Abstract
The MBL gene, encoding mannose-binding lectin, deter- mines interindividual variation in susceptibility to certain infectious agents, such as Chlamydia pneumoniae. We exam- ined whether infection-susceptibility alleles of MBL, called “non-A alleles,” would be associated with increased carotid plaque area (CPA), an intermediate phenotype of atheroscle- rosis. In 164 subjects, we measured CPA with 2-dimensional ultrasound. We also determined traditional atherosclerosis risk factors and genotyped all subjects for MBL codons 52, 54, and 57. We used ANOVA to determine sources of varia- tion for CPA and tested the hypothesis that the presence of a single MBL non-A “infection-susceptibility” allele was asso- ciated with increased CPA; 45.7% of subjects had at least one non-A allele. ANOVA showed that CPA was significantly associated with MBL genotype, age, smoking, hypertension, and hyperlipidemia (P<0.05). When MBL was used as the sole independent variable in the regression analysis, the as- sociation with CPA was even more significant (P=0.009). Subjects with at least one MBL non-A allele had significantly higher CPA than subjects homozygous for the MBL A allele and were significantly more likely to have CPA in excess of the sample median. Thus, infection-susceptibility alleles of MBL were associated with increased CPA in this study sam- ple; these alleles may be a determinant of interindividual differences in atherosclerosis risk.
Introduction
Atherogenesis results from a relatively large number of contributing factors but produces a relatively limited num- ber of clinical end points. Recently, there has been an increased appreciation of a role for inflammatory factors in atherogenesis1 and several lines of experimentation point to a possible link with certain infectious agents.2 One genetic determinant for interindividual differences in sus- ceptibility to infections is the common coding sequence variation in the MBL gene, which encodes mannose-bind- ing lectin (MBL).3,4 MBL is an immune defense protein that binds mannose and other sugars on the surface of infectious agents, thereby facilitating phagocytosis and activation of the complement cascade.3,4 MBL likely mod- ulates the severity of infection with Chlamydia pneu- moniae, a pathogen linked to the initiation and progres- sion of atherosclerosis.2 This might explain the association of genetic variation in MBL with severe atherosclerosis.6
There are three common polymorphic sites in MBL. The Gly->Asp variant at MBL codon 54 (G54D) in exon 1, also referred to as the “B allele,” has been associated with recurrent infections.3,4 This is probably because MBL G54D destabilizes the sixth collagen-like repeat of MBL, which results in a failure to activate complement for the B-type chains of MBL expressed in vitro.3,4 Furthermore, subjects who are heterozygotes for the wild-type MBL A allele (G54) and the poorly functioning MBL B allele (D54) have a 20-fold reduction in MBL levels compared to G54 homozygotes, suggesting that the MBL B allele product acts as a dominant negative.3,4 The two other MBL polymorphisms, namely G57E (also called the “C allele”) and R52C (also called the “D allele”) have also been associated with reduced levels of MBL.3,4 Also, patients with severe atherosclerosis had a lower frequency of the MBL A allele and a higher frequency of the MBL B, C, and D alleles compared with normal controls.6
Methods
We hypothesized that the MBL non-A alleles would be associated with an increase in a more proximal, interme- diate phenotype of atherosclerosis, such as carotid plaque area (CPA), as determined by 2-dimensional ultrasound. We evaluated the association of MBL genotype with CPA with multivariate regression, using classic atherosclerosis risk factors as covariates in the model.
Study Subjects
164 subjects ranging in age from 28-78 years were recruited from a vascular disease prevention clinic. These subjects had been referred to the clinic because of the early onset of clinical end points, defined as under the age of 55 in men and under the age of 60 in women. All subjects were white. Sixty-three percent of subjects had early pre- sentation of the end points of coronary heart disease (such as angina, myocardial infarction, and/or revascularization procedures), 27% had early presentation of the end points of cerebrovascular disease (such as transient ischemic attacks and/or stroke), and the remainder had early pre- sentation of the end points of both coronary heart and cerebrovascular disease. Sixty-one percent of subjects also had a strong family history of premature atherosclerosis, namely the onset of clinical end points under the age of 55 in male first-degree relatives and under the age of 60 in a female first-degree relative. All subjects gave their in- formed consent to participate. The protocol was approved by the Research Ethics Committee at the University of Western Ontario.
Clinical, Biochemical, and Genetic Determinations
Questionnaires relating to past medical history were administered to, and physical examinations were per- formed on, all study subjects. Variables included in sub- sequent analyses were: age, sex, smoking, diabetes, hy- pertension, and hyperlipidemia. Smoking was defined as a history of ever having smoked cigarettes for more than 6 months. Hypertension was defined as medically diagnosed hypertension and/or current treatment with an antihyper- tensive agent, which was usually an angiotensin-convert- ing enzyme inhibitor or β-blocker. Hyperlipidemia was defined as medically diagnosed hyperlipidemia and/or cur- rent treatment with a lipid lowering medication, which was usually a statin or fibric acid derivative. Diabetes was defined as medically diagnosed diabetes and/or current treatment with insulin or an oral hypoglycemic agent, usually glyburide or metformin. Established methods were used to genotype MBL codons 52, 54, and 57.7 Known DNA standard MBL genotypic controls were run with each genotyping reaction.
Ultrasound Evaluations
Ultrasound measurements were made from 2-dimen- sional B-mode images using a cursor and a microprocessor within the ultrasound equipment (Mark 9, ATL, Seattle, Wash).8,9 CPA was defined as the sum of the cross- sectional areas of all plaques seen in longitudinal views of the common, external, and internal carotid arteries. In- traobserver variability of CPA was assessed by having each of two technicians measure the CPA in 25 subjects on two separate occasions at least 1 week apart. These ultra- sound scans were then distributed in a blinded fashion within a large number of routine clinical scans, which resulted in them being scored again by the same individ- ual. The intraclass correlation coefficient for CPA by the same observer was 0.94. Between-observer reliability of CPA determination from the images was assessed by hav- ing two technicians exchange their tapes on 25 patients and measuring the plaque area. The intraclass correlation coefficient between observers was 0.99.
Statistical Analysis
SAS (version 6.12) was used for all statistical analy- ses.10 ANOVA, using the general linear models procedure, was performed to determine the sources of variation for 2-dimensional CPA. Significance was assessed from F tests computed from the type III sums of squares.10 This form of sums of squares is applicable to unbalanced study designs and reports the effect of an independent variable after adjusting for all other variables included in the model. The untransformed CPA had a distribution that was significantly non-normal, but a cube root transformation resulted in a variable whose distribution was not signifi- cantly different from normal (Wilk's W=0.96, P=NS).
The dependent variable in the ANOVA was the trans- formed CPA. Independent covariates were age, sex, his- tory of and/or treatment for diabetes, history of and/or treatment for hyperlipidemia, history of and/or treatment for hypertension, and a history of cigarette smoking. The independent variables in the model for each analysis in- cluded MBL genotype with assumption of a dominant effect of non-A MBL alleles on CPA. This was done by setting the MBL genotype variable at 0 for A/A subjects and at 1 for subjects with all other genotypes, including heterozygotes for non-A alleles, compound heterozygotes for the non-A alleles, and homozygotes for non-A alleles. We subsequently performed pairwise comparisons of least squares means of subjects with the A/A genotype compared to subjects with all other genotypes. Least squares means are also called population marginal means and reflect the mean values after adjustment for all the covariates used in the model. P≤0.05 was chosen as the nominal level of significance for pairwise comparisons of least squares means. The partial correlation coefficient from a post hoc multivariate regression analysis was used to estimate the proportion of variation in CPA that was due to MBL genotype. The sample was then divided according to subjects whose CPA was above the median for the sample and those whose CPA was below the median for the sample. χ2 tests were used to determine whether there was a between- group difference between low and high CPA subgroups in the proportion of subjects who had at least one non-A MBL allele. Estimates of relative risk between genotypes were determined using Mantel-Haenszel odds ratios.
Results
Characteristics of the Study Sample
The baseline clinical and biochemical features of the 164 study subjects, divided according to gender, are shown in Table 1. As expected, many subjects smoked and had a history of and/or received treatment for hyperlipid- emia and/or hypertension.
Clinical and biochemical features of the study sample.
MBL Allele and Genotype Frequencies
Alleles and genotypes were designated using the con- ventional nomenclature for MBL.3–7 The A allele had the DNA sequence that encoded R52, G54, and G57 together on a single allele. The B, C, and D alleles were determined on the basis of the presence of D54, E57, and C52, respectively. The frequencies of the B, C, and D alleles were 15.9, 6.4, and 7.0%, respectively. Therefore, the frequencies of the A and non-A alleles were 70.7 and 29.3%, respectively. Using this convention, 89 subjects (54.3%) had the A/A genotype, 54 subjects (32.9%) were heterozygous for one A allele and one non-A allele, also called the A/O genotype, and 24 subjects (12.8%) were simple or compound heterozygotes for non-A alleles, also called the O/O genotype. The frequency of the A allele and the A/A genotype were similar to those observed in sub- jects with atherosclerosis and were markedly lower than those frequencies reported in control white subjects, which typically range from 60-70%.6,7 Genotype frequencies did not deviate from Hardy-Weinberg expectations.
Sources of Variation for CPA
ANOVA showed that CPA was significantly associated with age, smoking, hypertension, and hyperlipidemia (P<0.05). There was no significant association with either gender or diabetes. The MBL genotype under a dominant model for non-A alleles was significantly associated with CPA (P=0.049) (Table 2). When MBL genotype was included as the only independent variable, the association was much stronger (P=0.009) (Table 3). Subsequent re- gression analysis indicated MBL genotype, under a dom- inant model for non-A alleles, accounted for ≈1.3% of the total, and ≈3.3% of the attributable variation in CPA.
Results of ANOVA for sources of variation in carotid plaque area.
NS indicates not significant with nominal P<0.05.
Mean (±SD) CPA according to MBL genotype.
P<0.05. 2-tailed t test.
CPA and MBL Genotype
When subjects were classified according to MBL geno- type, we found that the mean±SD of transformed CPA was significantly higher in subjects with at least one MBL susceptibility allele, namely A/O plus O/O genotypes, than in A/A homozygotes, namely 0.82±0.38 versus 0.73±0.37 (P=0.049). Of the subjects whose CPA was below the median, 36.9% had at least one non-A allele of MBL. Of the subjects whose CPA was above the median, 54.1% had at least one non-A allele of MBL. These pro- portions were significantly different (P=0.025). Further- more, subjects with at least one non-A allele had a relative risk of 1.38 (95% confidence interval [CI] 1.04 to 1.82) of being above the median for CPA compared to those sub- jects who were homozygous for MBL A/A.
Discussion
We report that CPA is significantly greater in subjects with at least one infection-susceptible MBL non-A allele compared to subjects homozygous for the infection-resis- tant MBL A allele. The MBL genotype, under a dominant model for non-A alleles, was a significant determinant of CPA, accounting for ≈3.3% of the attributable variation. Furthermore, subjects with at least one infection-suscep- tible MBL non-A allele had an odds ratio of 1.38 (95% CI 1.04 to 1.82) for having CPA greater than the sample median compared to subjects homozygous for the infec- tion-resistant MBL A allele. These findings thus suggest that interindividual genetic differences affecting suscepti- bility to certain infections are significantly associated with differences in the presence and extent of carotid arterial plaque.
The MBL gene product has functional features of some immunoglobulins and members of the complement cas- cade.3,4 In the presence of calcium, the MBL gene product can bind to a wide variety of oligosaccharides through multiple lectin domains.3,4 Such binding to the repeating carbohydrate lattices on the surface of microorganisms may result in their uptake directly by collectin receptors on macrophages or may trigger the activation of the classic complement pathway.3,4
The variation in MBL codons 52, 54, and 57 leads to secondary structural abnormalities of the MBL triple helix, which probably results in a failure to form functional higher order oligomers.3,4 Such variations have been shown to have dominant-negative effects on intermediate phenotypes of MBL activity.3,4 This is the likely func- tional basis for the observed association of MBL variation in codons 52, 54, and 57 with an increased susceptibility to certain infections, such as C. pneumoniae.3,4 Some exper- iments have implicated infection with C. pneumoniae as a factor that contributes to atherosclerosis.2,5 Our observa- tions are consistent with a dominant effect of the MBL non-A alleles leading to susceptibility to infection, that in turn leads to increased CPA. Alternatively, the associa- tions may have been due to linkage disequilibrium with unmeasured variation in MBL or another locus on chro- mosome 10.
Recent technical improvements in carotid ultrasound have widened its use as a research tool.11–22 There is a correlation between carotid ultrasound, usually of carotid intimal-medial wall thickness, and angiographic measure- ments.11–22 Previous concerns about discrepancies be- tween angiography and carotid ultrasound have been alle- viated by understanding that factors, which affect lesion development in the artery wall, do not always affect the appearance of the lumen on angiography.23,24 As plaque thickens, the artery enlarges so that the intima is exposed to a constant shear rate; thus, significant plaque can de- velop before stenosis occurs.23,24 The complexity of the atherogenesis could weaken the relationship between any single indirect variable, the underlying pathology, and the expression of disease. B-mode ultrasound measurement of CPA may thus be complementary to other measures, such as carotid intimal-medial wall thickness, in the assessment of the progression of preclinical atherosclerosis, and for monitoring its rate of progression.
Our data are consistent with the idea that variation in CPA is associated with variation in the MBL gene. Perhaps subjects who have increased susceptibility to infections are prone to develop more atherosclerosis as assessed noninvasively. Alternatively, homozygotes for MBL A/A are relatively more resistant to infection compared with subjects with other MBL genotypes, and across a sample of such homozygotes, there is relatively less burden of arte- rial disease that would result from an infectious etiology. Our study cohort was small and somewhat heterogeneous: it would be important to confirm these findings in samples taken from different patient groups. Furthermore, the mean difference between genotypes in CPA was very small, and the clinical relevance remains to be determined. It also remains to be prospectively shown whether the development atherosclerosis end points from infection de- pends upon MBL genotype, or whether knowledge of the MBL genotype can have an impact on specific interven- tions to reduce the risk of vascular disease.
Footnotes
Acknowledgments
This project was supported by operating grants from the Heart & Stroke Foundation of Canada (NA3628), by the Medical Research Coun- cil of Canada (MT13430), by the Heart and Stroke Foundation of Ontario (B3738), and by Factor Research Innovations Inc. Dr. Hegele is Career Investigator of the Heart and Stroke Foundation of Ontario (CI2979).