Author + information
- Received February 18, 2009
- Revision received May 4, 2009
- Accepted May 27, 2009
- Published online September 29, 2009.
- Jakub J. Regieli, MD*,†,
- J. Wouter Jukema, MD, PhD‡,§∥,* (, )
- Pieter A. Doevendans, MD, PhD*,
- Aeilko H. Zwinderman, PhD#,
- John J. Kastelein, MD, PhD**,
- Diederick E. Grobbee, MD, PhD† and
- Yolanda van der Graaf, MD, PhD†
- ↵*Reprint requests and correspondence:
Dr. J. Wouter Jukema, Department of Cardiology C5-P, Leiden University Medical Center, P.O. Box 9600, Leiden 2300 RC, the Netherlands
Objectives We investigated the effects of paraoxonase (PON)-1 variants on long-term clinical outcome in patients with coronary artery disease (CAD).
Background PON-1 is a potential therapeutic target to further reduce cardiovascular risk because it is a detoxifying esterase with antioxidant properties. The PON-1 knockout models result in higher susceptibility to atherosclerosis, and PON activity contributes to cardiovascular risk in humans. Human gene variants determine PON activity; however, the impact of these variants on recurrent cardiovascular events in vascular disease is as of yet unknown.
Methods We conducted a 10-year follow-up study of 793 CAD patients in the REGRESS (REgression GRowth Evaluation Statin Study) trial cohort, using nationwide registries. Genotypes were obtained of 2 PON-1 isotypes (L55M, rs854560, and Q192R, rs662), which were previously associated with PON activity. Absolute and relative risks by genotype were estimated using Kaplan-Meier and proportional hazards analyses.
Results Carriership of the PON-1 glutamine isotype at codon 192 and methionine at codon 55 was associated with a higher risk of death due to ischemic heart disease. Hazard ratios per allele copy were 1.71 (95% confidence interval: 1.0 to 2.8, p = 0.03) for the glutamine isotype at codon 192 and 1.56 (95% confidence interval: 1.1 to 2.3, p = 0.03) for methionine at codon 55. Both isotypes had previously been related to lower PON activity. No effect was observed on all-cause mortality.
Conclusions PON-1 gene variants influence the 10-year risk of fatal complications from CAD in male patients, despite no effect on all-cause mortality. These long-term findings confirm functional data on PON-1 activity, emphasize the relevance of this pathway in vascular disease, and enforce its putative role as a target to modify and estimate cardiovascular risk.
Low plasma levels of high-density lipoprotein cholesterol (HDL-C) or apolipoprotein AI constitute independent risk factors for the development of cardiovascular disease in all populations studied (1,2). Among patients with clinically apparent coronary artery disease (CAD), plasma HDL-C levels are inversely related to cardiovascular risk (3,4). The mechanisms underlying these relations are, at present, incompletely understood. Basic science has recognized a number of causal roles that elements of high-density lipoprotein (HDL) particles may play in the pathogenesis of new and recurrent cardiovascular events. First, proteins physically attached to HDL mediate reverse cholesterol transport from tissues toward the liver and may thereby modify the progression of atherosclerosis.
A second explanation for the link between HDL-C and cardiovascular outcome is the antioxidant capacity of HDL (5). The formation of foam cells is initiated by oxidization of low-density lipoprotein (LDL) particles induced by a number of processes in the arterial wall (6–10). Oxidized LDL is taken up by macrophage scavenger receptors and attracts peripheral monocytes by inducing chemoattracting factors (11,12). Antioxidant effects of HDL particles have been principally attributed to paraoxonase (PON)-1, a plasma protein residing on HDL. More specifically, this protein seems to be anchored to HDL particles containing apolipoprotein AI and clusterin (apolipoprotein J) (13,14). In fact, in mice lacking the PON-1 gene, more extensive aortic sclerosis develops when the mice are on a high-fat diet (15).
The spectrum of PON-1 enzymatic properties comprises the capacity to break down biologically active oxidized phospholipids and oxidized cholesteryl esters (12,16–18). Moreover, hydrolytic activity of PON-1 toward a number of other endogenous and exogenous esters has also been described, yet other substrates occurring in vivo may remain to be discovered (14). Although it has been proposed that the catalytic sites for different substrates vary, the body of evidence concerning the consequences of PON-1 activity in humans has predominantly focused on hydrolysis of the toxic organophosphates paraoxon and phenylacetate (19).
In a recent cohort of 1,353 men with and without pre-existing CAD, baseline serum hydrolytic activity toward paraoxon was inversely related to the occurrence of fatal or nonfatal myocardial infarction (MI) over 15 years of follow-up (20).
In vitro mutagenesis and expression of human PON-1 resolved the location of the active catalytic site toward paraoxon and 2-naphtyl-acetate, along with the PON-1 tertiary structure (13). The regions involved in this active site comprise a previously described nonsynonymous genetic (A/G) polymorphism (rs662), resulting in glutamine (Q) to arginine (R) substitution at codon 192. This 192Q isoform has been related to lower paraoxon-hydrolyzing and arylesterase (phenylacetate-hydrolyzing) activity (14,21,22). Another nonsynonymous genetic (T/A) variant (rs854560) encodes for a leucine (L) or methionine (M) residue at codon 55. The 55M isoform has also been related to lower paraoxon-hydrolyzing and arylesterase activity (14,21,22).
PON is, therefore, among the potential new targets for secondary prevention of recurrent cardiovascular disease. Nevertheless, a recent meta-analysis of several (nested) case-control studies on the effect of PON-1 genotype in coronary heart disease in the open population reported a lack of such an effect (23). However, to date, no prospective studies have established the impact of PON on cardiovascular risk in the context of patients with manifest vascular disease. We therefore investigated the relation between these well-described gene variants on the 10-year risk of cardiovascular morbidity and mortality in a well-established, prospectively defined cohort of patients with ischemic heart disease (IHD).
The study participants were derived from the REGRESS (REgression GRowth Evaluation Statin Study), which enrolled 884 male, nondiabetic patients with symptomatic CAD between 1989 and 1993. The trial design and main findings have been reported previously (24,25). In brief, the primary objectives of this angiographic trial were to evaluate the effects of 24 months of 40-mg pravastatin therapy on the evolution of atherosclerotic lesions in male patients with documented CAD. Within the framework of the trial, clinical and angiographic follow-up was documented after the initial 2-year trial. The clinical outcomes, assessed by an independent clinical events committee, were fatal or nonfatal MI, death due to CAD, repeated coronary revascularization (coronary artery bypass grafting or percutaneous transluminal coronary angioplasty), stroke, and death due to noncoronary causes. All participants gave written informed consent, and the clinical trial and subsequent deoxyribonucleic acid (DNA) studies were approved by all 7 institutional review boards of the participating centers and by the medical ethics committees of all centers.
Morbidity and mortality 10-year follow-up study
To obtain 10-year follow-up data of the participants, cause-specific mortality until January 1, 2001, was extracted from nationwide registries. All diagnoses in these registers are coded according to the International Classification of Diseases, Clinical Modification (ICD-9 and ICD-10). The research protocol was approved by the institutional review board and ethics committee of the coordinating center (University Medical Center Utrecht, Utrecht, the Netherlands).
Linkage process method
The study database, comprising all 884 REGRESS participants, was linked to the registers on the basis of birth date, sex, and postal address code (26). As is customary because of privacy legislation, patient names were omitted in the linkage process. On a per-patient basis, historical registers of the Dutch inhabitants were searched for this unique combination of characteristics, and once found, were automatically merged into migration history over the follow-up time. The vital status of the participants was then obtained through linking municipal administration registries using a 6-character postal code. Of the 884 participants, 861 (97%) could be uniquely traced with this method. The 23 patients who could not be uniquely traced were right-censored at the end of the 24-month follow-up in the mortality analyses. Information on the occurrence of nonfatal MI was obtained through linkage with the registry of the hospital discharge diagnoses, using the 4-digit part of the postal code. This registry files admissions of all general and university hospitals in the Netherlands. Of the 884 participants, 740 (84%) could be uniquely traced. The 144 patients who could not be uniquely traced were right-censored at the end of the 24-month follow-up for morbidity analyses.
In the outcome events analyses, we considered the primary causes of death and the primary clinical diagnosis recorded during hospitalization. The composite end point “death due to IHD” consisted of all primary causes of death within the ICD-9 codes 410 to 414 and ICD-10 codes I20 to I25. “Nonfatal MI or death due to IHD” additionally comprised the clinical occurrence of ICD-9-CM codes 4100 to 4109. The composite end point “death due to atherosclerotic disease” consisted of all primary causes of death within ICD-9 codes 410 to 414, 430 to 438, and 440 to 448, and ICD-10 codes I20 to I25, I60 to I69, I70 to I79, and F01.
Genomic DNA was extracted from blood collected at baseline according to standard procedures. Genotyping of the PON-1 L55M (dbSNP rs854560) and Q192R (dbSNP rs662) variants was performed using a multiplex polymerase chain reaction and immobilized probe-based assay (Roche Molecular Systems, Alameda, California), and read-outs were available from 791 and 793 participants, respectively (27).
Lipoprotein cholesterol content measurements
Total cholesterol was measured with an enzymatic kit (Boehringer Mannheim, Mannheim, Germany) and calibrated with a human serum calibrator. The HDL-C was measured after precipitation of apolipoprotein B–containing lipoproteins with a 4% tungstate solution and centrifugation, and the triglycerides (TG) were analyzed enzymatically (Bayer/Technicon, Pittsburgh, Pennsylvania) by a technique that included free glycerol (24). The LDL-C was calculated according to the Friedewald formula.
Within the framework of the original trial, patients underwent coronary arteriography upon enrollment. All angiograms were analyzed by quantitative coronary angiography, and the average mean segment diameter (reflecting the extent of diffuse atherosclerosis) and the average minimal obstruction diameter per patient (reflecting extent of focal atherosclerosis) were calculated as described previously (24).
Demographic, clinical, and angiographic characteristics and concentrations of lipids among the 3 genotypes of each polymorphism were tabulated (Table 1).For continuous values, mean values with standard deviations are displayed in principle, except for TG concentrations and minimal obstruction diameter, for which, due to non-normality, medians with range are displayed.
The relation between genotype and outcome events was analyzed through survival analyses using proportional hazards (Cox regression). The hazard ratios (HRs) were tabulated (Table 2).The mode of inheritance exerting the effects of these gene variants remains unknown. We considered codominant (additive) models in principle, and therefore, the HR of the defined end points was estimated by entering the number of rare allele copies as a linear covariate in the model. We further estimated risks while controlling for most recent HDL-C level measured within the randomized trial, LDL and TG levels, and current smoking, to provide more insights into clinical relevance of genotype as an independent risk factor and into underlying mechanisms. Absolute risk of the end points in the genotype groups were plotted using the Kaplan-Meier method, tested by the log-rank test, and p for trend values were displayed. All of the above mentioned statistical analyses were carried out using SPSS for Windows, release 14.0 (SPSS Inc., Chicago, Illinois).
Additionally, linkage between the L55M and Q192R variants was explored, and haplotype effects estimated using maximum likelihood estimation by means of the survival analysis function of the application Thesias (28), release 3.1. Throughout, a 2-tailed p value of 0.05 was interpreted as indicating a statistically significant difference. Analyses were performed by 2 of the authors (J.J.R. and A.H.Z.).
Distribution of allele frequencies
The study population comprised 325 (41%) leucine to leucine, 352 (45%) leucine to methionine, and 114 (14%) methionine to methionine subjects, and 420 (53%) glutamine to glutamine, 287 (36%) glutamine to arginine, and 87 (11%) arginine to arginine subjects. Participants' characteristics according to the L55M and Q192R variants are summarized in Table 1.
Genotype and cardiovascular risk
Figure 1displays Kaplan-Meier survival curves for the outcome events death due to IHD and all-cause mortality. After 10 years of follow-up, carriers of the 55M allele had a considerably increased risk of cardiovascular complications, compared with the L55 carriers. A marked allele–dose effect was visible from the survival curves. For instance, the 10-year absolute risk of death due to IHD was 4.6% (standard error 1.2%) in L55 homozygote patients, whereas it was 7.1% (standard error 1.5%) in heterozygote patients and 10.9% (standard error 3.0%) in 55M homozygote patients, as also displayed in Table 2. The M55 allele thus conferred a higher risk of death due to IHD (HR: 1.56, 95% confidence interval [CI]: 1.1 to 2.3, p = 0.03) and nonfatal MI or death due to IHD (HR: 1.34, 95% CI: 1.0 to 1.8, p = 0.04), but had no effect on all-cause mortality (HR: 1.00, 95% CI: 0.8 to 1.3, p = 0.99). Similarly, an elevated risk was found in 192Q variant carriers. The relations found remained unchanged upon correction for HDL-C, LDL, TG, and current smoking. Both polymorphisms were partly linked with statistical measures of linkage disequilibrium (D′ = 0.86 and r2= 0.17). No significant haplotype effects were observed among the >1% prevalent haplotypes, comprising LQ (36%), MQ (35%), and LR (28%).
The REGRESS 10-year follow-up study of 793 statin-treated male CAD patients provides evidence that genetic variation associated with lower PON activity is associated with increased long-term cardiovascular mortality and morbidity. Specifically, the 55M isotype of the PON-1 enzyme, present in 37% of our population as well as in other Caucasian samples, can be considered a risk factor for recurrent MI and death from IHD, with marked allele–dose effects observed (22). We can report similar findings for the 192Q isotype of PON-1. These genotype effects remained unchanged after adjustments, illustrating the clinical relevance of PON-1 as an independent risk factor.
Both of the PON-1 isotypes studied have been related to lower PON activity (22), which has been put forward as a causal factor that may provide 1 explanation for the inverse relation between HDL-C levels and cardiovascular risk. Recently, it was described that serum PON activity had an impact on the 15-year risk of fatal or nonfatal MI in 1,353 males from the general population (Caerphilly Prospective Study) (20). Of note, this effect seemed more pronounced in participants with pre-existing CAD (20).
Furthermore, the role of PON-1 isotypes in the development of cardiovascular disease has been addressed previously in several case-control studies. A systematic review of the case-control evidence seems at odds with the current findings, since the combined evidence indicates that the 192R allele occurs slightly more frequently in patients with cardiovascular disease than in control subjects (20,23). However, given the evidence of publication bias in the combined data, the authors reach the conclusion that PON-1 genotype distribution does not vary between cases and controls (20,23). Notably, the combined evidence in these analyses included various ethnic groups and patient categories in both cases and controls, including patients with diabetes mellitus and nephropathy. These differences in baseline characteristics may underlie the discrepancy with the current REGRESS study findings. The applicability of these case-control findings may be limited, furthermore, in context of the significance of PON in a population with established CAD. In fact, in the presence of clinically apparent cardiovascular disease, the determinants of cardiovascular prognosis may differ from those in the general population (29–31). Moreover, 1 potential limitation of case-control studies—those not nested in particular—is potential survival bias because genetic factors pre-disposing to high case-fatality might be under-represented among cases because of natural selection. To the best of our knowledge, the current study provides the first prospective data on the impact of the PON-1 L55M and Q192R variants in patients with clinically manifest vascular disease. These long-term findings confirm functional data on PON-1 activity and support a causal role of PON-1 in the pathogenesis of recurrent complications of CAD.
The exact mechanism mediating the effects of PON-1 isotypes in recurrent CAD cannot be deduced from the current data but should be considered in the context of current clinical evidence.
Regarding biological functionality of the described isoforms, considerable data are available on enzymatic activities (14,21,22). Seemingly at odds with the current long-term study are early data suggesting that the 55L and 192R isoforms of PON-1 have relatively low peroxidase activity toward copper-induced oxidated LDL particles (14). However, this difference appeared to be reversed when PON-1 was added at later time points, at which the 55M and 192Q isoforms were more capable of protecting against LDL oxidation (19). Although this evidence relies on a limited number of patients, the latter situation indeed rhymes with the cardioprotective effect of 192R and 55L found in the current long-term study.
Notably, quantitative coronary angiography indexes of coronary atherosclerotic burden did not vary by PON-1 genotypes (Table 1, angiographic characteristics). These findings are in synergy with the reports by Turban et al. (32) and Chen et al. (33), who did not observe an effect on angiographic end points. In view of the presently available evidence of no effects of PON-1 genotype on angiographic outcome, it could be argued that the extent of atherosclerotic burden might insufficiently explain the effect of PON-1 genotypes on outcome. Other mechanisms need to be considered in this respect. It has been suggested that high PON activity is related to slow coronary blood flow (34). Another explanation might emerge from findings on statin therapy for patients with hypercholesterolemia that indicate a pharmacogenetic interaction such that the 192R allele is associated with a higher HDL-C increment with statin therapy (35,36). In fact, statins influence PON-1 transcriptional activity, and that may also be partly relevant for normocholesterolemic CAD patients included in the current report (37–39), particularly since, given the positive effects of pravastatin therapy observed after the initial 24-month follow-up in the REGRESS trial, all participants and treating physicians were informed of the trial outcomes and explicitly recommended to start or continue statin therapy. Indeed, a survey at 5 years after completion of the trial provided the necessary data that 91% of patients were actually receiving statin therapy. Thus, as the great majority received statin therapy for the last 8 years of the follow-up period, it seems a plausible explanation that the PON-1 192R carriers may have benefited more from statin therapy in terms of clinical outcome (40,41). Such pharmacogenetic interaction might account for the long-term impact of the genotypes reported here, clearly meriting further exploration in this direction.
Moreover, it has been suggested from in vitro data that PON may affect cholesterol efflux from macrophages to copper-oxidized HDL (16). That could imply that HDL may mediate the effect of PON-1 genotype on cardiovascular outcome. However, the HDL-C–adjusted risk estimates presented in Table 2did not support such an explanation for the clinical outcomes observed. This finding is in line with results from the Caerphilly Prospective Study in which serum PON activity affected risk independent of baseline HDL-C level (20). In turn, such findings could be a reflection of the fact that PON-1 is present on only certain subfractions of HDL.
Study strengths and limitations
To appreciate these findings, some aspects of our study merit consideration. First, the data originate from a randomized trial in which assessment of the benefits of pravastatin treatment were the primary objective. Because the study medication taken during 2 years of the follow-up time was allocated at random, namely, irrespective of genotype, that would not have affected our findings. Second, our follow-up dataset was not complete for all patients: 3% and 16% of the full cohort could not be uniquely identified in the mortality and hospital registries, respectively. Because these patients were right-censored at lost-to-follow-up time, again it seems unlikely that this would have affected the primary outcome of the current study. We elected to calculate actuarial survival across genotypes, in contrast to the case-control design. Survival analysis enabled us to efficiently study all available information, including that of censored participants. Third, an important issue is that the results in this study were obtained in a cohort of male Caucasian patients with established CAD, and further research is needed to demonstrate whether the results of our study also apply to women and patients of different ethnicity. Nevertheless, these are the first prospective data describing the long-term effects of PON-1 on recurrent complications of CAD, studied in a substantial cohort of patients, and adding significant understanding of the role of PON-1 in cardiovascular disease.
Interventions to enhance PON activity
Several nutritional factors such as vitamin C and E may influence PON activity, albeit to a limited extent, and the therapeutic value has been questioned in clinical trials (42,43). In vitro data indicate that dietary polyphenols such as resveratrol may induce PON-1 gene expression (44,45). At present, it is known that statins to some extent modulate PON activity; however, no specific modulators of PON activity or PON-1 gene expression have been reported (37). Effects of nicotinic acid on PON activity have not been reported thus far; nevertheless, given the powerful HDL-C increasing effects of this compound, this relation merits further investigation.
Among symptomatic male CAD patients, carriership of the PON-1 192Q or 55M alleles appears to increase the 10-year risk of MI and mortality from IHD. All-cause mortality was not affected by these genotypes. The current data support the view that PON may be a potential target of secondary prevention.
Drs. Jukema and Kastelein are established clinical investigators of the Netherlands Heart Foundation. Dr. Kastelein is a consultant for all pharmaceutical companies that produce low-density lipoprotein lowering medication. The original REGRESS trial was sponsored by Bristol-Myers Squibb, New York. The current work was supported by grants from the Netherlands Organisation for Health Research and Development (AGIKO grant 920-03-367 to Dr. Regieli and program grant 904-65-095 to Drs. Regieli, Grobbee, and van der Graaf) and the Interuniversity Cardiology Institute of the Netherlands (project number 15 to Drs. Jukema and Zwinderman), and by the Bekalis Foundation (Dr. Doevendans). These funding sources had no involvement in the writing of this paper or its submission.
- Abbreviations and Acronyms
- coronary artery disease
- confidence interval
- high-density lipoprotein
- high-density lipoprotein cholesterol
- hazard ratio
- ischemic heart disease
- low-density lipoprotein
- low-density lipoprotein cholesterol
- myocardial infarction
- Received February 18, 2009.
- Revision received May 4, 2009.
- Accepted May 27, 2009.
- American College of Cardiology Foundation
- Sacks F.M.,
- Moye L.A.,
- Davis B.R.,
- et al.
- Sacks F.M.,
- Tonkin A.M.,
- Shepherd J.,
- et al.
- Palinski W.,
- Rosenfeld M.E.,
- Yla-Herttuala S.,
- et al.
- Henriksen T.,
- Mahoney E.M.,
- Steinberg D.
- Henriksen T.,
- Mahoney E.M.,
- Steinberg D.
- Navab M.,
- Imes S.S.,
- Hama S.Y.,
- et al.
- Navab M.,
- Berliner J.A.,
- Subbanagounder G.,
- et al.
- Durrington P.N.,
- Mackness B.,
- Mackness M.I.
- Aviram M.,
- Billecke S.,
- Sorenson R.,
- et al.
- Mackness B.,
- Durrington P.,
- McElduff P.,
- et al.
- van Himbergen T.M.,
- Roest M.,
- de Graaf J.,
- et al.
- Jukema J.W.,
- Bruschke A.V.,
- van Boven A.J.,
- et al.
- Herings R.M.,
- Bakker A.,
- Stricker B.H.,
- Nap G.
- Tregouet D.A.,
- Ricard S.,
- Nicaud V.,
- et al.
- Grobbee D.E.
- Koerselman J.,
- Van der Graaf Y.,
- de Jaegere P.P.,
- Grobbee D.E.
- Oliver M.F.
- Chen Q.,
- Reis S.E.,
- Kammerer C.M.,
- et al.
- Deakin S.,
- Leviev I.,
- Guernier S.,
- James R.W.
- Tomas M.,
- Senti M.,
- Garcia-Faria F.,
- et al.
- Jarvik G.P.,
- Tsai N.T.,
- McKinstry L.A.,
- et al.
- Gouedard C.,
- Barouki R.,
- Morel Y.
- Gouedard C.,
- Barouki R.,
- Morel Y.