Genetic Inflammatory Factors Predict Restenosis After Percutaneous Coronary Interventions
http://www.100md.com
《循环学杂志》
the Department of Cardiology (P.S.M., N.M.M.P., W.R.P.A., B.J.M.v.V., D.E.A., A.v.d.L., E.E.V.D.W., J.W.J.), the Department of Surgery (A.S., P.H.A.Q.), and the Department of Human Genetics, Center for Human and Clinical Genetics (R.R.F.), Leiden University Medical Center, Leiden
Interuniversity Cardiology Institute of the Netherlands (ICIN) (P.S.M., W.R.P.A., P.A.D., J.W.J.), Utrecht
Gaubius Laboratory, TNO PG (N.M.M.P., A.S., P.H.A.Q., B.J.M.v.V.), Leiden
the Department of Medical Statistics (A.H.Z.) and the Department of Cardiology (R.J.d.W.), Academic Medical Center, Amsterdam
the Department of Hematology, Erasmus University Medical Center (M.P.M.d.M.), Rotterdam
the Department of Cardiology, HLCU, University Medical Center Utrecht (P.A.D.)
the Department of Cardiology, University Medical Center Groningen (R.A.T.)
the Department of Cardiology, Academic Hospital Maastricht (J.W.), Maastricht, the Netherlands.
Abstract
Background— Restenosis is a negative effect of percutaneous coronary intervention (PCI). No clinical factors are available that allow good risk stratification. However, evidence exists that genetic factors are important in the restenotic process as well as in the process of inflammation, a pivotal factor in restenosis. Association studies have identified genes that may predispose to restenosis, but confirmation by large prospective studies is lacking. Our aim was to identify polymorphisms and haplotypes in genes involved in inflammatory pathways that predispose to restenosis.
Methods and Results— The GENetic DEterminants of Restenosis (GENDER) project is a multicenter prospective study, including 3104 consecutive patients after successful PCI. Forty-eight polymorphisms in 34 genes in pathways possibly involved in the inflammatory process were analyzed. The 16Gly variant of the 2-adrenergic receptor gave an increased risk of target vessel revascularization (TVR). The rare alleles of the CD14 gene (–260T/T), colony-stimulating factor 2 gene (117Thr/Thr), and eotaxin gene (–1328A/A) were associated with decreased risk of TVR. However, through the use of multiple testing corrections with permutation analysis, the probability of finding 4 significant markers by chance was 12%.
Conclusions— Polymorphisms in 4 genes considered involved in the inflammatory reaction showed an association with TVR after PCI. Our results may contribute to the unraveling of the restenotic process. Given the explorative nature of this analysis, our results need to be replicated in other studies.
Key Words: genetics restenosis risk factors inflammation angioplasty
Introduction
Restenosis is still the major limitation of percutaneous coronary interventions (PCI), resulting from injury of the vessel wall caused by balloon dilation and stent placement.1,2 The vascular damage is characterized by irritation of endothelial and subendothelial structures and injury of medial regions with rupture of the internal elastic lamina. This damage causes segmental thrombus formation and subsequent invasion of macrophages and polymorphonuclear leukocytes, followed by expression and release of numerous growth factors and cytokines from blood cells and stretched smooth muscle cells, leading to proliferation of smooth muscle cells.3,4 Vascular inflammation thus plays an important role in this complex multifactorial process.5–7
Editorial p 2378
Identifying patients at increased risk of restenosis may improve stratification of patients to individually tailored treatment. Thus far, however, it has proven difficult to stratify patients with regard to risk of coronary restenosis based only on clinical or procedural risk factors, because risk factors in relation to restenosis identified so far have not been consistently reported.8 There are indications that genetic factors explain part of the excessive risk of restenosis independent of conventional clinical variables. In patients with multivessel disease, the incidence of restenosis of a second lesion was 2.5 times higher if the first lesion had restenosis, even after adjustments for well-known patient-related risk factors, including diabetes and hypertension.9 Inflammatory responsiveness is highly genetically determined. Many studies have demonstrated genetic influences on the inflammatory response of an individual.10,11 Therefore, it is plausible that differences in genetic makeup of inflammatory genes between individuals may explain part of the risk of the at least partially inflammation-driven restenotic process. Association studies have identified several candidate genes that may predispose to restenosis, such as the genes for stromelysin-1, IL6, E-selectin, CD18, CD14, and IL1 receptor.12,13 However, these studies are mostly inconclusive because of a number of limitations, including limited study size and study design (eg, not prospective studies or studies using selected patient groups), limiting the clinical value of the observations.
The purpose of the present study was to evaluate in a large unselected study sample whether a variety of genetic determinants, considered to be involved in the inflammatory process, can predict the risk of clinical restenosis after PCI.
Methods
Study Design
The present study sample has been described previously.14 In brief, the GENetic DEterminants of Restenosis project (GENDER) was designed to study the association between various gene polymorphisms and clinical restenosis defined in our study by target vessel revascularization (TVR). Patients were eligible for inclusion if they were successfully treated for stable angina, non–ST-elevation acute coronary syndromes or silent ischemia by PCI. Patients treated for acute ST-elevation myocardial infarction were excluded. All patients were treated in 4 referral centers for interventional cardiology in the Netherlands (Academic Medical Center Amsterdam, University Medical Center Groningen, Leiden University Medical Center, and Academic Hospital Maastricht). The overall inclusion period lasted from March 1999 until June 2001. In total, 3104 consecutive patients were included in this prospective, multicenter, follow-up study.
The study protocol conforms to the Declaration of Helsinki and was approved by the medical ethics committees of each participating institution. Written informed consent was obtained from each participant before the PCI procedure.
PCI Procedure
Standard angioplasty and stent placement were performed by experienced operators, using a radial or femoral approach. Before the procedure, patients received 300 mg aspirin and 7500 IU heparin. The use of intracoronary stents and additional medication, such as glycoprotein IIb/IIIa inhibitors, was at the discretion of the operator. If a stent was implanted, patients received either ticlopidine or clopidogrel for at least 1 month after the procedure, depending on local practice. During the study, no drug-eluting stents were used.
Follow-Up and Study End Points
Follow-up lasted at least 9 months or until a coronary event occurred. Patients were either seen in the outpatient clinic or contacted by telephone. TVR, either by PCI or CABG, was considered as the primary end point because it is considered most relevant for clinical practice by regulatory agencies. An independent clinical events committee adjudicated the clinical events.
Events occurring within 1 month after the procedure were excluded from the analysis because these events were attributable to subacute stent thrombosis or occluding dissections.
Data were collected with standardized case report forms that were completed by the research coordinator at each site, who was blinded to the genotype of the patients. Representatives from the data coordinating center monitored the sites.
Genetic Methodology
Blood was collected in EDTA tubes at baseline, and genomic DNA was extracted by following standard procedures. Genotyping was performed by a validated multilocus genotyping assay to test several markers of inflammation (Roche Molecular Systems).15,16 Inflammation is known to play an important role in the development of restenosis, although it is not fully elucidated which factors are exactly involved in the process. Therefore, we intended to examine a very broad spectrum of factors that are considered to have an effect on the inflammatory response. The polymorphisms on the array were chosen on the basis of their relation to inflammation previously described in literature reports of gene associations with inflammatory diseases. Furthermore, polymorphisms were also selected on the basis of their known functionality or their known allele frequency. In addition, developing the assay for the selected polymorphism had to be possible. A total of 48 markers in 34 genes were examined by using these criteria (Table 1). Each DNA sample was amplified in multiple polymerase chain reactions (PCRs), using biotinylated primers. The PCR product was then hybridized to a corresponding panel of sequence-specific oligonucleotide probes that had been immobilized in a linear array on nylon membrane strips.17 A colorimetric detection method based on incubation with streptavidin–horseradish peroxidase conjugate, using hydrogen peroxide and 3,3,5,5-tetramethylbenzidine as substrates, was used. Operators blinded to patient data performed genotyping. To confirm genotype assignments, PCR analysis was randomly performed in replicate on 10% of the samples. Two independent observers carried out scoring. Disagreements (<1%) were resolved by further joint reading, and when necessary a repeat genotyping reaction was performed.
Statistical Methodology
Deviations of the genotype distribution from that expected for a sample in Hardy-Weinberg equilibrium were tested by using the 2 test with 1 degree of freedom. Allele frequencies were determined by gene counting. The 95% confidence intervals of the allele frequencies were calculated from sample allele frequencies, based on the approximation of the binomial and normal distributions in large sample sizes.
In the first stage, we determined the association between each of the 48 polymorphisms and TVR by using a Cox proportional regression model. We considered codominant, dominant, and recessive inheritance models, and the model with the lowest Akaike information criterion was used.18 If fewer than 10 patients were homozygous for a particular allele, the homozygotes and heterozygotes were taken together, thereby assuming a dominant model. No adjustment for covariates was performed at this stage to allow for the assessment of the possible involvement of the polymorphisms in the causal pathway for TVR. Haplotypes were constructed from polymorphisms known to be located in the same gene, or in genes located near each other, and in statistically significant linkage disequilibrium (all probability values of the Pearson 2 test <0.001). The effect of haplotypes on restenosis risk was estimated according to the methods developed by Tanck et al.19 We considered only one TVR per patient.
The robustness of these "individual" findings was investigated by a bootstrap study of 1000 bootstrap samples drawn from the original data set. In each bootstrap sample, the optimal inheritance model as well as the association between each of the polymorphisms and TVR was determined, and the percentage of bootstrap samples with a significant association was counted for each polymorphism.
By following the SNP selection method of Hoh et al,20 we performed multivariable regression analysis of the TVR risk with all polymorphisms and haplotypes having an individual probability value of 0.10 or less and being significant in at least 40% of the bootstrap samples. First, the association between each polymorphism and TVR was adjusted for the confounding effect of the clinical risk factors age and sex and other clinical and intervention-related risk factors that were significantly (P<0.10) associated with TVR, being diabetes, stenting, residual stenosis >20%, current smoking, total occlusion, and hypertension. If 2 or more arterial segments were treated, we only used intervention-related characteristics of the most severely affected segment.
Second, polymorphisms with independent prognostic value were selected through the use of a multivariable regression model, using a stepwise backward selection algorithm.
No multiple testing correction method was applied to keep power at an acceptable level, and we therefore performed a permutation study to assess the experiment-wide error rate. One thousand permutation samples were created by reshuffling at random the 304 observed TVRs over all available patients. In each permutated (reshuffled) sample, the same statistical analysis was performed as described above, and we counted the number of reshuffled samples with 0, 1, 2, 3, 4, or 5 significant polymorphisms in the final multivariable regression model. In reshuffled data, no significant association is expected, and the percentage of reshuffled samples with 1 or more significant polymorphisms is a quantification of the experiment-wide error rate.
Statistical analysis was carried out using SPSS 11.5 (SPSS Inc).
Results
Patient Characteristics
The characteristics of this patient sample have been described previously.14 In summary, a total of 3146 patients had a complete follow-up (99.3%), with a median duration of 9.6 months (interquartile range, 3.9). Of 3146 patients, 42 had an event in the first 30 days. These patients were excluded from further analysis, according to the protocol. The remaining 3104 patients had a mean age of 62.1±10.7 years. Of the patients, 888 (28.6%) were female and 453 (14.6%) had diabetes mellitus. The majority of the patients were white (97%); 812 patients (26.2%) received glycoprotein IIb/IIIa inhibitors. Stents were used in 2309 (74.4%) patients. A total of 4061 lesions were treated in this unselected patient sample. Complex (type C) lesions, classified according to the modified American College of Cardiology and American Heart Association Task Force classification, were treated in 802 patients (25.8%). Other patient characteristics, as well as details of the interventions, are summarized in Table 2.
Follow-Up
Of the 3104 patients, 304 (9.8%) patients underwent TVR during follow-up. Fifty-one patients died (1.6%) and 22 (0.7%) had a myocardial infarction.
Inflammation Array Results
In the present study, genotyping was performed in 3029 patients of the total sample. Results of the remaining patients (n=75, 2.4%) are lacking due to unavailable DNA or inconclusive genotyping. Patients who could not be genotyped did not differ in any characteristic from those who could be genotyped.
The following polymorphisms were associated with TVR (P<0.05): 2-adrenergic receptor (ADRB2) Arg16Gly, CD14–260C/T, colony stimulating factor 2 (CSF2) Ile117Thr, and small inducible cytokine subfamily A, member 11, alias Eotaxin (CCL11)-1328G/A (Table 3), being significant in 65.0%, 61.8%, 73.7%, and 69.3% of bootstrap samples, respectively. Interleukin-4 receptor (IL4R) Gln576Arg, and T-cell transcription factor (TCF7) Pro19Thr polymorphisms also showed a tendency of association with TVR (0.05
TVR occurred more often in ADRB2 Gly16 homozygotes (11.3%) than in Arg16Gly heterozygotes (8.7%) or Arg16 homozygotes (9.1%). In contrast, CD14 T/T-genotypes had decreased TVR risk (7.6%) compared with C/T genotypes (10.5%) and C/C genotypes (10.3%). CSF2 Thr homozygotes (6.6%) and Ile/Thr heterozygotes (8.5%) also showed a decreased risk of TVR compared with Ile genotypes (10.8%). CCL11–1328 A/A-homozygotes had decreased TVR risk (4.3%) compared with G/A-heterozygotes (8.6%) and G/G-homozygotes (10.5%) (Table 3). More details are provided in Data Supplement Table I (http://circ.ahajournals.org/cgi/content/full/112/16/2417/DC1). As arterial remodeling is more evident in plain balloon angioplasty and neointima formation is more pronounced in stenting,21 we examined whether the association between the polymorphisms and TVR differed between patients who received stents and patients who did not receive stents. Analyzing the interaction between balloon angioplasty/stenting and the significantly associated genes with TVR showed no significant difference (CD14, P=0.34; ADRB2, P=0.83; CSF2, P=0.30 and CCL11, P=0.67), nor was there a significant difference between angioplasty and stented patients with regard to the association of any of the other polymorphisms with TVR (P>0.09). Because the power to find an interaction between type of PCI (angioplasty versus stent) and associations is limited, such interactions cannot be fully excluded. Separately within the balloon angioplasty–treated patient group and the stented patient group, the hazard ratios are in the same direction, but many of the 95% confidence intervals cross the value of 1.0 (see Data Supplement Table II).
Linkage Disequilibrium and Haplotypes
The three IL4R polymorphisms were in linkage disequilibrium (LD) (P<0.0001), and the same applied to the polymorphisms of the ADRB2, CTLA4, NOS3, CCL11, IL1A, IL1B, and IL6 genes. Furthermore, LD was found between polymorphisms in the IL4, IL13, and CSF2 genes. Significant LD was observed between SELE and the SELP (Val640Leu) polymorphism and involving the two CCR5 and the CCR2 polymorphisms. Except for a haplotype in the ADRB2 gene, none of the haplotypes were significantly associated with TVR risk (P>0.13). ADRB2 haplotypes, including the ADRB2 16Gly allele, were all associated with increased TVR risk, whereas haplotypes without the ADRB2 16Gly allele had lower TVR risk. In subsequent analyses, we therefore used the ADRB2 16 genotype and not these haplotypes.
Multivariable Cox Regression
Adjustment of the association between the selected polymorphisms and TVR for the clinical risk factors age, sex, diabetes, stenting, residual stenosis >20%, current smoking, hypertension, and total occlusion changed little (last column of Table 3); ADRB2 Arg16Gly, CD14–260C/T, CSF2 Ile117Thr, and CCL11–1328G/A polymorphisms remained significantly related to TVR.
Subsequently, we included all selected polymorphisms in the multivariable regression analysis and included the aforementioned clinical risk factors. After stepwise backward selection among the selected polymorphisms, the same polymorphisms appeared to be significantly related to TVR (Table 4).
Finally, we considered the association of the 21 possible 2-way interaction terms between the 7 selected polymorphisms and TVR. Only 1 interaction term was found to be statistically significant (P=0.04) between TCF7 and CSF2. The relative risk of TVR in CSF2 Thr carriers versus noncarriers was 0.37 (95% CI, 0.17 to 0.79) in TCF7 Thr carriers and 0.83 (95% CI, 0.67 to 1.03) in TCF7 noncarriers.
To illustrate the clinical relevance of our finding, the log-relative risks of the genotypes defined by the 5 polymorphisms with a P<0.1 selected in the Cox model were divided into quartile groups. We calculated the Kaplan-Meier curves in the first quartile-group (low risk, n=746), the second and third quartiles (medium risk, n=1488), and the fourth quartile (high risk, n=749). At 9 months after intervention, TVR risks were 5.0%, 9.2%, and 12.9% in these respective groups, and at 12 months 5.3%, 11.0%, and 14.3%, respectively (Figure). The high-risk quartile consisted almost exclusively of 435 patients having the TCF7 19Pro/Pro, CSF2 117Ile/Ile, CD14–260CC/CT, and the ADRB2 16Gly/Gly genotype. The low- and medium-risk quartiles consisted of patients with several genotype combinations.
Cumulative target vessel revascularization (TVR) risks in low-, medium-, and high-risk groups, based on genotype. Kaplan-Meier curves in the first quartile group (low risk, n=746), the second and third quartiles (medium risk, n=1488), and the fourth quartile (high risk, n=749) were calculated. At 9 months after intervention, TVR risks were 5.0%, 9.2%, and 12.9% and at 12 months 5.3%, 11.0%, and 14.3%, respectively. The high-risk quartile consisted almost exclusively of the 435 patients having the TCF7 19Pro/Pro, CSF2 117Ile/Ile, CD14–260CC/CT, and the ADRB2 16Gly/Gly genotype. The low- and medium-risk quartiles consisted of patients with several genotype combinations.
Error Rate
The experiment-wide type I error rate of the present study was assessed with a permutation study. Of 1000 permutated (reshuffled) data set, there were 97 (10%) in which none of the 48 markers were significantly related to TVR. Thirty-one percent of the reshuffled data sets had 1 significant marker, 25% of the reshuffled data sets had 2, 22% had 3, and 12% had 4 or more significant markers. The median false detection rate was only 2 markers.
Discussion
In a prospective, multicenter, follow-up study, we investigated 48 polymorphisms from 34 different genes. Our assumption that a relation exists between these genes and the development of restenosis after PCI was based on observations suggesting a role of these genes in the process of inflammation, a well-known determinant in the development of restenosis. After multivariable analysis, we identified polymorphisms in the CD14, 2-adrenergic receptor (ADRB2), colony stimulating factor 2 (CSF2), and eotaxin (CCL11) genes that were significantly associated with restenosis after PCI. Although neointimal formation is more pronounced after stenting and remodeling is prominent after plain balloon angioplasty, stenting did not give change the associations between the 4 genes and restenosis.
CD14
The –260 T/T genotype of CD14 was found to be protective against restenosis after PCI. Two previous studies have investigated the role of CD14 in the development of restenosis, one being a prospective study by Zee et al22 in 779 patients and the other being a prospective study by Shamada et al23 in 129 patients. They found the –260 T/T genotype to be a risk factor for restenosis. Our data are in conflict with these findings, which may be explained by a biological significance of CD14 that differs between Japanese subjects and whites, as well as a small sample size. However, the discrepant results of Zee et al and ours are not yet explained and await further study.
ADRB2
The second gene we found to be associated with TVR is the 2-adrenergic receptor gene (ADRB2), located on chromosome 5q31-q32. ADRB2s are cell-surface receptors that on binding to norepinephrine activate cellular adenylylcyclase through coupling to G-proteins. The ADRB2 gene has a role in the inflammatory response, because adrenoceptors are present on human platelets, and ADRB2 stimulation activates platelet nitric oxide synthase (NOS).24 NOS catalyzes the formation of NO, which has an inhibitory role on leukocyte adhesion, platelet adhesion and aggregation, smooth muscle cell proliferation, and synthesis of matrix proteins, and it promotes endothelial survival and proliferation.25 In addition, ADRB2 has an effect on the immune system because lymphocytes express ADRB2s.26
The polymorphism in the ADRB2 gene that after adjusted analyses showed to be a risk factor for restenosis was the 16A/G polymorphism that results in an amino acid change of glycine to arginine at position 16 (Arg16Gly). Patients with homozygosity for the 16Gly variant had a higher risk of TVR compared with patients with the 16Arg variant (11.3% versus 9.1%, respectively). Previous in vivo and in vitro studies have suggested that this Arg16Gly variant may differently affect functional responses to adrenergic stimulation, thereby possibly modulating cardiovascular and metabolic phenotypes. It has been reported that the 16Gly variant of ADRB2 is associated with faster agonist-induced downregulation of the receptor, as compared with the 16Arg variant.27 The higher risk of TVR may be related to less vasodilatation as a result of the downregulation of the receptor containing 16Gly, as compared with the receptor containing 16Arg. Moreover, downregulation of the ADRB2 could result in impaired inhibition of platelet aggregation.24
CSF 2
The polymorphism in the colony stimulating factor 2 gene (CSF2) that is significantly associated with restenosis is the 117T/C polymorphism, which results in an isoleucin for threonin substitution on position 117. The Thr117 variant showed a protective association with TVR. The functional effect of this CSF2, also known as granulocyte-macrophage colony stimulating factor, polymorphism still has to be investigated.
CCL11
The fourth polymorphism that was associated with restenosis is eotaxin (CCL11), a CC chemokine that is localized on chromosome 17. The –1328A/A promoter variant of this gene demonstrated a protective association with TVR. Economou et al28 reported that eotaxin is elevated in plasma of patients with advanced atherosclerosis. The plasma level of eotaxin in their study rose in the first day after PCI and declined to baseline in the following 3 months. In what way the polymorphism determines the expression level on a protein basis is as yet unknown.
Recently, Humphries et al29 have put forward the criteria for a genetic variant to be included in clinical risk management of patients with cardiovascular disease. For candidate variants, there should be enough circumstantial evidence from literature or from theoretical points of view to be involved in the disease and the studies should have enough power to detect a relative risk of 1.25 or more. The genotype must then give a predictive value in carriers over and above established risk factors, and the final data set should show no significant evidence for heterogeneity of risk effect. Finally, for each selected gene locus only functional variants (ie, variants that alter an amino acid or a transcription factor–binding element in a promoter region demonstrated in vitro) should be included.29 With these criteria, Humphries et al referred to genotypes encoding a specific phenotype, such as Factor V Leiden, for venous thrombosis. The polymorphisms we examined are explorative and chosen on basis of their known involvement in the multiple pathways of inflammation, being potentially implicated in the development of restenosis. However, they have no strong direct a priori theoretical value in terms of biological plausibility as meant by Humphries et al. However, we believe that the 4 factors associated with restenosis identified in the present study meet these criteria considerably.
Because it is not only the inflammatory response that causes restenosis, more research and confirmation of our findings are needed before these genetic variants could be used for making a genetic risk profile for patients at increased risk of restenosis.
In addition, circulating protein levels were not assessed in the present study. Basal (pre-PCI) plasma levels of the gene product probably do not reflect genetically determined differences in reaction to a trauma such as PCI. Moreover, local differences in reactions (in the vessel wall at the place of PCI) may not be determined systemically. In the human situation, it is impossible to measure gene products locally in the acute phase of treatment or the following days, and several months later the causal trigger has probably already disappeared.
Limitations of the Study
The 48 polymorphisms examined in the present study represent only a small proportion of genetic information that is potentially associated with TVR. However, by looking at a broad spectrum of polymorphisms in genes that are considered to be involved in the inflammatory response, we did try to cover a large set of factors that may be associated with restenosis. Furthermore, the candidate gene approach currently remains the most practical approach. Second, one or more of the SNPs associated with TVR in our study may be in linkage disequilibrium with other polymorphisms in the gene or with other nearby genes that are actually responsible for the development of this condition.
We did not apply adjustment for multiple testing, which sometimes is considered appropriate for hypothesis-testing studies. Moreover, our experiment-wide error rate was found to be 12% in the permutation analysis. In addition, we performed additional hypothesis testing in our haplotype analysis, which further increases the multiple testing problem. Our results should be primarily seen as hypothesis generating and need independent validation.
In addition, data on our haplotype analysis is limited due to the fact that we do not comprehensively cover the haplotype structure of our selected genes. Thus, individuals who appear identical as our haplotypes are concerned may very well differ when more polymorphisms are taken into account.
Our study has insufficient statistical power to examine whether the genetic associations we observed differed by sex because there were only 888 female patients, of whom 84 had TVR.
Another possible limitation is that we examined TVR as our primary end point instead of angiographic outcomes such as late loss. This could have caused a problem with ascertainment. However, in clinical practice clinical restenosis is an end point much more valuable than angiographic restenosis.
Furthermore, the 4 factors we found show a small hazard ratio (0.7 to 1.3), but it should be taken into account that the process of restenosis is multifactorial, involving multiple genes. Thus, relatively small hazard ratios relating to contribution of a single gene to restenosis might be of paramount importance in the overall process, and even a small genetic risk may identify a gene with an important biological role that could reveal new mechanistic insights and provide novel therapeutic targets.
Finally, because our study was conducted in a sample of white patients, extrapolation of the data to other ethnic groups should be done with great caution.
Conclusions
We show that genetic variants in 4 different genes that are considered to be involved in the inflammatory response may play a role in the development of restenosis. Three of these genotypes have, to our knowledge, not been described before in relation to restenosis. Because of the explorative nature of this analysis, our results need to be replicated in other studies to confirm the data. After this, our data could contribute to the unraveling of the process of restenosis and thereby provide novel therapeutic targets as well as contribute to development of improved risk stratification of patients who are scheduled for elective PCI, thereby creating the opportunity to individualize treatment in the future.
Acknowledgments
P.S. Monraats and W.R.P. Agema were supported by grant 99.210 from the Netherlands Heart Foundation and a grant from the Interuniversity Cardiology Institute of the Netherlands (ICIN). J.W. Jukema is an Established Clinical Investigator of the Netherlands Heart Foundation (2001 D 032). N.M.M. Pires was supported by grant 2001 T 32 from the Netherlands Heart Foundation. A. Schepers and P.H.A Quax (Established Investigator) were supported by the Molecular Cardiology Program of the Netherlands Heart Foundation (M 93.001). The contribution of the members of the clinical event committee, J.J. Schipperheyn, MD, PhD, J.W. Viersma, MD, PhD, D. Düren, MD, PhD, and J. Vainer, MD, is greatly acknowledged. The authors thank Lori Steiner and Karen Walker for their efforts in developing the genotyping reagents used for this study. They also thank Paul Schiffers from the University of Maastricht for his assistance in the laboratory.
Footnotes
The online-only Data Supplement can be found at http://circ.ahajournals.org/cgi/content/full/112/16/2417/DC1.
References
Fattori R, Piva T. Drug-eluting stents in vascular intervention. Lancet. 2003; 361: 247–249.
Pache J, Kastrati A, Mehilli J, Schuhlen H, Dotzer F, Hausleiter J, Fleckenstein M, Neumann FJ, Sattelberger U, Schmitt C, Muller M, Dirschinger J, Schomig A. Intracoronary stenting and angiographic results: strut thickness effect on restenosis outcome (ISAR-STEREO-2) trial. J Am Coll Cardiol. 2003; 41: 1283–1288.
Agema WR, Jukema JW, Pimstone SN, Kastelein JJ. Genetic aspects of restenosis after percutaneous coronary interventions: towards more tailored therapy. Eur Heart J. 2001; 22: 2058–2074.
Bhargava B, Karthikeyan G, Abizaid AS, Mehran R. New approaches to preventing restenosis. BMJ. 2003; 327: 274–279.
Serrano CV Jr, Ramires JA, Venturinelli M, Arie S, D’Amico E, Zweier JL, Pileggi F, da Luz PL. Coronary angioplasty results in leukocyte and platelet activation with adhesion molecule expression: evidence of inflammatory responses in coronary angioplasty. J Am Coll Cardiol. 1997; 29: 1276–1283.
Hojo Y, Ikeda U, Katsuki T, Mizuno O, Fukazawa H, Fujikawa H, Shimada K. Chemokine expression in coronary circulation after coronary angioplasty as a prognostic factor for restenosis. Atherosclerosis. 2001; 156: 165–170.
Welt FG, Rogers C. Inflammation and restenosis in the stent era. Arterioscler Thromb Vasc Biol. 2002; 22: 1769–1776.
Weintraub WS, Kosinski AS, Brown CL III, King SB III. Can restenosis after coronary angioplasty be predicted from clinical variables J Am Coll Cardiol. 1993; 21: 6–14.
Kastrati A, Schomig A, Elezi S, Schuhlen H, Wilhelm M, Dirschinger J. Interlesion dependence of the risk for restenosis in patients with coronary stent placement in multiple lesions. Circulation. 1998; 97: 2396–2401.
Bidwell J, Keen L, Gallagher G, Kimberly R, Huizinga T, McDermott MF, Oksenberg J, McNicholl J, Pociot F, Hardt C, D’Alfonso S. Cytokine gene polymorphism in human disease: on-line databases, supplement 1. Genes Immunol. 2001; 2: 61–70.
Verschuur M, Van Der Beek MT, Tak HS, Visser LG, De Maat MP. Interindividual variation in the response by fibrinogen, C-reactive protein and interleukin-6 to yellow fever vaccination. Blood Coagul Fibrinolysis. 2004; 15: 399–404.
de Maat MP, Jukema JW, Ye S, Zwinderman AH, Moghaddam PH, Beekman M, Kastelein JJ, van Boven AJ, Bruschke AV, Humphries SE, Kluft C, Henney AM. Effect of the stromelysin-1 promoter on efficacy of pravastatin in coronary atherosclerosis and restenosis. Am J Cardiol. 1999; 83: 852–856.
Kastrati A, Koch W, Berger PB, Mehilli J, Stephenson K, Neumann FJ, von Beckerath N, Bottiger C, Duff GW, Schomig A. Protective role against restenosis from an interleukin-1 receptor antagonist gene polymorphism in patients treated with coronary stenting. J Am Coll Cardiol. 2000; 36: 2168–2173.
Agema WRP, Monraats PS, Zwinderman AH, de Winter RJ, Tio RA, Doevendans PAFM, Waltenberger J, de Maat MPM, Frants RR, Atsma DE, van der Laarse A, van der Wall EE, Jukema JW. Current PTCA practice and clinical outcomes in the Netherlands: the real world in the pre–drug-eluting stent era. Eur Heart J. 2004; 25: 1163–1170.
Cheng S, Grow MA, Pallaud C, Klitz W, Erlich HA, Visvikis S, Chen JJ, Pullinger CR, Malloy MJ, Siest G, Kane JP. A multilocus genotyping assay for candidate markers of cardiovascular disease risk. Genome Res. 1999; 9: 936–949.
Barcellos LF, Begovich AB, Reynolds RL, Caillier SJ, Brassat D, Schmidt S, Grams SE, Walker K, Steiner LL, Cree BA, Stillman A, Lincoln RR, Pericak-Vance MA, Haines JL, Erlich HA, Hauser SL, Oksenberg JR. Linkage and association with the NOS2A locus on chromosome 17q11 in multiple sclerosis. Ann Neurol. 2004; 55: 793–800.
Saiki RK, Walsh PS, Levenson CH, Erlich HA. Genetic analysis of amplified DNA with immobilized sequence-specific oligonucleotide probes. Proc Natl Acad Sci U S A. 1989; 86: 6230–6234.
Li W, Nyholt DR. Marker selection by Akaike information criterion and Bayesian information criterion. Genet Epidemiol. 2001; 21 (suppl 1): S272–S277.
Tanck MW, Klerkx AH, Jukema JW, De Knijff P, Kastelein JJ, Zwinderman AH. Estimation of multilocus haplotype effects using weighted penalized log-likelihood: analysis of five sequence variations at the cholesteryl ester transfer protein gene locus. Ann Hum Genet. 2003; 67: 175–184.
Hoh J, Wille A, Zee R, Cheng S, Reynolds R, Lindpaintner K, Ott J. Selecting SNPs in two-stage analysis of disease association data: a model-free approach. Ann Hum Genet. 2000; 64: 413–417.
Agema WR, Jukema JW, Pimstone SN, Kastelein JJ. Genetic aspects of restenosis after percutaneous coronary interventions: towards more tailored therapy. Eur Heart J. 2001; 22: 2058–2074.
Shimada K, Miyauchi K, Mokuno H, Watanabe Y, Iwama Y, Shigekiyo M, Matsumoto M, Okazaki S, Tanimoto K, Kurata T, Sato H, Daida H. Promoter polymorphism in the CD14 gene and concentration of soluble CD14 in patients with in-stent restenosis after elective coronary stenting. Int J Cardiol. 2004; 94: 87–92.
Zee RY, Hoh J, Cheng S, Reynolds R, Grow MA, Silbergleit A, Walker K, Steiner L, Zangenberg G, Fernandez-Ortiz A, Macaya C, Pintor E, Fernandez-Cruz A, Ott J, Lindpainter K. Multi-locus interactions predict risk for post-PTCA restenosis: an approach to the genetic analysis of common complex disease. Pharmacogenomics J. 2002; 2: 197–201.
Queen LR, Xu B, Horinouchi K, Fisher I, Ferro A. Beta(2)-adrenoceptors activate nitric oxide synthase in human platelets. Circ Res. 2000; 87: 39–44.
Chen AF, Ren J, Miao CY. Nitric oxide synthase gene therapy for cardiovascular disease. Jpn J Pharmacol. 2002; 89: 327–336.
Wahle M, Stachetzki U, Krause A, Pierer M, Hantzschel H, Baerwald CG. Regulation of beta2-adrenergic receptors on CD4 and CD8 positive lymphocytes by cytokines in vitro. Cytokine. 2001; 16: 205–209.
Hoit BD, Suresh DP, Craft L, Walsh RA, Liggett SB. Beta2-adrenergic receptor polymorphisms at amino acid 16 differentially influence agonist-stimulated blood pressure and peripheral blood flow in normal individuals. Am Heart J. 2000; 139: 537–542.
Economou E, Tousoulis D, Katinioti A, Stefanadis C, Trikas A, Pitsavos C, Tentolouris C, Toutouza MG, Toutouzas P. Chemokines in patients with ischaemic heart disease and the effect of coronary angioplasty. Int J Cardiol. 2001; 80: 55–60.
Humphries SE, Ridker PM, Talmud PJ. Genetic testing for cardiovascular disease susceptibility:a useful clinical management tool or possible misinformation Arterioscler Thromb Vasc Biol. 2004; 24: 628–636.(Pascalle S. Monraats, MSc)
Interuniversity Cardiology Institute of the Netherlands (ICIN) (P.S.M., W.R.P.A., P.A.D., J.W.J.), Utrecht
Gaubius Laboratory, TNO PG (N.M.M.P., A.S., P.H.A.Q., B.J.M.v.V.), Leiden
the Department of Medical Statistics (A.H.Z.) and the Department of Cardiology (R.J.d.W.), Academic Medical Center, Amsterdam
the Department of Hematology, Erasmus University Medical Center (M.P.M.d.M.), Rotterdam
the Department of Cardiology, HLCU, University Medical Center Utrecht (P.A.D.)
the Department of Cardiology, University Medical Center Groningen (R.A.T.)
the Department of Cardiology, Academic Hospital Maastricht (J.W.), Maastricht, the Netherlands.
Abstract
Background— Restenosis is a negative effect of percutaneous coronary intervention (PCI). No clinical factors are available that allow good risk stratification. However, evidence exists that genetic factors are important in the restenotic process as well as in the process of inflammation, a pivotal factor in restenosis. Association studies have identified genes that may predispose to restenosis, but confirmation by large prospective studies is lacking. Our aim was to identify polymorphisms and haplotypes in genes involved in inflammatory pathways that predispose to restenosis.
Methods and Results— The GENetic DEterminants of Restenosis (GENDER) project is a multicenter prospective study, including 3104 consecutive patients after successful PCI. Forty-eight polymorphisms in 34 genes in pathways possibly involved in the inflammatory process were analyzed. The 16Gly variant of the 2-adrenergic receptor gave an increased risk of target vessel revascularization (TVR). The rare alleles of the CD14 gene (–260T/T), colony-stimulating factor 2 gene (117Thr/Thr), and eotaxin gene (–1328A/A) were associated with decreased risk of TVR. However, through the use of multiple testing corrections with permutation analysis, the probability of finding 4 significant markers by chance was 12%.
Conclusions— Polymorphisms in 4 genes considered involved in the inflammatory reaction showed an association with TVR after PCI. Our results may contribute to the unraveling of the restenotic process. Given the explorative nature of this analysis, our results need to be replicated in other studies.
Key Words: genetics restenosis risk factors inflammation angioplasty
Introduction
Restenosis is still the major limitation of percutaneous coronary interventions (PCI), resulting from injury of the vessel wall caused by balloon dilation and stent placement.1,2 The vascular damage is characterized by irritation of endothelial and subendothelial structures and injury of medial regions with rupture of the internal elastic lamina. This damage causes segmental thrombus formation and subsequent invasion of macrophages and polymorphonuclear leukocytes, followed by expression and release of numerous growth factors and cytokines from blood cells and stretched smooth muscle cells, leading to proliferation of smooth muscle cells.3,4 Vascular inflammation thus plays an important role in this complex multifactorial process.5–7
Editorial p 2378
Identifying patients at increased risk of restenosis may improve stratification of patients to individually tailored treatment. Thus far, however, it has proven difficult to stratify patients with regard to risk of coronary restenosis based only on clinical or procedural risk factors, because risk factors in relation to restenosis identified so far have not been consistently reported.8 There are indications that genetic factors explain part of the excessive risk of restenosis independent of conventional clinical variables. In patients with multivessel disease, the incidence of restenosis of a second lesion was 2.5 times higher if the first lesion had restenosis, even after adjustments for well-known patient-related risk factors, including diabetes and hypertension.9 Inflammatory responsiveness is highly genetically determined. Many studies have demonstrated genetic influences on the inflammatory response of an individual.10,11 Therefore, it is plausible that differences in genetic makeup of inflammatory genes between individuals may explain part of the risk of the at least partially inflammation-driven restenotic process. Association studies have identified several candidate genes that may predispose to restenosis, such as the genes for stromelysin-1, IL6, E-selectin, CD18, CD14, and IL1 receptor.12,13 However, these studies are mostly inconclusive because of a number of limitations, including limited study size and study design (eg, not prospective studies or studies using selected patient groups), limiting the clinical value of the observations.
The purpose of the present study was to evaluate in a large unselected study sample whether a variety of genetic determinants, considered to be involved in the inflammatory process, can predict the risk of clinical restenosis after PCI.
Methods
Study Design
The present study sample has been described previously.14 In brief, the GENetic DEterminants of Restenosis project (GENDER) was designed to study the association between various gene polymorphisms and clinical restenosis defined in our study by target vessel revascularization (TVR). Patients were eligible for inclusion if they were successfully treated for stable angina, non–ST-elevation acute coronary syndromes or silent ischemia by PCI. Patients treated for acute ST-elevation myocardial infarction were excluded. All patients were treated in 4 referral centers for interventional cardiology in the Netherlands (Academic Medical Center Amsterdam, University Medical Center Groningen, Leiden University Medical Center, and Academic Hospital Maastricht). The overall inclusion period lasted from March 1999 until June 2001. In total, 3104 consecutive patients were included in this prospective, multicenter, follow-up study.
The study protocol conforms to the Declaration of Helsinki and was approved by the medical ethics committees of each participating institution. Written informed consent was obtained from each participant before the PCI procedure.
PCI Procedure
Standard angioplasty and stent placement were performed by experienced operators, using a radial or femoral approach. Before the procedure, patients received 300 mg aspirin and 7500 IU heparin. The use of intracoronary stents and additional medication, such as glycoprotein IIb/IIIa inhibitors, was at the discretion of the operator. If a stent was implanted, patients received either ticlopidine or clopidogrel for at least 1 month after the procedure, depending on local practice. During the study, no drug-eluting stents were used.
Follow-Up and Study End Points
Follow-up lasted at least 9 months or until a coronary event occurred. Patients were either seen in the outpatient clinic or contacted by telephone. TVR, either by PCI or CABG, was considered as the primary end point because it is considered most relevant for clinical practice by regulatory agencies. An independent clinical events committee adjudicated the clinical events.
Events occurring within 1 month after the procedure were excluded from the analysis because these events were attributable to subacute stent thrombosis or occluding dissections.
Data were collected with standardized case report forms that were completed by the research coordinator at each site, who was blinded to the genotype of the patients. Representatives from the data coordinating center monitored the sites.
Genetic Methodology
Blood was collected in EDTA tubes at baseline, and genomic DNA was extracted by following standard procedures. Genotyping was performed by a validated multilocus genotyping assay to test several markers of inflammation (Roche Molecular Systems).15,16 Inflammation is known to play an important role in the development of restenosis, although it is not fully elucidated which factors are exactly involved in the process. Therefore, we intended to examine a very broad spectrum of factors that are considered to have an effect on the inflammatory response. The polymorphisms on the array were chosen on the basis of their relation to inflammation previously described in literature reports of gene associations with inflammatory diseases. Furthermore, polymorphisms were also selected on the basis of their known functionality or their known allele frequency. In addition, developing the assay for the selected polymorphism had to be possible. A total of 48 markers in 34 genes were examined by using these criteria (Table 1). Each DNA sample was amplified in multiple polymerase chain reactions (PCRs), using biotinylated primers. The PCR product was then hybridized to a corresponding panel of sequence-specific oligonucleotide probes that had been immobilized in a linear array on nylon membrane strips.17 A colorimetric detection method based on incubation with streptavidin–horseradish peroxidase conjugate, using hydrogen peroxide and 3,3,5,5-tetramethylbenzidine as substrates, was used. Operators blinded to patient data performed genotyping. To confirm genotype assignments, PCR analysis was randomly performed in replicate on 10% of the samples. Two independent observers carried out scoring. Disagreements (<1%) were resolved by further joint reading, and when necessary a repeat genotyping reaction was performed.
Statistical Methodology
Deviations of the genotype distribution from that expected for a sample in Hardy-Weinberg equilibrium were tested by using the 2 test with 1 degree of freedom. Allele frequencies were determined by gene counting. The 95% confidence intervals of the allele frequencies were calculated from sample allele frequencies, based on the approximation of the binomial and normal distributions in large sample sizes.
In the first stage, we determined the association between each of the 48 polymorphisms and TVR by using a Cox proportional regression model. We considered codominant, dominant, and recessive inheritance models, and the model with the lowest Akaike information criterion was used.18 If fewer than 10 patients were homozygous for a particular allele, the homozygotes and heterozygotes were taken together, thereby assuming a dominant model. No adjustment for covariates was performed at this stage to allow for the assessment of the possible involvement of the polymorphisms in the causal pathway for TVR. Haplotypes were constructed from polymorphisms known to be located in the same gene, or in genes located near each other, and in statistically significant linkage disequilibrium (all probability values of the Pearson 2 test <0.001). The effect of haplotypes on restenosis risk was estimated according to the methods developed by Tanck et al.19 We considered only one TVR per patient.
The robustness of these "individual" findings was investigated by a bootstrap study of 1000 bootstrap samples drawn from the original data set. In each bootstrap sample, the optimal inheritance model as well as the association between each of the polymorphisms and TVR was determined, and the percentage of bootstrap samples with a significant association was counted for each polymorphism.
By following the SNP selection method of Hoh et al,20 we performed multivariable regression analysis of the TVR risk with all polymorphisms and haplotypes having an individual probability value of 0.10 or less and being significant in at least 40% of the bootstrap samples. First, the association between each polymorphism and TVR was adjusted for the confounding effect of the clinical risk factors age and sex and other clinical and intervention-related risk factors that were significantly (P<0.10) associated with TVR, being diabetes, stenting, residual stenosis >20%, current smoking, total occlusion, and hypertension. If 2 or more arterial segments were treated, we only used intervention-related characteristics of the most severely affected segment.
Second, polymorphisms with independent prognostic value were selected through the use of a multivariable regression model, using a stepwise backward selection algorithm.
No multiple testing correction method was applied to keep power at an acceptable level, and we therefore performed a permutation study to assess the experiment-wide error rate. One thousand permutation samples were created by reshuffling at random the 304 observed TVRs over all available patients. In each permutated (reshuffled) sample, the same statistical analysis was performed as described above, and we counted the number of reshuffled samples with 0, 1, 2, 3, 4, or 5 significant polymorphisms in the final multivariable regression model. In reshuffled data, no significant association is expected, and the percentage of reshuffled samples with 1 or more significant polymorphisms is a quantification of the experiment-wide error rate.
Statistical analysis was carried out using SPSS 11.5 (SPSS Inc).
Results
Patient Characteristics
The characteristics of this patient sample have been described previously.14 In summary, a total of 3146 patients had a complete follow-up (99.3%), with a median duration of 9.6 months (interquartile range, 3.9). Of 3146 patients, 42 had an event in the first 30 days. These patients were excluded from further analysis, according to the protocol. The remaining 3104 patients had a mean age of 62.1±10.7 years. Of the patients, 888 (28.6%) were female and 453 (14.6%) had diabetes mellitus. The majority of the patients were white (97%); 812 patients (26.2%) received glycoprotein IIb/IIIa inhibitors. Stents were used in 2309 (74.4%) patients. A total of 4061 lesions were treated in this unselected patient sample. Complex (type C) lesions, classified according to the modified American College of Cardiology and American Heart Association Task Force classification, were treated in 802 patients (25.8%). Other patient characteristics, as well as details of the interventions, are summarized in Table 2.
Follow-Up
Of the 3104 patients, 304 (9.8%) patients underwent TVR during follow-up. Fifty-one patients died (1.6%) and 22 (0.7%) had a myocardial infarction.
Inflammation Array Results
In the present study, genotyping was performed in 3029 patients of the total sample. Results of the remaining patients (n=75, 2.4%) are lacking due to unavailable DNA or inconclusive genotyping. Patients who could not be genotyped did not differ in any characteristic from those who could be genotyped.
The following polymorphisms were associated with TVR (P<0.05): 2-adrenergic receptor (ADRB2) Arg16Gly, CD14–260C/T, colony stimulating factor 2 (CSF2) Ile117Thr, and small inducible cytokine subfamily A, member 11, alias Eotaxin (CCL11)-1328G/A (Table 3), being significant in 65.0%, 61.8%, 73.7%, and 69.3% of bootstrap samples, respectively. Interleukin-4 receptor (IL4R) Gln576Arg, and T-cell transcription factor (TCF7) Pro19Thr polymorphisms also showed a tendency of association with TVR (0.05
TVR occurred more often in ADRB2 Gly16 homozygotes (11.3%) than in Arg16Gly heterozygotes (8.7%) or Arg16 homozygotes (9.1%). In contrast, CD14 T/T-genotypes had decreased TVR risk (7.6%) compared with C/T genotypes (10.5%) and C/C genotypes (10.3%). CSF2 Thr homozygotes (6.6%) and Ile/Thr heterozygotes (8.5%) also showed a decreased risk of TVR compared with Ile genotypes (10.8%). CCL11–1328 A/A-homozygotes had decreased TVR risk (4.3%) compared with G/A-heterozygotes (8.6%) and G/G-homozygotes (10.5%) (Table 3). More details are provided in Data Supplement Table I (http://circ.ahajournals.org/cgi/content/full/112/16/2417/DC1). As arterial remodeling is more evident in plain balloon angioplasty and neointima formation is more pronounced in stenting,21 we examined whether the association between the polymorphisms and TVR differed between patients who received stents and patients who did not receive stents. Analyzing the interaction between balloon angioplasty/stenting and the significantly associated genes with TVR showed no significant difference (CD14, P=0.34; ADRB2, P=0.83; CSF2, P=0.30 and CCL11, P=0.67), nor was there a significant difference between angioplasty and stented patients with regard to the association of any of the other polymorphisms with TVR (P>0.09). Because the power to find an interaction between type of PCI (angioplasty versus stent) and associations is limited, such interactions cannot be fully excluded. Separately within the balloon angioplasty–treated patient group and the stented patient group, the hazard ratios are in the same direction, but many of the 95% confidence intervals cross the value of 1.0 (see Data Supplement Table II).
Linkage Disequilibrium and Haplotypes
The three IL4R polymorphisms were in linkage disequilibrium (LD) (P<0.0001), and the same applied to the polymorphisms of the ADRB2, CTLA4, NOS3, CCL11, IL1A, IL1B, and IL6 genes. Furthermore, LD was found between polymorphisms in the IL4, IL13, and CSF2 genes. Significant LD was observed between SELE and the SELP (Val640Leu) polymorphism and involving the two CCR5 and the CCR2 polymorphisms. Except for a haplotype in the ADRB2 gene, none of the haplotypes were significantly associated with TVR risk (P>0.13). ADRB2 haplotypes, including the ADRB2 16Gly allele, were all associated with increased TVR risk, whereas haplotypes without the ADRB2 16Gly allele had lower TVR risk. In subsequent analyses, we therefore used the ADRB2 16 genotype and not these haplotypes.
Multivariable Cox Regression
Adjustment of the association between the selected polymorphisms and TVR for the clinical risk factors age, sex, diabetes, stenting, residual stenosis >20%, current smoking, hypertension, and total occlusion changed little (last column of Table 3); ADRB2 Arg16Gly, CD14–260C/T, CSF2 Ile117Thr, and CCL11–1328G/A polymorphisms remained significantly related to TVR.
Subsequently, we included all selected polymorphisms in the multivariable regression analysis and included the aforementioned clinical risk factors. After stepwise backward selection among the selected polymorphisms, the same polymorphisms appeared to be significantly related to TVR (Table 4).
Finally, we considered the association of the 21 possible 2-way interaction terms between the 7 selected polymorphisms and TVR. Only 1 interaction term was found to be statistically significant (P=0.04) between TCF7 and CSF2. The relative risk of TVR in CSF2 Thr carriers versus noncarriers was 0.37 (95% CI, 0.17 to 0.79) in TCF7 Thr carriers and 0.83 (95% CI, 0.67 to 1.03) in TCF7 noncarriers.
To illustrate the clinical relevance of our finding, the log-relative risks of the genotypes defined by the 5 polymorphisms with a P<0.1 selected in the Cox model were divided into quartile groups. We calculated the Kaplan-Meier curves in the first quartile-group (low risk, n=746), the second and third quartiles (medium risk, n=1488), and the fourth quartile (high risk, n=749). At 9 months after intervention, TVR risks were 5.0%, 9.2%, and 12.9% in these respective groups, and at 12 months 5.3%, 11.0%, and 14.3%, respectively (Figure). The high-risk quartile consisted almost exclusively of 435 patients having the TCF7 19Pro/Pro, CSF2 117Ile/Ile, CD14–260CC/CT, and the ADRB2 16Gly/Gly genotype. The low- and medium-risk quartiles consisted of patients with several genotype combinations.
Cumulative target vessel revascularization (TVR) risks in low-, medium-, and high-risk groups, based on genotype. Kaplan-Meier curves in the first quartile group (low risk, n=746), the second and third quartiles (medium risk, n=1488), and the fourth quartile (high risk, n=749) were calculated. At 9 months after intervention, TVR risks were 5.0%, 9.2%, and 12.9% and at 12 months 5.3%, 11.0%, and 14.3%, respectively. The high-risk quartile consisted almost exclusively of the 435 patients having the TCF7 19Pro/Pro, CSF2 117Ile/Ile, CD14–260CC/CT, and the ADRB2 16Gly/Gly genotype. The low- and medium-risk quartiles consisted of patients with several genotype combinations.
Error Rate
The experiment-wide type I error rate of the present study was assessed with a permutation study. Of 1000 permutated (reshuffled) data set, there were 97 (10%) in which none of the 48 markers were significantly related to TVR. Thirty-one percent of the reshuffled data sets had 1 significant marker, 25% of the reshuffled data sets had 2, 22% had 3, and 12% had 4 or more significant markers. The median false detection rate was only 2 markers.
Discussion
In a prospective, multicenter, follow-up study, we investigated 48 polymorphisms from 34 different genes. Our assumption that a relation exists between these genes and the development of restenosis after PCI was based on observations suggesting a role of these genes in the process of inflammation, a well-known determinant in the development of restenosis. After multivariable analysis, we identified polymorphisms in the CD14, 2-adrenergic receptor (ADRB2), colony stimulating factor 2 (CSF2), and eotaxin (CCL11) genes that were significantly associated with restenosis after PCI. Although neointimal formation is more pronounced after stenting and remodeling is prominent after plain balloon angioplasty, stenting did not give change the associations between the 4 genes and restenosis.
CD14
The –260 T/T genotype of CD14 was found to be protective against restenosis after PCI. Two previous studies have investigated the role of CD14 in the development of restenosis, one being a prospective study by Zee et al22 in 779 patients and the other being a prospective study by Shamada et al23 in 129 patients. They found the –260 T/T genotype to be a risk factor for restenosis. Our data are in conflict with these findings, which may be explained by a biological significance of CD14 that differs between Japanese subjects and whites, as well as a small sample size. However, the discrepant results of Zee et al and ours are not yet explained and await further study.
ADRB2
The second gene we found to be associated with TVR is the 2-adrenergic receptor gene (ADRB2), located on chromosome 5q31-q32. ADRB2s are cell-surface receptors that on binding to norepinephrine activate cellular adenylylcyclase through coupling to G-proteins. The ADRB2 gene has a role in the inflammatory response, because adrenoceptors are present on human platelets, and ADRB2 stimulation activates platelet nitric oxide synthase (NOS).24 NOS catalyzes the formation of NO, which has an inhibitory role on leukocyte adhesion, platelet adhesion and aggregation, smooth muscle cell proliferation, and synthesis of matrix proteins, and it promotes endothelial survival and proliferation.25 In addition, ADRB2 has an effect on the immune system because lymphocytes express ADRB2s.26
The polymorphism in the ADRB2 gene that after adjusted analyses showed to be a risk factor for restenosis was the 16A/G polymorphism that results in an amino acid change of glycine to arginine at position 16 (Arg16Gly). Patients with homozygosity for the 16Gly variant had a higher risk of TVR compared with patients with the 16Arg variant (11.3% versus 9.1%, respectively). Previous in vivo and in vitro studies have suggested that this Arg16Gly variant may differently affect functional responses to adrenergic stimulation, thereby possibly modulating cardiovascular and metabolic phenotypes. It has been reported that the 16Gly variant of ADRB2 is associated with faster agonist-induced downregulation of the receptor, as compared with the 16Arg variant.27 The higher risk of TVR may be related to less vasodilatation as a result of the downregulation of the receptor containing 16Gly, as compared with the receptor containing 16Arg. Moreover, downregulation of the ADRB2 could result in impaired inhibition of platelet aggregation.24
CSF 2
The polymorphism in the colony stimulating factor 2 gene (CSF2) that is significantly associated with restenosis is the 117T/C polymorphism, which results in an isoleucin for threonin substitution on position 117. The Thr117 variant showed a protective association with TVR. The functional effect of this CSF2, also known as granulocyte-macrophage colony stimulating factor, polymorphism still has to be investigated.
CCL11
The fourth polymorphism that was associated with restenosis is eotaxin (CCL11), a CC chemokine that is localized on chromosome 17. The –1328A/A promoter variant of this gene demonstrated a protective association with TVR. Economou et al28 reported that eotaxin is elevated in plasma of patients with advanced atherosclerosis. The plasma level of eotaxin in their study rose in the first day after PCI and declined to baseline in the following 3 months. In what way the polymorphism determines the expression level on a protein basis is as yet unknown.
Recently, Humphries et al29 have put forward the criteria for a genetic variant to be included in clinical risk management of patients with cardiovascular disease. For candidate variants, there should be enough circumstantial evidence from literature or from theoretical points of view to be involved in the disease and the studies should have enough power to detect a relative risk of 1.25 or more. The genotype must then give a predictive value in carriers over and above established risk factors, and the final data set should show no significant evidence for heterogeneity of risk effect. Finally, for each selected gene locus only functional variants (ie, variants that alter an amino acid or a transcription factor–binding element in a promoter region demonstrated in vitro) should be included.29 With these criteria, Humphries et al referred to genotypes encoding a specific phenotype, such as Factor V Leiden, for venous thrombosis. The polymorphisms we examined are explorative and chosen on basis of their known involvement in the multiple pathways of inflammation, being potentially implicated in the development of restenosis. However, they have no strong direct a priori theoretical value in terms of biological plausibility as meant by Humphries et al. However, we believe that the 4 factors associated with restenosis identified in the present study meet these criteria considerably.
Because it is not only the inflammatory response that causes restenosis, more research and confirmation of our findings are needed before these genetic variants could be used for making a genetic risk profile for patients at increased risk of restenosis.
In addition, circulating protein levels were not assessed in the present study. Basal (pre-PCI) plasma levels of the gene product probably do not reflect genetically determined differences in reaction to a trauma such as PCI. Moreover, local differences in reactions (in the vessel wall at the place of PCI) may not be determined systemically. In the human situation, it is impossible to measure gene products locally in the acute phase of treatment or the following days, and several months later the causal trigger has probably already disappeared.
Limitations of the Study
The 48 polymorphisms examined in the present study represent only a small proportion of genetic information that is potentially associated with TVR. However, by looking at a broad spectrum of polymorphisms in genes that are considered to be involved in the inflammatory response, we did try to cover a large set of factors that may be associated with restenosis. Furthermore, the candidate gene approach currently remains the most practical approach. Second, one or more of the SNPs associated with TVR in our study may be in linkage disequilibrium with other polymorphisms in the gene or with other nearby genes that are actually responsible for the development of this condition.
We did not apply adjustment for multiple testing, which sometimes is considered appropriate for hypothesis-testing studies. Moreover, our experiment-wide error rate was found to be 12% in the permutation analysis. In addition, we performed additional hypothesis testing in our haplotype analysis, which further increases the multiple testing problem. Our results should be primarily seen as hypothesis generating and need independent validation.
In addition, data on our haplotype analysis is limited due to the fact that we do not comprehensively cover the haplotype structure of our selected genes. Thus, individuals who appear identical as our haplotypes are concerned may very well differ when more polymorphisms are taken into account.
Our study has insufficient statistical power to examine whether the genetic associations we observed differed by sex because there were only 888 female patients, of whom 84 had TVR.
Another possible limitation is that we examined TVR as our primary end point instead of angiographic outcomes such as late loss. This could have caused a problem with ascertainment. However, in clinical practice clinical restenosis is an end point much more valuable than angiographic restenosis.
Furthermore, the 4 factors we found show a small hazard ratio (0.7 to 1.3), but it should be taken into account that the process of restenosis is multifactorial, involving multiple genes. Thus, relatively small hazard ratios relating to contribution of a single gene to restenosis might be of paramount importance in the overall process, and even a small genetic risk may identify a gene with an important biological role that could reveal new mechanistic insights and provide novel therapeutic targets.
Finally, because our study was conducted in a sample of white patients, extrapolation of the data to other ethnic groups should be done with great caution.
Conclusions
We show that genetic variants in 4 different genes that are considered to be involved in the inflammatory response may play a role in the development of restenosis. Three of these genotypes have, to our knowledge, not been described before in relation to restenosis. Because of the explorative nature of this analysis, our results need to be replicated in other studies to confirm the data. After this, our data could contribute to the unraveling of the process of restenosis and thereby provide novel therapeutic targets as well as contribute to development of improved risk stratification of patients who are scheduled for elective PCI, thereby creating the opportunity to individualize treatment in the future.
Acknowledgments
P.S. Monraats and W.R.P. Agema were supported by grant 99.210 from the Netherlands Heart Foundation and a grant from the Interuniversity Cardiology Institute of the Netherlands (ICIN). J.W. Jukema is an Established Clinical Investigator of the Netherlands Heart Foundation (2001 D 032). N.M.M. Pires was supported by grant 2001 T 32 from the Netherlands Heart Foundation. A. Schepers and P.H.A Quax (Established Investigator) were supported by the Molecular Cardiology Program of the Netherlands Heart Foundation (M 93.001). The contribution of the members of the clinical event committee, J.J. Schipperheyn, MD, PhD, J.W. Viersma, MD, PhD, D. Düren, MD, PhD, and J. Vainer, MD, is greatly acknowledged. The authors thank Lori Steiner and Karen Walker for their efforts in developing the genotyping reagents used for this study. They also thank Paul Schiffers from the University of Maastricht for his assistance in the laboratory.
Footnotes
The online-only Data Supplement can be found at http://circ.ahajournals.org/cgi/content/full/112/16/2417/DC1.
References
Fattori R, Piva T. Drug-eluting stents in vascular intervention. Lancet. 2003; 361: 247–249.
Pache J, Kastrati A, Mehilli J, Schuhlen H, Dotzer F, Hausleiter J, Fleckenstein M, Neumann FJ, Sattelberger U, Schmitt C, Muller M, Dirschinger J, Schomig A. Intracoronary stenting and angiographic results: strut thickness effect on restenosis outcome (ISAR-STEREO-2) trial. J Am Coll Cardiol. 2003; 41: 1283–1288.
Agema WR, Jukema JW, Pimstone SN, Kastelein JJ. Genetic aspects of restenosis after percutaneous coronary interventions: towards more tailored therapy. Eur Heart J. 2001; 22: 2058–2074.
Bhargava B, Karthikeyan G, Abizaid AS, Mehran R. New approaches to preventing restenosis. BMJ. 2003; 327: 274–279.
Serrano CV Jr, Ramires JA, Venturinelli M, Arie S, D’Amico E, Zweier JL, Pileggi F, da Luz PL. Coronary angioplasty results in leukocyte and platelet activation with adhesion molecule expression: evidence of inflammatory responses in coronary angioplasty. J Am Coll Cardiol. 1997; 29: 1276–1283.
Hojo Y, Ikeda U, Katsuki T, Mizuno O, Fukazawa H, Fujikawa H, Shimada K. Chemokine expression in coronary circulation after coronary angioplasty as a prognostic factor for restenosis. Atherosclerosis. 2001; 156: 165–170.
Welt FG, Rogers C. Inflammation and restenosis in the stent era. Arterioscler Thromb Vasc Biol. 2002; 22: 1769–1776.
Weintraub WS, Kosinski AS, Brown CL III, King SB III. Can restenosis after coronary angioplasty be predicted from clinical variables J Am Coll Cardiol. 1993; 21: 6–14.
Kastrati A, Schomig A, Elezi S, Schuhlen H, Wilhelm M, Dirschinger J. Interlesion dependence of the risk for restenosis in patients with coronary stent placement in multiple lesions. Circulation. 1998; 97: 2396–2401.
Bidwell J, Keen L, Gallagher G, Kimberly R, Huizinga T, McDermott MF, Oksenberg J, McNicholl J, Pociot F, Hardt C, D’Alfonso S. Cytokine gene polymorphism in human disease: on-line databases, supplement 1. Genes Immunol. 2001; 2: 61–70.
Verschuur M, Van Der Beek MT, Tak HS, Visser LG, De Maat MP. Interindividual variation in the response by fibrinogen, C-reactive protein and interleukin-6 to yellow fever vaccination. Blood Coagul Fibrinolysis. 2004; 15: 399–404.
de Maat MP, Jukema JW, Ye S, Zwinderman AH, Moghaddam PH, Beekman M, Kastelein JJ, van Boven AJ, Bruschke AV, Humphries SE, Kluft C, Henney AM. Effect of the stromelysin-1 promoter on efficacy of pravastatin in coronary atherosclerosis and restenosis. Am J Cardiol. 1999; 83: 852–856.
Kastrati A, Koch W, Berger PB, Mehilli J, Stephenson K, Neumann FJ, von Beckerath N, Bottiger C, Duff GW, Schomig A. Protective role against restenosis from an interleukin-1 receptor antagonist gene polymorphism in patients treated with coronary stenting. J Am Coll Cardiol. 2000; 36: 2168–2173.
Agema WRP, Monraats PS, Zwinderman AH, de Winter RJ, Tio RA, Doevendans PAFM, Waltenberger J, de Maat MPM, Frants RR, Atsma DE, van der Laarse A, van der Wall EE, Jukema JW. Current PTCA practice and clinical outcomes in the Netherlands: the real world in the pre–drug-eluting stent era. Eur Heart J. 2004; 25: 1163–1170.
Cheng S, Grow MA, Pallaud C, Klitz W, Erlich HA, Visvikis S, Chen JJ, Pullinger CR, Malloy MJ, Siest G, Kane JP. A multilocus genotyping assay for candidate markers of cardiovascular disease risk. Genome Res. 1999; 9: 936–949.
Barcellos LF, Begovich AB, Reynolds RL, Caillier SJ, Brassat D, Schmidt S, Grams SE, Walker K, Steiner LL, Cree BA, Stillman A, Lincoln RR, Pericak-Vance MA, Haines JL, Erlich HA, Hauser SL, Oksenberg JR. Linkage and association with the NOS2A locus on chromosome 17q11 in multiple sclerosis. Ann Neurol. 2004; 55: 793–800.
Saiki RK, Walsh PS, Levenson CH, Erlich HA. Genetic analysis of amplified DNA with immobilized sequence-specific oligonucleotide probes. Proc Natl Acad Sci U S A. 1989; 86: 6230–6234.
Li W, Nyholt DR. Marker selection by Akaike information criterion and Bayesian information criterion. Genet Epidemiol. 2001; 21 (suppl 1): S272–S277.
Tanck MW, Klerkx AH, Jukema JW, De Knijff P, Kastelein JJ, Zwinderman AH. Estimation of multilocus haplotype effects using weighted penalized log-likelihood: analysis of five sequence variations at the cholesteryl ester transfer protein gene locus. Ann Hum Genet. 2003; 67: 175–184.
Hoh J, Wille A, Zee R, Cheng S, Reynolds R, Lindpaintner K, Ott J. Selecting SNPs in two-stage analysis of disease association data: a model-free approach. Ann Hum Genet. 2000; 64: 413–417.
Agema WR, Jukema JW, Pimstone SN, Kastelein JJ. Genetic aspects of restenosis after percutaneous coronary interventions: towards more tailored therapy. Eur Heart J. 2001; 22: 2058–2074.
Shimada K, Miyauchi K, Mokuno H, Watanabe Y, Iwama Y, Shigekiyo M, Matsumoto M, Okazaki S, Tanimoto K, Kurata T, Sato H, Daida H. Promoter polymorphism in the CD14 gene and concentration of soluble CD14 in patients with in-stent restenosis after elective coronary stenting. Int J Cardiol. 2004; 94: 87–92.
Zee RY, Hoh J, Cheng S, Reynolds R, Grow MA, Silbergleit A, Walker K, Steiner L, Zangenberg G, Fernandez-Ortiz A, Macaya C, Pintor E, Fernandez-Cruz A, Ott J, Lindpainter K. Multi-locus interactions predict risk for post-PTCA restenosis: an approach to the genetic analysis of common complex disease. Pharmacogenomics J. 2002; 2: 197–201.
Queen LR, Xu B, Horinouchi K, Fisher I, Ferro A. Beta(2)-adrenoceptors activate nitric oxide synthase in human platelets. Circ Res. 2000; 87: 39–44.
Chen AF, Ren J, Miao CY. Nitric oxide synthase gene therapy for cardiovascular disease. Jpn J Pharmacol. 2002; 89: 327–336.
Wahle M, Stachetzki U, Krause A, Pierer M, Hantzschel H, Baerwald CG. Regulation of beta2-adrenergic receptors on CD4 and CD8 positive lymphocytes by cytokines in vitro. Cytokine. 2001; 16: 205–209.
Hoit BD, Suresh DP, Craft L, Walsh RA, Liggett SB. Beta2-adrenergic receptor polymorphisms at amino acid 16 differentially influence agonist-stimulated blood pressure and peripheral blood flow in normal individuals. Am Heart J. 2000; 139: 537–542.
Economou E, Tousoulis D, Katinioti A, Stefanadis C, Trikas A, Pitsavos C, Tentolouris C, Toutouza MG, Toutouzas P. Chemokines in patients with ischaemic heart disease and the effect of coronary angioplasty. Int J Cardiol. 2001; 80: 55–60.
Humphries SE, Ridker PM, Talmud PJ. Genetic testing for cardiovascular disease susceptibility:a useful clinical management tool or possible misinformation Arterioscler Thromb Vasc Biol. 2004; 24: 628–636.(Pascalle S. Monraats, MSc)