Angiotensin converting enzyme insertion or deletion polymorphism and coronary restenosis: meta-analysis of 16 studies
BMJ 2002; 325 doi: https://doi.org/10.1136/bmj.325.7363.517 (Published 07 September 2002) Cite this as: BMJ 2002;325:517- François Bonnici, Rhodes scholara,
- Bernard Keavney, senior lecturerb,
- Rory Collins, professor of medicine and epidemiologya,
- John Danesh, professor of epidemiology and medicinec
- aClinial Trial Service Unit and Epidemiological Studies Unit, University of Oxford, Radcliffe Infirmary, Oxford OX2 6HE
- bUniversity Department of Cardiology, Freeman Hospital, Newcastle upon Tyne NE7 7DN
- cDepartment of Public Health and Primary Care, University of Cambridge, Institute of Public Health, Cambridge CB2 2SR
- Correspondence to: J Danesh
- Accepted 17 April 2002
Abstract
Objective: To assess the association between genotype at the insertion or deletion polymorphism of the angiotensin converting enzyme gene and risk of coronary restenosis after percutaneous coronary intervention.
Design: Meta-analysis of studies before July 2001 that reported on these genotypes and risk of coronary restenosis after a percutaneous coronary intervention, with or without coronary stenting.
Results: 16 studies, involving 4631 patients undergoing a percutaneous coronary intervention, yielded 1683 patients with restenosis after a mean weighted follow up of 5.5 months. The combined odds ratio for restenosis in people with the DD genotype was 1.23 (99% confidence interval 1.03 to 1.46). When studies were grouped by size, however, the combined odds ratios for restenosis in people with the DD genotype were 1.94 (1.39 to 2.71) for studies with less than 100 cases, 1.33 (0.92 to 1.93) for studies with 100-200 cases, and 0.92 (0.72 to 1.18) for studies with more than 200 cases (trend P=0.02). Similarly, when studies were grouped by genotyping procedures, significantly larger odds ratios were found in the studies that did not conceal disease status from laboratory staff and in the studies that did not use a second polymerase chain reaction amplification to reduce genetic mistyping.
Conclusion: Compared with other studies, larger and more rigorous studies show a weaker association between the angiotensin converting enzyme gene DD genotype and restenosis. Publication bias or detection biases can produce artefactual associations at least as large as those that might be expected for common polymorphisms in complex diseases, suggesting the need for larger and more rigorous genetic epidemiological investigations than are now customary.
What is already known on this topic
Restenosis after a percutaneous coronary intervention is one of the principal limitations of the technique
Genotype at the angiotensin converting enzyme insertion or deletion polymorphism is proposed to be important in restenosis
What this study adds
Weaker associations between the angiotensin converting enzyme DD genotype and restenosis were found in larger and more rigorous studies than in other studies
Publication bias or detection biases, or both, can produce artefactual associations at least as large as those that might be expected for common polymorphisms in complex diseases
Introduction
Restenosis after a percutaneous coronary intervention is one of the principal limitations of this technique, occurring in up to 50% of patients undergoing the procedure without stenting and in about 20% of patients receiving stents.1 Despite a lack of good evidence that susceptibility to restenosis is genetically determined, several studies have investigated polymorphisms that might be associated with restenosis. As the angiotensin converting enzyme insertion or deletion (I/D) polymorphism is strongly associated with plasma and cellular angiotensin converting enzyme concentrations, it has been considered a strong candidate.2 It has been suggested that the incidence of coronary restenosis after a percutaneous coronary intervention is much higher in patients with the angiotensin converting enzyme DD genotype (which is associated with particularly high plasma angiotensin converting enzyme levels) than in others, but published observational studies are conflicting.3–18 To help clarify the evidence we considered all available relevant studies in a meta-analysis.
Methods
We sought studies published before July 2001 of the angiotensin converting enzyme insertion or deletion polymorphism and coronary restenosis after a percutaneous coronary intervention, with or without coronary stenting, by computer based searches (Medline, Embase, PubMed, Web of Science), reviews of reference lists, hand searching relevant journals, and correspondence with authors. For the electronic searches we used combinations of key words relating to the angiotensin converting enzyme gene (for example, angiotensin converting enzyme, ACE, polymorphism, insertion/deletion, I/D, D/I) and to restenosis (for example, coronary, restenosis, percutaneous, angioplasty, PTCA, stent, stenting). We included all identified studies. Articles in languages other than English were translated.
From each study we abstracted (supplemented by correspondence with investigators) geographical location, race of participants, numbers of cases and controls, the coronary intervention procedure, definition of restenosis, frequency of insertion or deletion genotypes, genotyping methods and laboratory procedures, mean age, and duration of follow up.
We estimated odds ratios by comparing cases who developed coronary restenosis after a percutaneous coronary intervention with controls who did not within the same study. We did this under the prior hypothesis that individuals who were homozygous for the angiotensin converting enzyme D allele were predisposed to restenosis compared with those with the ID or II genotypes.19
Results
We identified 16 relevant studies concerning a percutaneous coronary intervention, 11 without stenting (2535 patients) and five with stenting (2096 patients), yielding 4631 patients (94% white) undergoing such interventions.3–18 All the studies used quantitative computer assisted methods to define restenosis as a narrowing of the coronary diameter by 50% or more at follow up compared with the minimal luminal diameter immediately after intervention. Overall, 1683 of these patients (mean age 60 years) developed coronary restenosis after a mean weighted follow up of 5.5 months. The studies were conducted in Australia, Chile, France, Germany, Italy, Japan, Spain, Turkey, the United Kingdom, and the United States.3–18 Following correspondence with the authors, it was confirmed that genotyping had been performed by staff unaware of the disease status of the patients in four of the five studies with more than 100 cases,6 16–18 compared with only two of the 11 studies with less than 100 cases.4 8
Overall, the combined odds ratio for restenosis in people with the DD genotype was 1.23 (99% confidence interval 1.03 to 1.46; test for heterogeneity χ=24.1, df=15, P=0.06: fig 1). We found no significant heterogeneity between studies of percutaneous coronary intervention with stenting and those without stenting (combined odds ratios for the DD genotype of 1.17 v 1.27 respectively; χ=0.23, df=1, P=0.6). When studies were grouped by size, however, the combined odds ratios for restenosis were 1.94 (1.39 to 2.71) for the 11 studies with less than 100 cases,3–5 7–9 11–15 1.33 (0.92 to 1.93) for the three studies with 100-200 cases,6 10 16 and 0.92 (0.72 to 1.18) for the two studies with more than 200 cases (fig 2).17 18 A test for trend across these three groups of studies was significant (χ=5.6, df=1, P=0.02). Similar results were observed when odds ratios for restenosis were calculated per D allele in these three groups (1.74 v 1.00 v 0.98, respectively). When studies were grouped by genotyping procedures, significantly larger odds ratios were observed in the studies that did not conceal disease status from laboratory staff (combined odds ratios for the DD genotype of 1.87 v 1.02 respectively, χ=9.7, df=1, P=0.002) 3 5 7 9–15 and in the studies that did not use a second polymerase chain reaction amplification to reduce mistyping of angiotensin converting enzyme ID heterozygotes as DD (1.55 v 1.13, respectively; χ=4.05, df=1, P=0.03).3 5–7 10 12–14
Discussion
Weaker associations between the angiotensin converting enzyme DD genotype and restenosis were found in larger and more rigorous studies than in other studies. We had observed a similar trend in published studies of the angiotensin converting enzyme DD genotype and myocardial infarction.19 These findings have important implications for genetic epidemiology, suggesting that publication bias or detection biases can produce artefactual associations at least as large as those that might be realistically expected for common polymorphisms in complex diseases.
Acknowledgments
We thank the investigators for supplying additional information and Alison Palmer for plotting the figures. RC holds a British Heart Foundation personal chair. JD holds the Raymond and Beverly Sackler Research Award in the Medical Sciences.
Contributors: All authors contributed to the design, analysis, and interpretation of the study. FB and JD will act as guarantors for the paper.
Footnotes
-
Funding FB was supported by a Rhodes scholarship.
-
Competing interests None declared.