Associations of Iris Structural Measurements in a Chinese Population: The Singapore Chinese Eye Study

Abstract

Purpose.: We determined the ocular biometric and demographic factors associated with iris parameters in Singaporean Chinese persons from a population-based sample.

Methods.: Subjects were participants in the Singapore Chinese Eye Study, a population-based cross-sectional study of eye disease. Anterior segment optical coherence tomography images were analyzed using customized software to measure iris thickness at 750 μm from the scleral spur (IT750), iris area (I-Area), and iris curvature (I-Curv). Regression analyses were performed to assess the association between iris measurements with a range of demographic and ocular variables. The contribution of each independent variable to the iris parameter of interest was determined sequentially using a stepwise selection algorithm.

Results.: We included 1473 participants with a mean age of 57.7 + 8.68 years, and 50.6% were men. The mean IT750, I-Area, and I-Curv were 0.46 ± 0.10 mm, 1.49 ± 0.24 mm², and 0.25 ± 0.13 mm, respectively. Statistical regression models, including a range of demographic and ocular parameters, explained 59.3%, 41.9%, and 34.3% of the variability in I-Curv, IT750, and I-Area, respectively. Angle opening distance at 750 μm from the scleral spur (AOD750) was the single factor associated most strongly with I-Curv, and explained 46.6% of its variation.

Conclusions.: A significant proportion of the variation in iris area, curvature, and thickness was not explained by other ocular and demographic parameters. Iris curvature was associated strongly with angle width, and of all parameters investigated, AOD750 was most highly correlated with iris curvature.

Introduction

The iris has a key role in the pathogenesis of primary angle closure, in which the forward bowing and convex configuration of the iris results in pupil block and angle narrowing. ^1–5 Laser peripheral iridotomy (LPI) relieves pupil block and widens the anterior chamber angle by flattening the iris curvature. ^6–13 However, nonpupil block mechanisms are present in a significant proportion of eyes with angle closure, ¹⁴ and LPI does not widen the angle in all cases. ^14–16  

With the introduction of new imaging technology, such as anterior segment optical coherence tomography (ASOCT), precise and reproducible quantitative measurements of the iris can now be made using customized software. ^12,17–22 Several quantitatively measured iris parameters have now been identified as novel anatomic risk factors for primary angle closure. ^12,21 In a large community-based study in Singapore, we have shown that three of these iris parameters (greater iris curvature, area, and thickness) were independent factors associated with narrow angles. ²¹ We have reported further increased iris thickness in eyes with primary angle closure (PAC) and primary angle closure glaucoma (PACG) compared to normal controls. ¹² Dynamic changes in the iris configuration also may contribute to the pathogenesis of angle closure. ^23,24  

While iris parameters are associated with PACG, the influence of demographic factors, as well as other ocular parameters on iris parameters, has not been well studied. In a previous hospital-based study, increased iris convexity was found to correlate with older age and decreased anterior chamber depth (ACD) in patients with PAC and PACG. ²⁵ Wang et al. assessed the iris structural measurements in American Caucasian and Chinese persons, and reported that lens location, pupil diameter, and spherical equivalent refraction correlated with iris curvature, while iris thickness and area were associated significantly with sex, ethnicity, and pupil diameter. ²⁶ These studies, however, have been limited by the inclusion of highly selected clinic- or volunteer-based samples, which are prone to selection bias. To our knowledge, there are no population-based data evaluating the relationships and relative importance of different demographic and ocular correlates of iris measurements. To address this gap, we performed a population-based study to investigate associations of ocular biometric and demographic factors with iris thickness, area, and curvature in Singaporean Chinese persons. 

Methods

Participants

The objectives and methodology of the Singapore Chinese Eye Study (SCES) have been described previously. ²⁷ In brief, the SCES is a population-based study of Chinese persons aged 40 years and older residing in the southwestern part of Singapore examined between 2009 and 2011. The study was designed to ascertain the prevalence and impact of major eye disease in Chinese persons in Singapore. An age-stratified (by 10-year age group) random sampling strategy was used to select participants from a computer-generated list provided by the Ministry of Home Affairs, Singapore. Standardized interviewer-administered questionnaires were administered to all participants who underwent a detailed ocular examination and ASOCT imaging. The study adhered to the tenets of the Declaration of Helsinki, and ethical approval was obtained from the Institutional Review Board of the Singapore Eye Research Institute. All participants provided written informed consent. In this substudy, we examined subjects recruited from June 2009 to August 2010. We excluded subjects who were pseudophakic, and those who underwent previous intraocular surgery or laser peripheral iridotomy. None of the subjects received pilocarpine or cycloplegic eyedrops. 

Iris Parameters Measured From ASOCT

All participants had a standardized examination. A horizontal scan of the nasal and temporal quadrants of all subjects was obtained using ASOCT (Visante, software version 2.01.88; Carl Zeiss Meditec, Dublin, CA) by a single experienced operator who was masked to the clinical data. ASOCT imaging was performed in a dark room (0 lux) with the images centered on the pupil. The ASOCT obtains high resolution cross-sectional images of the anterior segment using infrared light of 1300 nm wavelength. ²⁸ ASOCT images were obtained using the standard anterior segment single-scan protocol, which produces 256 scans in 0.125 second. The reflex saturation beam was used to ensure scan alignment. The examiner optimized the polarization for each scan, and adjusted the image saturation and noise to obtain the best quality image. Poor quality images with artifacts resulting from the eyelids and movement artifacts were repeated. Several ASOCT images were obtained for each participant, and the image deemed to have the best quality with the fewest image artifacts was selected for analysis. 

Measurements of ASOCT parameters were obtained using the Zhongshan Angle Assessment Program (ZAAP, Guangzhou, China) ^{14,17–20,22,26} by a single observer (FLL), who was masked to the clinical data. Identification of the location of the two scleral spurs on the ASOCT images was the only observer input required, after which the algorithm calculated the following iris parameters automatically. Iris thickness at 750 μm from the scleral spur (IT750) was measured directly, iris area (I-Area) was calculated as the cross-sectional area of the iris from the scleral spur to the pupil, and iris curvature (I-Curv) was measured as the perpendicular distance from the posterior iris surface at the point of greatest convexity to a line between the most central to the most peripheral point of the iris pigment epithelium (Fig. 1A). ^12,21  

Figure 1. 

 

Measurement of anterior segment parameters on anterior segment optical coherence tomography images using customized software. (A) Measurement of iris parameters. (B) Measurement of other anterior segment parameters. SS, scleral spur.

Other ASOCT Parameters

Other measured parameters included angle opening distance at 750 μm from the scleral spur (AOD750), the distance from the corneal endothelium/trabecular meshwork to the anterior surface of the iris perpendicular to a line drawn along the trabecular meshwork at 750 μm anterior to the scleral spur ²⁰ ; anterior chamber width (ACW), the horizontal distance between the two scleral spurs ¹⁹ ; and lens vault (LV), the perpendicular distance from the horizontal line between the two scleral spurs to the anterior pole of the lens. ¹⁸ Posterior corneal arc length (PCAL) was measured as the arc length of the posterior corneal surface between the two scleral spurs. ^20,29 Anterior chamber area (ACA) was measured as the cross-sectional area of the anterior chamber bounded by the anterior surface of the iris, the anterior surface of the lens (within the pupil), and the posterior surface of the cornea (Fig. 1B). A vertical axis was plotted through the center of the ACA by the software, and anterior chamber volume (ACV) was calculated by rotating ACA 360° around this vertical axis. ²² Measurements of anterior segment parameters obtained using ZAAP have been shown to be highly reproducible, with intraclass correlation coefficient exceeding 0.88. ^19,21 The intraclass correlation coefficient was greater than 0.8 for all ASOCT measurements in our study. The average of temporal and nasal measurements was used in the analysis of individual quadrant parameters. Pupil diameter was defined as the shortest distance between the pupil edges of the iris ²¹ and measured manually using the ZAAP software calipers tool. 

Measurement of Other Variables

Lens thickness was measured by a trained ophthalmic technician using A-scan ultrasound (Model US-800; Nidek Co. Ltd., Tokyo, Japan) to ensure an SD among measurements of 0.12 mm or less. Measurements of ACD, axial length (AL), and radius of corneal curvature were obtained using the IOLMaster (software version 3.02; Carl Zeiss Meditec). ACD was measured from the corneal epithelium to the anterior lens surface. All readings had a signal-to-noise ratio of >2.0, indicating a clear signal when the measurement was performed. Five consecutive readings were taken for ACD and AL, and three consecutive readings were taken for corneal curvature, with the average values used in subsequent analyses. For each parameter, all readings were within 0.05 mm of the reading with the highest signal-to-noise ratio. IOP was measured using Goldmann applanation tonometry, and the vertical cup-disc ratio (CDR) was assessed clinically at the slit-lamp (Haag-Streit model BQ-900; Haag-Streit, Koeniz, Switzerland) using a 78 diopter lens. Lens position (LP) was defined as ACD + ½ lens thickness and relative lens position (RLP) as LP/AL. A wall-mounted measuring tape was used to measure height in centimeters (cm). 

Statistical Analysis

Demographic data and ocular characteristics from the right eye of the participants were described using the mean, SD, and minimum and maximum values. Characteristics of included versus excluded participants were compared using the t-test or the Wilcoxon rank sum test as appropriate. The Pearson correlation coefficient was determined for iris parameters (IT750, I-Area, I-Curv), and other ocular and demographic parameters. The t-test was used to compare I-Curv, IT750, and I-Area between male and female persons. Multivariate linear regression was performed on each iris parameter (IT750, I-Area, I-Curv) as the dependent variable with all demographic and ocular characteristics included as independent variables. To investigate a possible threshold effect for age, we defined age categories (<50, 50–59, 60–69, ≥70 years) and incorporated age as a categorical variable into the analysis of IT750, I-Area, and I-Curv. Stepwise regression (significance level to enter = 0.05) was used to select variables sequentially into an explanatory model based on the greatest improvement in R ². The degree of multicollinearity among the variables selected, which is associated with instability of the estimated coefficients, was assessed using the variance inflation factor (VIF). Among the variables selected for the model, those with VIF > 10 were identified and a minimal subset removed so as to achieve VIF < 10 for the remaining variables. The assumption of normality of residuals was assessed using graphical methods and found this to be tenable. P < 0.05 was considered the threshold for declaring statistical significance. All analyses were performed using SAS version 9.2 (SAS Institute, Inc., Cary, NC). 

Results

Data were available for 1707 participants, and 234 participants were excluded from the analysis, as 56 had poor quality ASOCT images, 77 had software delineation errors, and 93 had ASOCT images with indeterminate scleral spurs. The remaining 1473 participants (86.2%) were eligible for the final analysis, of which 746 (50.6%) were male. Comparison of persons included and excluded showed that there were no significant differences in sex or height between included and excluded subjects. However, subjects (mean, 95% confidence interval [CI]) who were included (In), compared to those excluded (Ex), were younger (year, P < 0.001; In 57.8, 57.3–58.2 versus Ex 61.1, 60.3–61.9), had higher IOP (mm Hg, P = 0.008; In 14.5, 14.3–14.6 versus Ex 14.1, 13.8–14.3), longer AL (mm, P = 0.008; In 24.0, 23.9–24.0 versus Ex 24.1, 24.0–24.3), and deeper ACD (mm, P < 0.001; In 3.22, 3.20–3.23 versus Ex 3.32, 3.28–3.35) compared to excluded subjects. 

In this population, the mean IT750, I-Area, and I-Curv were 0.46 ± 0.10 mm, 1.49 ± 0.24 mm², and 0.25 ± 0.13 mm, respectively. Figure 2 shows that there was little variation of IT750 with age, but I-Area and I-Curv increased with age. The demographic and ocular characteristics of the participants are summarized in Table 1, and Pearson correlations among ocular and demographic parameters are summarized in Table 2. 

Figure 2

 

Distribution of iris parameters measured using ASOCT with age. (A) Distribution of iris thickness at 750 μm from the sclera spur with age. (B) Distribution of iris area with age. (C) Distribution of iris curvature with age.

Table 1. 

 

Demographic and Ocular Characteristics of the Participants in the Singapore Chinese Eye Study

Parameter	Mean	SD	Minimum	Maximum
Age, y	57.8	8.68	46.0	84.0
Height, cm	163	8.35	140	190
IT750, mm	0.46	0.10	0.24	1.24
Iris area, mm2	1.49	0.24	0.74	2.25
Iris curvature, mm	0.25	0.13	−0.52	0.67
AOD750, μm	0.38	0.19	0.00	1.62
ACW, mm	11.5	0.42	9.30	12.7
ACA, mm2	20.3	3.56	9.33	34.1
ACV, mm3	135	29.8	49.9	256
LV, μm	386	292	−552	1475
PCAL, mm	13.4	0.55	10.6	15.2
ACD, mm	3.22	0.36	1.95	4.41
AL, mm	24.0	1.39	20.4	39.1
Lens thickness, mm	3.96	0.72	2.08	5.64
Corneal curvature, mm	6.48	0.36	5.01	9.33
Pupil diameter, mm	4.19	0.76	0.52	8.27
CDR	0.40	0.13	0.11	0.94
IOP, mm Hg	14.5	3.14	5.00	43.0
LP, mm	5.12	0.65	2.06	6.41
RLP	0.21	0.03	0.10	0.27

Table 2. 

 

Correlation Between Iris Parameters and Other Ocular and Demographic Parameters in the Singapore Chinese Eye Study

Parameter	Pearson Correlation/ P Value	IT750	Iris Area	Iris Curvature
Age, y	Correlation	−0.04	0.11	0.30
	P	0.142	<0.001*	<0.001*
Height, cm	Correlation	0.08	0.07	−0.20
	P	0.002*	0.005*	<0.001*
IT750, mm	Correlation	–	0.30	−0.03
	P	–	<0.001*	0.244
Iris area, mm2	Correlation	0.30	–	0.27
	P	<0.001*	–	<0.001*
Iris curvature, mm	Correlation	−0.04	0.27	–
	P	0.171	<0.001*	–
AOD750, μm	Correlation	−0.25	−0.14	−0.69
	P	<0.001*	<0.001*	<0.001*
ACW, mm	Correlation	−0.06	0.19	−0.12
	P	0.020*	<0.001*	<0.001*
ACA, mm2	Correlation	0.06	−0.10	−0.64
	P	0.028*	<0.001*	<0.001*
ACV, mm3	Correlation	0.05	−0.09	−0.63
	P	0.075	<0.001*	<0.001*
LV, μm	Correlation	−0.06	0.08	0.67
	P	0.021*	0.002*	<0.001*
PCAL, mm	Correlation	−0.13	0.16	−0.09
	P	<0.001*	<0.001*	<0.001*
ACD, mm	Correlation	0.04	−0.03	−0.59
	P	0.141	0.223	<0.001*
AL, mm	Correlation	0.03	0.05	−0.50
	P	0.200	0.042*	<0.001*
Lens thickness, mm2	Correlation	−0.02	−0.04	−0.08
	P	0.542	0.134	0.003*
Corneal curvature, mm	Correlation	0.14	0.15	−0.06
	P	<0.001*	<0.001*	0.031*
Pupil diameter, mm	Correlation	0.182	−0.03	0.02
	P	<0.001*	0.196	0.485
CDR	Correlation	0.04	−0.01	−0.03
	P	0.155	0.842	0.232
IOP, mm Hg	Correlation	−0.001	−0.02	−0.02
	P	0.970	0.416	0.517
LP, mm	Correlation	0.03	−0.07	−0.39
	P	0.248	0.009*	<0.001*
RLP	Correlation	0.009	−0.10	−0.19
	P	0.726	<0.001*	<0.001*

*   P < 0.05.

The results of multivariate analysis of regression models for IT750, I-Area, and I-Curv are summarized in Table 3. Increased IT750 was associated with the male sex, smaller AOD750, larger ACV, shorter PCAL, larger ACD, shorter AL, larger lens thickness, larger corneal curvature, and smaller RLP. Increased I-Area was associated with smaller AOD750, larger ACW, larger ACA, smaller ACV, larger ACD, shorter AL, larger lens thickness, larger corneal curvature, smaller pupil diameter, and smaller RLP. Increased I-Curv was associated with older age, smaller AOD750, smaller ACW, smaller ACA, smaller LV, longer PCAL, smaller pupil diameter, and larger RLP. For the analyses of IT750 and I-Area, age as a categorical variable was not statistically significant (P = 0.118 and P = 0.694, respectively). Incorporating age as a categorical variable into the multivariate analysis model of I-Curv showed that age remained associated significantly with I-Curv (P < 0.001). The mean I-Curv values corresponding to the three older age categories (50–59, 60–69, ≥70 years) did not differ significantly from each other, but all three mean I-Curv values were significantly higher than that of the <50 years age category (P < 0.001). This indicates that there was a threshold effect in I-Curv from <50 to ≥50 years of age. 

Table 3. 

 

Multivariate Analysis of the Associations of Iris Thickness at 750 μm From the Scleral Spur, Iris Area, and Iris Curvature With the Ocular and General Parameters in the Singapore Chinese Eye Study

Parameter	IT750		I-Area		I-Curv
	Regression Coefficient (95% CI)	P Value	Regression Coefficient (95% CI)	P Value	Regression Coefficient (95% CI)	P Value
Age, y	0.0005 (−0.0001, 0.0010)	0.081	0.0003 (−0.0010, 0.0016)	0.666	0.0009 (0.0003, 0.0014)	0.004*
Sex, ref = male	−0.021 (−0.0325, −0.00895)	<0.001*	−0.018 (−0.047, 0.012)	0.240	−0.005 (−0.018, 0.008)	0.439
Height, cm	−0.0001 (−0.0008, 0.0006)	0.759	0.0014 (−0.0004, 0.0032)	0.136	−0.0002 (−0.0010, 0.0006)	0.694
AOD750, μm	−0.543 (−0.590, −0.495)	<0.001*	−0.175 (−0.295, −0.055)	0.004*	−0.240 (−0.292, −0.187)	<0.001*
ACW, mm	0.008 (−0.054, 0.038)	0.727	0.254 (0.139, 0.369)	<0.001*	−0.092 (−0.143, −0.042)	0.003*
ACA, mm2	0.002 (−0.016, 0.019)	0.863	0.123 (0.078, 0.167)	<0.001*	−0.026 (−0.045, −0.006)	0.009*
ACV, mm3	0.004 (0.002, 0.006)	<0.001*	−0.017 (−0.022, −0.013)	<0.001*	0.0005 (−0.0016, 0.0026)	0.627
LV, μm	0.00002 (−0.00002, 0.00007)	0.355	−0.00008 (−0.00020, 0.00004)	0.198	−0.000053 (−0.000105, −0.000001)	0.046*
PCAL, mm	−0.128 (−0.165, −0.091)	<0.001*	0.027 (−0.065, 0.119)	0.564	0.149 (0.108, 0.189)	<0.001*
ACD, mm	0.182 (0.079, 0.284)	<0.001*	0.424 (0.1650, 682)	0.001*	−0.061 (−0.174, 0.052)	0.289
AL, mm	−0.032 (−0.052, −0.011)	0.002*	−0.070 (−0.121, −0.019)	0.007*	0.001 (−0.021, 0.024)	0.907
Lens thickness, m	0.095 (0.043, 0.147)	<0.001*	0.208 (0.077, 0.339)	0.002*	−0.038 (−0.096, 0.019)	0.191
Corneal curvature, mm	0.017 (0.003, 0.031)	0.019*	0.076 (0.041, 0.111)	<0.001*	0.014 (−0.001, 0.029)	0.072
Pupil diameter, mm	−0.002 (−0.008, 0.005)	0.627	−0.155 (−0.171, −0.139)	<0.001*	−0.013 (−0.021, −0.006)	0.002*
CDR	−0.009 (−0.039, 0.021)	0.560	−0.045 (−0.120, 0.031)	0.247	−0.030 (−0.064, 0.003)	0.071
IOP, mm Hg	0.0001 (−0.001, 0.0015)	0.841	0.002 (−0.002, 0.005)	0.325	−0.0006 (−0.0021, 0.0008)	0.425
RLP	−4.80 (−7.04, −2.16)	<0.001*	−9.56 (−15.7, −3.41)	0.002*	1.83 (−0.87, 4.52)	0.004*

*   P < 0.05.

The results of the stepwise linear regression analysis for IT750, I-Area, and I-Curv are summarized in Table 4. Five variables (AOD750, PCAL, ACV, sex, and AL) explained 41.9% of the variability in IT750. In the stepwise selection regression analysis on I-Area, a model consisting of 8 variables (pupil diameter, ACW, AOD750, LV, ACV, and height) explained only 34.3% of the I-Area variability. The stepwise linear regression analysis for I-Curv showed that AOD750 explained most of the variation (partial R ² 0.466, P < 0.001). The 7 selected variables (AOD750, LV, pupil diameter, AL, ACW, ACV, and age) explained 59.3% of the variability in I-Curv. Adding more variables to these three models did not substantially increase R ². 

Table 4. 

 

Stepwise Multiple Linear Regression Analysis on Iris Thickness at 750 μm From the Scleral Spur, Iris Area, and Iris Curvature in the Singapore Chinese Eye Study

Dependent Variable	N of Variables In Model	Variable	Regression Coefficient (95% CI)*	Partial R 2	Model R 2
IT750*	1	AOD750, μm	−0.541 (−0.577, −0.506)	0.059	0.059
	2	ACV, mm3	−0.0039 (−0.0036, −0.0042)	0.120	0.179
	3	PCAL, mm	−0.122 (−0.132, −0.112)	0.218	0.398
	4	Sex, ref = male	−0.024 (−0.032, −0.016)	0.017	0.414
	5	AL, mm	0.006 (0.002, 0.009)	0.004	0.419
I-Area†	1	Pupil diameter, mm	−0.147 (−0.161, −0.133)	0.189	0.189
	2	ACW, mm	0.276 (0.235, 0.317)	0.073	0.262
	3	AOD750, μm	−0.307 (−0.407, −0.207)	0.030	0.292
	4	LV, μm	−0.00031 (−0.00037, −0.00024)	0.030	0.322
	5	ACV, mm3	−0.0028 (−0.0038, −0.0019)	0.017	0.338
	6	Height, cm	0.0019 (0.0007, 0.0032)	0.004	0.343
I-Curv‡	1	AOD750, μm	−0.193 (−0.238, −0.149)	0.466	0.466
	2	LV, μm	0.00005 (0.00002, −0.00008)	0.057	0.523
	3	Pupil diameter, mm	−0.020 (−0.026, −0.013)	0.024	0.547
	4	AL, mm	−0.016 (−0.020, −0.012)	0.013	0.560
	5	Age, y	0.001 (0.0007, 0.0018)	0.007	0.567
	6	ACW, mm	0.071 (0.053, 0.090)	0.005	0.571
	7	ACV, mm3	−0.0014 (−0.0018, −0.001)	0.026	0.593

*  All regression coefficients were significant at P ≤ 0.001.

†  All regression coefficients were significant at P < 0.001 except height (P = 0.002).

‡  All regression coefficients were significant at P < 0.001.

Discussion

In our study, we have provided new population-based data on the distribution of quantitatively measured iris parameters, and their associations with a range of demographic and ocular parameters. We showed that statistical models, consisting of demographic and ocular parameters, explained approximately half of the variation in I-Curv and IT750, but only approximately one-third of the variability in I-Area. In contrast, a similar set of 8 demographic and ocular variables explained 80% of the variability in ACD, and 87% of the variability in AOD750 was explained by 6 variables in the same cohort. ^17,20  

Our finding that AOD750 was a major correlate of I-Curv has not been reported previously to our knowledge. AOD750 was shown previously to have a high discriminative ability for identifying the presence of narrow angles on gonioscopy. ³⁰ A community-based study that included 2047 Singaporean subjects > 50 years old showed that increased I-Curv, I-Area, and IT750 were associated significantly with narrow angles (odds ratio [OR] 2.5 and 95% CI 1.3–5.1, OR 2.7 and 95% CI 1.6–4.8, and OR 2.6 and 95% CI 1.6–4.1, respectively, comparing fourth with first quartile for each parameter) after adjusting for age, sex, ACD, AL, and pupil size. ²¹ In our study, AOD750 accounted for nearly half of the variation in I-Curv, supporting the hypothesis that angle width and the extent of pupil block are highly correlated, as I-Curv can be regarded as a surrogate measure of pupil block. Pupil block creates a pressure differential between the anterior and posterior chambers, which causes the iris to adopt a convex forward bowing configuration, and this results in a narrow or closed angle. ^1–5,31 In contrast, AOD750 only explained 5.9% of the variation in IT750 and 3.0% of the variation in I-Area. These findings are consistent with an earlier study by Wang et al., which compared 167 participants with angle closure and 1153 normal participants, and reported that larger IT750 was associated with a small, but significant increase in the risk of angle closure (multivariate-adjusted OR 1.7, 95% CI 1.1–2.7, P = 0.032, per 0.1 unit increase), while I-Area was not associated independently with angle closure, after adjusting for age, sex, pupil size, and ACD. ¹² A thicker peripheral iris would contribute to angle crowding and subsequent angle closure as the peripheral iris would be in closer proximity with the trabecular meshwork. However, IT750 was not found to be an independent determinant of AOD750 or TISA750 in an optimal model determined by a stepwise selection algorithm. In contrast, other anterior segment parameters are much more important determinants of AOD750 and TISA750. ¹⁷  

Age and sex were associated poorly with iris measurements. Older age and female sex are strong risk factors for primary angle closure glaucoma. ^32,33 Our findings showed that age was not associated significantly with IT750 and I-Area. While age was associated significantly with I-Curv, it only explained approximately 0.7% of the variation in I-Curv. This is consistent with the results of Nonaka et al., who found that increased iris convexity correlated only weakly, but significantly with age (r = 0.22, P < 0.01, Spearman rank correlation). ²⁵ Interestingly, we showed that I-Curv increased with age till 50 years, after which the correlation between I-Curv and age was poor. This threshold effect in I-Curv from <50 to ≥50 years of age has not been reported previously to our knowledge. We also found that IT750 was larger in males. This stands in contrast to the findings in an earlier hospital-based study, which showed that iris thickness was not associated significantly with sex in 365 Chinese participants. ²⁶ The smaller sample size and possible selection bias in the hospital-based study could explain this discrepancy with our results. Nevertheless, our results indicate that sex was a poor independent correlate for IT750, despite the fact that an association was seen. We also have shown that measurements of ocular dimensions, including AL and ACW, were associated poorly with iris measurements, accounting for less than 10% of the variation in IT750, I-Area, and I-Curv. 

Our finding that corneal measurements were correlated significantly with iris parameters has not been reported previously to our knowledge. PCAL, a measurement of the length of the posterior corneal surface, accounted for 21.8% of the variation in IT750. PCAL also was an important determinant of ACD in Singaporean Chinese persons, ^20,29 explaining almost 20% of the variation in ACD. ²⁰ These findings show that the shape of the posterior corneal surface is associated significantly with other ocular biometric parameters of the anterior segment, and future studies are necessary to evaluate the significance of these associations. 

Pupil diameter accounted for almost 20% of the variation in I-Area, and smaller I-Area was associated with a larger pupil diameter. Aptel et al. found that after pharmacologic mydriasis, iris volume decreased significantly in open-angle eyes (from 44.29 ± 3.9 to 37.88 ± 2.2 mm³, P < 0.01), while it increased significantly in fellow eyes of patients who had an episode of primary acute angle closure (from 44.94 ± 2.1–49.92 ± 2.9 mm³, P < 0.01). ²³ Subgroup analysis in our population-based study to determine whether this relationship was present in subjects with angle closure would be interesting, but the number of subjects with angle closure would not be sufficient for this analysis. 

The differential factors influencing iris parameters and other anterior segment measurements are interesting. Predictive modeling explained approximately 60% of the variation in I-Curv, 40% of the variation in IT750, and only approximately 35% of the variation in I-Area. In contrast, a statistical model consisting of eight variables explained 80% of the variability in ACD and a model consisting of six variables explained nearly 90% of the variability in AOD750 in the Singapore Chinese Eye Study. ^17,20 This suggests that, unlike ACD and AOD750, a significant number of factors that affect iris measurements have not been identified. These may include physiologic factors, such as the permeability of the iris to water, ³¹ or other measures of iris material properties (such as elasticity) that currently cannot be measured. Some of these factors are likely heritable as the genetic basis for iris features is thought to be high. ^34,35 A study of 309 monozygotic and 165 dizygotic twins aged 8 to 16 years found that the heritability for iris thickness was approximately 60%. ³⁵ Our finding that a significant proportion of the variation in iris measurements was not explained by ocular and demographic parameters suggests that the underlying unmeasured genetic determinants may be different for structural features of the iris. 

The strengths of our study included the population-based design, which minimizes the selection bias inherent in previous hospital- and community-based studies ^12,21,25,26 ; the analysis of ocular biometric parameters measured using ASOCT and IOLMaster, which were not available in previous studies; and the establishment of the relative importance of these correlates of iris measurements. However, some limitations should be considered. First, a large number of participants were excluded for poor quality ASOCT images, indeterminate scleral spurs, and ZAAP software delineation errors. This was comparable with other studies in which ASOCT was used to image the anterior segment. ^12,17,20 It is possible that bias was introduced due to the large number of ungradable images, and that the associations found and the magnitude of these associations may have differed if all data had been available. Second, a single scan was used to measure iris parameters, and regional variability in anterior chamber angle characteristics and iris parameters could account for the lack of an association. Third, data on the use of drugs, such as α₁-adrenergic receptor antagonists, that may affect iris structure were not available in our study. Lastly, our results may be applicable only to the Chinese population, and iris parameters in other ethnic groups may have different associations. Comparison of the iris structural measurements between Caucasian and Chinese persons showed that Chinese persons have thicker irides and greater I-Area than Caucasians. ^26,36  

In summary, currently known factors explain only a portion of iris parameter measurements. Age, sex, and measurements of ocular dimension, including AL, ACD, and ACW, were poor independent correlates of iris parameters. AOD750 was the single factor associated most strongly with I-Curv. Future studies will be needed to identify the currently unrecognized correlates of iris parameters, which will provide greater insights into the role of the iris in the pathophysiology of primary angle closure glaucoma. 

Acknowledgments

Supported by grants from the National Medical Research Council, Singapore and the National Research Foundation, Singapore. The authors alone are responsible for the content and writing of the paper. 

Disclosure: C.C. Sng, None; J.C. Allen, None; M.E. Nongpiur, None; L.-L. Foo, None; Y. Zheng, None; C.Y. Cheung, None; M. He, None; D.S. Friedman, None; T.Y. Wong, None; T. Aung, None