studies have been inconsistent, perhaps because of racial differences, small sample sizes or other design
features. Pooling the results of case-control studies provided a means of exploring the basis for heterogene-
ity in study outcomes.
The objective of this study was to quantitatively summarize the evidence for a relationship between the
MTHFR gene, coronary artery disease and myocardial infarction.
2. Methods
2
.
1
. Study selection Relevant studies in humans were identified by search-
ing the MEDLINE database for the years 1988 through 1998. The search combined the following terms;
methylenetetrahydrofolate reductase MTHFR, homo- cysteine and cardiovascular disease. In addition, refer-
ence lists of identified studies and review articles were examined for other relevant studies. Finally, experts in
the field were surveyed to identify additional studies. Twenty-three case-control studies [14 – 35] were iden-
tified. These articles were reviewed by two authors Jee and Yoon to determine whether they met predeter-
mined criteria for inclusion in our subsequent analysis. Areas of disagreement or uncertainty were adjudicated
by consensus. To be included, a study must have in- cluded the following criteria: [1] been conducted in
humans; [2] assessed MTHFR genotype determination as an exposure using standardized laboratory methods;
[12] and [3] reported the number of cases and controls with the different MTHFR genotypes. Studies [15,18 –
21,23 – 27] of CAD used angiographically — confirmed occlusion as outcomes, while studies of myocardial
infarction MI used WHO criteria for MI [16 – 18] which relies on symptoms, enzyme elevations or electro-
cardiographic
changes, and
clinicalpast history
[14,22,23].
2
.
2
. Data abstraction Information was abstracted on general characteristics
name of first author, year of publication, country of origin, sample size, mean age, gender of participants
and body mass index, study design case and control selection criteria, folate intake, plasma folate and tHcy
in the study population, method of MTHFR genotype determination including information on the polymerase
chain reaction and enzyme used, and information on Hardy – Weinberg equilibrium. If different outcomes
were employed in the same report, they were analyzed as separate studies [18,23]. When cases of CAD or MI
could not be separated, these studies were not included [27 – 34].
2
.
3
. Statistical analysis Alleic and genotypic frequencies were determined
from observed genotype counts, and the expectations of the Hardy – Weinberg equilibrium were evaluated by x
2
analysis. Comparisons between genotypic frequencies were done using x
2
analysis. A meta-analysis consists of tests for association and
homogeneity. Homogeneity testing assesses the homo- geneity of the different odds ratios determined in re-
spective studies. The overall test for association then assesses the significance of the association between the
C677T mutation and CAD or MI for all studies com- bined. To calculate the pooled effects of the MTHFR
gene, each study was assigned a weight consisting of the reciprocal of variance of its odds-ratio estimate [36].
Estimations of the mean odds ratio of cardiovascular disease associated with those having the MTHFR gene
and those with 95 confidence intervals were calculated using the Mantel – Haenszel method [26].
3. Results
Table 1 shows the definition or criteria of cases and controls. The case-control studies were conducted be-
tween 1996 and 1998, and varied in size from 84 to 735 cases and from 73 to 1250 controls. The total number
of participants was 9855 4594 cases and 5261 controls. In 16 studies, healthy individuals were used as controls.
In the remaining two studies, those subjects angio- graphically-confirmed as normal were used as controls.
CAD patients with stenosis greater than 50 in at least one coronary artery were used as the CAD case group
in eight studies [15] [18 – 21,23,25] and [27] whereas the four remaining studies [14,26,32,33] did not indicate
which criteria of angiography was employed. In two studies for CAD and three studies for MI, the control
group was matched for age and sex, whereas the re- maining 13 studies did not indicate whether controls
were matched for age and sex. All of the studies used a standardized laboratory method [12] to determine the
MTHFR genotype. Three studies were conducted in Japan [20,25,33]; the others were conducted in Aus-
tralia [15,19], the United States [18,23], Netherlands [21,24], Italy [26,32], and Germany [27].
Participant characteristics for the 18 case-control studies included in this meta-analysis are presented in
Table 2. All of the studies were conducted in adults. Among cases, the mean age of participants was 60
range of 53 – 65 across studies. The corresponding mean age among controls was 56 range of 53 – 65. Ten
of the 18 studies included both men and women, while men were the sole participants in two studies. The
remaining five studies did not specify the percentage of men.
Table 3 shows the genotypic and allelic distribution of the MTHFR gene and its Hardy – Weinberg equilibrium.
In all the studies included, the distributions of the genotypes were in the Hardy – Weinberg equilibrium for
both patients and controls. From 12 studies with angio- graphically-confirmed coronary artery disease CAD,
the frequency of the three genotypes among controls was AA homozygous normal, 43.6; VA heterozygous,
45.2; and VV homozygous mutant, 677 C T, 11.2. These frequencies in cases were 42.1 for AA,
43.9 for VA, and 13.2 for VV.
3
.
1
. Coronary artery disease Twelve case-control studies examined the risk of CAD
associated with those having the MTHFR gene. Geno- type VV increased in the risk of CAD compared to the
corresponding AA group in eight 66.7 of the 12 studies. In just two of these studies, the relationship was
statistically significant.
Table 4 provides the overall odds ratios of CAD across these 12 case-control studies. Compared to the AA
genotype, the overall odds ratio for CAD was 1.3 95 CI. 1.1 – 1.5 for the VA genotype and 1.4 1.2 – 1.6 for
the VV genotype. However, after excluding the three Japanese studies [20,25,33] which had a strong associa-
tion between MTHFR and CAD [20,25], the correspond- ing CAD odds ratios for the VV or VA genotypes were
not statistically significant 1.1, 0.9 – 1.3 and 1.1, 0.9 – 1.3, respectively. For the three Japanese studies
[20,25,33] the overall odds ratios were 1.6 1.2 – 2.0 for the VA genotype and 2.0 1.6 – 2.7 for the VV genotype.
Table 1 Case and control characteristics of 18 case-control studies
Authors year Controls
Cases Selection criteria
Selection criteria Number
Number, outcome
250 CAD 201
Subjects without ischemic heart disease Izumi 1996
Angiographically demonstrated ischemic heart disease
Clinical history and creatinine kinase rise 190 MI
188 Random from residents, age and sex matched
Schmitz 1996 \
50 occlusion in at least in one coronary 225
Volunteers participating in a heart health 109 CAD
Wilcken 1996 education program
artery 310 MI
Adams 1996 WHO criteria
222 Adults visitors to patients with
non-cardiovascular illness 293 MI
WHO criteria Ma 1996
290 Participants who were free of MI; matching
of age, sex, and smoking \
50 occlusion in at least in one coronary 155 CAD
155 Healthy Caucasian individuals, matched for
Brugada
a
1997a artery
age and sex 79 MI
Brugada 1997b WHO criteria
155 Healthy Caucasian individuals, matched for
age and sex Residents of metropolitan Perth, with no
73 139 CAD
\ 50 occlusion in at least in one coronary
van Bockxmeer history
1997 artery
362 CAD Morita 1997
\ 50 occlusion in at least in one coronary
Healthy volunteers from annual health 778
examination artery
131 CAD Veroef 1997
100 \
90 occlusion in one and \40 occlusion Population-based controls
in one additional coronary artery 79 MI
Cardiovascular health study Schwartz 1997
386 Residents of King, Pierce or Snohomish
counties 168
510 CAD Anderson
b
\ 60 occlusion in at least in one coronary
B 10 occlusion in all major vessels
1997a artery
Patients selected on basis of being healthy Anderson
200 MI History of MI
554 1997b
Angiographically assessed CAD 735 CAD
Healthy people Kluijtmas 1997
1250 168
Patients with B10 luminal obstruction 510 CAD
Ou 1998 \
60 occlusion in at least in one coronary artery
Girelli 1998 278 CAD
Angiographically documented multivessel Angiographically documented normal
137 coronary arteries
CAD \
50 occlusion in at least in one coronary 180 CAD
105 Reinhardt 1998
Healthy volunteers artery
84 CAD Angiographically documented CAD
Abbate 1998 106
Healthy subjects in the same age bracket
a
Brugada 1997: a, outcome was a CAD; b, outcome was a MI.
b
Anderson 1997: a, outcome was a CAD; b, outcome was a MI.
S .H
. Jee
et al
. Atherosclerosis
153 2000
161 –
168
Table 2 Participants and study design characteristics of 18 case-control studies
Plasma folate nmoll Body mass index kgm
2
Author year Country
Age years Percentage of males
Range Mean
Case Control
Case Control
Case Control
Control Case
Case Control
Japan 43–76
74 54
NA NA
23.3 23.3
60 59
Izumi 1996 41–80
USA B
76 NA
NA 9.1
9.8 25.9
25.7 58
59 Schmitz 1996
B 76
72 46
NA NA
28.2 18–65
28.2
c
42 65
Wilcken 1996 Australia
NA NA
UK 65
64 NA
NA 25.8
25.4 65
56 NA
Adams 1996 100
NA 3.3
3.7 NA
Ma 1996 NA
USA 62
NA 40–84
NA 72
72 NA
NA NA
NA NA
USA Brugada
a
1997a NA
60 60
NA USA
NA 72
NA NA
NA NA
NA 60
NA Brugada 1997b
van Bockxmeer 1997 NA
Australia NA
6.4 7.6
NA NA
NA NA
B 50
B 50
100 100
NA NA
23.8 26–86
23.8
c
NA Morita 1997
Japan 62
48 85
60 NA
NA 26.6
Veroef 1997 26.0
Netherlands 53
50 NA
NA NA
NA NA
NA NA
18–44 NA
Schwartz 1997 18–44
NA NA
USA 17–89
USA 79
67 NA
NA NA
NA 63
65 31–89
Anderson 1997a
b
79 48
NA NA
NA 17–84
NA USA
Anderson 1997b 36–89
62 65
NA Netherlands
100 100
NA NA
NA NA
NA NA
NA Kluijtmas 1997
NA NA
86 87
NA NA
23.2 23.9
Japan Ou 1998
55 55
87 58
11.3 14.4
26.3 NA
24.9 61
NA Girelli 1988
Italy 59
NA Germany
75 67
NA NA
26.9 27.0
61 52
NA Reinhardt 1998
NA Abbate 1998
NA Italy
NA NA
NA NA
NA NA
NA NA
83 69
7.5 8.9
25.5 25.4
17–89 18–89
56 Overall
60
a
Brugada 1997: a, outcome was a CAD; b, outcome was a MI.
b
Anderson 1997: a, outcome was a CAD; b, outcome was a MI. Note: NA means data was not available.
c
We assumed body mass index of case group was as same as that of control group.
Table 3 Genotypic and alleic distributions of the MTHFR polymorphism and Hardy–Weinberg equilibrium of 18 case-control studies
C677TV allelic frequency Authors year
Genotype frequency n Hardy–Weinberg equilibrium
Control Cases
Controls Case
Cases Cases
VV VA
AA VV
VA AA
x
2
P-value x
2
P-value NA
50 110
90 25
Izumi 1996 102
NA 74
NA NA
NA NA
38 29
66 95
27 Schmitz 1996
90 33
71 0.014
0.99 4.37
0.11 36
12 46
51 24
113 35
88 Wilcken 1996
0.387 0.824
1.813 0.404
34 Adams 1996
35 32
145 133
29 97
96 0.377
0.828 0.173
0.917 NA
33 124
136 39
116 NA
135 Ma 1996
NA NA
NA NA
31 10
69 76
12 73
Brugada
a
1997a 70
29 0.605
0.739 0.797
0.671 31
6 32
41 12
73 NA
70 Brugada 1997b
NA NA
NA NA
NA van Bockxmeer 1997
NA 15
63 61
8 30
35 NA
NA NA
NA 33
57 188
117 79
361 42
338 Morita 1997
0.905 0.636
0.806 0.668
NA Veroef 1997
NA 13
59 59
7 48
45 NA
NA NA
NA NA
7 34
28 43
141 154
NA NA
Schwartz 1997 NA
NA NA
32 57
212 241
22 73
33 73
Anderson
b
1997a 0.762
0.683 0.807
0.668 35
23 87
90 59
238 Anderson 1997b
257 32
0.229 0.892
2.115 0.347
30 70
328 337
106 527
32 617
Kluijtmas 1997 0.284
0.867 0.240
0.887 39
61 84
69 42
158 Ou 1998
110 48
5.031 0.081
0.787 0.675
44 25
70 42
40 148
41 90
Girelli 1988 0.601
0.740 2.561
0.278 34
23 66
91 9
46 Reinhardt 1998
49 31
1.887 0.389
0.610 0.737
53 25
38 21
30 52
53 24
Abbate 1998 0.022
0.989 0.052
0.951 33
Overall 473
33 1673
1667 558
2332 2298
2.134 0.344
0.080 0.961
a
Brugada 1997: a, outcome was a CAD; b, outcome was a MI.
b
Anderson 1997: a, outcome was a CAD; b, outcome was a MI. Note: NA means data was not available. Table 4
Odds ratios OR of coronary artery disease associated with MTHFR gene in 12 case-control studies VV vs AA
a
No. of studies VA vs AA
a
OR 95 CI
P-value OR
95 CI P-value
1.4 1.2–1.6
0.0002 Overall
1.3 12
1.1–1.5 0.0056
Japanese studies [20,25,33] excluded 9
1.1 0.9–1.3
0.42 1.1
0.9–1.3 0.49
Japanese studies [20,25,33] only 3
2.0 1.6–2.7
B 0.00001
1.6 1.2–2.0
0.00047
a
VV, homozygous for the mutant; VA, heterozygous; AA, homozygous normal.
3
.
2
. Myocardial infarction Six studies examined the relationship between the
risk of myocardial infarction MI and the MTHFR gene. Genotype VV increased the risk of MI compared
to the corresponding AA group in four 66.7 of six studies; while in none of these studies was the lower
bound of the 95 confidence interval greater than one. Compared to those with the AA genotype, the overall
odds ratio for MI associated with the MTHFR gene was 1.0 0.8 – 1.1 among those with the VA genotype
and 0.9 0.7 – 1.2 among those with the VV genotype each P \ 0.05.
3
.
3
. Coronary artery disease or myocardial infarction Using data from all 18 studies, the occurrence of
CAD or MI was assessed by the MTHFR genotype. Compared to the AA genotype, the overall odds ratio
for the combination of CAD and MI was 1.1 95 CI, 1.0 – 1.2 for the VA genotype and 1.2 1.1 – 1.4 for the
VV genotype. After excluding the three Japanese stud- ies, the corresponding odds ratios were 1.0 0.9 – 1.2
and 1.0 0.9 – 1.1.
4. Discussion