Association of MYOC and APOE promoter polymorphisms and primary open-angle glaucoma: a meta-analysis

Abstract

Background: Primary open-angle glaucoma (POAG) is the most common form of glaucoma with a genetic predisposition. The relationship between polymorphisms in MYOC or APOE promoter region and POAG has been addressed in many case-control studies, but the published results were not consistent. Methods: A meta-analysis assessing the association between five single nucleotide polymorphisms (SNPs) (in MYOC promoter: rs12035719 and rs2075648; in APOE promoter: rs405509, rs769446 and rs449647) and the risk of POAG was performed based on included studies from literature research. In fixed effect model or random effect model, the Mantel-Haenszel (M-H) pooled odds ratios (ORs) and 95% confidence intervals (95% CIs) were used to evaluate the genetic association. Stratification analysis was also conducted to test the association within Asian or Caucasian populations. Results: Twenty five case-control studies within multiple populations were identified and no publish bias was observed. Significant association was detected between POAG risk and MYOC rs2075648 in Caucasian (GA+AA vs. GG, OR=0.587, 95% CI=0.437-0.788, P < 0.001). For other SNPs and in other ethnic populations, no statistic evidence was detected for significant association between them and the development of POAG. Conclusions: This meta-analysis suggested a genetic association between one of MYOC polymorphism (rs2075648) and the risk of POAG only in Caucasian population. The significant heterogeneity for this locus might imply the different POAG genetic basis among different populations.

Introduction

Glaucoma is a group of optic diseases characterized by optic disc cupping and loss of visual caused by elevated intra-ocular pressure (IOP) and is predicted to involve 79.6 million patients by 2020 [1]. Primary open-angle glaucoma (POAG, OMIM137760) is a common form of glaucoma, which could be subdivided into chronic open-angle glaucoma (COAG) and juvenile open-angle glaucoma (JOAG). Family aggregation and high concordance in monozygotic twin pairs (98.0% in monozygotic twins pairs and 72.0% in twin/spouse pairs) implied strong genetic basis for POAG [2,3]. The first predisposing gene for POAG, MYOC (myocilin, also called trabecular meshwork-induced glucocorticoid response protein, TIGR) was identified in 1997 at the GLC1A locus [4], and latter identified POAG genes include OPTN [5] and WDR36 [6]. However currently only about 5% of POAG could be attributed to Mendelian form mutation, other POAG cases may have a complex genetic basis involving multiple minor-effect genes [7].

As the first recognized POAG gene, MYOC mutations cause most cases of autosomal dominant JOAG and lead to 4.6% of adult-onset POAG [8]. While some mutations in MYOC (such as GLN368STOP) have been detected in many association studies, the normal function of its encoding protein remains unclear [7,9]. In 2001, Colomb, E. et al. [10] found that a single nucleotide polymorphism (SNP) in MYOC promoter, -1000C/G (also designated as MYOC.mt1 or rs12035719), was associated with increased IOP and severity of damaged visual field in POAG. Subsequent studies (Table 1) further explored the association between this SNP and the development or clinical features of POAG, but their findings were inconsistent. In 2008, a systematic review and meta-analysis [11] evaluating the association between rs12035719 and the risk of POAG was published, with only four studies included. Another SNP in the promoter region of MYOC, -83G>A (rs2075648), was also analyzed in many POAG genetic association studies (Table 1). These case-control studies were limited in their sample size and there haven’t been meta-analysis performed to examine the relationship between POAG and MYOC rs2075648 to date.

Table 1.

Characteristics of the included studies

Study	Population	Ethnicity	SNPa	POAGs:Controls	Genotyping method
Fingert, J. H. 1999b [24]	USA Australian and Canadian	Caucasian	rs2075648	1284:91	SSCP
	African American	Negroid	rs2075648	312:40	SSCP
Suzuki, R. 2000 [57]	Japan	Asian	rs2075648	30:36	unspecified
Colomb, E. 2001 [10]	French	Caucasian	rs12035719	142:94	ASO
Mabuchi, F. 2001 [58]	Japanese	Asian	rs2075648	119:100	SSCP
Alward, W. L. 2002 [25]	USA	Caucasian	rs12035719	393:92	SSCP and RFLP
Copin, B. 2002 [18]	French	Caucasian	rs405509, rs769446, rs449647	191:102	unspecified
Hulsman, C. A. 2002 [59]	Dutch	Caucasian	rs2075648	40:94	SSCP
Mukhopadhyay, A. 2002 [60]	Indian	Asian	rs2075648	56:51	PCR-RFLP
Melki, R. 2003 [61]	French	Caucasian	rs2075648	237:108	DHPLC
Fan, B. J. 2004 [36]	Chinese	Asian	rs2075648	88:94	HTCSGE
			rs12035719	212:221	HTCSGE and RFLP
Fan, B. J.2005 [28]	Chinese	Asian	rs405509, rs769446, rs449647	400:281	PCR-RFLP
			rs2075648		HTCSGE
Ozgul, R. K. 2005 [62]	Turkish	Caucasian	rs12035719	88:123	PCR-RFLP
Saura, M. 2005 [63]	Galician	Caucasian	rs2075648	79:109	SSCP and RFLP
Bhattacharjee, A. 2007 [64]	Indian	Asian	rs2075648	315:100	PCR-sequencing
Kumar, A. 2007 [65]	Indian	Asian	rs2075648	116:98	PCR-SSCP
Lopez-Martinez, F. 2007 [66]	Spanish	Caucasian	rs2075648, rs12035719	110:98	PCR-sequencing
Yen, Y. C. 2007 [67]	Taiwanese	Asian	rs12035719	48:100	PCR-sequencing
Jia, L. Y. 2009 [38]	Chinese	Asian	rs2075648	175:200	PCR-sequencing
			rs405509, rs769446, rs449647		TaqMan
Sohn, S. 2010 [68]	Korean	Asian	rs2075648	60:74	PCR-sequencing
Kasahara, N. 2011 [69]	Brazilian	Caucasian	rs12035719	167:130	TaqMan
Whigham, B. T. 2011 [70]	Southern African	Negroid	rs2075648	113:131	PCR-sequencing
Banerjee, D. 2012 [71]	Indian	Asian	rs2075648	250:100	PCR-sequencing
Buentello-Volante, B. 2013 [72]	Mexican	Caucasian	rs12035719	118:100	PCR-sequencing
Nowak, A. 2013 [72]	Polish	Caucasian	rs449647	183:209	PCR-RFLP
Saglar, E. 2014 [73]	Turkish	Caucasian	rs405509	75:122	PCR-RFLP

PCR: polymerase chain reaction; RFLP: restricted fragment length polymorphism; SSCP: single strand conformation polymorphism; ASO: allele-specific oligonucleotide; DHPLC: denaturing high performance liquid chromatography; HTCSGE: high throughput conformation sensitive gel electrophoresis; POAG: primary open-angle glaucoma.

^aHere we only listed our concerned SNPs in the included studies, which might discussed SNPs other than the five SNPs we focused.

^bThe association study in this literature was subdivided into two dependent studies (in Caucasian and in Negroid) given the ethnic heterogeneity of its samples.

APOE encodes the major apolipoprotein in central nerve system and its polymorphism plays a role in Alzheimer’s disease (AD) and coronary heart disease [12,13]. The optic nerve injury in POAG has some similarity with the involved neuron in AD [14] and the association studies between APOE polymorphisms and AD has proved evidence for genetic predisposition [15], suggesting a potential role of APOE in the development of POAG. The association between APOE polymorphisms and POAG has been discussed in a number of case-control studies, but the results remained controversial [16,17]. In 2002, Copin, B. et al. found that the APOE promoter SNPs, which were valued in AD, also modified the POAG phenotype and might have an interaction with a SNP (rs12035719) in the MYOC promoter [18]. The subsequent studies (listed in Table 1) focused on the relation between three APOE promoter SNPs, namely -219T>G (rs405509), -427T/C (rs769446) and -491A>T (rs449647), and POAG, but their conclusions were not in agreement.

To give a relatively generalized and precise estimation of the association between the genetic polymorphisms in MYOC promoter and APOE promoter with the risk of POAG, we performed this literature based meta-analysis, which include five SNPs: rs2075648, rs12035719, rs405509, rs769446 and rs449647.

Methods

Search strategy

We conducted literature searches in PubMed, Web of Science, and China Biological Medicine Database (CBMD) to indentify related published articles up to September 2014. The following terms and their possible combinations were used: MYOC (or myocilin, trabecular meshwork-induced glucocorticoid response protein, TIGR, GLC1A), APOE (or apolipoprotein E), promoter, SNP (or respective name of the five included SNPs: rs2075648, rs12035719, rs405509, rs769446, rs449647 or their traditional names) and primary open angle glaucoma (or POAG).

Inclusion and exclusion criteria

The included studies must meet the following criteria: (1), evaluation the associations between the promoter polymorphisms of MYOC or APOE and POAG; (2), original and independent case-control study; (3), containing sufficient data for extraction for meta-analysis; (4), if genotype data were provided, there should be no significant deviation from Hardy-Weinberg equilibrium (HWE) in χ² test. Studies were excluded if they were only in vitro or in vivo mechanism studies or family linkage studies without population studies. Reviews or other studies based on earlier published data were also excluded. The literature language is restricted to English.

Data extraction

Data extraction was performed independently by 2 coauthors. Disagreements were resolved by discussion. The following information, in addition to published statistical data, was extracted: first author, publication year, population ethnicity, number of cases and controls, genotyping methods.

Statistical analysis

Heterogeneity among studies was assessed by Cochran’s chi-square test and by the I² statistic: I² < 25% implies slight heterogeneity while I² > 50% implies notable heterogeneity [19]. In the test of heterogeneity, P < 0.10 indicating significant heterogeneity; otherwise, P < 0.05 was considered statistically significant. If no significant heterogeneity detected among studies, the fixed-effects model was used, which assumes that the variability is due to random variation; otherwise, the random-effects model is used [20]. Random-effect model could accommodate the possible different environment effect among studies. The pooled odds ratios (ORs) and 95% confidence intervals (95% CIs) were calculated by the Mantel-Haenszel (M-H) method [21]. Subgroup analyses were performed by ethnicity and other environmental effects were ignored considering the relatively great heterogeneity. Begg’s funnel plots and Egger’s tests were used to detect publication bias: the more symmetrical the funnel plot, the less probability of publication bias. Egger’s linear regression method was used to measure the asymmetry of Begg’s funnel plots: the larger the deviation of the intercept from zero, the more asymmetrical the funnel plot [22]. Statistical analyses involved were performed using the statistic software STATA version 12.0 (STATA Corporation, College Station, TX, USA).

Results

Characteristics of included studies

Among the 252 electronically or manually identified articles, 25 articles were eventually included in our meta-analysis. The POAG patients in the study by Alward, W. L. et al. [23] were included in the study by Fingert, J. H. et al. [24] and only the latter were included in this study; the study by Alward, W. L. et al. published in 2002 [25] also included some patients in previous studies [23,24], but it was included here for the evaluation of another SNP (rs12035719, Table 1). The 91 patients in the study by Lam, D. S. C. et al. [26] were included in the study by Pang, C. P. et al. [27], and the 187 POAG patients in the latter study were again included in the study by Fan, B. J. et al. [28]; another study by Lam, C. Y. and Fan, B. J. et al. published next year [29] also provided the same genotype data of 400 POAG patients in rs12035719. Among the four researches of overlapping populations, only Fan’s study published in 2005 [28] were included in this meta-analysis. The detailed process was displayed in Figure 1 and the characteristics of finally included studies were listed in Table 1.

Figure 1.

Flow chart of literature search and study selection.

Estimation of association

The main calculation results were shown in Table 2. We first assessed the publication bias of included studies for each SNP. The Begg’s funnel plots for rs12035719 and rs2075648 were quite symmetrical (Figure 4) and the p values of Begg’s test and Egger’s test were all larger than 0.1 (Table 2), which suggested no publication bias in any discussed SNP.

Table 2.

Results of meta-analysis for MYOC and APOE promoter polymorphisms and risk of POAG

Polymorphism (comparison)	Population	No. of studies	Test of publish bias (p-value)		Test of association			Test of heterogeneity
			Egger’s test	Begg’s testa	OR (95% CI)	P-valueb	Effect Model	P-value	I 2
rs2075648 (GA+AA vs. GG)	Overall	17	0.554	0.592	0.849 (0.662-1.088)	0.196	random	0.019	46.4%
	Asian	10	0.328	0.371	0.996 (0.744-1.333)	0.978	random	0.059	45.1%
	Negroid	2	-	1.000	2.118 (0.080-56.354)	0.654	random	0.124	57.8%
	Caucasian	5	0.309	0.806	0.587 (0.437-0.788)	0.000	fixed	0.419	0.0%
rs12035719 (C vs. G)	Overall	8	0.777	0.536	1.017 (0.840-1.231)	0.184	fixed	0.184	30.5%
	Caucasian	6	0.516	1.000	0.991 (0.775-1.269)	0.946	fixed	0.140	39.8%
	Asian	2	-	1.000	1.056 (0.781-1.429)	0.772	fixed	0.189	42.0%
rs405509 (G vs. T)	Overall	4	0.384	1.000	1.022 (0.878-1.188)	0.782	fixed	0.503	0.0%
	Caucasian	2	-	1.000	0.976 (0.750-1.270)	0.856	fixed	0.614	0.0%
	Asian	2	-	1.000	1.045 (0.869-1.257)	0.642	fixed	0.166	48.0%
rs769446 (C vs. T)	overall	3	0.985	1.000	0.962 (0.658-1.408)	0.842	fixed	0.392	0.0%
rs449647 (T vs. A)	overall	4	0.442	0.308	1.257 (0.991-1.594)	0.059	fixed	0.225	31.2%
	Caucasian	2	-	1.000	1.272 (0.970-1.668)	0.081	fixed	0.143	53.5%
	Asian	2	-	1.000	1.207 (0.735-1.982)	0.458	fixed	0.140	54.1%

POAG: primary open-angle glaucoma.

^aContinuity-corrected P-value.

^bSignificance test of OR=1.

Figure 4.

Begg’s funnel plots with pseudo 95% confidence limits. A: rs12035719, overall; B: rs2075648, overall. The circles represent separate studies and its size represents the sample size of corresponding study.

For rs12035719, 8 studies (1278 POAG patients and 958 controls in total) were involved without publication bias (Figure 4A). The association was not statistically significant in addictive genetic model (C vs. G pooled OR=1.017, P=0.184; Table 1 and Figure 2). After stratified by ethnicity, the association was still not statistically proved in either Caucasian (P=0.946) and Asian (P=0.772). The association was also not observed in dominant or recessive model (data not shown).

Figure 2.

Forest plots of odds ratios (ORs) and 95% confidence intervals (95% CIs) of rs12035719 in MYOC promoter in the case-control studies. The size of the gray square represents the relative weight of each study.

For rs2075648, 17 studies (3719 patients and 1806 controls) were included for calculation. Since many of published literatures only provide the number of mutation without exact genotype data, the comparison were conducted only in dominant genetic model, namely GA+AA vs. GG in patients and controls. No publication bias was detected overall or in each ethnic subgroup (Figure 4B) but notable or moderate heterogeneity within studies exists except in Caucasian. The significant association was only found in Caucasian (OR=0.587, 95% CI: 0.437-0.788, P < 0.001; Table 1 and Figure 3). No association was found in Asian subgroups (OR=0.996, P=0.978) although data from ten studies were pooled.

Figure 3.

Forest plots of odds ratios (ORs) and 95% confidence intervals (95% CIs) of rs2075648 in MYOC promoter in the case-control studies. The size of the gray square represents the relative weight of each study. AA: African American.

For the three SNPs in the promoter region of APOE, rs405509 (four included studies, 842 patients and 705 controls), rs769446 (3 included studies, 767 patients and 583 controls) and rs449647 (four included studies, 950 patients and 792 controls), the meta-analysis results showed no evidence for significant association between any of them and the risk of POAG (rs405509 G vs. T OR=1.022 P=0.782; rs769446 C vs. T OR=0.962 P=0.842; rs449647 T vs. A OR=1.257 P=0.059; Table 2). The results calculated in dominant or recessive genetic model were similar with addictive genetics model (data not shown). The ethnic stratification meta-analysis still statistically suggested no significant association (Table 2).

Discussion

As the most common type of glaucoma, POAG is contributed by complex genetic factors [30]. In this meta-analysis, to assess the association between MYOC or APOE promoter variants and the risk of POAG in Asian and Caucasian populations, twenty five studies addressing one or some of two SNPs in MYOC promoter (rs2075648 and rs12035719) and three SNPs in APOE promoter (rs405509, rs769446 and rs449647) were included. While significant association was detected for rs2075648 in the Caucasian population, the results showed no evidence for other SNPs or in other populations.

MYOC is the first identified POAG gene and it encodes a functional unclear protein, myocilin, which is expressed in most tissues of the body and in almost every ocular tissue [31,32]. The mutations in MYOC coding region have already been showed strong relationship with the development of both JOAG and COAG. In 2012, a meta-analysis evaluated the associations between POAG risk and myocilin polymorphisms, which conducted calculations on five mutations (R46X, R76K, Y347Y, T353I, and Q368X) and found two of them (Q368X and T353I) significantly associated with POAG [33]. The expression of myocilin in trabecular meshwork was valued because it might be the cause for elevation of IOP in POAG by resisting aqueous humor outflow [9,34]. Immunohistochemical study showed an increased expression of myocilin in the trabecular meshwork of patients with COAG [34]. However, the expression is not altered in the blood of POAG patients, suggesting that the specific altered expression might contribute to POAG pathogenesis in related tissues [35].

The importance of MYOC transcription regulation in POAG was revealed by some studies [10,36] which evaluated the role of rs12035719 (MYOC.mt1) in MYOC promoter and obtained conflicting results. The minor allele (C) is more common in Asian than in European (20% vs. 10%, 1000 genomes project). In the former meta-analysis for rs12035719 including researches up to 2005 [11], the results suggested no significant association and the publication bias was found using Egger’s test. In this updated meta-analysis for rs12035719, the result were still not significant but no publication bias detected (C vs. G, in Egger’s test, intercept=1.615, 95%=-3.74-4.43, P=0.777; the results in other genetic models were similar).

Another MYOC promoter SNP rs2075648 (-83G>A) is located nearer to the transcription start site and involved in MYOC basal transcription. According to 1000 genomes project, the minor allele (A) frequency is 5% in Asian and 15% in European. Promoter study has proved that the SNP is within a canonical E-box sequence, which is responsible for the binding of transcription factors [37]. The rs2075648 is also in a linkage disequilibrium (LD) block and showed strong LD with an exon polymorphism c.227G>A [38]. Our meta-analysis showed significant heterogeneity within the effect size of overall 17 studies on this locus, but the heterogeneity was eliminated in Caucasian subgroup, suggesting the genetic heterogeneity in POAG between different ethnic populations. The M-H pooled OR in Caucasian is 0.587, which statistically suggests that rs2075648 mutation carriers Caucasian (GA or AA) have lower chance to develop POAG. Nevertheless, the heterogeneity remains significant and the pooled OR is not significantly different from 1 in Asian subgroup, which might be contributed to the complicated genetic backgrounds within the Asian cohorts. The heterogeneous genetic basis of POAG among geographic or ethnic populations has been suggested in many studies, for examples, many MYOC mutations were only observed in specific populations and founder effects might contribute to the heterogeneity [39].

The APOE encoding protein, apolipoprotein E, is mainly expressed in central nervous system and plays a role in the development of neurodegenerative diseases, such as AD. In fact, accumulating evidence suggested a profound relationship between AD and glaucoma [14,40,41]. Animal experiments have shown apolipoprotein E could be synthesized by Muller cells, the important glial cell of retina [42,43]. The three polymorphisms in APOE promoter region, rs405509 (-219G/T), rs769446 (-427T/C) and rs449647 (-491A/T), were first analyzed in 1998 by Artiga, M. J. et al. [44] and rs405509 and rs449647 were suggested influence on transcriptional activity. Rs405509 and rs449647 also showed significant associations with the development of AD [45]. In 2002, Copin, B. et al. reported that rs405509 and rs449647 had influence on the POAG clinical features and rs449647 might have an interaction with rs12035719 (MYOC.mt1) [18]. This report revealed the potential role of APOE and MYOC promoter SNP in POAG and firstly linked them together. In the 2005 study by Fan, B. J. et al., another two polymorphism interaction between APOE and MYOC (rs2075648 and APOE ε2/ε3/ε4; MYOC IVS2+35A>G and rs405509) was identified significant in high tension glaucoma patients [28]. This meta-analysis failed to provide any evidence for the association between APOE promoter SNPs and the risk of POAG. The relatively small numbers of the included studies for the three SNPs might influence the evaluation.

Until now, eleven GWAS has been conducted to explore the genetic basis of POAG [46-56]. To our knowledge, the five SNPs discussed here were not reported in any of them except rs2075648. The study by Wael Osman et al. performed in 2012 revealed that rs2075648 was not associated with POAG in a Japanese cohort (OR=0.99, 95% CI=0.79-1.24). Due to the lack of information relating to the detailed genotypes, we cannot include this study in this meta-analysis. However, the conclusion of this study was consistent with the Japanese study. Although significant association was detected between POAG risk and rs2075648 in Caucasian, we didn’t detect association between POAG and Asian populations.

Our study contains some limitations. In promoter region of MYOC, we only selected two SNPs in consideration of sufficient studies included. The ethnicity stratification is also limited because it ignores the different genetic backgrounds within the Caucasian and Asian cohorts. We could not further examine the associations between these SNPs and POAG clinical features, which are very meaningful, because of limited data to extract.

Conclusion

This meta-analysis assessed the association between five SNPs in MYOC or APOE promoter regions and the risk of POAG and firstly provide the statistic evidence for the association between rs2075648 (GA+AA vs. GG) and POAG in Caucasian population, which might imply mechanisms in POAG pathogenesis and molecular diagnosis values. This result should be interpreted with caution because of some limitations. Well-designed studies with more ethnic groups are required to further validate the results.

Disclosure of conflict of interest

None.