0.97, 1.51)
0.02 (47)
5 (848)
1.18 (0.76, 1.84)
0.24 (27)
0.58
-
0.98 (0.59, 1.60)
8 (8817)
1.05 (0.86, 1.27)
0.20 (29)
7 (5174)
0.92 (0.78, 1.08)
0.77 (0)
0.29
-
0.87 (0.68, 1.12)
12
(22447)
24
(21186)
12
(9142)
0.99 (0.90, 1.09)
0.06 (43)
8 (8443)
0.97 (0.90, 1.04)
0.76 (0)
0.02
-
0.97 (0.86, 1.10)
0.88 (0.74, 1.05)
18
(11014)
12
(8492)
1.16 (1.01, 1.32)
<0.01 (61)
0.41
Yes
1.36 (1.04, 1.77)
0.93 (0.80, 1.07)
<0.01
(79)
0.17 (28)
1.00 (0.91, 1.10)
0.90 (0)
0.46
-
1.08 (0.90, 1.28)
1.01 (0.91, 1.13)
0.06 (37)
5 (2516)
1.12 (0.89, 1.42)
0.78 (0)
0.48
No
1.10 (0.85, 1.43)
1.19 (0.96, 1.48)
<0.01
(66)
<0.01
(82)
5
(11448)
19
(18381)
0.98 (0.90, 1.07)
0.72 (0)
0.19
No
0.82 (0.65, 1.03)
1.10 (0.97, 1.25)
<0.01 (80)
0.63
No
1.10 (0.91, 1.33)
0.01 (51)
5 (6973)
1.22 (0.94, 1.59)
0.06 (57)
0.03
Yes
1.22 (0.90, 1.66)
Lipid metabolism
APOB (rs11279109)
LPL (rs268)
LPL (rs285)
PON1 (rs854560)
APOB (rs693)
ThrombosisHemostasis
ITGA2 (rs1126643)
FGB
(rs1800787/rs1800790)
ITGB3 (rs5918)
F7 (rs6046)
Endothelial
dysfunction
HFE (rs1800562)
AGT (rs4762)
AGT (rs699)
18
(48463)
11
(8258)
21
(14285)
1.00 (0.86, 1.15)
Inflammation
LTA (rs909253)
15
(23437)
0.99 (0.85, 1.16)
Supplementary Table 13. Results from subgroup and meta-regression analyses for angiographic controls.
Subgroup meta-analyses are stratified into analyses for studies with angiographic controls and studies with non-angiographic
controls. Summary estimates of the association (OR and 95%CI) and heterogeneity statistics (PQ and I2) are shown for each
genetic variant separately. P-values for the statistical significance of the between-subgroup heterogeneity (from the Q-partitioning
technique) are also provided.
Meta-regression analyses results are derived by regressing the observed genetic effects against an indicator variable representing
the angiographic controls and with non-angiographic controls subgroups. The total between study variance (τ2total) from the
intercept model only and the unexplained between-study variance (τ2unexplained) from the model with the “angiographic controls”
indicator variables are compared. The relative magnitude of effects of “angiographic” vs “non-angiographic controls” is assessed
with the RORs and 95%CI.
All results are displayed separately: a. for associations with statistically significant summary ORs in the all-inclusive meta-analyses,
and b. for associations with statistically non-significant summary ORs in the all-inclusive meta-analyses. Significant results are shown
in bold.
a.
Gene (rsnumber)
N of
studies (N
of
patients)
PQnon-
ORnonangiographic
(95%CI)
angiographic
2
(I )
Subgroup analyses
N of
studies
(N of
ORangiographic(95%
patients)
CI)
PQangiographi
2
c (I )
Heterogeneit
y between
subgroupsp-value
Meta-regression analyses
ROR (95%CI)
angiographic controls
τ2unexplained
versus non-angiographic
2
controls
< τ total
Lipid metabolism
LPL (rs1801177)
17 (3663)
1.29 (1.10, 1.51)
LPL (rs328)
22 (7223)
0.83 (0.73, 0.93)
PON1 (rs662)
32
(17148)
17
(43701)
1.15 (1.05, 1.27)
16
(11913)
37
(23951)
0.85 (0.72, 0.99)
APOE (rs7412)
0.78 (0.68, 0.89)
0.30
(13)
<0.01
(50)
<0.01
(70)
<0.01
(68)
2 (632)
1.39 (0.64, 3.02)
0.43
-
1.07 (0.48, 2.37)
4 (819)
1.07 (0.79, 1.45)
0.94 (0)
0.19
No
1.29 (0.92, 1.80)
4 (3598)
1.15 (1.01, 1.31)
0.35 (9)
0.52
Yes
0.99 (0.85, 1.17)
3 (3206)
1.07 (0.83, 1.37)
0.24 (31)
0.02
Yes
1.37 (0.92, 2.03)
<0.01
(71)
0.08
(26)
3 (2194)
0.92 (0.71, 1.19)
0.19 (40)
0.70
No
1.08 (0.80, 1.46)
2 (769)
1.67 (0.74, 3.77)
0.84 (0)
0.48
No
1.21 (0.43, 3.41)
<0.01
(71)
<0.01
(40)
<0.01
(61)
<0.01
(59)
0.01
(46)
6 (5312)
1.13 (0.96, 1.33)
0.13 (42)
0.70
No
0.95 (0.78, 1.16)
11
(4864)
4 (2766)
0.98 (0.83, 1.15)
0.67 (0)
0.16
No
0.85 (0.71, 1.03)
2.19 (0.94, 5.09)
<0.01 (84)
0.38
Yes
2.00 (0.85, 4.73)
20
(4802)
8 (5521)
1.18 (0.96, 1.44)
0.01 (49)
0.59
No
0.92 (0.74, 1.15)
1.09 (0.85, 1.40)
<0.01 (69)
0.02
No
0.93 (0.71, 1.23)
<0.01
(67)
4 (8043)
1.04 (0.88, 1.22)
0.22 (32)
0.03
Yes
0.82 (0.64, 1.06)
PQangiographi
2
c (I )
Heterogeneit
y between
subgroupsp-value
Thrombosis-Hemostasis
F13A1 (rs5985)
F2 (rs1799963)
1.37 (1.11, 1.69)
Endothelial dysfunction
NOS3 (rs1799983)
MTHFR (rs1801133)
ESR1 (rs2234693)
ACE (rs4340)
NOS3 (4a/b)
35
(20586)
64
(18665)
12
(13582)
91
(61613)
23
(13731)
1.18 (1.06, 1.30)
18 (8114)
1.25 (1.04, 1.50)
1.14 (1.04, 1.25)
1.09 (0.93, 1.28)
1.26 (1.17, 1.37)
1.16 (1.05, 1.29)
Inflammation
CD14 (rs2569190)
b.
Gene (rsnumber)
Lipid metabolism
LPL (rs285)
PON1 (rs854560)
Thrombosis-Hemostasis
N of
studies (N
of
patients)
13
(10416)
17 (8968)
ORnon-angiographic
(95%CI)
0.93 (0.85, 1.02)
0.96 (0.89, 1.04)
PQnonangiographi
2
c
(I )
0.39
(6)
0.40
(5)
Subgroup analyses
N of
studies (N
of
ORangiographic(95%
patients)
CI)
Meta-regression analyses
ROR (95%CI)
angiographic controls
τ2unexplained
versus non-angiographic
2
controls
< τ total
5 (1440)
1.10 (0.87, 1.40)
0.72 (0)
0.21
-
1.18 (0.91, 1.52)
3 (3420)
1.00 (0.90, 1.12)
0.82 (0)
0.60
-
1.04 (0.90, 1.19)
ITGA2 (rs1126643)
11 (7252)
0.95 (0.79, 1.15)
ITGB3 (rs5918)
38
(28946)
1.00 (0.89, 1.13)
13
(13545)
32
(24132)
1.18 (0.98, 1.41)
0.17
(29)
<0.01
(76)
4 (6517)
0.99 (0.83, 1.17)
0.99 (0)
0.80
-
1.03 (0.80, 1.33)
4 (3330)
1.07 (0.91, 1.26)
0.55 (0)
0.34
Yes
1.06 (0.86, 1.30)
<0.01
(62)
<0.01
(77)
3 (6161)
0.95 (0.84, 1.08)
0.73 (0)
0.24
No
0.81 (0.64, 1.01)
8 (8534)
0.90 (0.70, 1.16)
<0.01 (89)
0.04
No
0.70 (0.42, 1.15)
Endothelial dysfunction
AGT (rs4762)
AGT (rs699)
1.09 (0.98, 1.20)
Supplementary Table 14. Results from subgroup and meta-regression analyses for study design.
Subgroup meta-analyses are stratified into analyses for studies with retrospective design (case-control and cross-sectional studies)
and studies with prospective design (cohort and nested case-control). Summary estimates of the association (OR and 95%CI) and
heterogeneity statistics (PQ and I2) are shown for each genetic variant separately. P-values for the statistical significance of the
between-subgroup heterogeneity (from the Q-partitioning technique) are also provided.
Meta-regression analyses results are derived by regressing the observed genetic effects against an indicator variable representing
the retrospective and prospective subgroups. The total between study variance (τ2total) from the intercept model only and the
unexplained between-study variance (τ2unexplained) from the model with the “prospective” indicator variables are compared. The
relative magnitude of effects of “prospective” vs “retrospective” is assessed with the RORs and 95%CI.
All results are displayed separately: a. for associations with statistically significant summary ORs in the all-inclusive meta-analyses,
and b. for associations with statistically non-significant summary ORs in the all-inclusive meta-analyses. Significant results are shown
in bold.
a.
N of
studies (N
of
patients)
Lipid metabolism
ORretrospective
(95%CI)
PQretrospective
(I2)
N of
studies (N
of
patients)
ORprospective (95%CI)
PQprospective(I2)
Heterogeneity
between
subgroupsp-value
τ2unexplained < τ2total
ROR (95%CI)
prospective vs
retrospective
LPL (rs1801177)
17 (3485)
1.28 (1.06, 1.55)
0.24 (18)
2 (897)
1.32 (0.97, 1.80)
0.96 (0)
0.83
No
1.03 (0.71, 1.47)
LPL (rs328)
21 (5634)
34
(19944)
17
(27223)
17
(30167)
0.83 (0.74, 0.93)
0.15 (24)
5 (2378)
0.81 (0.62, 1.05)
0.01 (69)
0.27
No
1.12 (0.95, 1.32)
1.15 (1.05, 1.25)
<0.01 (69)
2 (830)
1.14 (0.82, 1.59)
0.30 (5)
0.96
No
0.99 (0.71, 1.38)
0.94 (0.90, 0.97)
0.44 (1)
8 (17248)
0.96 (0.84, 1.10)
0.02 (59)
0.53
Yes
1.05 (0.96, 1.15)
0.77 (0.66, 0.89)
<0.01 (66)
3 (16740)
1.01 (0.84, 1.23)
0.09 (58)
0.02
Yes
1.28 (1.06, 1.55)
1.09 (1.02, 1.18)
<0.01 (60)
4 (3998)
1.04 (0.95, 1.15)
1.00 (0)
0.98
No
0.95 (0.84, 1.07)
0.85 (0.74, 0.98)
<0.01 (71)
2 (902)
1.05 (0.80, 1.37)
0.48 (0)
0.77
No
1.23 (0.90, 1.67)
1.26 (1.14, 1.40)
0.33 (7)
6 (19447)
0.90 (0.73, 1.11)
0.70 (0)
0.02
Yes
0.71 (0.56, 0.90)
1.18 (1.08, 1.29)
<0.01 (69)
2 (2646)
0.89 (0.69, 1.14)
0.87 (0)
0.23
Yes
0.74 (0.57, 0.97)
1.11 (1.02, 1.22)
<0.01 (37)
6 (784)
1.02 (0.75, 1.39)
0.15 (38)
0.53
No
0.87 (0.70, 1.07)
1.13 (0.93, 1.37)
<0.01 (62)
4 (7465)
1.29 (0.89, 1.87)
<0.01 (83)
0.59
No
0.95 (0.75, 1.21)
1.25 (1.16, 1.35)
<0.01 (57)
4 (7392)
1.03 (0.78, 1.35)
0.29 (19)
0.74
No
0.82 (0.67, 1.01)
1.14 (1.02, 1.26)
<0.01 (56)
2 (2931)
1.24 (0.95, 1.61)
0.45 (0)
0.77
No
1.08 (0.81, 1.44)
20
(14889)
1.24 (1.06, 1.45)
<0.01 (66)
2 (1268)
0.90 (0.69, 1.17)
0.49 (0)
0.15
Yes
0.72 (0.53, 0.98)
13
(19185)
1.31 (1.26, 1.36)
0.99 (0)
4 (21973)
1.20 (1.13, 1.27)
0.31 (17)
0.03
Yes
0.91 (0.85, 0.97)
PON1 (rs662)
CETP (rs708272)
APOE (rs7412)
ThrombosisHemostasis
SERPINE1
(rs1799889)
F13A1 (rs5985)
F5 (rs6025)
31
(19304)
17
(14125)
56
(45397)
Endothelial
dysfunction
NOS3 (rs1799983)
MTHFR
(rs1801133)
ESR1 (rs2234693)
ACE (rs4340)
NOS3 (4a/b)
39
(23153)
69
(22979)
12 (8902)
107
(59186)
29
(16221)
Inflammation
CD14 (rs2569190)
Unknown
9p21 (rs1333049)
b.
N of
studies (N
of
patients)
Lipid metabolism
ORretrospective (95%CI)
PQretrospective
(I2)
N of
studies (N
of
patients)
ORprospective (95%CI)
PQprospective(I2)
Heterogeneity
between
subgroupsp-value
τ2unexplained
2
< τ total
ROR (95%CI)
prospective vs
retrospective
LPL (rs268)
PON1 (rs854560)
ThrombosisHemostasis
FGB
(rs1800787/rs1800790)
ITGB3 (rs5918)
F7 (rs6046)
Endothelial
dysfunction
HFE (rs1800562)
AGT (rs699)
18
(12154)
18
(11585)
16
(21101)
39
(20473)
21
(11896)
17
(39824)
36
(30165)
1.01 (0.89, 1.14)
0.92 (0)
3 (10162)
1.74 (0.92, 3.29)
0.05 (67)
0.98 (0.92, 1.04)
0.60 (0)
2 (831)
1.20 (0.85, 1.69)
0.26 (21)
0.98 (0.91, 1.05)
0.10 (33)
4 (9789)
0.98 (0.88, 1.09)
0.66 (0)
1.02 (0.90, 1.16)
<0.01 (75)
3 (11727)
0.96 (0.79, 1.15)
0.15 (48)
0.96 (0.88, 1.04)
0.48 (0)
3 (5738)
1.03 (0.89, 1.20)
0.56 (0)
0.98 (0.89, 1.09)
0.20 (21)
6 (11155)
1.19 (0.98, 1.44)
0.21 (30)
1.03 (0.94, 1.14)
<0.01 (82)
4 (2501)
1.17 (1.01, 1.35)
0.97 (0)
0.08
Yes
1.37 (1.03, 1.83)
0.46
No
1.21 (0.89, 1.66)
0.97
No
0.99 (0.87, 1.13)
0.92
No
0.99 (0.84, 1.16)
0.70
No
1.07 (0.90, 1.28)
0.38
No
1.15 (0.96, 1.38)
0.09
No
1.12 (0.94, 1.34)
Supplementary Table 15. Power estimates for various assumptions of relative odds ratios (ROR; range 1.2-2.5), within study variance
(0.2, 0.08, 0.007) and between study variance (range from 0.007(low) – 0.059(high)). Estimates were calculated separately for a
hypothetical synthesis of 15 and 30 studies. The calculations were performed according to the method described in Cafri et al. Behav
Res Methods 2009 for the calculation of contrasts of group mean effect sizes. The ROR assumptions were selected to represent a
range of plausible effect sizes of effect modification. The input parameters for the within study variance were selected based on
representative values of variances observed in odds ratios for genotypic distributions of samples sizes representing the median, the 5th
and the 95th percentile of study sample size in our database of 965 genetic association studies. The between study variance parameters
were selected based on the distribution of τ2 (median, interquartile range) observed in the all inclusive meta-analyses conducted in this
project.
ROR
Within
study
variance
Between
study
variance
(low)
Estimated
power
ROR
Within study
variance
Between
study
variance
(median)
Estimated
power
ROR
Within
study
variance
Between
study
variance
(high)
Estimated
power
N=30 studies
1.2
1.5
2.5
N=15 studies
1.2
0.2
0.08
0.007
0.2
0.08
0.007
0.2
0.08
0.007
0.2
0.007
0.007
0.007
0.007
34%
66%
99%
93%
99%
99%
99%
99%
99%
19%
1.2
1.5
2.5
1.2
0.2
0.08
0.007
0.2
0.08
0.007
0.2
0.08
0.007
0.2
0.037
0.037
0.037
0.037
30%
54%
91%
89%
99%
99%
99%
99%
99%
17%
1.2
1.5
2.5
1.2
0.2
0.08
0.007
0.2
0.08
0.007
0.2
0.08
0.007
0.2
0.059
0.059
0.059
0.059
28%
47%
78%
86%
99%
99%
99%
99%
99%
16%
1.5
2.5
0.08
0.007
0.2
0.08
0.007
0.2
0.08
0.007
0.007
0.007
39%
98%
68%
96%
99%
99%
99%
99%
1.5
2.5
0.08
0.007
0.2
0.08
0.007
0.2
0.08
0.007
0.037
0.037
31%
66%
62%
90%
99%
99%
99%
99%
1.5
2.5
0.08
0.007
0.2
0.08
0.007
0.2
0.08
0.007
0.059
0.059
26%
49%
58%
84%
99%
99%
99%
99%