Genetically Reduced Antioxidative Protection and Increased Ischemic Heart Disease Risk
The Copenhagen City Heart Study
Background— Extracellular superoxide dismutase (EC-SOD) is an antioxidative enzyme found in high concentrations in the arterial wall. Two to three percent of all people in Denmark carry an R213G substitution, which increases plasma concentration 10-fold. This may reduce arterial wall EC-SOD concentrations, increase intimal LDL oxidation, and therefore may accelerate atherogenesis. Our primary hypothesis was that EC-SOD-R213G predisposes to ischemic heart disease (IHD). The secondary hypothesis was that EC-SOD-R213G offers predictive ability with respect to IHD beyond that offered by measurements of plasma EC-SOD and autoantibodies against oxidized LDL (oxLDL).
Methods and Results— The primary hypothesis was tested in a prospective, population-based study of 9188 participants from The Copenhagen City Heart Study with 956 incident IHD events during 23 years of follow-up and retested cross-sectionally with independent case populations of patients with IHD (n=943) or ischemic cerebrovascular disease (ICVD) (n=617). Case populations were compared with unmatched IHD/ICVD-free control subjects from The Copenhagen City Heart Study (n=7992). The secondary hypothesis was tested by using a nested case-control study comparing patients with IHD (n=956) with age- and gender-matched control subjects (n=956). Age- and gender-adjusted relative risk for IHD in heterozygotes (n=221, 2.4%) versus noncarriers (n=8965, 97.6%) was 1.5 (95% CI, 1.1 to 2.1). Retesting confirmed this: Age- and gender-adjusted odds ratios for IHD was 1.4 (1.0 to 2.0) and for ICVD 1.7 (1.1 to 2.7). Additional adjustment for plasma EC-SOD produced an odds ratio for IHD in heterozygotes versus noncarriers of 9.2 (1.2 to 72), whereas adjustment for autoantibodies against oxLDL produced an odds ratio of 2.5 (1.2 to 5.3).
Conclusions— Heterozygosity for EC-SOD-R213G is associated with increased IHD risk. Genotyping offers predictive ability with respect to IHD beyond that offered by plasma EC-SOD and autoantibodies against oxLDL.
Received September 24, 2003; de novo received July 15, 2003; revision received September 16, 2003; accepted September 18, 2003.
Oxidative processes are involved in atherogenesis,1 and the superoxide anion (O2−·) is an important prooxidative molecule.
The major extracellular scavenger of superoxide anions is extracellular superoxide dismutase (EC-SOD).2 More than 90% of EC-SOD is located in the extravascular space bound to heparan sulfate proteoglycans in the glycocalyx of endothelial cell surfaces and in connective tissue matrix, especially in the arterial wall.3
We studied a relatively common missense mutation (nucleotide 760 G>C) in codon 213 in exon 3 of the EC-SOD gene (EC-SOD-R213G). Heterozygous carriers of EC-SOD-R213G have an ≈10-fold plasma concentration of functional EC-SOD compared with noncarriers.4 It has been speculated that this increase is due to accelerated release of EC-SOD from the interstitial matrix.4,5 Consequently, despite high plasma concentration, the arterial wall may be deficient in EC-SOD and thus may have insufficient antioxidative capacity against superoxide anions and LDL oxidation, which could lead to increased atherogenesis.
This study was performed to test the following hypotheses: (1) EC-SOD-R213G heterozygosity is associated with increased ischemic heart disease (IHD) risk; (2) EC-SOD-R213G genotype offers predictive ability with respect to IHD beyond that offered by measurements of plasma EC-SOD concentration and autoantibodies against oxLDL; and 3) EC-SOD-R213G heterozygosity and homozygosity is associated with plasma EC-SOD and autoantibodies against oxLDL.
The Copenhagen City Heart Study
This is a prospective study of the Danish general population initiated in 1976 to 1978 with follow-up examinations in 1981 to 1983 and 1991 to 1994. In 1991 to 1994, blood samples for DNA extraction were drawn; 16 563 were invited for the 1991 to 1994 examination. The response rate was 61.2%. Hence, 10 135 individuals were examined, of whom 9259 gave blood for DNA analysis.
The study was approved by the Danish Ethics Committee for the City of Copenhagen and Frederiksberg (No. 100.2039/91). Informed consent was obtained from all participants.
Copenhagen University Hospital
Patients with IHD were identified among 992 consecutive patients from the Greater Copenhagen area referred for coronary angiography because of angina pectoris from 1991 through 1993. Of these patients, 943 with coronary atherosclerosis and definite IHD were genotyped for EC-SOD-R213G.
Patients with ischemic cerebrovascular disease (ICVD) were identified among >3000 patients referred in 1994 to 1999 for ultrasonic evaluation of the carotid arteries because of focal neurological symptoms suggestive of ICVD (stroke, transient ischemic attack, or amaurosis fugax). Of these patients, 617 with >50% stenosis of the carotid artery and ICVD were genotyped for EC-SOD-R213G.
Details on selection procedures, demographics, and clinical characteristics of all three populations have previously been published6,7; in all populations, >99% were white and of Danish descent. A search using the Central Population Register Code (which unambiguously identifies a person) ruled out overlap of cases between populations.
To test the primary hypothesis, 9188 participants were followed prospectively from entry in the period from 1976 to 1978, 1981 to 1983, or 1991 to 1994 until 1999 (Figure 1). Information on diagnoses of IHD and ICVD (World Health Organization International Classification of Diseases, 8th edition, codes 410 to 414 and 432 to 435; 10th edition, I20 to I25 and I63 to I64) were gathered until 1999 from the Danish National Hospital Discharge Register, from the Danish National Register of Causes of Death, and from medical records of general practitioners and hospitals. Of the 1021 participants recorded with IHD, 65 were diagnosed before entry into the Copenhagen City Heart Study; these individuals were excluded, leaving 956 incident IHD cases. Follow-up was >99% complete. Median follow-up time was 21 years (range, 0.04 to 23 years).
To retest the primary hypothesis by using independent case populations, 943 patients with IHD and 617 patients with ICVD from Copenhagen University Hospital were compared with 7992 unmatched control subjects from the Copenhagen City Heart Study free from IHD and ICVD (Figure 1).
Nested Case-Control Studies
To test the secondary hypothesis, the 956 incident IHD cases in The Copenhagen City Heart Study were matched 1:1 on age and gender to IHD/ICVD-free control subjects from within The Copenhagen City Heart Study (Figure 1).
To test the tertiary hypothesis, all EC-SOD-R213G heterozygotes and homozygotes were matched 1:1 on age, gender, and IHD-status to EC-SOD-R213G noncarriers (Figure 1).
A 145-bp sequence spanning the G>C mutation at nucleotide position 760 was amplified with the use of sense primer 5′GCGGCAACCAGGCCAGCGTGGA3′ and antisense primer 5′TTGCACTCGCTCCGCCGGC 3′ (GenBank accession No. NM003102). A mismatch (underlined) was introduced, creating a restriction site for HaeII, which was abolished by the presence of the R213G mutation. A naturally occurring control cut site ruled out false-negative results. Thus, noncarriers displayed bands of 90, 33, and 22 bp, heterozygotes bands of 90, 55, 33, and 22 bp, and homozygotes bands of 90 and 55 bp. To confirm genotypes and confidently distinguish between heterozygotes and homozygotes, all heterozygotes and homozygotes by RFLP were reanalyzed through the use of DNA sequencing. All RFLP results were confirmed by this method.
An 863-bp PCR product covering the entire protein coding region of the EC-SOD gene was sequenced using sense primer 5′-GTGACTAAGCCTCACTCTGCCC-3′ and antisense primer 5′-CTGTTGGAGCAGAGGAGAT-3′.
Colorimetric assays were used to measure plasma levels of total cholesterol, HDL cholesterol, and triglycerides (all Boehringer Mannheim). Plasma EC-SOD and autoantibodies against oxidized (ox)LDL were determined in duplicate by means of enzyme-linked immunosorbent assays, as previously described.8,9 Plasma samples were stored at −80°C until analyzed. Coefficients of variation on duplicate measurements were 4% for plasma EC-SOD and 17% for autoantibodies against oxLDL.
We used the statistical software package Stata. Two-sided probability values <0.05 were significant. We used the Pearson χ2 test, Fisher exact test, Student t test, and the Mann-Whitney U test in 2-group comparisons and Kruskal-Wallis test and 1-way ANOVA in 3-group comparisons. Log-rank tests and Kaplan-Meier curves are presented for prospective data. With the use of left truncation (ie, delayed entry) Cox proportional hazards regression models with age as time scale were used to estimate relative risk (RR) for IHD, which implies that age is automatically adjusted for. Unconditional logistic regression models estimated odds ratios (ORs) for coronary atherosclerosis/IHD and carotid atherosclerosis/ICVD in unmatched case-control studies. Conditional logistic regression estimated ORs for IHD in nested case-control studies. Except for plasma EC-SOD and autoantibodies against oxLDL, which were included as continuous variables, all covariates were dichotomized, as listed in Table 1 and forced into the regression models. To test for bivariate interaction between EC-SOD-R213G and the covariates listed in Table 1 on IHD and ICVD end points, 2-factor interaction terms were individually included in the different models and tested for significance using a likelihood ratio test. No statistically significant interactions were observed, and no data are therefore presented on interactions. RR and OR are presented for heterozygotes versus noncarriers only because there were only 2 homozygous individuals.
Among participants in The Copenhagen City Heart Study, 8965 (97.6%), 221 (2.4%), and 2 (0.02%) were noncarriers, heterozygous, and homozygous for EC-SOD-R213G, respectively. This distribution did not differ from that predicted by the Hardy-Weinberg equilibrium (P=0.59). Cardiovascular risk factors did not differ between genotypes, except for hypertension, which was more frequent among homozygotes than among noncarriers (P=0.01) (Table 1). Correction for multiple comparisons, however, renders this difference nonsignificant.
During 23 years of follow-up and 162 071 person-years, the cohort had 956 IHD events producing IHD incidence rates in noncarriers, heterozygotes, and homozygotes of 58, 90, and 323 events per 10 000 person-years, respectively (Table 2) (Figure 2, P=0.02).
Age- and gender-adjusted RR for IHD in heterozygotes versus noncarriers was 1.5 (95% CI, 1.1 to 2.1) (Table 2). Additional adjustment for total cholesterol, HDL cholesterol, triglycerides, hypertension, diabetes mellitus, and body mass index resulted in an OR of 1.5 (1.0 to 2.1).
Gender- and age-adjusted OR for IHD in heterozygotes versus noncarriers was 1.4 (1.0 to 2.1) (Table 3). Multifactorial adjustment for the same covariates as mentioned above produced an OR of 1.4 (0.9 to 2.2).
Gender- and age-adjusted and multifactorially adjusted ORs for ICVD in heterozygotes versus noncarriers were 1.7 (1.1 to 2.7) and 1.9 (1.2 to 3.0).
Nested Case-Control Studies
The crude OR for IHD in heterozygotes versus noncarriers in this gender- and age-matched study was 2.1 (1.1 to 4.5). Multifactorial adjustment produced an OR of 2.3 (1.2 to 4.5) (Figure 3). Additional adjustment for plasma EC-SOD produced an OR in heterozygotes versus noncarriers of 9.2 (1.2 to 72). Adjustment for autoantibodies against oxLDL produced an OR of 2.5 (1.2 to 5.3).
Among EC-SOD-R213G noncarriers, participants with a plasma EC-SOD of 150 to 300 and >300 ng/mL had multifactorially adjusted ORs for IHD of 0.9 (0.7 to 1.2) and 0.5 (0.2 to 1.6) relative to participants with a concentration <150 ng/mL(Figure 3). Furthermore, participants with autoantibodies against oxLDL of 50 to 250 and >250 U had multifactorially adjusted ORs for IHD of 1.0 (0.8 to 1.3) and 1.5 (0.6 to 3.5) relative to participants with a concentration of <50 U (Figure 3). Plasma EC-SOD concentrations in heterozygotes and homozygotes were 1278±27 ng/mL (mean±SEM) and 4148±266 ng/mL, compared with 142±1 ng/mL in age-, gender-, and IHD status–matched noncarriers (P<0.001) (Figure 4). In contrast, EC-SOD-R213G genotype was not associated with autoantibodies against oxLDL (P=0.38).
This study demonstrates for the first time that EC-SOD-R213G is associated with increased risk of IHD in the general population and that EC-SOD-R213G offers predictive ability with respect to IHD beyond that offered by measurements of EC-SOD plasma levels and autoantibodies against oxLDL. EC-SOD heterozygosity and homozygosity were associated with 9-fold and 31-fold increase in plasma EC-SOD levels. However, plasma EC-SOD levels in noncarriers were not associated with risk of IHD. Finally, no significant associations were observed between EC-SOD-R213G genotype and autoantibodies against oxLDL or between these autoantibodies and risk of IHD.
Previous studies have indicated biological relevance for EC-SOD in atherogenesis. Thus, Luoma et al10 demonstrated increased expression of EC-SOD in both human and rabbit atherosclerotic lesions. Furthermore, Wang et al11 demonstrated that in patients with previous myocardial infarction, plasma EC-SOD was ≈30% lower than in patients with non–myocardial infarction IHD, suggesting that increased plasma EC-SOD protects against myocardial infarction in individuals with established coronary atherosclerosis. What is important is that this finding does not contradict our observation, as we demonstrate that EC-SOD-R213G heterozygotes with increased plasma EC-SOD levels have increased IHD risk, a question not addressed by Wang et al.11 In our study of patients with IHD, patients with a previous myocardial infarction (n=469) had plasma EC-SOD levels that were ≈8% lower than those in patients with non–myocardial infarction IHD (n=469) (P=0.14).
Biologically, we believe our observation of an increased IHD risk in EC-SOD-R213G heterozygotes makes sense. The R213G substitution in EC-SOD disrupts a cluster of 6 positively charged amino acids responsible for the interaction of EC-SOD with negatively charged sulfated glycosaminoglycans. An EC-SOD variant with a heparin affinity similar to that of EC-SOD-R213G is 4-fold more rapidly lost from binding to heparan sulfate in the tissue interstitial matrix than is the wild-type enzyme.12 It has therefore been speculated that the impaired binding of EC-SOD to the interstitial matrix would result in a decreased arterial wall concentration, although this has not been demonstrated. Assuming that this important but unproven assumption is true, a genotype associated with a local reduced antioxidative capacity in the arterial wall would be expected to increase arterial wall oxidation of LDL, and this in turn would promote atherogenesis and IHD, exactly as we observed. Also, because both EC-SOD and autoantibodies against oxLDL are plasma measurements that supposedly do not reflect conditions locally in the arterial wall in EC-SOD-R213G carriers, this mechanism may explain why the association between EC-SOD-R213G and IHD is independent of plasma EC-SOD and autoantibodies against oxLDL.
Our demonstration of a 9-fold elevated plasma EC-SOD level in heterozygotes and a 31-fold elevated level in homozygotes agrees with previous studies.11,13 Also in accordance with previous reports, we found that plasma EC-SOD increased with age and hypertension and decreased with male gender, smoking, and diabetes (data not shown).11,13
To investigate whether EC-SOD-R213G genotype offered additional predictive ability with respect to IHD risk beyond that offered by plasma EC-SOD, we estimated the OR for IHD in heterozygotes relative to noncarriers before and after adjustment for plasma EC-SOD. Such adjustment ensures that any direct association between plasma levels and IHD will not mask an association between genotype and IHD, particularly if genotype does not exert its action through plasma levels, as is likely for EC-SOD-R213G. Thus, if EC-SOD-R213G genotype held the same information as plasma EC-SOD, adjustment for plasma EC-SOD would have resulted in an OR close to unity. The fact that the OR increased from 2.3 to 9.2, however, supports the idea that despite high plasma EC-SOD, EC-SOD-R213G heterozygotes may have an arterial wall deficient in EC-SOD and therefore an increased risk of IHD. It also documents that genotype has predictive ability with respect to IHD beyond that offered by plasma EC-SOD levels.
Likewise, the fact that the OR for IHD in EC-SOD-R213G versus noncarriers before and after adjustment for autoantibodies against oxLDL was 2.3 and 2.6 suggests that genotype has predictive ability with respect to IHD beyond that offered by a measure of oxidative stress in plasma. Only one study has investigated EC-SOD-R213G in relation to cardiovascular disease. This study, by Yamada et al,14 was conducted in a highly selected population of hemodialysis patients, a situation known to be associated with increased oxidative stress. In their study, the 5-year survival rate of diabetic EC-SOD-R213G heterozygotes was significantly lower than that of diabetic noncarriers. The excess deaths among EC-SOD-R213G heterozygotes were due to IHD or ICVD, in accordance with our demonstration of increased risk of IHD and ICVD in heterozygotes versus noncarriers.
Other reports seemingly contradict this conclusion. Gene transfer of EC-SOD into LDL receptor–null mice did not affect the extent of lesion formation even though plasma EC-SOD was increased 4- to 7-fold.15 It should, however, be kept in mind that the extra EC-SOD expression transfected to these mice was added onto a normal physiological level. If physiological EC-SOD levels are sufficient to balance out oxidative stress from superoxide anions, increasing EC-SOD concentrations, even to excessive levels, cannot be expected to reduce atherogenesis. Another study by Sentman et al16 showed that EC-SOD null mice, crossed onto apoE-null mice, had no evidence for an enhancement of atherogenesis, increase of lipid peroxidation, or change in the content of oxidized LDL detected immunochemically in the artery wall of the double-mutant mice compared with apoE-null mice. There are, however, several reasons to question extrapolations from animal models of oxidative stress in rodents to humans, as follows.1 Atherosclerosis developing over a relatively short time span (8 months) as investigated by Sentman et al16 may not be comparable to clinical IHD events developed over several decades as investigated by us. Finally, the null finding communicated by Sentman et al16 may represent a type II error because of the limited sample size. Thus, the statistical power in that study to detect a 25% increase in the aortic lesion surface area at 8 months is only 21%.
Diagnostic information on IHD in The Copenhagen City Heart Study originates from public registries on hospitalizations and deaths. The major problem of using such data is the possibility of misclassification of IHD status. However, for such misclassification to systematically bias our results, it must be associated with the exposure variable of interest. Because this study examines a genotype, the status of which is unknown to patients and their physicians, we believe that such differential misclassification is unlikely.
Nondifferential misclassification as a source of bias is another possible limitation. To limit this possibility, patient records on 200 participants with a diagnosis of myocardial infarction were systematically reviewed through records of hospitals and general practitioners. The World Health Organization’s diagnostic criteria were met in 99% of patients (data not published). A similar validation of non–myocardial infarction IHD has not been performed. However, the fact that prospective results on IHD were subsequently confirmed by using an independent population of IHD patients diagnosed using coronary angiography makes us believe that nondifferential misclassification is of limited importance in this study. Importantly, even if nondifferential misclassification were pronounced, such misclassification would result in risk estimates approaching unity. It therefore cannot explain the results of this study.
Nonresponse also represents a potential bias. Of the 16 563 participants invited to participate in the 1991 to 1994 examination, 9258 were genotyped for EC-SOD-R213G. Among nonresponders, 56% were women, compared with 55% among responders (P=0.37). The average age among nonresponders was 61±0.2 years (mean±SEM), compared with 58±0.2 years among responders (P<0.001). As expected, IHD morbidity and all-cause mortality rates were higher among nonresponders than among responders. Thus, gender- and age-adjusted RR of IHD in nonresponders versus responders was 1.2 (1.1 to 1.3), and RR for all-cause death was 1.9 (1.8 to 2.1), which demonstrates that responders on average are more healthy than nonresponders. However, for the IHD risk estimates by EC-SOD-R213G genotype to be biased, nonresponse must be differential with respect to EC-SOD-R213G genotype. We believe that an association between genotype and response/nonresponse status is unlikely, as neither the participants nor their physicians were aware of the EC-SOD-R213G genotype status.
Less than 1% of participants were lost to follow-up in this study, and such loss is unlikely to seriously affect the conclusions of this study. Nevertheless, assuming the extreme situation where 50% of EC-SOD-R213G noncarriers lost to follow-up subsequently developed IHD whereas all heterozygotes lost to follow-up remained healthy, would only change the adjusted RR for IHD in heterozygotes relative to noncarriers from the observed 1.5 (1.0 to 2.1) to 1.4 (1.0 to 2.0).
If the EC-SOD-R213G genotype were associated with either increased or decreased survival, selection bias could affect the conclusions of this study because only those who survived to attend the 1991 to 1994 examination were included in the study. Arguing against such bias, however, is the fact that death from all causes, examined after the 1991 to 1994 examination, was not affected by EC-SOD-R213G genotype (log-rank test, P=0.64). To further reduce the likelihood of the conclusions being due to selection bias, we reanalyzed the data limiting the study to the period after the 1991 to 1994 examination. The relative risk for IHD in heterozygotes versus noncarriers in this study was 1.7 (1.1 to 2.6).
Although there is a large amount of intervention data from studies using hypercholesterolemic animal models and antioxidants that support the idea that oxidative processes are important in atherogenesis, such prospective intervention in humans has been disappointing.1 However, it is possible that only a subgroup of people with increased “oxidative stress” will benefit from antioxidant intervention. The fact that the most positive study in support of the oxidative hypothesis comes from the demonstration that vitamin E dramatically inhibited clinical coronary heart disease events in patients undergoing hemodialysis17 makes this an attractive possibility. Therefore, a study such as ours, which defines a population at increased risk for atherosclerosis because of a genetically determined decrease in a relevant antioxidant enzyme, is potentially important.
In conclusion, we find that EC-SOD-R213G heterozygosity increases IHD risk in the general population. Furthermore, determining EC-SOD-R213G genotype adds predictive ability with respect to IHD beyond that afforded by plasma measurement of EC-SOD and autoantibodies against oxLDL.
This study was supported by the Danish Heart Foundation, Chief Physician Johan Boserup and Lise Boserup’s Fund, Lykfeldt’s Fund, Dagmar Marshall’s Fund, Wedelborg’s Fund, Lily Benthine Lund’s Fund, the Beckett Fund, and P. Carl Petersen’s Fund. The skillful technical assistance of biotechnicians Nina Dahl, Karin Hjertkvist, and Ewa Kogutowska is gratefully acknowledged.
Stralin P, Karlsson K, Johansson BO, et al. The interstitium of the human arterial wall contains very large amounts of extracellular superoxide dismutase. Arterioscler Thromb Vasc Biol. 1995; 15: 2032–2036.
Sandstrom J, Nilsson P, Karlsson K, et al. Ten-fold increase in human plasma extracellular superoxide dismutase content caused by a mutation in heparin-binding domain. J Biol Chem. 1994; 269: 19163–19166.
Adachi T, Yamada H, Yamada Y, et al. Substitution of glycine for arginine-213 in extracellular-superoxide dismutase impairs affinity for heparin and endothelial cell surface. Biochem J. 1996; 313: 235–239.
Luoma JS, Stralin P, Marklund SL, et al. Expression of extracellular SOD and iNOS in macrophages and smooth muscle cells in human and rabbit atherosclerotic lesions: colocalization with epitopes characteristic of oxidized LDL and peroxynitrite-modified proteins. Arterioscler Thromb Vasc Biol. 1998; 18: 157–167.
Wang XL, Adachi T, Sim AS, et al. Plasma extracellular superoxide dismutase levels in an Australian population with coronary artery disease. Arterioscler Thromb Vasc Biol. 1998; 18: 1915–1921.
Sentman ML, Brannstrom T, Westerlund S, et al. Extracellular superoxide dismutase deficiency and atherosclerosis in mice. Arterioscler Thromb Vasc Biol. 2001; 21: 1477–1482.