Advertisement
Browse Subject Areas
?

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Angiotensin-Converting Enzyme Insertion/Deletion Polymorphism and Susceptibility to Osteoarthritis of the Knee: A Case-Control Study and Meta-Analysis

  • Chin Lin ,

    Contributed equally to this work with: Chin Lin, Hsiang-Cheng Chen, Wen-Hui Fang

    Affiliation School of Public Health, National Defense Medical Center, Taipei, Taiwan, ROC

  • Hsiang-Cheng Chen ,

    Contributed equally to this work with: Chin Lin, Hsiang-Cheng Chen, Wen-Hui Fang

    Affiliation Division of Rheumatology/Immunology/Allergy, Department of Internal Medicine, Tri-Service General Hospital, National Defense Medical Center, Taipei, Taiwan, ROC

  • Wen-Hui Fang ,

    Contributed equally to this work with: Chin Lin, Hsiang-Cheng Chen, Wen-Hui Fang

    Affiliation Department of Family and Community Medicine, Tri-Service General Hospital, National Defense Medical Center, Taipei, Taiwan, ROC

  • Chih-Chien Wang,

    Affiliation Department of Orthopedics, Tri-Service General Hospital, National Defense Medical Center, Taipei, Taiwan, ROC

  • Yi-Jen Peng,

    Affiliation Department of Pathology, Tri-Service General Hospital, National Defense Medical Center, Taipei, Taiwan, ROC

  • Herng-Sheng Lee,

    Affiliation Department of Pathology and Laboratory Medicine, Kaohsiung Veterans General Hospital, Kaohsiung, Taiwan, ROC

  • Hung Chang,

    Affiliations Department of Physiology and Biophysics, National Defense Medical Center, Taipei, Taiwan‏, ROC, Division of Thoracic Surgery, Tri-Service General Hospital, National Defense Medical Center, Taipei, Taiwan, ROC

  • Chi-Ming Chu,

    Affiliation School of Public Health, National Defense Medical Center, Taipei, Taiwan, ROC

  • Guo-Shu Huang,

    Affiliation Department of Radiology, Tri-Service General Hospital, National Defense Medical Center, Taipei, Taiwan, ROC

  • Wei-Teing Chen,

    Affiliations Department of Medicine, Cheng Hsin General Hospital, Taipei, Taiwan, ROC, Department of Medicine, Tri-Service General Hospital, National Defense Medical Center, Taipei, Taiwan, ROC

  • Yu-Jui Tsai,

    Affiliation School of Public Health, National Defense Medical Center, Taipei, Taiwan, ROC

  • Hong-Ling Lin,

    Affiliation School of Public Health, National Defense Medical Center, Taipei, Taiwan, ROC

  • Fu-Huang Lin ,

    noldling@ms10.hinet.net (FHL); a131419@gmail.com (SLS)

    Affiliation School of Public Health, National Defense Medical Center, Taipei, Taiwan, ROC

  • Sui-Lung Su

    noldling@ms10.hinet.net (FHL); a131419@gmail.com (SLS)

    Affiliation School of Public Health, National Defense Medical Center, Taipei, Taiwan, ROC

Angiotensin-Converting Enzyme Insertion/Deletion Polymorphism and Susceptibility to Osteoarthritis of the Knee: A Case-Control Study and Meta-Analysis

  • Chin Lin, 
  • Hsiang-Cheng Chen, 
  • Wen-Hui Fang, 
  • Chih-Chien Wang, 
  • Yi-Jen Peng, 
  • Herng-Sheng Lee, 
  • Hung Chang, 
  • Chi-Ming Chu, 
  • Guo-Shu Huang, 
  • Wei-Teing Chen
PLOS
x

Abstract

Background

Studies of angiotensin-converting enzyme insertion/deletion (ACE I/D) polymorphisms and the risks of knee osteoarthritis (OA) have yielded conflicting results.

Objective

To determine the association between ACE I/D and knee OA, we conducted a combined case-control study and meta-analysis.

Methods

For the case-control study, 447 knee OA cases and 423 healthy controls were recruited between March 2010 and July 2011. Knee OA cases were defined using the Kellgren-Lawrence grading system, and the ACE I/D genotype was determined using a standard polymerase chain reaction. The association between ACE I/D and knee OA was detected using allele, genotype, dominant, and recessive models. For the meta-analysis, PubMed and Embase databases were systematically searched for prospective observational studies published up until August 2015. Studies of ACE I/D and knee OA with sufficient data were selected. Pooled results were expressed as odds ratios (ORs) with corresponding 95% confidence intervals (CI) for the D versus I allele with regard to knee OA risk.

Results

We found no significant association between the D allele and knee OA [OR: 1.09 (95% CI: 0.76–1.89)] in the present case-control study, and the results of other genetic models were also nonsignificant. Five current studies were included, and there were a total of six study populations after including our case-control study (1165 cases and 1029 controls). In the meta-analysis, the allele model also yielded nonsignificant results [OR: 1.37 (95% CI: 0.95–1.99)] and a high heterogeneity (I2: 87.2%).

Conclusions

The association between ACE I/D and knee OA tended to yield negative results. High heterogeneity suggests a complex, multifactorial mechanism, and an epistasis analysis of ACE I/D and knee OA should therefore be conducted.

1. Introduction

Knee osteoarthritis (OA) is characterized by a highly catabolic state, chondrocyte apoptosis, articular cartilage degeneration, morphologic changes to the subchondral bone, and damage to the surrounding synovium [15]. Multiple factors, such as ageing, genetic, hormonal, and mechanical factors, contribute to OA onset and progression [59]. Prior research suggests that OA is primarily influenced by genetic risk factors due to common population polymorphisms in multiple genes [1016]. This heritability of OA development was once estimated to be as high as 65% [17,18].

Bradykinin is an inflammatory nonapeptide vasodilator and has a role in pain and inflammation mainly mediated via its receptor [19]. Previous evidence suggested an important role of bradykinin in the generation of pain, swelling, and cellular damage associated with inflammatory joint disease, including OA [20,21]. The possible mechanism may be via a decrease in subchondral bone remodeling and an increase in cartilage thickness. Moreover, it can increase levels of cartilage proteoglycans and type II collagen [21]. Bellucci et al. (2009, 2013) further proposed a correlation between the presence of bradykinin in the synovial fluid of OA knees and cartilage degradation and a participatory role of bradykinin in OA pathology [19,22]. Finally, the epidemiological evidence demonstrated a positive correlation between bradykinin and the synovitis score and a higher detection rate of bradykinin among several pain-related mediators [23].

Angiotensin-converting enzyme (ACE, EC3.4.15.1) is a membrane metallopeptidase that converts angiotensin I to the potent vasoconstrictor, angiotensin II [24,25]. ACE plays a role in the cross talk between the renin–angiotensin system (RAS) and the kallikrein–kinin system. ACE not only converts angiotensin I to angiotensin II but also metabolizes bradykinin, which is a strong vasodilator, to form inactive bradykinin 1–5. This phenomenon has been demonstrated through in vivo experimental studies, which demonstrated that ACE inhibitor treatment decreased the blood angiotensin II concentration but increased the blood bradykinin concentration in normal human subjects and dogs [2629].

Given the key role of ACE in the RAS, sufficient evidence has led to suspicion of a relationship between ACE polymorphisms and OA. One of the most important ACE polymorphisms is a 287-bp insertion/deletion in intron 16 (ACE I/D), with this angiotensin-ACE I/D genotype associated with plasma, cellular, and tissue ACE levels. Plasma ACE levels are highest in subjects with the DD genotype, followed by subjects with the ID genotype and lowest in subjects with the II genotype [3033]. According to the above mechanism, we supposed the D allele was thought to be a protective factor against knee OA, because D allele carriers have higher ACE levels and therefore lower bradykinin concentrations, whereas the I allele is considered a risk factor.

Although many genome wide association studies investigated SNP–disease association in OA [34,35], they were unable to find any evidence of ACE I/D because this locus is a structural variant. Moreover, no meta-analyses have been carried out on this subject. Only a limited number of studies have investigated the link between OA and the ACE I/D polymorphism [3640]. In reports by Bayram et al. (2011) and Inanir et al. (2013), the DD genotype of the ACE gene I/D polymorphism was associated with knee OA in a Turkish study population [38,39]. Poornima et al. (2015) observed the same phenomenon in an Asian Indian population [40]. In contrast with our hypothesis, the above three studies considered that D allele carriers had the greater risk of OA. However, although Hong et al. (2003) and Shehab et al. (2008) found a negative association between ACE alleles and knee OA, carriers of the I allele had the greater risk of OA [36,37]. Given these inconclusive findings in the literature, we attributed the high heterogeneity to small sample sizes (sample sizes of those studies ranged from 200 to 421 [3640]). To resolve this dispute with regard to consistent evidence, we proposed a case–control study with a sample size of >800 subjects. Moreover, our study was combined with a concurrent meta-analysis of the present and previous studies to investigate whether the I allele at intron 16 in ACE contributes to knee OA susceptibility.

2. Materials and Methods

2.1 Case–control study

2.1.1 Sample size and ethical issues.

Before starting the study, an appropriate sample size was estimated using formulas developed by Fleiss et al. [41]. The settings used were as follows: a two-sided test with a power (1 − β) of 0.8 at a significance level of 0.05, and a ratio of controls to cases equal to 1; the hypothetical proportion of controls with exposure was 30 [42] with at least an odds ratio (OR) to be detected of 1.5 [43]. Based on these settings, the minimum study sample size required was 850 subjects.

Using this number, we initiated a population-based study at the Tri-Service General Hospital (TSGH), a medical teaching hospital of the National Defence Medical Centre in Taipei, Taiwan. The project was reviewed and approved by the institutional ethical committee of the Tri-Service General Hospital (TSGH-100-05-023). Individuals willing to participate in this study after receiving a full explanation from investigators were enrolled, and all participants included in this project provided signed, informed consent. This project was described in our previous report [44].

2.1.2 Subjects.

We recruited potential participants at the Health Management Centre of TSGH from those participating in a check-up program. The Taipei city senior medical check-up program is a governmental welfare program provided for people aged 65 years or older and who have been registered residents in Taipei city for more than 1 year. Accordingly, we began to release related recruitment information in March 2010. Participants who met the following criteria were excluded from this study: (1) patients who had undergone knee surgery (e.g., total knee arthroplasty) and (2) those unable to provide a sufficient blood sample. Demographic data included age, gender, and body mass index (BMI: kg/m2) and were collected from medical records. A total of 870 independent subjects (400 men and 470 women) aged 65 years or older [mean (standard deviation; SD) age = 74.1 (6.9) years] participated in this study up to July 2011.

2.1.3 Radiographic assessment.

All participants underwent a radiographic examination of both knees with anterior–posterior and lateral views analyzed as well as weight bearing and foot-map positioning recorded. Knee radiographs were read and scored by two readers, a radiologist and rheumatologist blinded to the patients’ clinical information, using the Kellgren–Lawrence (KL) grading system [45]. In the KL system, radiographs receive scores of 0–4 points. If the readers assigned different KL grades, we recruited a third interpreter to confirm the final grade. For patients with different KL grades in each knee, the more advanced grade was used for evaluation. We used a radiographic KL grade of ≥2 to define knee OA. According to the above classification, the study included 447 knee OA patients and 423 healthy controls.

2.1.4 Genomic DNA extraction and genotyping.

Genomic DNA was extracted from peripheral blood samples using standard procedures for proteinase K (Invitrogen, Carlsbad, CA, USA) digestion and phenol/chloroform extraction. ACE I/D polymorphisms were screened using the polymerase chain reaction (PCR) according to a previously described protocol [46]. The primers were also described previously [47]. To exclude this possibility, all DD homozygotes were retyped using an I-specific sense primer as previously described [48]. The PCR program included the following steps: initial denaturation at 95°C for 5 min; 35 cycles of denaturation at 95°C for 30 s, annealing at 58°C for 30 s, and extension at 72°C for 30 s and a final extension at 72°C for 10 min. Genotyping was performed under blinding of the case or control status. Two independent investigators interpreted images of each gel, and all ambiguous samples were analyzed twice. To validate the genotyping results, at least 10% of samples were randomly selected for repeated genotyping.

2.1.5 Statistical analysis.

Continuous demographic variables were evaluated using a Student’s t test and reported as means ± SDs. The Hardy–Weinberg equilibrium (HWE) was assessed using a goodness-of-fit χ2 test performed to identify possible genotyping errors among the controls of each study. Genotypes and allelic frequencies were compared between patients with knee OA and healthy controls using the χ2 test or Fisher’s exact test where appropriate. Logistic regression was used to estimate ORs and 95% confidence intervals (CIs) as a measure of the association with the risk of knee OA. Allele type, genotype, and dominant/recessive models were used to calculate the association between genetic polymorphism and knee OA risk. To avoid the multiple comparison problem, we tested the global p value via two robust tests, MAX3 [49,50] and the genetic model selection (GMS) [51]. This study considered a p value of <0.05 as significant for all analyses. Statistical analyses were carried out using R software, version 3.2.3 (R Project for Statistical Computing, Vienna, Austria) with the “Rassoc” package [52].

2.2 Meta-analysis

2.2.1 Search methods and criteria for study consideration.

The PRISMA checklist and Meta-analysis on Genetic Association Studies Checklist is described in S1 and S2 Tables [53]. This study focused on a general population and aimed to compare OA risks between individuals carrying the major (I) and minor (D) alleles of ACE I/D. To identify relevant studies, English-language articles in PubMed and Embase were searched using relevant text words and medical subject headings that included all spellings of ACE I/D and OA (detailed search strategy and records are shown in S3 Table). All articles published from the dates of inception of these medical databases to January 2016 were included.

All related studies that assessed the association between ACE I/D polymorphisms and the risk of knee OA were considered for inclusion. The criteria for study inclusion were as follows: (1) cross-sectional surveys or case–control studies, (2) OA defined as a KL grade of ≥2, and (3) a detailed distribution of ACE genotypes. If the published data was incomplete, we made attempts to contact the authors for further information.

2.2.2 Data extraction and quality assessment.

Two reviewers (Chin Lin and Wen-Hui Fang) independently extracted data and assessed the risk of bias. For each article, we recorded the first author’s name, year of publication, ethnicity of the study population, definition of the case group, and population characteristics (mean age, proportion of male subjects, BMI, and ACE I/D genotype distribution). All extracted papers were assessed using the Newcastle–Ottawa Scale (NOS) [54], and all received scores >5 points.

2.2.3 Statistical analysis.

The population characteristics of each included study are presented as means or proportions where appropriate. Our meta-analysis examined the association between ACE I/D polymorphisms and the risk of knee OA in each study using ORs with 95% CIs. The τ2 statistic, which was estimated using the DerSimonian–Laird method, was used to assess heterogeneity, and a random-effects model was used to calculate the weighed effect size. Three common genetic models, including allele type, dominant, and recessive models, were used to calculate the association between genetic polymorphism and the risk of knee OA. Moreover, we performed a genotype model with three comparisons (ID vs. II, DD vs. II, and DD vs. ID) for comprehensively examining the association between ACE I/D and knee OA. Because the genotype model tested three times in an association, the Bonferroni method was used to correct these for significant values (divided by 3).

Egger’s regression and a funnel plot were used to test the symmetry of pooled results [55]. I2 was calculated with the Cochrane Q test and used to quantify heterogeneity; an I2 value >50% indicated a moderate to high heterogeneity [56]. Moreover, meta-analyses commonly remove studies with HWE violations [57,58]; we therefore performed a sensitivity analysis according to this rule.

Meta-regression using an average summary value is used to explore the source of heterogeneity. According to our previous studies, the average summary value of the case group can be used to build a model and can help to estimate the interaction effect [59,60]. An interaction effect is determined using the OR and defined as the ratio between ORs per 1 unit. For example, if the OR of the association between ACE I/D and knee OA risk is 3 in the Arab subgroup and 2 in the Asian subgroup, the moderate effect of ethnicity would be 3/2 or 1.5. Possible moderators (ethnicity, age, gender, and BMI) were tested to explore heterogeneity.

This study considered a p value of <0.05 to be significant for all analyses. Statistical analyses were conducted using the “metafor” [61] and “meta” [62] packages of R software, version 3.2.3.

3. Results

3.1 Case–control study

There were 870 samples in this study, and 83 (9.5%) ambiguous calls were identified and re-genotyped. After the first round of genotyping, we randomly selected 96 (11.0%) repeated samples for re-genotyping. The reproducibility rate of analyzing ACE I/D in this study was 100%.

The characteristics of subjects according to the severity of knee OA are shown in Table 1. We included 447 cases with a mean age of 74.9 ± 7.1 years (253 men and 194 women) and 423 controls with a mean age of 73.3 ± 6.6 years (217 men and 206 women). The mean age was significantly higher among the cases relative to the controls (p = 0.001), whereas the gender distributions did not differ significantly (p = 0.117). Approximately 80% of cases had moderate knee OA (KL grade = 2), whereas approximately 20% had serious knee OA (KL grade > 2). Moreover, the case group had a significantly higher BMI relative to the control group (p = 0.005).

thumbnail
Table 1. Characteristics of subjects with knee osteoarthritis and control subjects.

https://doi.org/10.1371/journal.pone.0161754.t001

We further used the global test and four types of genetic models to test the association between ACE I/D and knee OA; the results are shown in Table 2. The D allele frequencies were 35.9% and 33.9% among cases and controls, respectively. The global test of the association of ACE I/D and knee OA was not significant [p value = 0.3167 (MAX3) and 0.4370 (GMS)], and the association between the D allele and knee OA was also nonsignificant [OR: 1.09 (95% CI: 0.76–1.89)]. Moreover, we also evaluated the results from the codominant, dominant, and recessive models and only obtained nonsignificant results. Therefore, we found no association between ACE I/D and knee OA in our case–control study. We also stratified patients by the severity of osteoarthritis and found no significant findings in any of the models (details were shown in S4 Table). To further enhance the level of evidence, we performed a meta-analysis that included our case–control study.

thumbnail
Table 2. Angiotensin-converting enzyme insertion/deletion (I/D) genotype frequencies in cases and controls.

https://doi.org/10.1371/journal.pone.0161754.t002

3.2 Meta-analysis

Fig 1 presents the overall study identification process. Our search strategy returned 17 and 9 records from PubMed and Embase, respectively. Twenty records remained after removing duplicates during the title and abstract review, and 14 of the 20 records were found to be unrelated to the topic (details were shown in S3 Table). Of the remaining six papers, one paper did not have sufficiently detailed data for the analysis. Accordingly, five studies were included in our meta-analysis [3640], and detailed data are presented in Table 3. Our meta-analysis therefore comprised three Asian studies and three Arab studies, including our case–control study. It is worth mentioning that our study population had the highest mean age, highest proportion of men, and lowest mean BMI. We subjected each control group to HWE testing. Interestingly, none of the Arab studies passed this test. Therefore, the results from the Asian subgroup will represent the results of our sensitivity analysis. Accordingly, we will not separately present the results of the sensitivity analysis.

thumbnail
Fig 1. Flow diagram of the identification process for eligible studies.

https://doi.org/10.1371/journal.pone.0161754.g001

Fig 2 presents selected results from our meta-analysis. None of the ACE alleles were associated with a significantly increased knee OA risk (p value = 0.092), and this result remained consistent for both subgroups (p values for Asians and Arabs: 0.445 and 0.148, respectively). Funnel plots were used to demonstrate the association between OR and standard error in the allele model, with each point representing a study. We found no evidence of asymmetry from a visual observation, and Egger’s regression yielded the same result (p value of Egger’s regression: 0.628). Other selected results were based on dominant and recessive models. In all classical genetic model analysis, the risk of knee OA for individuals carrying different ACE I/D genotype was not significantly higher than other genotypes in either the entire meta-analysis or the subgroup analyses. Other symmetrical assessments of the genotype, including dominant and recessive models, are shown in S1 Fig. Egger’s regression test indicated no evidence of publication bias among the included studies and all genetic models in this meta-analysis. A summary of the all results is shown in Table 4. Post hoc analysis shows a negative finding in all comparisons after multiple comparison correction. Moreover, all I2 values of classical genetic models exceeded 57%. This indicated a high heterogeneity on the association between ACE I/D on knee OA.

thumbnail
Fig 2. Selected results from the meta-analysis of angiotensin-converting enzyme insertion/deletion (ACE I/D) and knee osteoarthritis (OA).

The top left subplot is a forest plot based on an allele model assumption (reference: I allele), and the top right subplot is a funnel plot based on the allele model assumption. The allele model is the most common method for detecting gene–disease associations; however, we found no significant signal in the allele model. However, the funnel plot indicates good symmetry in this meta-analysis. Results obtained with the dominant and recessive models are presented at the bottom. All results were nonsignificant.

https://doi.org/10.1371/journal.pone.0161754.g002

thumbnail
Table 4. Odds ratios of angiotensin-converting enzyme insertion/deletion (I/D) and knee osteoarthritis using assumptions from allele type, genotype, dominant, and recessive models.

https://doi.org/10.1371/journal.pone.0161754.t004

Table 5 shows the results of a meta-regression that explored the source of the high heterogeneity. All of the potential factors could not significantly explain heterogeneity in the meta-analyses of all classical genetic models and post hoc analysis.

4. Discussion

Our case-control study results showed no significant associations between ACE I/D and knee OA. A meta-analysis of the associations in six studies yielded the same results. In our hypothesis, the ACE I/D DD genotype was associated with higher gene expression [63] and serum ACE levels [64] than the ID genotype, which was then followed by the II genotype. High blood ACE levels might increase bradykinin metabolism [2629], thereby reducing the risk of knee OA because bradykinin involves the generation of pain, swelling, and cellular damage associated with joint disease [1922]. According to this mechanism, the D allele may therefore be a protective factor against knee OA, although this was not confirmed in our study. In short, we considered our analysis provided a negative finding.

The main function of ACE is to convert angiotensin I to the potent vasoconstrictor angiotensin II [24,25], and we considered that this pathway might also impact susceptibility to knee OA. Knee OA is also considered an inflammation-related disease [9], and previous studies have confirmed a relationship between inflammation-related gene polymorphisms and knee OA [34,35,65,66]. Angiotensin II regulates the synthesis of proinflammatory cytokines, including TNF-alpha, IL-6, MCP-1, and NF-kappaB, and these proinflammatory cytokines play a key role in OA progression [67]. According to this mechanism, a higher ACE concentration might also increase the risk of knee OA. ACE expression is upregulated in the synovial stroma in rheumatoid arthritis, thereby contributing to synovial hypoxia and proliferation [68,69]; rheumatoid arthritis patients have higher ACE concentrations when compared with healthy controls [70]. As observed in patients with rheumatoid arthritis, patients with OA have higher levels of ACE activity in synovial fluid than controls [71]. The above evidence expounds upon the relationship between ACE and knee OA and includes at least two molecular pathways with opposite effects. Therefore, the effect of the ACE I/D D allele on knee OA might exert two opposing forces, and many possible factors could modify its effect. According to this complex mechanism, we considered the high heterogeneity in our meta-analysis to be reasonable, although further research is needed.

We performed a Trial Sequential Analysis (TSA) for calculating the required sample size in this issue [72]. The settings used were as follows: a two-sided test with a power = 0.95 (because our result is negative, so we used a higher power to avoid false negatives) at a significance level of 0.05, ratio of controls to cases = 1, hypothetical proportion of controls with exposure = 49 (based on data), the lowest extreme OR to be detected = 1.5, and I2 (heterogeneity) = 90%. Following these settings, the number of needed samples is equal to 2575 and we collected 2194 samples in this meta-analysis. Although this was still insufficient, Fig 3 shows the last point is in the invalid area. Thus, we considered this meta-analysis provided negative evidence of ACE I/D on knee OA. High heterogeneity in the meta-analysis suggests that complex molecular pathways in ACE I/D and knee OA likely exist. Epistasis analysis is therefore necessary for explaining this heterogeneity.

thumbnail
Fig 3. Trial Sequential Analysis (TSA) in this meta-analysis.

TSA is a methodology that includes a sample size calculation for a meta-analysis with the threshold of statistical significance. We performed a TAS using an allele model assumption, but replaced the allele count with the sample size (divided by 2). Detailed settings: Significance level = 0.05; Power = 0.95; ratio of controls to cases = 1; hypothetical proportion of controls with D allele = 49; least extreme OR to be detected = 1.5; I2 (heterogeneity) = 90%.

https://doi.org/10.1371/journal.pone.0161754.g003

Despite the high heterogeneity in the current study, our case–control study results were most similar to those reported by Hong et al. [36]. We considered this to be due to the high genetic homogeneity. Korea and Taiwan are both East Asian countries and share a similar ancestral population. We used the Cochrane Q test to compare these two studies in all genetic models and only obtained nonsignificant results. However, the Asian subgroup also exhibited high heterogeneity in the meta-analysis. This heterogeneity might have been due to a study of an Indian population [40]. The 1,000-Genome Project considered South Asians and East Asians to comprise different racial categories [42]. Because the relationship between ACE I/D and knee OA involves a complex mechanism, a small racial difference might also increase heterogeneity. We think that additional future studies conducted in neighboring regions will help us to control the population stratification.

Compared with the other included populations, our study population had the highest mean age, highest proportion of men, and lowest mean BMI. However, we considered that these differences in population characteristics might not affect the association between ACE I/D and knee OA because the results of a meta-regression showed that age, gender, and BMI could not explain the heterogeneity. Therefore, a high heterogeneity, regardless of the inclusion of Asian studies or Arab studies, might be due to another unknown factor. Because the role of ACE in knee OA may involve many possible pathways that even include opposing effects, we considered that the source of heterogeneity might include molecular factors such as gene–gene interactions. In fact, complex gene–gene interactions in the RAS were reported in the context of other diseases [73], which involved gene polymorphisms of ACE I/D. Future studies should further investigate the epistasis analysis of ACE I/D and knee OA [74].

In sensitivity analysis, we found none of the Arab population studies passed the HWE test. Previous studies reported that genotyping error was the most common reason for this HWE failure [7577]. Although we could not confirm this reason, it suggests a lower evidence level in the Arab population studies. A meta-analysis remaining Asian studies yielded a nonsignificant result [OR: 1.22 (95% CI: 0.96–1.54, p value: 0.101)], this is a type of sensitivity analysis conducted by removing studies without HWE [57,58]. In summary, the sensitivity analysis shows the same results compared with original result.

Although we investigated the most common polymorphism in ACE, a previous study suggested that this might not be a functional locus. The functional polymorphism is most likely located between intron 18 and the 3′ UTR [78], and other commonly loci or unobserved ACE variants may capture the effects of functional variants on such local ACE actions more effectively. However, a very few genetic association studies have been carried out to investigate the association between these loci and OA, so we were unable to collect enough information in our meta-analysis. In addition to epistasis analysis, future efforts should be made for collecting a large sample to allow testing of more candidate genes, including the structural variation, to find additional putative functional variants, or an analysis at the genome-wide level.

Several potential limitations should be acknowledged. First, we relied on the tabular data for the meta-analysis, rather than on the individual patient data. However, we included a case–control study to improve the sample size and evidence level. Second, in addition to our case–control study, only five studies had investigated the link between OA and the ACE I/D polymorphism. The relationship between ACE I/D and knee OA involves a complex mechanism, thus indicating the need for additional studies. Our case–control study involves the largest sample to date and thus provides more evidence for a clear understanding of this association.

In conclusion, our meta-analysis provides evidence important to an exploration of the association between ACE I/D and knee OA. The ACE I/D D allele might not be protective against knee OA, and negative results revealed ACE I/D was not directly associated with OA. The high level of heterogeneity suggests a complex mechanism, and several possible environmental factors, including gender, age, and BMI, cannot explain this heterogeneity. This unexplainable heterogeneity might be related to gene–gene interactions although additional well-designed studies to address the relationship between ACE I/D and knee OA are needed to explore the underlying mechanism. An epistasis analysis of ACE I/D and knee OA should be conducted. A potential gene–environmental interaction should not be excluded, and additional environmental factors should also be investigated.

Supporting Information

S1 Fig. Funnel plots of the genotype, dominant, and recessive models.

https://doi.org/10.1371/journal.pone.0161754.s001

(PDF)

S2 Table. Meta-analysis on Genetic Association Studies Checklist | PLOS ONE.

https://doi.org/10.1371/journal.pone.0161754.s003

(DOCX)

S3 Table. Search strategies and detailed records.

https://doi.org/10.1371/journal.pone.0161754.s004

(DOCX)

S4 Table. Angiotensin-converting enzyme insertion/deletion (I/D) genotype frequencies in advanced OA, early OA, and controls.

https://doi.org/10.1371/journal.pone.0161754.s005

(DOC)

Author Contributions

  1. Conceived and designed the experiments: CL WHF CMC FHL SLS.
  2. Performed the experiments: WHF CCW YJP GSH FHL SLS.
  3. Analyzed the data: CL WHF.
  4. Contributed reagents/materials/analysis tools: CMC CCW YJP GSH YJT HLL FHL SLS.
  5. Wrote the paper: CL WHF.
  6. Critical revise: HCC HSL HC CMC WTC FHL SLS.

References

  1. 1. Altman RD. Classification of disease: osteoarthritis. Seminars in arthritis and rheumatism. 1991;20(6 Suppl 2):40–7. Epub 1991/06/01. pmid:1866629
  2. 2. Felson DT, Lawrence RC, Dieppe PA, Hirsch R, Helmick CG, Jordan JM, et al. Osteoarthritis: new insights. Part 1: the disease and its risk factors. Annals of internal medicine. 2000;133(8):635–46. Epub 2000/10/18. pmid:11033593
  3. 3. Ciombor DM, Aaron RK, Wang S, Simon B. Modification of osteoarthritis by pulsed electromagnetic field—a morphological study. Osteoarthritis and cartilage / OARS, Osteoarthritis Research Society. 2003;11(6):455–62. Epub 2003/06/13. pmid:12801485
  4. 4. Dieppe PA, Lohmander LS. Pathogenesis and management of pain in osteoarthritis. Lancet. 2005;365(9463):965–73. Epub 2005/03/16.
  5. 5. Michael JW, Schluter-Brust KU, Eysel P. The epidemiology, etiology, diagnosis, and treatment of osteoarthritis of the knee. Deutsches Arzteblatt international. 2010;107(9):152–62. Epub 2010/03/23. pmid:20305774
  6. 6. Ghosh P, Cheras PA. Vascular mechanisms in osteoarthritis. Best practice & research Clinical rheumatology. 2001;15(5):693–709. Epub 2002/01/29.
  7. 7. Buchanan WW, Kean WF, Kean R. History and current status of osteoarthritis in the population. Inflammopharmacology. 2003;11(4):301–16. Epub 2004/03/24. pmid:15035784
  8. 8. Goldring MB, Otero M. Inflammation in osteoarthritis. Current opinion in rheumatology. 2011;23(5):471–8. Epub 2011/07/27.
  9. 9. Berenbaum F. Osteoarthritis as an inflammatory disease (osteoarthritis is not osteoarthrosis!). Osteoarthritis and cartilage / OARS, Osteoarthritis Research Society. 2013;21(1):16–21. Epub 2012/12/01. pmid:23194896
  10. 10. Reginato AM, Olsen BR. The role of structural genes in the pathogenesis of osteoarthritic disorders. Arthritis research. 2002;4(6):337–45. Epub 2002/11/28. pmid:12453309
  11. 11. Snelling S, Sinsheimer JS, Carr A, Loughlin J. Genetic association analysis of LRCH1 as an osteoarthritis susceptibility locus. Rheumatology (Oxford, England). 2007;46(2):250–2. Epub 2006/08/08.
  12. 12. Evangelou E, Chapman K, Meulenbelt I, Karassa FB, Loughlin J, Carr A, et al. Large-scale analysis of association between GDF5 and FRZB variants and osteoarthritis of the hip, knee, and hand. Arthritis and rheumatism. 2009;60(6):1710–21. Epub 2009/05/30. pmid:19479880
  13. 13. Kerkhof HJ, Lories RJ, Meulenbelt I, Jonsdottir I, Valdes AM, Arp P, et al. A genome-wide association study identifies an osteoarthritis susceptibility locus on chromosome 7q22. Arthritis and rheumatism. 2010;62(2):499–510. Epub 2010/01/30. pmid:20112360
  14. 14. Valdes AM, Spector TD. Genetic epidemiology of hip and knee osteoarthritis. Nature reviews Rheumatology. 2011;7(1):23–32. Epub 2010/11/17. pmid:21079645
  15. 15. Waarsing JH, Kloppenburg M, Slagboom PE, Kroon HM, Houwing-Duistermaat JJ, Weinans H, et al. Osteoarthritis susceptibility genes influence the association between hip morphology and osteoarthritis. Arthritis and rheumatism. 2011;63(5):1349–54. Epub 2011/03/15. pmid:21400473
  16. 16. Chen S, Zhou Y, Li J, Shan LQ, Fan QY. The effect of bradykinin B2 receptor polymorphisms on the susceptibility and severity of osteoarthritis in a Chinese cohort. Journal of biomedicine & biotechnology. 2012;2012:597637. Epub 2012/10/25.
  17. 17. MacGregor AJ, Spector TD. Twins and the genetic architecture of osteoarthritis. Rheumatology (Oxford, England). 1999;38(7):583–8. Epub 1999/08/26.
  18. 18. Valdes AM, Van Oene M, Hart DJ, Surdulescu GL, Loughlin J, Doherty M, et al. Reproducible genetic associations between candidate genes and clinical knee osteoarthritis in men and women. Arthritis and rheumatism. 2006;54(2):533–9. Epub 2006/02/03. pmid:16453284
  19. 19. Bellucci F, Cucchi P, Catalani C, Giuliani S, Meini S, Maggi CA. Novel effects mediated by bradykinin and pharmacological characterization of bradykinin B2 receptor antagonism in human synovial fibroblasts. British journal of pharmacology. 2009;158(8):1996–2004. Epub 2010/01/06. pmid:20050188
  20. 20. Meini S, Cucchi P, Catalani C, Bellucci F, Giuliani S, Maggi CA. Bradykinin and B(2) receptor antagonism in rat and human articular chondrocytes. British journal of pharmacology. 2011;162(3):611–22. Epub 2010/10/16. pmid:20946124
  21. 21. Warde N. Osteoarthritis: Local antagonism of endothelin-1 and bradykinin receptors improves OA pain and joint morphology in rats. Nature reviews Rheumatology. 2011;7(7):375. Epub 2011/06/22.
  22. 22. Bellucci F, Meini S, Cucchi P, Catalani C, Nizzardo A, Riva A, et al. Synovial fluid levels of bradykinin correlate with biochemical markers for cartilage degradation and inflammation in knee osteoarthritis. Osteoarthritis and cartilage / OARS, Osteoarthritis Research Society. 2013;21(11):1774–80. Epub 2013/08/15. pmid:23942063
  23. 23. Nishimura M, Segami N, Kaneyama K, Suzuki T, Miyamaru M. Relationships between pain-related mediators and both synovitis and joint pain in patients with internal derangements and osteoarthritis of the temporomandibular joint. Oral surgery, oral medicine, oral pathology, oral radiology, and endodontics. 2002;94(3):328–32. Epub 2002/09/27. pmid:12324788
  24. 24. Riordan JF. Angiotensin-I-converting enzyme and its relatives. Genome biology. 2003;4(8):225. Epub 2003/08/14. pmid:12914653
  25. 25. Dalbeth N, Edwards J, Fairchild S, Callan M, Hall FC. The non-thiol angiotensin-converting enzyme inhibitor quinapril suppresses inflammatory arthritis. Rheumatology (Oxford, England). 2005;44(1):24–31. Epub 2004/09/09.
  26. 26. Pellacani A, Brunner HR, Nussberger J. Plasma kinins increase after angiotensin-converting enzyme inhibition in human subjects. Clinical science (London, England: 1979). 1994;87(5):567–74. Epub 1994/11/01.
  27. 27. Barbe F, Su JB, Guyene TT, Crozatier B, Menard J, Hittinger L. Bradykinin pathway is involved in acute hemodynamic effects of enalaprilat in dogs with heart failure. The American journal of physiology. 1996;270(6 Pt 2):H1985–92. Epub 1996/06/01. pmid:8764248
  28. 28. Su JB, Barbe F, Crozatier B, Campbell DJ, Hittinger L. Increased bradykinin levels accompany the hemodynamic response to acute inhibition of angiotensin-converting enzyme in dogs with heart failure. Journal of cardiovascular pharmacology. 1999;34(5):700–10. Epub 1999/11/05. pmid:10547087
  29. 29. Witherow FN, Helmy A, Webb DJ, Fox KA, Newby DE. Bradykinin contributes to the vasodilator effects of chronic angiotensin-converting enzyme inhibition in patients with heart failure. Circulation. 2001;104(18):2177–81. Epub 2001/10/31. pmid:11684627
  30. 30. Tiret L, Rigat B, Visvikis S, Breda C, Corvol P, Cambien F, et al. Evidence, from combined segregation and linkage analysis, that a variant of the angiotensin I-converting enzyme (ACE) gene controls plasma ACE levels. American journal of human genetics. 1992;51(1):197–205. Epub 1992/07/01. pmid:1319114
  31. 31. Chung CM, Wang RY, Fann CS, Chen JW, Jong YS, Jou YS, et al. Fine-mapping angiotensin-converting enzyme gene: separate QTLs identified for hypertension and for ACE activity. PloS one. 2013;8(3):e56119. Epub 2013/03/08. pmid:23469169
  32. 32. Gunel T, Gumusoglu E, Hosseini MK, Yilmazyildirim E, Dolekcap I, Aydinli K. Effect of angiotensin I-converting enzyme and alpha-actinin-3 gene polymorphisms on sport performance. Molecular medicine reports. 2014;9(4):1422–6. Epub 2014/02/26. pmid:24566537
  33. 33. Shinjo SK, Uno M, Oba-Shinjo SM, Marie SK. Angiotensin-converting enzyme insertion/deletion gene polymorphism is associated with dermatomyositis. Journal of the renin-angiotensin-aldosterone system: JRAAS. 2015;16(3):666–71. Epub 2014/03/05. pmid:24591530
  34. 34. Abel K, Reneland R, Kammerer S, Mah S, Hoyal C, Cantor CR, et al. Genome-wide SNP association: identification of susceptibility alleles for osteoarthritis. Autoimmunity reviews. 2006;5(4):258–63. Epub 2006/05/16. pmid:16697966
  35. 35. Evangelou E, Valdes AM, Kerkhof HJ, Styrkarsdottir U, Zhu Y, Meulenbelt I, et al. Meta-analysis of genome-wide association studies confirms a susceptibility locus for knee osteoarthritis on chromosome 7q22. Annals of the rheumatic diseases. 2011;70(2):349–55. Epub 2010/11/12. pmid:21068099
  36. 36. Hong SJ, Yang HI, Yoo MC, In CS, Yim SV, Jin SY, et al. Angiotensin converting enzyme gene polymorphism in Korean patients with primary knee osteoarthritis. Experimental & molecular medicine. 2003;35(3):189–95. Epub 2003/07/15.
  37. 37. Shehab DK, Al-Jarallah KF, Alawadhi AM, Al-Herz A, Nahar I, Haider MZ. Prevalence of angiotensin-converting enzyme gene insertion-deletion polymorphism in patients with primary knee osteoarthritis. Clinical and experimental rheumatology. 2008;26(2):305–10. Epub 2008/06/21. pmid:18565253
  38. 38. Bayram B, Sayin E, Gunes HV, Degirmenci I, Turkoglu Z, Doganer F, et al. DD genotype of ace gene I/D polymorphism is associated in a Turkish study population with osteoarthritis. Molecular biology reports. 2011;38(3):1713–6. Epub 2010/09/17. pmid:20845076
  39. 39. Inanir A, Yigit S, Tural S, Cecen O, Yildirim E. MTHFR gene C677T mutation and ACE gene I/D polymorphism in Turkish patients with osteoarthritis. Disease markers. 2013;34(1):17–22. Epub 2012/10/24. pmid:23089924
  40. 40. Poornima S, Subramanyam K, Khan IA, Hasan Q. The insertion and deletion (I28005D) polymorphism of the angiotensin I converting enzyme gene is a risk factor for osteoarthritis in an Asian Indian population. Journal of the renin-angiotensin-aldosterone system: JRAAS. 2015;16(4):1281–7. Epub 2014/09/03. pmid:25178458
  41. 41. Fleiss JL, Levin B, Paik MC. Statistical methods for rates and proportions: John Wiley & Sons; 2013.
  42. 42. Auton A, Brooks LD, Durbin RM, Garrison EP, Kang HM, Korbel JO, et al. A global reference for human genetic variation. Nature. 2015;526(7571):68–74. Epub 2015/10/04. pmid:26432245
  43. 43. Wacholder S, Chanock S, Garcia-Closas M, El Ghormli L, Rothman N. Assessing the probability that a positive report is false: an approach for molecular epidemiology studies. Journal of the National Cancer Institute. 2004;96(6):434–42. Epub 2004/03/18. pmid:15026468
  44. 44. Fang WH, Huang GS, Chang HF, Chen CY, Kang CY, Wang CC, et al. Gender differences between WOMAC index scores, health-related quality of life and physical performance in an elderly Taiwanese population with knee osteoarthritis. BMJ open. 2015;5(9):e008542. Epub 2015/09/17. pmid:26373405
  45. 45. Kellgren JH, Lawrence JS. Radiological assessment of rheumatoid arthritis. Annals of the rheumatic diseases. 1957;16(4):485–93. Epub 1957/12/01. pmid:13498603
  46. 46. Yang HY, Lu KC, Fang WH, Lee HS, Wu CC, Huang YH, et al. Impact of interaction of cigarette smoking with angiotensin-converting enzyme polymorphisms on end-stage renal disease risk in a Han Chinese population. Journal of the renin-angiotensin-aldosterone system: JRAAS. 2015;16(1):203–10. Epub 2013/03/13. pmid:23477970
  47. 47. Rigat B, Hubert C, Corvol P, Soubrier F. PCR detection of the insertion/deletion polymorphism of the human angiotensin converting enzyme gene (DCP1) (dipeptidyl carboxypeptidase 1). Nucleic acids research. 1992;20(6):1433. Epub 1992/03/25.
  48. 48. Shanmugam V, Sell KW, Saha BK. Mistyping ACE heterozygotes. PCR methods and applications. 1993;3(2):120–1. Epub 1993/10/01. pmid:8268786
  49. 49. Freidlin B, Zheng G, Li Z, Gastwirth JL. Trend tests for case-control studies of genetic markers: power, sample size and robustness. Human heredity. 2002;53(3):146–52. Epub 2002/07/30.
  50. 50. Gonzalez JR, Carrasco JL, Dudbridge F, Armengol L, Estivill X, Moreno V. Maximizing association statistics over genetic models. Genetic epidemiology. 2008;32(3):246–54. Epub 2008/01/30. pmid:18228557
  51. 51. Zheng G, Ng HK. Genetic model selection in two-phase analysis for case-control association studies. Biostatistics (Oxford, England). 2008;9(3):391–9. Epub 2007/11/16.
  52. 52. Zang Y, Fung WK, Zheng G. Simple algorithms to calculate the asymptotic null distributions of robust tests in case-control genetic association studies in R. Journal of Statistical software. 2010;33(8).
  53. 53. Moher D, Liberati A, Tetzlaff J, Altman DG. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. PLoS medicine. 2009;6(7):e1000097. Epub 2009/07/22. pmid:19621072
  54. 54. Wells G, Shea B, O’connell D, Peterson J, Welch V, Losos M, et al. The Newcastle-Ottawa Scale (NOS) for assessing the quality of nonrandomised studies in meta-analyses 2000 [cited 2013 24 August]. Available from: http://www.ohri.ca/programs/clinical_epidemiology/oxford.asp.
  55. 55. Egger M, Davey Smith G, Schneider M, Minder C. Bias in meta-analysis detected by a simple, graphical test. BMJ (Clinical research ed). 1997;315(7109):629–34. Epub 1997/10/06.
  56. 56. Higgins JP, Thompson SG, Deeks JJ, Altman DG. Measuring inconsistency in meta-analyses. BMJ (Clinical research ed). 2003;327(7414):557–60. Epub 2003/09/06.
  57. 57. Trikalinos TA, Salanti G, Khoury MJ, Ioannidis JP. Impact of violations and deviations in Hardy-Weinberg equilibrium on postulated gene-disease associations. American journal of epidemiology. 2006;163(4):300–9. Epub 2006/01/18. pmid:16410351
  58. 58. Ziegler A, Van Steen K, Wellek S. Investigating Hardy-Weinberg equilibrium in case-control or cohort studies or meta-analysis. Breast cancer research and treatment. 2011;128(1):197–201. Epub 2010/12/25. pmid:21184275
  59. 59. Lin C, Chu CM, Lin J, Yang HY, Su SL. Gene-gene and gene-environment interactions in meta-analysis of genetic association studies. PloS one. 2015;10(4):e0124967. Epub 2015/04/30. pmid:25923960
  60. 60. Lin C, Yang HY, Wu CC, Lee HS, Lin YF, Lu KC, et al. Angiotensin-converting enzyme insertion/deletion polymorphism contributes high risk for chronic kidney disease in asian male with hypertension-a meta-regression analysis of 98 observational studies. PloS one. 2014;9(1):e87604. pmid:24498151
  61. 61. Viechtbauer W. Conducting meta-analyses in R with the metafor package. Journal of Statistical Software. 2010;36(3):1–48.
  62. 62. Schwarzer G. meta: Meta-Analysis with R. 2012.
  63. 63. Mizuiri S, Hemmi H, Kumanomidou H, Iwamoto M, Miyagi M, Sakai K, et al. Angiotensin-converting enzyme (ACE) I/D genotype and renal ACE gene expression. Kidney international. 2001;60(3):1124–30. Epub 2001/09/05. pmid:11532108
  64. 64. Rigat B, Hubert C, Alhenc-Gelas F, Cambien F, Corvol P, Soubrier F. An insertion/deletion polymorphism in the angiotensin I-converting enzyme gene accounting for half the variance of serum enzyme levels. The Journal of clinical investigation. 1990;86(4):1343–6. Epub 1990/10/01. pmid:1976655
  65. 65. Su SL, Yang HY, Lee CH, Huang GS, Salter DM, Lee HS. The (-1486T/C) promoter polymorphism of the TLR-9 gene is associated with end-stage knee osteoarthritis in a Chinese population. Journal of orthopaedic research: official publication of the Orthopaedic Research Society. 2012;30(1):9–14. Epub 2011/07/13.
  66. 66. Yang HY, Lee HS, Lee CH, Fang WH, Chen HC, Salter DM, et al. Association of a functional polymorphism in the promoter region of TLR-3 with osteoarthritis: a two-stage case-control study. Journal of orthopaedic research: official publication of the Orthopaedic Research Society. 2013;31(5):680–5. Epub 2012/12/21.
  67. 67. Ruiz-Ortega M, Ruperez M, Lorenzo O, Esteban V, Blanco J, Mezzano S, et al. Angiotensin II regulates the synthesis of proinflammatory cytokines and chemokines in the kidney. Kidney international Supplement. 2002;(82):S12–22. Epub 2002/11/02. pmid:12410849
  68. 68. Walsh DA, Catravas J, Wharton J. Angiotensin converting enzyme in human synovium: increased stromal [(125)I]351A binding in rheumatoid arthritis. Annals of the rheumatic diseases. 2000;59(2):125–31. Epub 2000/02/09. pmid:10666168
  69. 69. Min BH, Kim HJ, Lim H, Park CS, Park SR. Effects of ageing and arthritic disease on nitric oxide production by human articular chondrocytes. Experimental & molecular medicine. 2001;33(4):299–302. Epub 2002/01/25.
  70. 70. Cobankara V, Ozturk MA, Kiraz S, Ertenli I, Haznedaroglu IC, Pay S, et al. Renin and angiotensin-converting enzyme (ACE) as active components of the local synovial renin-angiotensin system in rheumatoid arthritis. Rheumatology international. 2005;25(4):285–91. Epub 2005/03/12. pmid:15761728
  71. 71. Blann AD. von Willebrand factor antigen and angiotensin converting enzyme in synovial fluid. Scandinavian journal of rheumatology. 1991;20(3):213–4. Epub 1991/01/01. pmid:1648787
  72. 72. Wetterslev J, Thorlund K, Brok J, Gluud C. Trial sequential analysis may establish when firm evidence is reached in cumulative meta-analysis. Journal of clinical epidemiology. 2008;61(1):64–75. Epub 2007/12/18. pmid:18083463
  73. 73. Su SL, Yang HY, Wu CC, Lee HS, Lin YF, Hsu CA, et al. Gene-gene interactions in renin-angiotensin-aldosterone system contributes to end-stage renal disease susceptibility in a Han Chinese population. Scientific World Journal. 2014;2014:169798. pmid:24977181
  74. 74. Lin C, Chu CM, Su SL. Epistasis Test in Meta-Analysis: A Multi-Parameter Markov Chain Monte Carlo Model for Consistency of Evidence. PloS one. 2016;11(4):e0152891. Epub 2016/04/06. pmid:27045371
  75. 75. Xu J, Turner A, Little J, Bleecker ER, Meyers DA. Positive results in association studies are associated with departure from Hardy-Weinberg equilibrium: hint for genotyping error? Human genetics. 2002;111(6):573–4. Epub 2003/01/09. pmid:12516594
  76. 76. Hosking L, Lumsden S, Lewis K, Yeo A, McCarthy L, Bansal A, et al. Detection of genotyping errors by Hardy-Weinberg equilibrium testing. European journal of human genetics: EJHG. 2004;12(5):395–9. Epub 2004/02/12. pmid:14872201
  77. 77. Salanti G, Amountza G, Ntzani EE, Ioannidis JP. Hardy-Weinberg equilibrium in genetic association studies: an empirical evaluation of reporting, deviations, and power. European journal of human genetics: EJHG. 2005;13(7):840–8. Epub 2005/04/14. pmid:15827565
  78. 78. Sayed-Tabatabaei FA, Oostra BA, Isaacs A, van Duijn CM, Witteman JC. ACE polymorphisms. Circulation research. 2006;98(9):1123–33. Epub 2006/05/13. pmid:16690893