To perform a quantitative trait locus (QTL) analysis and evaluate whether a locus between SIX1 and SIX6 is associated with retinal nerve fiber layer (RNFL) thickness in individuals of European descent.
Observational, multicenter, cross-sectional study.
A total of 231 participants were recruited from the Diagnostic Innovations in Glaucoma Study and the African Descent and Glaucoma Evaluation Study. Association of rs10483727 in SIX1-SIX6 with global and sectoral RNFL thickness was performed. Quantitative trait analysis with the additive model of inheritance was analyzed using linear regression. Trend analysis was performed to evaluate the mean global and sectoral RNFL thickness with 3 genotypes of interest (T/T, C/T, C/C). All models were adjusted for age and sex.
Direction of association between T allele and RNFL thickness was consistent in the global and different sectoral RNFL regions. Each copy of the T risk allele in rs10483727 was associated with −0.16 μm thinner global RNFL thickness ( β = −0.16, 95% confidence interval: −0.28 to −0.03; P = .01). Similar patterns were found for the sectoral regions, including inferior ( P = .03), inferior-nasal ( P = .017), superior-nasal ( P = .0025), superior ( P = .002) and superior-temporal ( P = .008). The greatest differences were observed in the superior and inferior quadrants, supporting clinical observations for RNFL thinning in glaucoma. Thinner global RNFL was found in subjects with T/T genotypes compared to subjects with C/T and C/C genotypes ( P = .044).
Each copy of the T risk allele has an additive effect and was associated with thinner global and sectoral RNFL. Findings from this QTL analysis further support a genetic contribution to glaucoma pathophysiology.
Primary open-angle glaucoma (POAG), characterized by progressive neuronal loss and retinal nerve fiber layer (RNFL) thinning, is a common complex disorder for which a number of susceptible loci have been identified. Evidence of a genetic contribution to the pathogenesis of POAG has been well established from twin and family studies and, more recently, by the findings in large genome-wide association studies (GWAS). These studies have identified a number of susceptible loci with POAG, of which the association of a locus in the SIX1 – SIX6 region, 2 homeobox genes on chromosome 14 shown to regulate eye development, was found to have an increased risk for POAG in 2 large GWAS studies (top locus: rs10483727, T allele, odds ratio [OR] = 1.27 and 1.32, respectively). Furthermore, the association of rs10483727 in SIX1 – SIX6 was also corroborated in glaucoma quantitative trait locus (QTL) studies, including vertical cup-to-disc ratio and optic disc size.
Quantitative traits, or “endophenotypes,” are intermediate phenotypes that may affect genetic susceptibility. Approaches through the investigation of intermediate phenotypes have been successful, analogous to studies examining glucose-related traits with type 2 diabetes and lipids with myocardial infarction. QTL analysis offers a powerful tool to detect genetic contribution and has several advantages over case-control studies. These include an objective classification of phenotypes, a reduction in genetic heterogeneity owing to no misclassification of “normal” controls, and a more sensitive approach. This method has been successfully examined in other glaucoma traits, such as intraocular pressure (IOP), central corneal thickness, vertical cup-to-disc ratio, and disc area. However, QTL analysis has not been evaluated with RNFL in a large group of individuals of European descent with well-characterized ocular phenotypes.
The retinal nerve fiber layer is a relatively heritable glaucoma trait, as has been demonstrated by twin and family studies, and thus is an ideal candidate to perform a QTL study. After adjusting for age, Hougaard and associates estimated the heritability score to be as high as 0.82 for RNFL thickness in a twin study. Van Koolwijk and associates found the heritability estimate to be closer to 0.48 in a family study of 2620 participants. The latter study found RNFL thickness to be the most heritable trait, compared to IOP (heritability = 0.35) and neuroretinal rim (heritability = 0.39). Despite the variability between these 2 studies, likely owing to different study designs and different instruments used in RNFL measurement, findings from both studies demonstrate that RNFL is a heritable glaucoma trait.
Thus, the purpose of this study is to perform a QTL analysis and to evaluate the relationship between rs10483727 in SIX1-SIX6 with RNFL thickness in a cohort of individuals of European descent.
This was an observational, multicenter, cross-sectional study of individuals of European descent from the Diagnostic Innovations in Glaucoma Study (DIGS) and the African Descent and Glaucoma Evaluation Study (ADAGES). The ADAGES cohort has individuals of both European- and African-descent populations, as previously described.
Only individuals of European descent were studied in order to minimize genetic heterogeneity. Furthermore, prior studies demonstrating the association of SIX1-SIX6 to POAG were conducted only in Europeans; thus it is not known if this association is trans-ethnically shared in populations of African descent. Thus, only individuals of European descent were included in this study.
Participants were recruited from 3 participating sites: the Hamilton Glaucoma Center at the Department of Ophthalmology, University of California, San Diego (UCSD); the New York Eye and Ear Infirmary; and the Department of Ophthalmology, University of Alabama, Birmingham. The study protocols for both studies are identical and details of the methodological approach were described elsewhere. All participants gave written informed consent to participate in this research study. This study was approved by the institutional review boards of all 3 participating sites and was performed in accordance with the tenets of the Declaration of Helsinki. DIGS and ADAGES were registered at www.clinicaltrial.gov ( NCT00221897 and NCT00221923 , respectively).
Subjects and Ocular Examination
The participants’ demographics, medical history, family history, and current medication were collected. At each visit, participants underwent a comprehensive ophthalmologic examination, including best-corrected visual acuity, slit-lamp biomicroscopy, IOP measurement by Goldmann applanation tonometry (GAT), gonioscopy, dilated fundus examination, standard automated perimetry, and stereoscopic optic disc photography. Central corneal thickness (CCT) was measured by ultrasound pachymetry (Pachette GDH 500; DGH Technology, Inc, Philadelphia, Pennsylvania, USA) and calculated as an average from 3 readings. Rim area and disc area measurements were obtained using the Heidelberg Retina Tomograph (HRT II) and analyzed with software version 3.1 (Heidelberg Engineering, Inc, Heidelberg, Germany). Stereoscopic optic disc photography (Nidek Stereo Camera Model 3-DX; Nidek Inc, Palo Alto, California, USA) and standard automated perimetry with the 24-2 Swedish Interactive Threshold Algorithm (SAP-SITA, Humphrey Field Analyzer; Carl Zeiss Meditec, Dublin, California, USA) were obtained.
Only subjects with open angles as presented on gonioscopy examination were included in this study. At study entry, subjects were excluded if they had a best-corrected visual acuity <20/40, spherical refraction outside ±5.0 diopters (D), cylinder correction outside 3.0 D, or any other ocular or systemic disease that could affect the optic nerve or the visual field.
Participants were classified into 3 diagnostic groups based on the following criteria. Healthy eyes were defined as having a normal appearance on stereoscopic optic disc photographs, IOP of less than 22 mm Hg, no history of elevated IOP, and at least 2 reliable normal visual fields, defined as a pattern standard deviation (PSD) within 95% confidence limits and a Glaucoma Hemifield Test (GHT) result within normal limits. Glaucoma suspect eyes were defined as eyes having glaucomatous or suspicious-appearing optic discs based on stereoscopic photograph review by 2 experienced graders or ocular hypertension (OHT, IOP >22 mm Hg), without evidence of repeatable glaucomatous visual field defects (VFD) at baseline. Glaucoma eyes were defined as eyes that have glaucomatous-appearing optic discs (neuroretinal rim thinning, excavation, or RNFL defect) and repeatable visual field damage (PSD outside 95% confidence limits or GHT outside normal limits). This classification was only used to calculate the allele frequency between the cases (suspects and glaucoma subjects) vs controls (healthy subjects). Quantitative trait analysis using RNFL as a continuous trait does not require any diagnostic classification, minimizing misclassification in this situation of potential phenotypic overlap among the 3 categories.
Measurement of Retinal Nerve Fiber Layer
Spectralis spectral-domain optical coherence tomography (SD OCT) (software version 184.108.40.206; Heidelberg Engineering, Inc, Heidelberg, Germany) was used to image and measure RNFL thickness using previously described protocols. The RNFL circle scan consists of 1536 A-scan points projected onto a 12-degree circle centered on the optic disc corresponding to a 3.5-mm-diameter circle. A real-time eye-tracking system was used to compensate for any eye movements.
All images underwent quality control review, according to a standard protocol developed by the Imaging Data Evaluation and Assessment (IDEA) Reading Center at UCSD. Images were reviewed and RNFL segmentation manually corrected if needed. Images were excluded if the scan was not centered, the signal quality was less than 15 dB, or image artifacts affected RNFL measurements (eg, floater located on the RNFL measurement circle). A total of 6 RNFL sectors (temporal, nasal, superior-temporal, superior-nasal, inferior-temporal, and inferior-nasal) were automatically calculated and included in the analysis. The 2 superior (superior-temporal and superior-nasal) and 2 inferior (inferior-temporal and inferior-nasal) sectors were averaged to obtain the superior and inferior sectors, respectively. A total of 8 sectoral RNFL and global RNFL were used in this analysis. RNFL thickness were measured and calculated as an average between both eyes. The average from both eyes was used in the analysis. This approach has been shown to have higher homogeneity estimates and results in more power than obtaining data from only 1 eye for genetic studies.
DNA Extraction and Genotyping
Blood samples were obtained from participants and DNA samples were extracted from Epstein-Barr virus (EBV)-transformed lymphoblastoid cell lines using Gentra Puregene Blood Kit (Qiagen, Valencia, California, USA). Only DNA quality scores, as measured by absorbance (A 260/280 ) > 1.7, were genotyped.
Single nucleotide polymorphism (SNP) genotyping was performed at Washington University using the Sequenom MassARRAY iPLEX following the manufacturer’s protocol (Sequenom, San Diego, CA). A total of 53 SNPs, selected from lead SNPs in prior glaucoma studies (n = 19), ancestry informative markers (n = 31), and sex-specific markers (n = 3), were genotyped. The average genotyping call rate was 85.3%–87.9%. This is a custom-designed genotyping array; therefore, it is less robust and may have a lower call rate compared to a predesigned array that has gone through extensive quality control measures.
Subjects were excluded for the following reasons: sample genotyping call rate <70%, demonstrated evidence of sex discrepancy, unintended duplicates, inconsistency with self-reported ancestry through BioGeographical Ancestry (BGA) study, missing calls for rs10483727, or individuals were of non-European descent through BGA study. A total of 231 subjects passed quality control measures and are used in this study. In this study, only the genotype at the SIX1-SIX6 locus [rs10483727; Chr: 14; Position (build 37): 61072875] was analyzed together with RNFL thickness.
Linear regression with an additive genetic model was performed for global and sectoral RNFL thickness as a quantitative trait using PLINK. Briefly, this model assumes an additive effect of allele dosage, where the SNP genotypes were coded according to the number of risk alleles ranging from 0 to 2. Associations between genotypes and RNFL thickness were performed with linear regressions, adjusting for age and sex. Furthermore, the mean levels of global and sectoral RNFL thickness were examined with 3 genotype groups (homozygous T/T, heterozygous C/T, and homozygous C/C) using a trend test.
All statistical analyses were performed with PLINK and JMP Pro 11.0 (SAS Institute Inc, Cary, North Carolina, USA). A P value <.05 was considered statistically significant.
A total of 231 subjects passing genotyping quality control measures were included in this study. All participants are individuals of European descent, confirmed by both self-reported ancestry and genotyped ancestry through the BGA study. In the BGA study, ancestry informative markers were used to estimate individual admixture proportions. Participants that were included in this study have over 90% European origin (mean 94.2%; 95% confidence interval [CI]: 92.7%–95.9%). The mean age was 65.1 years (95% CI: 63.5–66.7 years). Approximately 80% of participants are either glaucoma suspects or confirmed glaucoma patients, defined as cases in this study. The demographics and ocular characteristics are presented in Table 1 .
|Mean or Percentage||95% CI|
|Subjects (n = 231)|
|Sex, % (female)||56.2%||–|
|Ethnicity (% European origin) b||94.3%||92.7–95.9|
|Diagnostic category c|
|Normal (n = 46)||19.9%||–|
|Suspects (n = 101)||43.7%||–|
|Glaucoma (n = 84)||36.4%||–|
|IOP (mm Hg) d||16.3||15.9–16.8|
|Disc area (mm 2 )||1.99||1.93–2.04|
|Rim area (mm 2 )||1.24||1.20–1.27|
We first examined the frequency of the T risk allele of rs10483727 between the controls (healthy subjects) and cases (suspects and glaucoma subjects). The cases have a higher proportion of the T risk allele compared to the controls (45.1% vs 39.1%). As expected, the frequency of the T allele of rs10483727 in the controls was fairly similar to the CEU (European) populations in the HapMap Phase 3 project (37.9%), but the cases exhibited higher frequency of the risk allele.
QTL analysis was performed for rs10483727 and RNFL thickness, in both the global and sectoral regions on all participants. We found that the direction of effect was consistent in all regions; the T risk allele corresponded to a thinner RNFL ( Table 2 ). For every copy of the T risk allele, the global RNFL is thinner by −0.16 ± 0.06 μm after adjusting for age and sex (95% CI: −0.28 to −0.03 μm; P = .01). A similar pattern was observed for the sectoral analysis. For every copy of the T risk allele, we found −0.14 ± 0.06 μm thinner inferior RNFL ( P = .03), a −0.16 ± 0.07 μm thinner inferior-nasal RNFL ( P = .017), a −0.20 ± 0.06 μm thinner superior-nasal RNFL ( P = .0025), a −0.20 ± 0.06 μm thinner superior RNFL ( P = .002), and a −0.17 ± 0.06 μm thinner superior-temporal RNFL ( P = .008), respectively, after adjusting for age and sex ( Table 2 ). The greatest difference was seen in the superior quadrant. Though the direction was consistent, the difference in the nasal, temporal, and inferior-temporal RNFL quadrants did not independently reach statistical significance. If a strict Bonferroni approach were taken ( P = .0055 attributable to 9 outcomes), only the superior and superior-nasal quadrant would be statistically significant ( Table 2 ). With additional analysis, there were similar findings adjusting for age, sex, IOP, and rs4977756 ( CDKN2B-AS1 ) ( Table 3 ).
|RNFL||Beta||SE||L95||U95||P Value a|
a Analyzed with linear regression with the additive genetic model, adjusted for age and sex. If a strict Bonferroni approach were taken ( P = .0055 attributable to 9 outcomes), only the superior and superior-nasal quadrant would be statistically significant. Results are presented as the effect per minor allele of rs10483727 [minor allele, T; minor allele frequency, 0.4; Chr 14; Position (build 37): 61072875].
|RNFL||Beta||SE||L95||U95||P Value a|