Nine Loci for Ocular Axial Length Identified through Genome-wide Association Studies, Including Shared Loci with Refractive Error

Abstract

Refractive errors are common eye disorders of public health importance worldwide. Ocular axial length (AL) is the major determinant of refraction and thus of myopia and hyperopia. We conducted a meta-analysis of genome-wide association studies for AL, combining 12,531 Europeans and 8,216 Asians. We identified eight genome-wide significant loci for AL (RSPO1, C3orf26, LAMA2, GJD2, ZNRF3, CD55, MIP, and ALPPL2) and confirmed one previously reported AL locus (ZC3H11B). Of the nine loci, five (LAMA2, GJD2, CD55, ALPPL2, and ZC3H11B) were associated with refraction in 18 independent cohorts (n = 23,591). Differential gene expression was observed for these loci in minus-lens-induced myopia mouse experiments and human ocular tissues. Two of the AL genes, RSPO1 and ZNRF3, are involved in Wnt signaling, a pathway playing a major role in the regulation of eyeball size. This study provides evidence of shared genes between AL and refraction, but importantly also suggests that these traits may have unique pathways.

Introduction

Myopia (nearsightedness), the most common form of refractive errors, is an ocular disorder of major public health importance worldwide, particularly in Asia. About 40% of adults and 80%–90% of children completing high school are myopic in urban areas in East Asian countries, and 10%–20% of them have high myopia.^1,2 Uncorrected myopia and refractive errors are leading causes of visual impairment.^3–6 Furthermore, adults with high myopia are at a substantially higher risk of potentially blinding pathologies, including glaucoma, retinal detachment, and myopic maculopathy.⁷ The correction of myopia and refractive errors in general by spectacles, contact lenses, or refractive surgery can entail substantial socioeconomic costs^8,9 and does not treat the underlying mechanism of disease.

Myopia develops primarily from an eye that is excessively elongated axially and thus ocular axial length (AL) is an attractive endophenotype to investigate for several reasons. First, AL alone accounts for more than 40% of variation in refractive errors.^10–12 MRI studies of the orbit have also demonstrated that extremely highly myopic eyes are generally prolate in shape with unusually long ALs, leading to associated visually disabling complications such as posterior staphylomas.^13,14 Second, the heritability of AL (67% to 94%) is consistently higher than that for refraction.^15–18 Furthermore, the measurement of AL (in mm) is more objective, precise, and reproducible compared to assessments of refractive status.

Although more than 30 myopia loci have been implicated in previous linkage and genome-wide association studies (GWASs), there have been few reports of AL-specific loci. A recent GWAS identified an association at ZC3H11B for both AL and high myopia in Asians.¹⁹ To identify additional genetic variants that modulate AL, we conducted the largest international GWAS meta-analysis of AL to date in cohorts participating in the Consortium for Refractive Error and Myopia (CREAM).^20,21

Subjects and Methods

We used a three-stage approach.²⁰ First, we performed a GWAS meta-analysis in 12,531 European ancestry individuals (stage 1). Second, we tested the cross-ethnic transferability of the associations from this first stage in 8,216 Asian ancestry individuals (stage 2). Lastly, we conducted a meta-analysis combining individuals of European and Asian ancestry, totaling 20,747 individuals (stage 3). We subsequently examined the effect of the associated AL loci on spherical equivalent (SE) in 23,591 individuals from 18 other independent cohorts.

Study Populations in CREAM

All studies participating in this meta-analysis are part of the CREAM.^20,21 The discovery cohorts included 12,531 European ancestry individuals from 18 studies (Table 1), including ALSPAC Children,²² BATS/TEST,²³ BMES,^24,25 Croatia-Korcula, Croatia-Split, Croatia-Vis,²⁶ ERF,^27,28 RS1, RS2, RS3,²⁹ ORCADES,³⁰ and RAINE.^31–33 In addition, 8,216 Asian ancestry individuals from six cohorts (Table 1) (BES,³⁴ SCES,³⁵ SCORM,³⁶ SiMES,³⁷ SINDI,³⁵ and STARS Parents³⁸) were included in the replication stage. General methods, demographics, and phenotyping of the study cohorts have previously been described extensively and are provided in brief in Table 1. All studies were performed with the approval of their local Medical Ethics Committee, and written informed consent was obtained from all participants in accordance with the Declaration of Helsinki.

Table 1.

Study Cohorts and Summary of Axial Length Measures

Ethnicity	n	Study	Mean Age (SD), Years	Men, %	Axial Length
					Mean (SD), mm	Range, mm	Methods of Measurement
European	2,069	ALSPAC Children	15.5 (0.3)	46.5	23.41 (0.87)	20.49–26.57	IOLmaster
	1,316	BATS/TEST	24.6 (11.9)	43.2	23.25 (0.87)	20.03–28.25	IOLmaster
	1,030	BMES	73.8 (7.8)	59.5	23.45 (1.04)	19.94–29.86	IOLmaster
	826	Croatia-Korcula	55.8 (13.4)	35.1	23.19 (1.06)	18.55–28.24	Echoscan US-1800
	352	Croatia-Split	50.0 (14.2)	44.3	23.39 (0.90)	20.98–27.3	Echoscan US-1800
	552	Croatia-Vis	56.0 (14.0)	39.7	23.08 (0.90)	20.09–26.48	Echoscan US-1800
	2,397	ERF4	48.7 (14.2)	55.5	23.22 (1.04)	19.79–27.30	A scan
	503	ORCADES	57.6 (13.7)	43.3	23.70 (1.08)	20.69–28.00	IOLmaster
	1,011	Raine	20.1 (0.4)	51.6	23.56 (0.89)	20.36–27.94	IOLmaster
	676	RS1	78.4 (4.4)	49.0	23.52 (1.06)	20.44–27.72	Lenstar LS900
	1,085	RS2	72.0 (4.7)	47.2	23.50 (1.14)	19.87–28.00	Lenstar LS900
	714	RS3	59.3 (5.8)	42.6	23.56 (1.27)	19.79–28.45	Lenstar LS900 and A scan
Asian	564	BES	62.05 (8.4)	35.5	23.07 (1.15)	19.90–30.36	Lenstar LS900
	1,720	SCES	57.6 (9.0)	51.7	23.95 (1.31)	20.87–32.66	IOLmaster
	926	SCORM	10.8 (0.8)	51.7	24.13 (1.12)	21.05–28.20	Echoscan US-800
	2,141	SiMES	57.6 (10.7)	49.3	23.57 (1.04)	20.48–31.11	IOLmaster
	2,120	SINDI	55.9 (8.8)	51.4	23.41 (1.08)	19.07–31.59	IOLmaster
	745	STARS Parents	38.8 (5.3)	51.0	24.64 (1.51)	21.66–31.57	IOLmaster

Abbreviations are as follows: ALSPAC, Avon Longitudinal Study of Parents and Children; BATS, Brisbane Adolescent Twins Study; TEST, Twins Eye Study in Tasmania; BMES, Blue Mountains Eye Study; ERF, Erasmus Rucphen Family Study; ORCADES, Orkney Complex Disease Study; RS, Rotterdam Study; BES, Beijing Eye Study; SCES, Singapore Chinese Eye Study Singapore; SCORM, Singapore Cohort Study of the Risk Factors for Myopia; SiMES, Singapore Malay Eye Study; SINDI, Singapore Indian Eye Study; STARS, Strabismus, Amblyopia, and Refractive Error Study of Preschool Children; SD, standard deviation.

Independent Populations in CREAM

To examine whether the loci affecting AL contributed to SE, we studied associations with SE in an additional 18 studies (Table S1 available online): 1958 British Birth Cohort,³⁹ ALSPAC Mothers,⁴⁰ ANZRAG,⁴¹ AREDS 1a1b, AREDS 1c,^15,16 DCCT,⁴² EGCUT,⁴³ FECD,⁴⁴ FES,⁴⁵ FITSA,⁴⁶ GHS 1, GHS 2, KORA,^47–50 OGP Talana,⁵¹ SP2,⁵² TwinsUK,⁵³ WESDR,⁵⁴ and Young Finns Study.⁵⁵ Only SE (not AL) measures were available in these additional 18 CREAM studies. Detailed study design and methodology of these studies have been published elsewhere. Descriptive data on demographics and phenotypes of these cohorts are shown in brief in Table S1.

Phenotype Measurements

All studies used a similar protocol for ocular phenotype measurements. Eligible participants underwent an ophthalmologic examination including measurements of AL and refraction of both eyes. AL was measured with either optical laser interferometry or A-scan ultrasound biometry (Table 1). Refraction was measured by autorefractor and/or subjective refraction (Table S1). SE was calculated according to the standard formula (SE = sphere+1/2 cylinder).

Genotyping and Imputation

The study samples were genotyped on either the Illumina or Affymetrix platforms. Each study performed SNP imputation with the genotype data, together with the HapMap Phase II ethnically matched reference panels (CEU, JPT+CHB, or the four HapMap populations) on the basis of HapMap build 36 databases (release 22 or 24). The Markov Chain Haplotyping software, IMPUTE^56,57 or MACH,⁵⁸ were adopted for imputation. A detailed description regarding genotyping platforms and imputation procedures for each study is provided in Tables S2 and S3.

Stringent quality control of genotype data was applied in each cohort. Samples with low call rates (<95%) or with gender discrepancies were excluded. Cryptically related samples and outliers in population structure from principal component analyses were also excluded. SNPs flagged with missingness >5%, gross departure from Hardy-Weinberg equilibrium (p value < 10⁻⁶, except in the ALSPAC study where a threshold of < 10⁻⁷ was used), and minor allele frequency (MAF) <1% were removed from further analyses.

Statistical Analysis

For each study, an allele-dosage regression model at each genotyped or imputed SNP was conducted to determine its association with AL as a quantitative trait as well as its association with SE. Individuals with prior refractive or cataract surgery or other intraocular procedures that could alter refraction were excluded. The mean of the right and left eyes was taken. When data from only one eye were available, the AL or SE of this eye was used. Sample outliers with AL value exceeding four standard deviations from the mean were excluded at the study level. We assumed an additive genetic model where the dosage of each SNP is a continuous variable ranging from 0 to 2 for minor alleles carried. Primary analysis for AL was adjusted for age, sex, and height (because height was consistently correlated with AL^59,60) and in the case of SE for age and sex. Additional adjustment for principal components was carried out according to the population substructure in each individual study.

The per-SNP meta-analyses were performed by METAL software with weighted inverse-variance approach, assuming fixed effects, because for initial discovery purposes, the fixed-effects model is preferred for increased statistical power.⁶¹ A Cochran’s Q test was used to assess heterogeneity across studies.⁶² Imputation quality scores were reviewed for the top SNPs reported to ensure good imputation quality (proper-info of IMPUTE or R² of MACH > 0.3).

Gene-based testing was conducted with VEGAS software⁶³ on the European ancestry and Asian ancestry meta-analysis results separately. VEGAS incorporates information from the full set of markers within a gene and thus can be more powerful than tests of individual SNPs if there are multiple risk variants within a gene. VEGAS corrects for LD and gene size by conducting simulations based on the LD structure in the population of interest (here, European or Asian ancestry). VEGAS was therefore run separately on all the European and Asian GWAS data, with results for each gene combined at the end by meta-analysis on the two sets of gene-based p values by Fisher’s methods. For samples of European descent, we used the HapMap 2 CEU population as the reference to estimate patterns of LD. For Asian ancestry groups, we used the combined HapMap 2 JPT and CHB populations as the reference population to approximate linkage disequilibrium (LD) patterns. To include gene regulatory regions, SNPs were included if they fell within 50 kb of a gene.

VEGAS-Pathway analysis^63,64 was carried out with prespecified pathways from Gene Ontology. Pathways with 10 to 1,000 components were selected, yielding 4,628 pathways. Pathway analysis was based on combining gene-based test results from VEGAS. Pathway p values were computed by summing χ² test statistics derived from VEGAS p values. Empirical VEGAS-Pathway p values for each pathway were computed by comparing the summed χ² test statistics from real data with those generated in 500,000 simulations where the relevant number (according to the size of the pathway) of randomly drawn χ² test statistics was summed. To ensure that clusters of genes did not adversely affect results, within each pathway, gene sets were pruned such that each gene was >500 kb away from all other genes in the pathway. Where required, all but one of the clustered genes was dropped at random when genes were clustered. We performed meta-analysis on the two sets of pathway p values by Fisher’s method.

Differential Gene Expression in a Mouse Model of Myopia

Animal study approval was obtained from the SingHealth Institutional Animal Care and Use Committee (AAALAC accredited). All procedures performed in this study complied with the Association of Research in Vision and Ophthalmology Statement for the Use of Animals in Ophthalmology and Vision Research. Experimental myopia was induced in B6 wild-type (WT) mice (n = 36) by applying a −15.0 diopter spectacle lens on the right eye (experimental eye) for 6 weeks from postnatal day 10. The left uncovered eye served as the contra-lateral control eye. Age-matched naive mice eyes were also used as independent control eyes (n = 36).^65,66 Eye biometry, refraction, tissue collection, RNA extraction, real-time polymerase chain reaction (PCR) qRT-PCR methods, and analysis were followed as described previously.¹⁹ qRT-PCR primers (Table S4) were designed with ProbeFinder 2.45 (Roche Applied Science) and performed with a Lightcycler 480 Probe Master (Roche Applied Science). The experiments were repeated in triplicate. Gene expression of all identified genes in the control and experimental groups was quantified by the 2^−ΔΔCt method.⁶⁷ Student’s t test was performed to determine the significance of the relative fold difference of mRNA between the myopic eyes of the experimental mice and the age-matched controls.

Gene Expression in Human Tissues

Adult ocular samples were obtained from normal eyes of an 82-year-old female of European ancestry from the North Carolina Eye Bank (Winston-Salem, NC). All adult ocular samples were stored in QIAGEN’s RNAlater within 6.5 hr of collection and shipped on dry ice overnight to the lab. Isolated tissues were snap-frozen and stored at −280°C until RNA extraction. RNA was extracted from each tissue sample independently by the Ambion mirVana total RNA extraction kit. The tissue samples were homogenized in Ambion lysis buffer with an Omni Bead Ruptor Tissue Homogenizer per protocol. Reverse transcription reactions were performed with Invitrogen SuperScript III First-Strand Synthesis kit. The expression of the identified genes was assessed by running 10 μl reactions with QIAGEN’s PCR products consisting of 1.26 μl H₂O, 1.0 μl 10× buffer, 1.0 μl dNTPs, 0.3 μl MgCl, 2.0 μl Q-Solution, 0.06 μl taq polymerase, 1.0 μl forward primer, 1.0 μl reverse primer, and 1.5.0 μl cDNA. The reactions were run on a Eppendorf Mastercycler Pro S thermocycler with touchdown PCR ramping down 1°C per cycle from 72°C to 55°C followed by 50 cycles of 94°C for 30 s, 55°C for 30 s, and 72°C for 30 s with a final elongation of 7 min at 72°C. All primer sets were designed by Primer3.⁶⁸ Products were run on a 2% agarose gel at 70 V for 35 min. Primer sets were run on a custom tissue panel including Clontech’s Human MTC Panel I, Fetal MTC Panel I, and an ocular tissue panel.

Results

We analyzed 2.5 million genotyped and imputed SNPs (Table S2). The genomic control inflation factor (λ) for individual studies (Table S2) as well as for the meta-analysis (λ_GC = 1.06) and quantile-quantile plots (Figure S1) showed little evidence for inflation.

Per-SNP Meta-analysis

In the first stage, a total of 177 SNPs, representing 24 physically distinct loci, were associated with p < 1 × 10⁻⁵ in the European ancestry discovery cohort (Table S5). Of them, we identified one locus at chromosome 15q14 in the proximity of GJD2 (MIM 607058; rs11073058, p = 2.0 × 10⁻⁸) exceeding genome-wide significance level (p < 5 × 10⁻⁸; Table 2), which was previously reported to be associated with refractive errors.⁶⁹ We took the 177 SNPs forward for replication in the Asian cohorts (stage 2). Five regions showed significant evidence of replication (1.12 × 10⁻⁹ ≤ p ≤ 1.18 × 10⁻²; Table 2): RSPO1 (MIM 609595), C3orf26, LAMA2 (MIM 156225), and regions close to ZC3H11B and GJD2. In the combined meta-analysis of all 18 European and Asian cohorts (stage 3, n = 20,747), all five loci surpassed genome-wide significance level (3.97 × 10⁻¹³ ≤ p ≤ 1.24 × 10⁻⁸; Table 2 and Figure 1). Furthermore, in stage 3 we detected an additional genome-wide significant locus at ZNRF3 (MIM 612062, p = 4.08 × 10⁻⁸; Table 2).

Table 2.

Associations with Ocular Axial Length in the European Ancestry Cohorts with Results in the Asian Cohorts and Combined Analysis

Lead SNPa	Chr	Positionb	EA	European Ancestry Cohorts (Stage 1, n = 12,531)				Asian Cohorts (Stage 2, n = 8,216)				Combined (Stage 3, n = 20,747)				Localization Relative to Protein-Coding Genesb
				EAF	Betac	SEM	p Value	EAF	Betac	SEM	p Value	EAF	Betac	SEM	p Value
rs4074961	1	37865310	T	0.43	0.06	0.01	6.6 × 10−6	0.45	0.10	0.02	1.1 × 10−9	0.44	0.07	0.01	4.0 × 10−13	intron 4 of RSPO1 (MIM 609595)
rs994767	1	217842055	A	0.45	−0.06	0.01	1.2 × 10−6	0.32	−0.10	0.02	4.4 × 10−7	0.41	−0.07	0.01	9.6 × 10−12	7 kb upstream of ZC3H11B
rs9811920	3	101326983	A	0.41	0.07	0.01	3.0 × 10−7	0.36	0.13	0.03	6.0 × 10−6	0.40	0.08	0.01	4.9 × 10−11	intron 1 of C3orf26
rs12193446	6	129861731	A	0.91	0.12	0.02	1.1 × 10−7	0.98	0.28	0.11	1.2 × 10−2	0.91	0.12	0.02	1.2 × 10−8	intron 58 of LAMA2 (MIM 156225)
rs11073058	15	32776918	T	0.43	0.07	0.01	2.0 × 10−8	0.50	0.06	0.02	4.7 × 10−4	0.45	0.07	0.01	4.3 × 10−11	57 kb upstream of GJD2 (MIM 607058)
Additional locus identified through the combined analysis of European and Asian cohorts
rs12321	22	27783183	C	0.44	−0.05	0.01	1.1 × 10−5	0.49	−0.06	0.02	9.9 × 10−4	0.46	−0.05	0.01	4.1 × 10−8	3′ UTR of ZNRF3 (MIM 612062)

SNPs with p < 1 × 10⁻⁵ in European ancestry cohorts were brought for replication in Asians. Genome-wide significance is defined as p < 5.0 × 10⁻⁸. The following abbreviations are used: SNP, single-nucleotide polymorphism; Chr, chromosome; EA, effect allele; EAF, effect allele frequency; SEM, standard error of the mean.

^aLead SNPs of each locus identified in the combined meta-analysis (stage 3) are presented. The lead SNPs in the combined meta-analysis are the same as the lead SNPs in the European-only analysis (stage 1) for all loci, except for the 1q41 locus near ZC3H11B, where the lead SNP in European-only analysis is rs10863469 (position, 217844091; frequency of effect allele T = 0.47; Beta = 0.47, p = 1.2 × 10⁻⁶), being in high LD (r² = 0.84) with rs994767.

^bPosition is based on NCBI human genome build 36.

^cEffect sizes on axial length are in mm.

Figure 1.

Summary of Meta-analysis Results for Genome-wide Association to Ocular Axial Length

Data of both directly genotyped and imputed SNPs are presented in the Manhattan plot. The y axis represents –log₁₀ p values for association with axial length, and the x axis represents chromosomes and base-pair positions based on human genome build 36. The horizontal red line indicates the genome-wide significance level of p < 5.0 × 10⁻⁸. The horizontal blue line indicates the suggestive significance level of p < 1.0 × 10⁻⁵. The previously described locus for axial length is labeled in black. Other loci reaching genome-wide significance identified from the per-SNP meta-analysis are labeled in red. The genes identified in gene-based tests are labeled in blue.

Overall, the significant regions included six loci for AL: RSPO1, C3orf26, LAMA2, GJD2, ZNRF3, and one previously identified locus for AL at 1q41 close to ZC3H11B.¹⁹ A common SNP in RSPO1 displayed the strongest evidence for association (rs4074961, β = 0.07 mm per copy of risk allele, p = 3.97 × 10⁻¹³), with no evidence of heterogeneity (I² = 0%, p = 0.78) across the 18 AL cohorts (Table S6), although the strongest effect was observed for the rarer intronic variant in LAMA2 (rs12193446, β = 0.12 mm, p = 1.24 × 10⁻⁸). Figure 2 shows the regional association plots for the six loci significant in single SNP tests. Forest plots showing the effect sizes across cohorts are provided in Figure S2. We constructed a multilocus genetic risk score to evaluate the combined effects of the AL SNPs in the Blue Mountains Eye Study^24,25 and the Singapore Chinese Eye Study,³⁵ both of which were part of the 18 AL discovery cohorts. Figure S3 shows that the odd ratios for longer AL (Tertile 3 versus Tertile 1) were higher with increasing genetic risk scores.

Figure 2.

Regional Association Plots and Recombination Rates of the Loci Associated with Ocular Axial Length

Data are shown for association at chromosome (A) 1p34.3 (RSPO1), (B) 1q41 (ZC3H11B), (C) 3q12.1 (C3orf26), (D) 6q22.33 (LAMA2), (E) 15q14 (GJD2), and (F) 22q12.1 (ZNRF3) in the combined meta-analysis. Data of both directly genotyped and imputed SNPs are presented. In each panel, the genotyped SNP with the most significant association is denoted with a purple diamond. The color coding of all other SNPs indicates LD with the lead SNP, estimated by CEU r² from phase II HapMap: red, r² ≥ 0.8; yellow, 0.6 ≤ r² < 0.8; green, 0.4 ≤ r² < 0.6; cyan, 0.2 ≤ r² < 0.4; blue, r² < 0.2; and gray, r² unknown. The left y axis represents –log₁₀ p values for association with axial length, the right y axis represents the recombination rate, estimated from the International HapMap Project, and the x axis represents base-pair positions along the chromosome based on human genome build 36. Gene annotations are taken from the University of California Santa Cruz (UCSC) genome browser. The plots were created with LocusZoom.

Gene-Based Meta-analysis

In addition to per-SNP meta-analysis, we applied gene-based tests with VEGAS,²⁵ with genome-wide significance declared if p_gene-based < 0.05/17,872 = 2.8 × 10⁻⁶ (17,872 genes tested). Over and above the loci found in per-SNP tests, three additional genomic regions were genome-wide significantly associated with AL via gene-based tests (Table 3): CD55 (MIM 125240), ALPPL2 (MIM 171810), and TIMELESS/MIP/SPRYD4/GLS2 (MIM 603887 for TIMELESS). Figure S4 shows the regional association for the three loci significant in gene-based tests.

Table 3.

Loci Associated with Ocular Axial Length in Gene-Based Tests

Gene	MIM Number	Chr	Start Positiona	End Positiona	pgene-basedValue
					European Ancestry Cohorts	Asian Cohorts	Combinedb
CD55	125240	1	205561439	205600934	1.3 × 10−5	9.6 × 10−4	2.3 × 10−7
ALPPL2	171810	2	232979795	232983669	6.4 × 10−5	1.7 × 10−3	1.8 × 10−6
TIMELESS/MIP/SPRYD4/GLS2c	603887	12	55097173	55168448	2.0 × 10−7	7.3 × 10−2	2.8 × 10−7

The following abbreviation is used: Chr, chromosome.

^aPosition is based on NCBI human genome build 36. Note this is the start and stop position of the gene. For gene-based tests, 50 kb was added to either side to account for possible regulatory variants that fall outside the gene boundaries.

^bGene-based genome-wide significance was defined as p < 2.80 × 10⁻⁶. Only loci that were genome-wide significant in gene-based testing but not genome-wide significant in per-SNP testing are shown.

^cTIMELESS was the most significant gene in the region. Because of the ±50 kb added to the definition for each gene and the close proximity of the genes, MIP, SPRYD4, GLS2, and TIMELESS all had similar gene-based p values (ranged from 1.4 × 10⁻⁶ to 2.8 × 10⁻⁷ for the combined analysis), and thus p value and MIM number for only TIMELESS is presented.

Association with Refraction

We subsequently assessed the association of these AL SNPs and genes with SE in 23,591 individuals from 18 independent studies in CREAM that had SE but no AL measures (Tables S1 and S3). We found associations (p < 0.05) with SE for three of the six AL SNPs (Table 4 and Figure S5) (rs994767 [ZC3H11B, p = 0.013], rs11073058 [GJD2, p = 1.66 × 10⁻⁸], and rs12193446 [LAMA2, p = 3.58 × 10⁻¹⁰]), with directions of the SE association being consistent with AL. For example, the risk allele T of rs11073058 in GJD2 was associated with both longer AL and more myopia (more negative SE). In gene-based tests, only CD55 (p = 4.5 × 10⁻⁶) and ALPPL2 (p = 8.3 × 10⁻³) were associated with SE (Table 5).

Table 4.

Association with Spherical Equivalent of the SNPs Most Strongly Associated with Axial Length in Each Genomic Locus in Independent Cohorts

Lead SNP	Nearest Gene	Effect Allele	Betaa	SEM	p Value
rs4074961	RSPO1 (MIM 609595)	T	0.004	0.023	0.84
rs994767	ZC3H11B	A	0.054	0.022	1.3 × 10−2
rs9811920	C3orf26	A	−0.022	0.022	0.31
rs12193446	LAMA2 (MIM 156225)	A	−0.242	0.039	3.6 × 10−10
rs11073058	GJD2 (MIM 607058)	T	−0.121	0.022	1.7 × 10−8
rs12321	ZNRF3 (MIM 612062)	C	−0.004	0.021	0.86

Abbreviations are as follows: SNP, single-nucleotide polymorphism; SEM, standard error of the mean.

^aEffect sizes on spherical equivalent are in diopters.

Table 5.

Association of the Axial Length Genes Identified in Gene-Based Tests with Spherical Equivalent in Independent Cohorts

Gene	MIM Number	Chr	pgene-basedValuea
CD55	125240	1	4.5 × 10−6
ALPPL2	171810	2	8.3 × 10−3
TIMELESS/MIP/SPRYD4/GLS2b	603887	12	0.14

Abbreviation is as follows: Chr, chromosome.

^aThe association with spherical equivalent was assessed in 17 European ancestry cohorts of the 18 independent cohorts, with the HapMap 2 CEU population as the reference to estimate patterns of LD.

^bBecause of the ±50 kb added to the definition for each gene and the close proximity of the genes, MIP, SPRYD4, GLS2, and TIMELESS all had similar gene-based p values (ranged from 0.14 to 0.20 for the combined analysis), and thus p value and MIM number for TIMELESS only is presented.

SNPs close to CD55 had reached genome-wide significant association with SE in the meta-analysis of all CREAM cohorts (i.e., with and without AL measures).²⁰ There was an association with SE at CHRNG, along with a less significant independent hit near ALPPL2 (125 kb away).²⁰ Our AL gene-based results showed a genome-wide significant signal at ALPPL2 but not at CHRNG. There was also an association with SE at RDH5,²⁰ on the same chromosomal band as the AL signal at MIP (MIM 154050), but RDH5 and MIP are 727 kb apart without LD between them, suggesting that they are independent signals.

Pathway Analysis

We conducted pathway analysis with VEGAS-Pathway^63,64 by combining the gene-based p values for 4,628 prespecified pathways. The most significant pathway was the “Wnt receptor signaling” pathway (p = 2.9 × 10⁻⁵). The Bonferroni corrected p value was 0.13 (for the total number of 4,628 pathways tested). However, Bonferroni correction is an overcorrection, because many of the pathways have overlapping genes. The identification of the Wnt signaling pathway, even if only nominally associated, is of interest because the pathway involves two genes identified from the per-SNP tests. Also among the top ten pathways were “lens development in camera-type eye” (p = 2.4 × 10⁻⁴) and “collagen” (p = 5.1 × 10⁻⁴) pathways (Table S7). The collagen pathway was implicated in a recent meta-analysis of corneal thickness.⁶⁴

Gene Expression

Differential expression of the nearest genes in the six implicated loci from per-SNP meta-analysis (Table S4) was assessed by measuring mRNA levels in minus-lens-induced myopia mouse models.^65,66 The mRNA levels of all six genes had a 2-fold difference in the induced myopic eyes as compared to the control eyes in most of the tissues tested: sclera, retinal pigment epithelium (RPE), and neural retina (Figure S6).

In human ocular tissue, we have previously shown that ZC3H11B is expressed in neural retina, RPE, and sclera,¹⁹ LAMA2 is expressed in sclera and optic nerve, and CD55 is expressed in retina, choroid, and cornea, and GJD2 is less abundant in sclera and other ocular tissues.²⁰ In this study, we measured the mRNA expression levels of the other genes in adult ocular tissues via reverse-transcriptase PCR. We found that C3orf26, ZNRF3, and TIMELESS were expressed in most ocular tissues and the expression of RSPO1, ALPPL2, and MIP was less strong and/or more restricted (Table S8).

Discussion

We identified five AL loci (RSPO1, C3orf26, LAMA2, GJD2, and ZNRF3) and confirmed the previously described locus (ZC3H11B) via per-SNP tests. In addition, three loci (CD55, ALPPL2, and TIMELESS/MIP/SPRYD4/GLS2) were identified by gene-based tests. Therefore, a total of nine AL loci were identified in this meta-analysis. Seven of the nine AL loci are located within the genomic region of protein-coding genes (Tables 2 and 3). Of note, two of them (RSPO1 and ZNRF3) encode proteins that are directly involved in the Wnt signaling pathway. RSPO1 is a member of a family of secreted proteins that act as stem-cell growth factors by enhancing the Wnt signaling pathway.⁷⁰ On the other hand, ZNRF3 is a membrane-bound protein that acts as a negative regulator of the Wnt signaling pathway by mediating degradation of the Wnt receptor complex components Frizzled and LRP6.⁷¹ The two proteins have recently been shown to interact, RSPO1 enhancing Wnt signaling through inhibition of ZNRF3.⁷¹ The Wnt signaling was the most significant pathway in our analysis, further supporting its prominent role in vertebrate eye development.⁷² Indeed, overexpression of a dominant-negative variant of human ZNRF3 in zebrafish embryos induces small eye or loss of eyes.⁷¹

Remodeling of extracellular matrix in sclera plays an important role in changes of eye size during myopia development. LAMA2 encodes the alpha 2 chain of laminin, a major extracellular protein of the basement membrane. We used HaploReg⁷³ to search for evidence of a functional role for variants at the LAMA2 locus, because it has the largest per-allele effect on AL. The intronic lead SNP rs12193446 lies within the promoter and enhancer histone marks as well as DNase hypersensitive sites. Analysis with RegulomeDB2⁷⁴ suggested that rs12193446 occurs in a region that binds EP300, TCF4, STAT3, GATA2, and RFX4. Four of these interactions (EP300, TCF4, STAT3, and GATA2) were predicted by HaploReg⁷³ to be affected by the genotype at rs12193446. Mutations in the cognate gene for TCF4 cause Pitt-Hopkins syndrome (PTHS [MIM 610954]), the predominant ocular feature of which is high-grade myopia.⁷⁵ Interestingly, common genetic variants in TCF4 (MIM 602272) have also been associated with Fuchs corneal dystrophy, suggesting the pleiotropic effects of TCF4 on ocular diseases.⁷⁶

Gene-based testing implicated the TIMELESS/MIP/SPRYD4/GLS2 region, although determining which of these genes are functionally relevant is difficult because there are multiple association signals in the region. MIP is an interesting candidate gene because it is expressed in the ocular lens and is required for correct lens function.⁷⁷ CD55, implicated here in AL and previously in SE,²⁰ is known to elevate cytosolic calcium ion concentration.

For all six of the genes identified in our per-SNP meta-analysis, we found evidence for differential expression in a mouse model of myopia. Differential expression was observed in the mouse sclera and retina as well as RPE cells, suggesting a role for these genes in myopia. Further strengthening our results, the expression data showed that all but one of these genes expressed in the adult human eye. These data potentially provide insights into the complexity of AL elongation and myopia at the biological level. Some genes, namely ZC3H11A, GJD2, and LAMA2, showed changes in expression that are consistently in the same direction across the different eye sections analyzed, whereas others, namely RSPO1, C3orf26, and ZNRF3, showed variable directions of differential expression. These results, together with the pathway analysis results, suggest that the genetic mechanisms of myopia are complex, involving more than one eye component.

We have previously shown that up to 50% of the variation in SE is due to shared genetic factors with AL.⁷⁸ Thus, we undertook further analyses and found that five of the nine AL loci are also associated with SE. Furthermore, we looked up the association of AL with the SNPs discovered from the recent CREAM GWAS meta-analysis on SE in 32 cohorts²⁰ and observed that 23 of the 29 SNPs identified with SE have significant effects on AL (p < 0.05; Table S9). This has important implications. First, it confirms the previous findings in twins⁷⁸ that there are common genetic determinants of the two traits, such as variants in GJD2, LAMA2, CD55, and ALPPL2. Second, it indicates that some genetic variants for AL do not influence SE, suggesting that they regulate the coordinated scaling of eye size.⁷⁹ For example, the SNP in RSPO1 showed the strongest evidence of association with AL, yet it had no association with refractive error. In eyes without refractive error, AL and corneal curvature are carefully scaled relative to one another and have a high phenotypic correlation between them.⁸⁰ Therefore, genes like RSPO1 might mediate a compensatory mechanism through changes in corneal curvature or optical power, thereby balancing their effects on SE.

Shorter axial length is a major risk factor for angle closure glaucoma. A recent GWAS on primary angle closure glaucoma identified three genome-wide significant loci located at PLEKHA7 (MIM 612686), COL11A1 (MIM 120280), and PCMTD1-ST18.⁸¹ However, none of the common variants in the three loci were significantly associated with AL in our meta-analysis (Table S10). This suggests that susceptibility genes do not overlap between primary angle closure glaucoma and eyes with shorter axial length.

In summary, we identified nine loci associated with AL. They fall into two groups, one also influencing common refractive error variation, and the other, which includes two genes in the Wnt signaling pathway, uniquely determining eye size with little effect on natural refractive status. Further elucidation and characterization of the causal variants underlying the growth of ocular component dimensions and the development of myopia may enable new pathway and target identification, leading to potential prevention and treatment development.

Web Resources

The URLs for data presented herein are as follows:

• HaploReg, http://www.broadinstitute.org/mammals/haploreg/haploreg.php

• IMPUTE2, http://mathgen.stats.ox.ac.uk/impute/impute_v2.html

• International HapMap Project, http://hapmap.ncbi.nlm.nih.gov/

• LocusZoom, http://csg.sph.umich.edu/locuszoom/

• MACH, http://www.sph.umich.edu/csg/abecasis/MACH/

• METAL, http://www.sph.umich.edu/csg/abecasis/metal/

• Online Mendelian Inheritance in Man (OMIM), http://www.omim.org/

• RegulomeDB, http://RegulomeDB.org/

• UCSC Genome Browser, http://genome.ucsc.edu

• VEGAS, http://gump.qimr.edu.au/VEGAS/

Consortia

Consortium for Refractive Error And Myopia (CREAM): 1958 British Birth Cohort: Jugnoo S. Rahi, Pirro G. Hysi; Aichi cohort: Nagahisa Yoshimura, Kenji Yamashiro, Masahiro Miyake; ALIENOR: Cécile Delcourt, Cecilia Maubaret; ALSPAC: Cathy Williams, Jeremy A. Guggenheim, Kate Northstone, Susan M. Ring, George Davey-Smith; ANZRAG: Jamie E. Craig, Kathryn P. Burdon, Rhys D. Fogarty; AREDS1a: Sudha K. Iyengar, Robert P. Igo, Jr., Emily Chew, Sarayut Janmahasathian; AREDS1b: Sudha K. Iyengar, Robert P. Igo, Jr., Emily Chew, Sarayut Janmahasathian; AREDS1c: Dwight Stambolian, Joan E. Bailey Wilson; BATS: Stuart MacGregor, Yi Lu; Beijing Eye Study: Jost B. Jonas, Liang Xu, Seang-Mei Saw; BMES: Paul N. Baird, Elena Rochtchina, Paul Mitchell, Jie Jin Wang; CIEMS: Jost B. Jonas, Vinay Nangia; CROATIA-Korčula: Caroline Hayward, Alan F. Wright, Veronique Vitart; CROATIA-Split: Ozren Polasek, Harry Campbell, Veronique Vitart; CROATIA-Vis: Igor Rudan, Zoran Vatavuk, Veronique Vitart; DCCT: Andrew D. Paterson, S. Mohsen Hosseini; Duke FECD Fuchs Dystrophy GWAS: Sudha K. Iyengar, Robert P. Igo Jr, Jeremy R. Fondran; Duke Myopia Study: Terri L. Young, Sheng Feng; Erasmus Rucphen Family Study: Virginie J.M. Verhoeven, Caroline C. Klaver, Cornelia M. van Duijn; Estonian Genome Project / EGCUT: Andres Metspalu, Toomas Haller, Evelin Mihailov; FITSA: Olavi Pärssinen, Juho Wedenoja; Framingham Eye Study: Joan E. Bailey Wilson, Robert Wojciechowski; GEMT: Paul N. Baird, Maria Schache; Gutenberg Health Study: Norbert Pfeiffer, René Höhn; Hong Kong cohort study: Chi Pui Pang, Peng Chen; KORA: Thomas Meitinger, Konrad Oexle, Aharon Wegner; Kyoto high myopia: Nagahisa Yoshimura, Kenji Yamashiro, Masahiro Miyake; LIKI: Olavi Pärssinen; Myopia Genomics Study (Hong Kong HTI): Shea Ping Yip, Daniel W. H. Ho; Ogliastra Genetic Park Study: Mario Pirastu, Federico Murgia, Laura Portas, Genevra Biino; ORCADES: James F. Wilson, Brian Fleck, Veronique Vitart; Penn Family Studies: Dwight Stambolian, Joan E. Bailey Wilson; RAINE: Alex W. Hewitt, Wei Ang; Rotterdam Study: Virginie J.M. Verhoeven, Caroline C. Klaver, Cornelia M. van Duijn; SCES: Seang-Mei Saw, Tien-Yin Wong, Yik-Ying Teo, Qiao Fan, Ching-Yu Cheng, Xin Zhou, M. Kamran Ikram; SCORM: Seang-Mei Saw, Yik-Ying Teo, Qiao Fan, Ching-Yu Cheng, Xin Zhou, M. Kamran Ikram; SIMES: Seang-Mei Saw, Tien-Yin Wong, Yik-Ying Teo, Qiao Fan, Ching-Yu Cheng, Xin Zhou, M. Kamran Ikram; SINDI: Seang-Mei Saw, Tien-Yin Wong, Yik-Ying Teo, Qiao Fan, Ching-Yu Cheng, Xin Zhou, M. Kamran Ikram; SP2: Seang-Mei Saw, E-Shyong Tai, Yik-Ying Teo, Qiao Fan, Ching-Yu Cheng, Xin Zhou, M. Kamran Ikram; STARS: Seang-Mei Saw, Yik-Ying Teo, Qiao Fan, Ching-Yu Cheng, Xin Zhou, M. Kamran Ikram; TEST: David A. Mackey, Stuart MacGregor; TwinsUK: Christopher J. Hammond, Pirro G. Hysi; Utah Timorese: Margaret M. Deangelis, Margaux Morrison; Wenzhou: Xiangtian Zhou, Wei Chen; WESDR: Andrew D. Paterson, S. Mohsen Hosseini; Yokohama Study: Nobuhisa Mizuki, Akira Meguro; Young Finns Study: Terho Lehtimäki, Kari-Matti Mäkelä, Olli Raitakari, Mika Kähönen

The Fuchs’ Genetics Multi-Center Study Group: William Reinhart, Michael W. Belin, Robert L. Schultze, Todd Morason, Alan Sugar, Shahzad Mian, Hunson Kaz Soong, Kathryn Colby, Ula Jurkunas, Richard Yee, Mark Vital, Eduardo Alfonso, Carol Karp, Yunhee Lee, Sonia Yoo, Kristin Hammersmith, Elisabeth Cohen, Peter Laibson, Christopher Rapuano, Brandon Ayres, Christopher Croasdale, James Caudill, Sanjay Patel, Keith Baratz, William Bourne, Leo Maguire, Joel Sugar, Elmer Tu, Ali Djalilian, Vinod Mootha, James McCulley, Wayne Bowman, H. Dwight Cavanaugh, Steven Verity, David Verdier, Ann Renucci, Matt Oliva, Walter Rotkis, David R. Hardten, Ahmad Fahmy, Marlene Brown, Sherman Reeves, Elizabeth A. Davis, Richard Lindstrom, Scott Hauswirth, Stephen Hamilton, W. Barry Lee, Francis Price, Marianne Price, Kathleen Kelly, Faye Peters, Michael Shaughnessy, Thomas Steinemann, B.J. Dupps, David M. Meisler, Mark Mifflin, Randal Olson, Anthony Aldave, Gary Holland, Bartly J. Mondino, George Rosenwasser, Mark Gorovoy, Steven P. Dunn, David G. Heidemann, Mark Terry, Neda Shamie, Steven I. Rosenfeld, Brandon Suedekum, David Hwang, Donald Stone, James Chodosh, Paul G. Galentine, David Bardenstein, Katrina Goddard, Hemin Chin, Mark Mannis, Rohit Varma, and Ingrid Borecki

Wellcome Trust Case Control Consortium 2: Management Committee: Peter Donnelly (Chair), Ines Barroso (Deputy Chair), Jenefer M. Blackwell, Elvira Bramon, Matthew A. Brown, Juan P. Casas, Aiden Corvin, Panos Deloukas, Audrey Duncanson, Janusz Jankowski, Hugh S. Markus, Christopher G Mathew, Colin N.A. Palmer, Robert Plomin, Anna Rautanen, Stephen J. Sawcer, Richard C. Trembath, Ananth C. Viswanathan, and Nicholas W. Wood; Data and Analysis Group: Chris C.A. Spencer, Gavin Band, Céline Bellenguez, Colin Freeman, Garrett Hellenthal, Eleni Giannoulatou, Matti Pirinen, Richard Pearson, Amy Strange, Zhan Su, Damjan Vukcevic, and Peter Donnelly; DNA, Genotyping, Data QC and Informatics Group: Cordelia Langford, Sarah E. Hunt, Sarah Edkins, Rhian Gwilliam, Hannah Blackburn, Suzannah J. Bumpstead, Serge Dronov, Matthew Gillman, Emma Gray, Naomi Hammond, Alagurevathi Jayakumar, Owen T. McCann, Jennifer Liddle, Simon C. Potter, Radhi Ravindrarajah, Michelle Ricketts, Matthew Waller, Paul Weston, Sara Widaa, Pamela Whittaker, Ines Barroso, and Panos Deloukas; Publications Committee: Christopher G. Mathew (Chair), Jenefer M. Blackwell, Matthew A. Brown, Aiden Corvin, and Chris C.A. Spencer

The Diabetes Control and Complications Trail/Epidemiology of Diabetes Interventions and Complications Research Group: Study Chairmen: S. Genuth, D.M. Nathan, B. Zinman (vice-chair), and O. Crofford (past); Members: J. Crandall, M. Reid, J. Brown-Friday, S. Engel, J. Sheindlin, H. Martinez (past), H. Shamoon (past), H. Engel (past), M. Phillips R. Gubitosi-Klug, L. Mayer, S. Pendegast, H. Zegarra, D. Miller, L. Singerman, S. Smith-Brewer, M. Novak, J. Quin (past), W. Dahms (deceased), Saul Genuth (past), M. Palmert (past), D. Brillon, M. E. Lackaye, S. Kiss, R. Chan, V. Reppucci (past), T. Lee (past), M. Heinemann (past), F. Whitehouse, D. Kruger, J.K. Jones, M. McLellan (past), J. D. Carey, E. Angus, A. Thomas, A. Galprin (past), R. Bergenstal, M. Johnson, M. Spencer (past), K. Morgan, D. Etzwiler, (deceased), D. Kendall (past), Lloyd Paul Aiello, E. Golden, A. Jacobson (past), R. Beaser, O. Ganda, O. Hamdy, H. Wolpert, G. Sharuk, P. Arrigg, D. Schlossman, J. Rosenzwieg (past), L. Rand (past), D.M. Nathan, M. Larkin, M. Ong, J. Godine, E. Cagliero, P. Lou, K. Folino, S. Fritz (past), S. Crowell (past), K. Hansen (past), C. Gauthier-Kelly (past), J. Service, G. Ziegler, L. Luttrell, S. Caulder, M. Lopes-Virella (past), J. Colwell (past), J. Soule (past), J. Fernandes, K. Hermayer, S. Kwon, M. Brabham (past), A. Blevins, J. Parker, D. Lee (past), N. Patel, C. Pittman, P. Lindsey (past), M. Bracey (past), K. Lee, M. Nutaitis, A. Farr (past), S. Elsing (past), T. Thompson (past), J. Selby (past), T. Lyons (past), S. Yacoub-Wasef (past), M. Szpiech (past), D. Wood (past), R. Mayfield (past), M. Molitch, B. Schaefer, L. Jampol, A. Lyon, M. Gill, Z. Strugula, L. Kaminski, R. Mirza, E. Simjanoski, D. Ryan, O. Kolterman, G. Lorenzi, M. Goldbaum, W. Sivitz, M. Bayless, D. Counts, S. Johnsonbaugh, M. Hebdon (past), P. Salemi, R. Liss, T. Donner (past), J. Gordon (past), R. Hemady (past), A. Kowarski (past), D. Ostrowski (past) S. Steidl (past), B. Jones (past), W.H. Herman, C.L. Martin, R. Pop-Busui, A. Sarma, J. Albers, E. Feldman, K. Kim, S. Elner, G. Comer, T. Gardner, R. Hackel, R. Prusak, L. Goings, A. Smith, J. Gothrup, P. Titus, J. Lee, M., Brandle, L. Prosser, D.A. Greene (past), M.J. Stevens (past), A.K. Vine (past), J. Bantle, N. Wimmergren, A. Cochrane, T. Olsen (past), E. Steuer (past), P. Rath (past), B. Rogness (past), D. Hainsworth, D. Goldstein, S. Hitt, J. Giangiacomo, D.S. Schade, J.L., Canady, J.E. Chapin, L.H. Ketai C, S. Braunstein, P.A. Bourne, S. Schwartz (past), A. Brucker, B.J. Maschak-Carey (past), L. Baker (deceased), T. Orchard, N. Silvers, C. Ryan, T. Songer, B. Doft, S. Olson, R.L. Bergren, L. Lobes, P. Paczan Rath, D. Becker, D. Rubinstein, P.W. Conrad, S. Yalamanchi, A. Drash (past), A. Morrison, M.L. Bernal, J. Vaccaro-Kish (past), J. Malone, P.R. Pavan, N. Grove, M.N. Iyer, A.F. Burrows, E.A. Tanaka (past), R. Gstalder (past), S. Dagogo-Jack, C. Wigley, H. Ricks, A. Kitabchi, M.B. Murphy (past), S. Moser (past), D. Meyer (past), A. Iannacone (past), E. Chaum, S. Yoser (past), M. Bryer-Ash (past), S. Schussler (past), H. Lambeth (past), P. Raskin, S. Strowig, B. Zinman, A. Barnie, R. Devenyi, M. Mandelcorn, M. Brent, S. Rogers, A. Gordon, J. Palmer, S. Catton, J. Brunzell, H. Wessells, I.H. de Boer, J. Hokanson, J. Purnell, J. Ginsberg, J. Kinyoun, S. Deeb, M. Weiss, G. Meekins, J. Distad, L. Van Ottingham (past), J. Dupre, J. Harth, D. Nicolle, M. Driscoll, J. Mahon, C. Canny, M. May, J. Lipps, A. Agarwal, T. Adkins, L. Survant, R.L. Pate, G.E. Munn, R. Lorenz (past), S. Feman (past) N. White, L. Levandoski, I. Boniuk, G. Grand, M. Thomas, D.D. Joseph, K. Blinder, G. Shah, Boniuk (past), Burgess (past), J. Santiago (deceased), W. Tamborlane, P. Gatcomb, K. Stoessel, K. Taylor (past), J. Goldstein (past), S. Novella (past), H. Mojibian (past), D. Cornfeld (past), J. Lima, D. Bluemke, E. Turkbey, R. J. van der Geest, C. Liu, A. Malayeri, A. Jain, C. Miao (past), H. Chahal (past), R. Jarboe (past), and J. Maynard; Clinical Coordinating Center: R. Gubitosi-Klug, J. Quin, P. Gaston, M. Palmert (past), R. Trail (past), and W. Dahms (deceased); Data Coordinating Center: J. Lachin, P. Cleary, J. Backlund, W. Sun, B. Braffett, K. Klumpp, K. Chan (past), L. Diminick, D. Rosenberg (past), B. Petty (past), A. Determan (past), D. Kenny (past), B. Rutledge (past), Naji Younes (past), L. Dews, and M. Hawkins; National Institute of Diabetes and Digestive and Kidney Disease Program Office: C. Cowie, J. Fradkin, C. Siebert (past), and R. Eastman (past); Central Fundus Photograph Reading Center: R. Danis, S. Gangaputra, S. Neill, M. Davis (past), L. Hubbard (past), H. Wabers, M. Burger, J. Dingledine, V. Gama, and R. Sussman; Central Biochemistry Laboratory: M. Steffes, J. Bucksa, M. Nowicki, and B. Chavers; Central Carotid Ultrasound Unit: D. O’Leary, J. Polak, A. Harrington, and L. Funk (past); Central ECG Reading Unit: R. Crow (past), B. Gloeb (past), S. Thomas (past), and C. O’Donnell (past); Central ECG Reading Unit: E. Soliman, Z.M. Zhang, R. Prineas (past), and C. Campbell; Central Neuropsychological Coding Unit: C. Ryan, D. Sandstrom, T. Williams, M. Geckle, E. Cupelli, F. Thoma, B. Burzuk, and T. Woodfill; Central ANS Reading Unit: P. Low, C. Sommer, and K. Nickander; Computed Tomography Reading Center: M. Budoff, R. Detrano (past), N. Wong, M. Fox, L. Kim (past), and R. Oudiz; External Evaluation Committee: G. Weir (Chairman), M. Espeland, T Manolio, L. Rand, D. Singer, M. Stern, A.E. Boulton, C. Clark (past), and R. D’Agostino (past); Molecular Risk Factors Program Project: M. Lopes-Virella, W.T. Garvey (past), T.J. Lyons, A. Jenkins, G. Virella, A. Jaffa, Rickey Carter, D. Lackland (past), M. Brabham (past), D. McGee (past), D. Zheng (past), and R.K. Mayfield (past); Genetic Studies Group: A. Boright, S. Bull, L. Sun, S. Scherer (past), and B. Zinman (past); Epigenetics: R. Natarajan, F. Miao, L. Zhang, and Z. Chen; Editor, EDIC Publications: D.M. Nathan