Paraoxonase Activity, But Not Haplotype Utilizing the Linkage Disequilibrium Structure, Predicts Vascular Disease
Objective— The effects of paraoxonase (PON1) activity and of genetic variation in the PON1 promoter and coding region on carotid artery disease (CAAD) were investigated.
Methods and Results— We identified functional promoter polymorphisms and examined their effects in a cohort with and without CAAD. We used the full sequences in 23 white subjects to determine the linkage disequilibrium (LD) structure of the PON1 region and to direct the grouping of haplotypes for disease association testing. There are several discrete regions of the PON1 gene with strong local LD, but the useful levels of LD do not extend across the entire gene. Indeed, PON1−162/−108/55/192 haplotype did not predict additional variation in PON1 activities compared with the 4 genotypes separately. PON1 hydrolysis activity predicted CAAD status, but this was not attributable to the promoter or coding region polymorphisms or haplotype or to the effects of smoking or statin use on PON1 activity.
Conclusions— PON1 does not have LD across the gene, and use of haplotypes in association studies should consider the LD structure. PON1 activity predicts CAAD, yet 4 functional polymorphisms do not. Additional investigations of genetic and environmental factors that influence PON1 activity as a risk factor for vascular disease are warranted.
We have previously reported that paraoxonase (PON1) activity for the hydrolysis of paraoxon (POase) or diazoxon (DZOase) predicted carotid artery disease (CAAD) status, even though PON1Q192R, the Gln(Q) to Arg(R) substitution at amino acid 192, and PON1L55M, the Leu(L) to Met(M) substitution at amino acid 55, coding region polymorphisms were not marginally predictive of disease.1 This result has recently been confirmed to be true for coronary heart disease.2 The inhibition or reduction of atherogenic LDL oxidation by HDL seems, at least in part, to be a function of PON1, which is physically associated with the HDL particle.3–7⇓⇓⇓⇓ Thus, the PON1-CVD association is expected to result from the role of PON1 in the metabolism of bioactive lipid molecules and protection against damage caused by oxidized low-density lipoprotein (LDL). PON1 knockout mice are devoid of PON1 hydrolysis activity in plasma and liver8 and are more susceptible to atherosclerosis.9
See page 1317
Association studies of the PON1192 or PON155 polymorphisms with vascular disease have had mixed results, with only some finding an excess of the PON1192R or PON155L alleles associated with cardiovascular disease.10–25⇓⇓⇓⇓⇓⇓⇓⇓⇓⇓⇓⇓⇓⇓⇓ Our previous work suggested that unmeasured factors that alter PON1 hydrolysis activity were important in heart disease.1 We hypothesized that lowered protein concentration attributable to PON1 5′ regulatory region polymorphisms might predict both the reduced activity and disease status. We subsequently identified 5 promoter region polymorphisms26 and used in vitro luciferase expression studies to determine that 2 polymorphisms, PON1−108 and PON1−162, influenced PON1 expression levels.27 The PON1−108 polymorphism accounted for 23% of the variance in PON1 expression.27 Additionally, as detailed below, smoking may depress PON1 activity and statin use may elevate PON1 activity, so PON1 activity effects in vascular disease may also reflect these environmental exposures.
The goal of this study was to further the understanding of the role of PON1 activity in CAAD and determine if a fraction of the reduced activity in cases was attributable to differences in PON1 5′ region variations or to differences in smoking or statin use. We also considered haplotype effects, which offer the potential advantage of higher heterozygosity28 and may capture the effects of even unsampled disease-causing mutations.29 Gabriel et al30 identified sizable regions with little recombination and with only a few common haplotypes, sometimes referred to as haplotype blocks. These authors estimated that most of the human genome is contained in large blocks and that within a block, 90% of all chromosomes are captured by only 3 to 5 haplotypes. However, utility of haplotype depends on the LD structure of the gene of interest. Therefore, this was investigated for PON1 in 23 white subjects. The LD structure then determined the strategy for the testing of PON1 polymorphism effects in vascular disease.
In addition to paraoxon and diazoxon, PON1 hydrolyzes phenylacetate and the nerve agents soman and sarin. PON1 arylesterase (Aase) activity in the hydrolysis of phenylacetate was known to be only slightly impacted by the PON1192 polymorphism27,31⇓ and thus more reflective of the concentration of PON1 protein present. Because of our interest in the changes in the amount of protein that are expected by the functional 5′ region variants, we added Aase as a PON1 activity measure.
The present sample of 302 white men used to evaluate the role of PON1 in vascular disease was expanded from the 212 men previously reported to lack marginal effects of the PON1 coding regions on CAAD prediction.1 All subjects were collected from Puget Sound Veterans Affairs Health Care System (PSVAHCS). The cases had greater than 80% internal carotid artery stenosis, unilaterally or bilaterally, on angiography or had undergone a carotid endarterectomy. All controls were drawn from patients without codes for vascular disease who subsequently were shown to have had less than 15% internal carotid artery stenosis, bilaterally, on carotid duplex ultrasound. Subjects with total serum cholesterol greater than 400 mg/dL or with known coagulopathies were excluded. Subjects were matched by censored age (within 12 months). Race was ascertained from both PSVAHCS record and self-report. Censored age matching was based on the age at time of blood draw for controls and age at diagnoses of carotid artery disease for cases. The mean duration of documented vascular disease before sampling of cases was 2.6 years. The study was approved by both the University of Washington and the PSVAHCS human subject review processes. Subjects gave written informed consent. These subjects had a mean censored age of 65.8 years (range, 41 to 84 years). Of the 302 white male subjects, 38% were taking lipid-lowering medications (from the patients’ pharmacy medication history) and 68% were current smokers or former smokers (dichotomous traits, by self-report).
Twenty-three white individuals (46 chromosomes) from the Center d’Etude du Polymorphisme Humain (CEPH) genomics repository were fully sequenced for the PON1 region. This separate sample was used to generate the linkage disequilibrium data summarized in Figure 1.
PON1 Genotype and Activity Phenotype Methods
DNA was prepared from buffy coat preparations by a modification of the procedure of Miller et al32 using Puregene reagents (Gentra). All genotyping was conducted by polymerase chain reaction amplification followed by polymorphism-specific restriction digestion and gel electrophoresis. The PON1Q192R polymorphism was detected by Alw I digestion and the PON1L55M polymorphism was detected by Nla III digestion as published previously.33 The genotype of the PON1C-108T polymorphism and the PON1A-162G polymorphism were determined by Bst UI digestion as described26,27⇓ Genotype distributions did not significantly differ from Hardy-Weinberg equilibrium expectations.
PON1 paraoxon (POase activity) and diazoxon (DZOase activity) hydrolysis rates were measured spectrophotometrically with lithium heparin plasma, as described.34 All samples were run in duplicate; the averaged value was used for analysis. PON1192 genotype can be predicted with high accuracy from examination of the 2D plot of paraoxon and diazoxon hydrolysis rates.34 When assignments did not match, both genotyping and phenotyping studies were repeated. Three hundred subjects had genotype-phenotype agreement. Two subjects genotyped as PON1192QR and phenotyped as PON1192RR, indicating a nonfunctional Q allele. Sequencing of the coding regions confirmed the PON1192QR genotype but failed to detect a coding region change that would account for the inactive allele. Arylesterase activities were performed in triplicate as described.27
Lipid measurements were performed on fasting whole plasma. Standard enzymatic methods were used to determine levels of total cholesterol, triglycerides, and HDL cholesterol on an Abbott Spectrum analyzer.35–37⇓⇓ LDL cholesterol was calculated.38 Apolipoprotein AI measurement methods were as previously reported.39
Linkage disequilibrium across the 26-kb PON1 region was assessed in the 23 white subjects from the CEPH sample that were fully sequenced for this region by the SeattleSNPs program. These sequences are publicly available at http://pga.mbt.washington.edu. The linkage disequilibrium statistics D′ and r2 were both computed for each pair of single nucleotide polymorphisms (SNPs) with an observed minor allele frequency (MAF) of greater than or equal to 10%. Given 2 SNPs (A and B) with 2 alleles each (A1/A2 and B1/B2), the basic metric of LD is D, the difference between the observed number of A1B1 haplotypes and the expected number of A1B1 haplotypes if genotypes were independent at A and B (D=pA1B1−pA1*pB1). The range of D depends on the allele frequencies at A and B, so it is difficult to compare D between marker pairs. D′ normalizes D to the maximum possible D given the allele frequencies at A and B, so the range of D′ is (−1,1) for all pairs of SNPs. |D′|=1 if only 3 of the 4 possible haplotypes are observed, so it is a useful statistic for detecting recombination between A and B. r2 is the Pearson correlation coefficient for alleles at A and B. The range of r2 depends on how similar allele frequencies are at A and B. If A and B have the same allele frequency, then the range of r2 is (0,1); otherwise the range is (0,<1). r2=1 if only 2 of the 4 possible haplotypes are observed, meaning that genotype at A and B is perfectly correlated, so r2 is a useful statistic for determining whether an effect associated with B could be detected by genotyping at A. Haplotypes were inferred for all sites with MAF >5% using the program PHASE.40 This allowed us to examine how well haplotypes constructed from the 4 genotyped polymorphisms capture the overall haplotype structure of the gene.
POase (sqrt-POase), DZOase (sqrt-DZOase), and Arylesterase (sqrt-Aase) were square root transformed to reduce positive skewness. Preliminary analyses included the estimation of the variance in sqrt-POase, sqrt-DZOase, and sqrt-Aase attributable to PON1192, PON155, PON1−108, or PON1−162 genotypes and haplotypes using the r2 from linear regression. This was done separately for cases and controls. The additional variance in PON1 activity measures explained by current smoking or pack years smoked, statin use, and levels of HDL, HDL2, and HDL3 were sequentially evaluated.
Matched logistic regression was used to test for effects in the prediction of CAAD cases (coded as 1) versus controls (coded as 0). The matched analyses were done using the Cox regression function in SPSS with CAAD status as the outcome, the pair ID as the strata, and a constant survival time of 1 for all subjects and included the covariates to be tested. Separate regressions were tested for a PON1−108 or PON1−162 genotype effect or a POase or DZOase effect on the prediction of CAAD status. Another logistic regression tested whether the addition of genotype information for all 4 polymorphisms altered the significance of sqrt-POase, sqrt-DZOase, or sqrt-Aase in the prediction of CAAD status. All regression analyses used SPSS 8.0 for Windows.41
Patterns of Linkage Disequilibrium Between Common Polymorphic Sites in the PON1 Gene Region
The linkage disequilibrium structure of the PON1 gene region is shown graphically in Figure 1. The LD statistic D′ (shown below the diagonal) is useful in detecting historical recombination events. A D′ value less than 1 for a pair of SNPs means that all 4 possible 2 SNP haplotypes exist, suggesting either recombination between the SNPs, recurrent mutation, or gene conversion. As shown, the PON1 gene is not characterized by a single large nonrecombinant haplotype block but rather by several smaller relatively nonrecombinant regions. Relatively nonrecombinant blocks span the regions of the promoter (SNPs 63 to 4754, positions reported relative to Genbank record AF539592), the amino acid 55 substitution (SNPs 6807 to 12194), the amino acid 192 substitution (SNPs 12826 to 25203), and the 3′ untranslated region (SNPs 26625 to 29021), but significant numbers of recombination events have occurred at the boundaries between these regions.
Where D′ is useful in detecting recombination, r2 describes how closely correlated the genotype is between a pair of SNPs, which is critical for association studies. Considering an r2 above 0.5 as potentially useful for detecting untyped sites,42 it is clear that typing a single SNP in any given region yields information about some nearby sites, but power to detect untyped sites in other regions of PON1 is dramatically lower. For example, D′=0.775 between PON155Leu and PON1192Arg, suggesting that a modest number of recombinant chromosomes exist. However, the correlation between PON155Leu and PON1192Arg is weak (r2=0.087) because the rare allele at each SNP is associated with the common allele at the other SNP, and therefore typing either SNP alone would provide little power to detect risks associated with the other polymorphism.
The observed patterns of D′ and r2 suggest that a modest number of sites need to be typed within each nonrecombinant region, consistent with a small number of haplotypes across each region. However, recombination between these relatively tractable regions has generated dramatic haplotype diversity across PON1. Indeed, using only 2 promoter and 2 coding region polymorphisms, the PHASE program inferred that 12 of 16 possible haplotypes were observed, with 8 haplotypes occurring at a frequency of >0.3% (Table 1). In regions with low levels of recombination, cladistic analysis would allow for grouping of haplotypes based on evolutionary relationships. Given the high frequency of recombinant haplotypes across PON1, we treated the 2 promoter area SNPs (PON1−162/−108, D′=1) as a 2-site haplotype and analyzed the PON155 and PON1192 sites separately.
Finally, we were able to estimate the amount of common genetic (MAF >0.1) variation in the PON1. Sixty SNPs were identified at greater than 10% MAF in the white CEPH population. In the present study, 4 of these sites were typed and 56 were unassayed. Of those 56, 44 (78.6%) exceed r2=0.25 with 1 or more of the 4 assayed SNPs; 29 (51.8%) exceed r2=0.5; 16 (28.6%) exceed r2=0.75; and 12 (21.4%) are perfectly associated (r2=1) with 1 or more of the assayed SNPs.
Relationship Between PON1 Genotypes and Phenotypes
As expected, the bivariate Pearson correlation between sqrt-POase and sqrt-DZOase was −0.01 (NS), although the correlation was high within PON1192 genotypes. The correlation between sqrt-POase and sqrt-Aase was 0.22 (P<0.01), and the correlation between sqrt-DZOase and sqrt-Aase was 0.70 (P<0.01). Pack-years smoked was not significantly correlated with sqrt-DZOase (correlation=−0.052, P=0.42), sqrt-Aase (−0.067, P=0.30), or POase (−0.019, P=0.77). Current smokers had significantly lower PON1 sqrt-POase (P=0.019) but not sqrt-DZOase or sqrt-Aase (P=0.43) activity levels, although all activities tended to be lower in smokers.
Table 2 summarizes the portion of variances in each PON1 activity measure attributable to current age, all 4 PON1 genotypes measured, statin drug use, smoking behavior, and levels of HDL, HDL2, and HDL3. Smoking behavior was evaluated both as square root of pack-years smoked and current smoking. Accurately reported, current smoking is expected to impact PON1 activity more than smoking history.43 Cases had a larger total variance and larger portion of unexplained variance relative to controls, except for sqrt-POase, which was largely explained by PON1192 genotype in both cases and controls (Table 2). When jointly considered, all 4 genotypes had statistically significant effects on DZOase in controls at the P<0.05 level, and only the PON1−162 polymorphism did not significantly add to the prediction of DZOase in CAAD cases. When jointly considered, the PON1192 and PON1−108 polymorphisms predicted variation in POase, as expected, whereas PON1−162 and PON155 did not in either cases or controls. The fact that PON1 haplotypes constructed using all 4 sites did not explain a greater fraction of variance in the phenotypes than an analysis including all 4 sites separately seems to confirm that the haplotype construction did not provide additional information about unassayed functional sites in the PON1 gene.
Phenotype and Genotype Effects on Vascular Disease Prediction
As a single predictor, sqrt-DZOase (P=0.016) was a significant predictor of CAAD status using matched logistic regression, with cases having lowered activity, as expected (Table 3). Sqrt-POase (P=0.355) and sqrt-Aase (0.200) were not significant for the prediction of CAAD status, although sqrt-POase became highly predictive when genotypes were jointly considered (Table 3, model 3). All 3 PON1 activities were lower in cases than controls within the PON1192QQ or PON1192QR genotypes, although this was not seen in the small number of RR subjects. Separate matched analysis rejected significant marginal effects of PON1−108 (P=0.30) or PON1−162 (P=0.49) genotype, as well as PON1−108/−162 haplotype (P=0.54) in the prediction of CAAD status (Table 3). Figure 2 demonstrates the genotype or haplotype frequencies for both cases and controls. When all PON1−108/−162 haplotypes and PON155 and PON1192 genotypes were jointly considered, none showed marginal significance for the prediction of CAAD status (Table 3, model 1).
PON1−162/−108, PON155, and PON1192 did not become significant (all P>0.2) when sqrt-DZOase was considered along with the 4 genotypes in the regression model (Table 3, model 2). However, sqrt-POase (P=0.001) and PON1192 genotype (P=0.02, degrees of fraction=2) did become significant in the prediction of CAAD but the PON1−162/−108 (P>0.2) and PON155 (P>0.2) did not when sqrt-POase was considered along with the 4 genotypes in the regression model (Table 3, model 3).
Evidence of historical recombination events at several positions within the PON1 gene was observed, which restricts LD to multiple small areas within and surrounding the gene. Within each small, nonrecombinant region, a limited number of local haplotypes was observed, but across the entire gene, recombination between regions has generated a large number of haplotypes. Although the recombinant structure of the PON1 locus is most obvious from the sequencing of the full gene in a group of individuals, it can also be inferred from the large number of haplotypes predicted from just 4 genotyped sites. Thus, cladistic analysis of the entire gene is not appropriate, because significant numbers of recombinant haplotypes exist but might be useful in smaller nonrecombinant regions if multiple SNPs were assayed within each region. However, whole-gene haplotypes might be useful for the prediction of disease if functional variants existed in multiple nonrecombinant regions, if the effects of these variants were not simply additive, and if appropriate groupings of haplotypes could be made to allow for reasonable power.
The finding of lowered PON1 activity in vascular disease without evident PON1 coding region genotype effects1 has been recently confirmed in a cohort with coronary heart disease2 and had previously been reported in myocardial infarction survivors.44 We previously hypothesized that the lower PON1 POase and DZOase activity found in CAAD cases versus controls, not attributable to PON1 coding polymorphisms, may be due to lowered PON1 serum concentrations.1 One speculation was that frequent PON1 promoter region polymorphisms might, at least in part, account for that result. We subsequently identified functional PON1−162 and PON1−108 polymorphisms26,27⇓ and apparently nonfunctional PON1−909, PON1-832, and PON1−126 polymorphisms. These polymorphisms do not account for reduced PON1 activity in CAAD versus controls, despite their large marginal effects on PON1 activity, nor do they improve disease prediction in this cohort. Leviev et al45 reported that PON1−909GG subjects had a marginally lower risk of myocardial infarction, particularly in subjects <50 years old. They detected no effect in subjects older than 60 years, similar to the result in our older subjects. Additionally, PON1 arylesterase activity, which is more closely related to protein level than POase or DZOase activity, performed more poorly as a predictor of CAAD status than DZOase activity. These data suggest that PON1 substrate specific activity, not simply quantity, may be important in the determination of vascular disease risk. These data continue to support the importance of PON1 activity measurement, rather than only PON1 genotype, for studies of cardiovascular risk. Indeed, we have recently shown that PON1 phenotype-genotype discordance can be used to detect novel PON1 nonsense and missense mutations.46
What factors may account for the lower PON1 activity in the cases remains unclear. Cases have a higher variance in activity than controls. It is noteworthy that 65% of the variance in PON1 DZOase activity in controls is accounted for by the 4 genotypes measured here, but only 43% of that in the cases is explained along these lines (Table 2). The LD data show substantial sequence variation in the PON1 region that is not captured by disequilibrium to typed sites. Considering the unassayed sites with MAF greater than 0.1, 48% had r2<0.5, with 1 of the 4 sites assayed in this cohort. The extent to which such unmeasured genetic variation contributes to the difference in PON1 activity between cases and controls remains to be determined. It will also be important to identify other genes that influence PON1 levels. Intervention and animal model studies will be of use in understanding the role of environmental factors in PON1 expression.
Addition of smoking and statin use and HDL level–related factors did not explain the difference in explained PON1 activity between cases and controls. This still leaves a large gap in our understanding of the sources of variation in cases versus controls, which requires additional investigation. PON1 activity may be altered by environmental factors, including tobacco smoking. Tobacco smoke depressed PON1 activity in ex vivo assays,47 and current tobacco use has been reported to lower POase activity.43 In a Costa Rican sample, PON1192 genotype predicted myocardial infarctions only in nonsmokers,48 suggesting the importance of jointly considering environmental factors that modify PON1 activity. Additionally, elevation of PON1 activity in statin drug users was consistent with the reported effects of Simvastatin on PON1 activity.49 PON1 POase activity has been found to increase with pharmacological therapy with Simvastatin, possibly by reducing oxidative stress.49 Dietary fat may also affect PON1 activity,50 because the concentration of PON1 mRNA has been shown to increase or decrease in a strain-dependent fashion when mice are given an atherogenic diet.51 Lipid, lipoprotein, or apolipoprotein levels have been weakly associated with PON1 activity or genotype in some20,52–54⇓⇓⇓ but not all12,55⇓ studies.
In conclusion, neither the common PON1 5′ functional regulatory variants PON1−162 and PON1−108 nor the environmental impact of smoking and statin use explain the relationship of PON1 activity to disease status or the large unexplained PON1 activity variance in cases. The PON1 gene is not contained in a single haplotype block, which suggests that promoter and coding region variants should be treated as independent factors and not grouped by haplotype for disease association studies. However, consideration of PON1 haplotype effects did not improve CAAD case-control prediction. This work continues to support the importance of measuring expression of genes and activity of enzymes in addition to genetic polymorphisms within and regulating a given gene.
This work was funded by the National Institutes of Health (NIH) grant HL67406 and the Veteran Affairs Epidemiology Research and Information Center Program (No. CSP 701S), with additional funding from an American Heart Association Physician Scientist Award (G.P.J.) and NIH awards T32AG00057, P30ESO7033, ES09883, HL66682, and HL66642. The authors would like to thank the subjects for their participation and thank the following people for their technical assistance: Laura McKinstry, Jeff Rodenbaugh, Andy Louie, and Tianji Yu.
Paraoxonase (PON1) does not have linkage disequilibrium across the gene but does have localized linkage disequilibrium. Thus, use of haplotypes in association studies should be informed by the linkage disequilibrium structure. PON1 activity predicts carotid artery disease, yet 4 functional polymorphisms, PON1−162PON1-108PON155 and PON1192, and related haplotypes do not.
- Received February 12, 2003.
- Accepted May 19, 2003.
- ↵Jarvik G, Rozek LS, Brophy VH, Hatsukami TS, Richter RJ, Schellenberg GD, Furlong CE. Paraoxonase (PON1) phenotype is a better predictor of vascular disease than is PON1(192) or PON1(55) genotype. Arterioscler Thromb Vasc Biol. 2000; 20: 2441–2447.
- ↵Mackness B, Gershan KD, Turkie W, Lee E, Roberts DH, Hill E, Roberts C, Durrington P, Mackness M. Paraoxonase status in coronary heart disease: are activity and concentration more important than genotype? Arterioscler Thromb Vasc Biol. 2001; 21: 1451–1457.
- ↵Mackness M. The human paraoxonase polymorphism and atherosclerosis. In: Mackness MI, Cleric M, eds. Esterases, Lipases, and Phospholipases. New York: Plenum Press; 1994.
- ↵Watson A, Berliner JA, Hama SY, La Du BN, Fuall KF, Fogelman AM, Navab M. Protective effect of high density lipoprotein associated paraoxonase: inhibition of the biological activity of minimally oxidized low density lipoprotein. J Clin Invest. 1995; 96: 2882–2891.
- ↵Serrato M, Marian AJ. A variant of human paraoxonase/arylesterase (HUMPONA) gene is a risk factor for coronary artery disease. J Clin Invest. 1995; 96: 3005–3008.
- ↵Zama T, Murata M, Matsubara Y, Kawano K, Aoki N, Yoshino H, Watanabe G, Ishikawa K, Ikeda Y. A 192Arg variant of the human paraoxonase (HUMPONA) gene polymorphism is associated with an increased risk for coronary artery disease in the Japanese. Arterioscler Thromb Vasc Biol. 1997; 17: 3565–3569.
- ↵Sanghera D, Saha N, Aston CE, Kamboh MI. Genetic polymorphism of paraoxonase and the risk of coronary heart disease. Arterioscler Thromb Vasc Biol. 1997; 17: 1067–1073.
- ↵Pfohl M, Koch M, Enderle MD, Kuhn R, Fullhase J, Karsch KR, Haring HU. Paraoxonase 192 Gln/Arg gene polymorphism, coronary artery disease, and myocardial infarction in type 2 diabetes. Diabetes. 1999; 48: 623–627.
- ↵Antikainen M, Murtomaki S, Syvanne M, Pahlman R, Tahvanainen E, Jauhiainen M, Frick MH, Ehnholm C. The Gln-Arg191 polymorphism of the human paraoxonase gene (HUMPONA) is not associated with the risk of coronary artery disease in Finns. J Clin Invest. 1996; 98: 833–835.
- ↵Schmidt H, Schmidt R, Niederkorn K, Gradert A, Schumacher M, Watzinger N, Hartung HP, Kostner GM. Paraoxonase PON1 polymorphism leu-met54 is associated with carotid atherosclerosis: results of the Austrian Stroke Prevention Study. Stroke. 1998; 29: 2043–2048.
- ↵Ombres D, Pannitteri G, Montali A, Candeloro A, Seccareccia F, Campagna F, Cantini R, Campa PP, Ricci G, Arca M. The Gln-Arg192 polymorphism of human paraoxonase gene is not associated with coronary artery disease in Italian patients. Arterioscler Thromb Vasc Biol. 1998; 18: 1611–1616.
- ↵Chen Q, Reis SE, Kammerer CM, McNamara DM, Holubkov R, Sharaf BL, Sopko G, Pauly DF, Bairey Merz CN, Kamboh MI for the WISE Study Group. Association between the severity of angiographic coronary artery disease and paraoxonase gene polymorphisms in the National Heart, Lung, and Blood Institute: Sponsored Women’s Ischemia Syndrome Evaluation (WISE) Study. Am J Hum Genet. 2003; 72: 13–22.
- ↵Stephens J, Schneider JA, Tanguay DA, Choi J, Acharya T, Stanley SE, Jiang R, Messer CJ, Chew A, Han JH, Duan J, Carr JL, Lee MS, Koshy B, Kumar AM, Zhang G, Newell WR, Windemuth A, Xu C, Kalbfleisch TS, Shaner L, Arnold K, Schulz V, Drysdale CM, Nandabalan K, Judson RS, Ruano G, Vovis GF. Haplotype variation and linkage disequilibrium in 313 human genes. Science. 2001; 293: 489–493.
- ↵Gabriel SB, Schaffner SF, Nguyen H, Moore JM, Roy J, Blumenstiel B, Higgins J, DeFelice M, Lochner A, Faggart M, Liu-Cordero SN, Rotimi C, Adeyemo A, Cooper R, Ward R, Lander ES, Daly MJ, Altshuler D. The structure of haplotype blocks in the human genome. Science. 2002; 296: 2225–2229.
- ↵Miller S, Dykes DD, Polesky HF. A simple salting out procedure for extracting DNA from human nucleated cells. Nucl Acids Res. 1988; 16: 1215.
- ↵Warnick G, Benderson J, Albers JJ. Dextran sulfate-Mg2+ precipitation procedure for quantitation of high-density-lipoprotein cholesterol. Clin Chem. 1982; 28: 1379–1388.
- ↵Friedewald W, Levy RI, Fredrickson DS. Estimation of the concentration of low-density lipoprotein cholesterol in plasma, without use of the preparative ultracentrifuge. Clin Chem. 1972; 18: 499–502.
- ↵Marcovina S, Albers JJ, Henderson LO, Hannon WH. International Federation of Clinical Chemistry standardization project for measurements of apolipoproteins A-I and B, III: comparability of apolipoprotein A-I values by use of international reference material. Clin Chem. 1993; 39: 773–781.
- ↵SPSS. SPSS Statistical Algorithims. Chicago, IL: SPSS, Inc; 1991.
- ↵James R, Leviev I, Righetti A. Smoking is associated with reduced serum paraoxonase activity and concentration in patients with coronary heart disease. Circulation. 2000; 101: 2252–2257.
- ↵Ayub A, Mackness MI, Arrol S, Mackness B, Patel J, Durrington PN. Serum paraoxonase after myocardial infarction. Arterioscler Thromb Vasc Biol. 1999; 19: 330–335.
- ↵Sen-Banerjee S, Siles X, Campos H. Tobacco smoking modifies association between Gln-Arg192 polymorphism of human paraoxonase gene and risk of myocardial infarction. Arterioscler Thromb Vasc Biol. 2000; 20: 2120–2126.
- ↵Tomas M, Senti M, Garcia-Faria F, Vila J, Torrents A, Covas M, Marrugat J. Effect of simvastatin therapy on paraoxonase activity and related properties in familial hypercholesterolemic patients. Arterioscler Thromb Vasc Biol. 2000; 20: 2113–2119.
- ↵Sutherland W, Walker RJ, de Jong SA, van Rij AM, Phillips V, Walker HL. Reduced postprandial serum paraoxonase activity after a meal rich in used cooking fat. Arterioscler Thromb Vasc Biol. 1999; 19: 1340–1347.
- ↵Hegele R, Brunt JH, Connelly PW. Multiple genetic determinants of variation of plasma lipoproteins in Alberta Hutterites. Arterioscler Thromb Vasc Biol. 1995; 15: 861–871.
- ↵Nevin D, Zambon A, Furlong CE, Richter R, Humbert R, Hokanson JE, Brunzell JD. Paraoxonase genotypes, lipoprotein lipase activity and HDL. Arterioscler Thromb Vasc Biol. 1996; 16: 1243–1249.