Cornea Biomechanical Characteristics and Their Correlates with Refractive Error in Singaporean Children

Abstract

purpose. To determine corneal biomechanical parameters measured with the Reichert Ocular Response Analyser (ORA) in Singaporean children, and to assess their possible correlations with refractive error and biometry.

methods. This was a cross-sectional study of 271 subjects from the Singapore Cohort Study of Risk Factors for Myopia (SCORM). Corneal hysteresis (CH), corneal resistance factor (CRF), central corneal thickness (CCT), and cornea-compensated intraocular pressure (IOPcc) were measured with the ORA. Spherical equivalent refraction was assessed with an autokeratorefractometer and axial length by contact ultrasound A-scan biometry. Height, weight, and blood pressure were measured.

results. The mean age of the study population was 13.97 ± 0.89 years, the distribution of the sexes was almost equal (138 boys, 50.9%), and most were Chinese (186 subjects, 68.6%). The mean (±SD) CH and CRF were 11.78 ± 1.55 (range, 6.93–16.53) and 11.81 ± 1.71 (range, 7.83–16.83) mm Hg. CH and CRF did not vary significantly with age (P = 0.24; 0.61), sex (P = 0.21; 0.08), or race (P = 0.23; 0.36). CH and CRF did not vary with myopia status (P = 0.79; 0.83) or axial length (Pearson correlation coefficient [r] = −0.11 and −0.05, P = 0.08 and 0.40). Multivariate analyses were performed with CH, CRF, or CCT as the dependent variable and age, sex, race, weight, IOPcc, CCT, SE refraction, and corneal curvature as covariates. CH was significantly associated with IOP (regression coefficients [β] = −0.22 [95% confidence interval = −0.27 to −0.17]), CCT (β = 0.03 [0.02–0.03]) and corneal curvature (β = −1.13 [−2.08 to −0.19]). CRF was significantly associated with IOP, CCT, and corneal curvature (β = 0.08 [0.02–0.14]; 0.03 [0.03–0.04], and −1.39 [−2.54 to −0.23], respectively). The only factor that was predictive of decreased CCT was Malay or Indian race (P = 0.03 and <0.001), compared with Chinese.

conclusions. The CH and CRF values in our study on Singaporean children are slightly higher than in adult studies. CH and CRF are not associated with refractive error or axial length. Flatter corneas are associated with lower CH and CRF readings.

Central corneal thickness (CCT) influences intraocular pressure (IOP) measurements with applanation tonometry, ^{1 2} and it has also served as an in vivo surrogate marker of corneal rigidity. Previously, assessment of the biomechanical properties of the cornea was only possible with theoretical and laboratory models. ^{3 4 5} With the recent introduction of the Ocular Response Analyser (ORA; Reichert Ophthalmic Instruments, Depew, NY), ^{6 7} direct clinical assessment of the biomechanical properties of the cornea has become possible. 

The ORA functions by emitting an air jet to deform the cornea, and the response of the cornea to deformation gives information on both the IOP and corneal biomechanical properties. By correlating the instantaneous air pressure with the time of occurrence of applanation events, the ORA is able to dissociate the influence of corneal biomechanical properties on corneal deformation from that of IOP. The principal biomechanical parameter measured by the ORA is corneal hysteresis (CH), which is defined as the difference between the force-in applanation pressure (P1) and the force-out applanation pressure (P2). CH is believed to be a reflection of the viscoelastic or damping properties of the cornea ⁸ and forms the basis of a derived parameter, the corneal resistance factor (CRF). ^{9 10} CH has a strong positive correlation with CRF, and both CH and CCT have moderately strong positive correlations with CCT. ⁷ The ORA also utilizes the CH to calculate a cornea-compensated IOP (IOPcc) that is purportedly free of the influence of CCT or corneal biomechanical properties on IOP measurement. 

To our knowledge, there are no published results on the possible association between CH and refractive error or axial length (AL). Axial elongation associated with myopia is known to be associated with structural changes in the cornea. Experimentally induced ametropias are known to induce corneal astigmatism and changes in corneal curvature in monkeys and chickens, ^{11 12 13} with changes in corneal fibril orientation suggesting regional variations in corneal biomechanics. Clinically, flatter corneal curvature and decreased corneal thickness have also been described in association with myopia. ^{14 15} Scleral remodeling occurs in myopia, with altered content and orientation of collagen fibrils. ^{16 17} The identification of a common 23.5-kDa serine proteinase that is upregulated in both the cornea and sclera with globe enlargement in avian models of myopia ¹⁸ suggests that corneal remodeling may occur in tandem with scleral changes. ¹⁸  

There is also limited literature on other factors that may affect CH and CRF, particularly in children. As children are generally free of other disease, they are ideal for studying potential relationships. As CH, CRF, and CCT are known to correlate highly, ⁶ factors that affect CCT including age, sex, race, ^{19 20 21 22} and corneal curvature, ^{23 24} may also affect CH and CRF. Age and sex have been addressed in the literature in mainly adult populations to date. ^{25 26 27 28} Kirwan et al. ⁹ performed CH measurements in 81 normal eyes of 42 children and reported that CH does not correlate with age. 

The purpose of this study was to determine the corneal biomechanical parameters measured with the ORA in Singaporean children and to assess their possible correlations with refractive error and biometry measures. 

Methods

Study Population

This cross-sectional study was on part of a cohort, the Singapore Cohort Study of Risk Factors for Myopia (SCORM), which examined 1979 children aged 7 to 9 years at baseline in three local schools in Singapore. The study methodology and details of the study population have been published. ^{29 30} Exclusion criteria included significant systemic illnesses and ocular conditions including media opacity, uveitis or a history of intraocular surgery, refractive surgery, glaucoma, or retinal disease. Two hundred seventy-one subjects from one participating school were systematically sampled for ORA measures during the 2007 visit. For the purposes of the study, all ocular measurements from the right eye were included in the analysis. 

All study procedures were performed in accordance with the tenets of the Declaration of Helsinki, as revised in 1989. Written informed consent was obtained from the parents of subjects, assent from the children, and the study was approved by the Institutional Review Board of the Singapore Eye Research Institute. 

The Ocular Response Analyser

The ORA is a noncontact tonometer with automated eye centration alignment. Subjects were seated on a chair and instructed to place their foreheads on the headrest of the ORA device that was made to match their height by adjusting the height of a table. To avoid startling the subjects, they were first briefed about a noncontact probe that would move toward the eye and emit a sudden but gentle puff of air. Subjects were told to focus on a blinking red light in the device. Thereafter, the ORA was activated and the air puff was emitted onto the center of the cornea. A typical applanation-pressure plot generated by the ORA (Fig. 1)for each eye shows two well-defined applanation peaks corresponding to inward and outward applanation. The intersection of a vertical line drawn from each applanation peak with the pressure curve defines two applanation pressures, the difference in magnitude of which is due to energy absorption during corneal deformation and forms the basis for the CH measurement. The ORA readings were obtained consecutively and only good-quality readings were analyzed, as defined by both the force-in and force-out applanation signal peaks on the ORA waveforms being symmetrical in height. The average of three readings was taken. The ORA software utilizes the CH to generate two additional parameters: the corneal-compensated IOP (IOPcc) and the corneal resistance factor (CRF). ⁷ A Goldmann-correlated IOP (IOPg) is also provided by the machine. No cycloplegic eyedrop or topical anesthetic was administered before the ORA measures, but a topical anesthetic was instilled before CCT measurement with the contact ultrasound pachymetry probe included with the ORA machine. The probe was placed perpendicular to the midpupillary axis, and the mean of three measurements was taken. Measurements were repeated two to three times for each eye. 

Other Study Procedures

After the ORA examinations, cycloplegic refraction was performed with an autokeratorefractometer (model RK5; Canon, Inc. Ltd., Tochigiken, Japan). Cycloplegia was achieved with three drops of 1% cyclopentolate 5 minutes apart. After an interval of at least 30 minutes after the third drop, five consecutive readings were obtained with one of two calibrated autokeratorefractometers. The mean corneal radius of curvature in each of the primary meridians was also recorded. AL measurements were performed with a contact ultrasound A-scan biometry machine (Echoscan model US-800, probe frequency of 10 mHz; Nidek Co., Ltd., Tokyo, Japan), with 1 drop of 0.5% proparacaine for topical anesthesia. The average of six measurements was taken if the SD was <0.12 mm. If the SD was ≥0.12 mm, measurements were repeated until the SD was <0.12 mm. 

Statistical Analyses and Definitions

For the purposes of the study, mild myopia was defined as between −0.5 and −3.0 D, moderate myopia was between −3.0 and −6.0 D, and high myopia was myopia of more than −6 D. Multivariable linear regression models were constructed with CH, CRF, or CCT as the dependent variable and the relevant predictive factors as covariates. All probabilities quoted are two-sided (Statistical Analysis System, ver. 8.0; SAS, Cary, NC). 

Results

Two hundred seventy-one eyes of 271 subjects were included in the study. The subjects’ mean age was 13.97 ± 0.89 years (range, 12–15), the distribution of the sexes was almost equal (138 boys, 50.9%), and the majority were Chinese (186 Chinese, 68.6%; 50 Malay, 18.5%; 33 Indian, 12.2%; and 2 other, 0.7%). 

Compared with children from the same school who did not undergo ORA measurements, the children included in the study were significantly older (mean age 13.96 ± 0.88 years vs. 13.80 ± 0.88 years; P = 0.03) and had a lower percentage of Chinese ethnicity (68.6% vs. 80.18%, P = 0.001). There were no significant differences in sex (percentage of boys 50.9% vs. 51.52%, P = 0.47), mean SE refraction (−2.35 ± 2.49 D vs. −2.68 ± 2.59 D, P = 0.12), or axial length (24.52 ± 1.15 mm vs. 24.63 ± 1.21 mm, P = 0.26). 

The mean (±SD) CH, CRF and CCT were 11.78 ± 1.55 mm Hg (range, 6.93–16.53), 11.81 ± 1.71 mm Hg (range, 7.83–16.83), and 578.67 ± 33.86 μm (range, 495.00–698.00). 

The Kolmogorov-Smirnov one-sample test was applied, and CH and CRF were fairly normally distributed (P = 0.73 and 0.31, respectively; Fig. 2 ). 

CH correlated significantly with CRF (Pearson correlation coefficient [r] = 0.86, P < 0.001), IOPcc (r = −0.44, P < 0.001), and CCT (r = 0.52, P < 0.001), whereas IOPcc did not correlate with CCT or CRF (r = −0.04, P = 0.53 and r = 0.09, P = 0.15, respectively; Table 1 ). 

CH and CRF did not vary significantly with age (P = 0.24 and 0.61, respectively) or sex (P = 0.21 and 0.08, respectively). There were no racial variations in CH or CRF either (P = 0.23 and 0.36, respectively). 

CCT did not vary with age, sex, or myopia status (P = 0.99, 0.31, and 0.85, respectively), but was found to be significantly affected by race. In the post hoc analysis with Bonferroni adjustment, Chinese subjects were found to have thicker CCT than did the Indian subjects. The CCT was highest in Chinese (584.1 ± 33.3 μm) compared with Malays (573.4 ± 32.6 μm), Indians (557.5 ± 30.5 μm), and other (558.0 ± 7.1 μm; P < 0.001). 

CH and CRF did not vary with myopia status (P = 0.79 and 0.83, respectively). CH and CRF did not correlate significantly with SE refraction (r = 0.04 and −0.03, P = 0.48 and 0.67, respectively) or AL (r = −0.11 and −0.05, P = 0.08 and 0.40, respectively). CCT did not vary with myopia status (P = 0.85) and did not correlate with SE refraction (r = −0.01, P = 0.82), but showed a weak correlation with AL (r = 0.14, P = 0.02). 

Three multiple linear regression models were constructed with CH, CRF, or CCT as the dependent variable and age, sex, race, IOPcc, CCT, SE refraction, and corneal curvature as covariates. For every 1-mm increase in radius of cornea curvature, the CH decreased by 1.28 mm Hg (P = <0.001). CH also correlated significantly with CCT (P < 0.001), after adjustment for all other factors (Table 2) . 

For every 1-mm Hg increase in IOP, the CRF increased by 0.30 mm Hg (P < 0.001), while every 1 μm increase in CCT was associated with an increase in CRF of 0.023 mm Hg (P < 0.001), and for every 1-mm increase in radius of corneal curvature, the CRF decreased by 1.08 mm Hg (P = <0.001; Table 3 ). The only factor that was predictive of decreased CCT was Malay or Indian race (P = 0.03 and <0.001, respectively). 

Discussion

The mean CH, CRF, and CCT were 11.78 mm Hg, 11.81 mm Hg, and 578.67 μm, respectively, in our study sample of normal Singaporean children. CH values in children have been reported in only one other study to date by Kirwan et al., ⁹ in which the ORA was used to assay corneal biomechanical properties in 81 normal eyes of 42 children. The age range of their sample was 4 to 18 years, which included children younger than those in our study, and the mean (±SD) CH was 12.5 ± 1.35 mm Hg. CH values reported in normal adults are generally lower. Shah et al. ²⁵ reported a mean CH of 10.7 ± 2.0 mm Hg in normal adults while Kirwan et al. ⁹ reported 10.8 ± 1.4 mm Hg. Most reports have confirmed that CH varies over a wide range in normal individuals, and the range of measurements we obtained of 6.93 to 16.53 mm Hg is comparable to the widest reported range of 6.1 to 17.6 mm Hg. ²⁵  

CH and CRF correlated negatively with corneal curvature in our study, with longer radii of curvature (flatter corneas) associated with lower CH and CRF. Corneal curvature in turn showed a moderate correlation with AL (r = 0.36; P < 0.001). Results in studies in which dynamic contour tonometry was used have suggested that corneal curvature affects corneal rigidity, with flatter corneas being less rigid, ^{31 32} and lower CH and CRF values are thus at least partially indicative of a less rigid corneal structure. 

Age and sex have been analyzed for their impact on CH and CRF, but little else is known about other physiological factors that may affect CH and CRF measurements. As CH and CRF correlated least moderately with CCT, factors that may influence CCT measurements may also impact on CH and CRF measurements. From the literature, the associations between CCT and age, sex, race, ^{19 20 21 22} corneal curvature, ^{23 24} refractive error, ³³ stature, ³⁴ and systemic vascular risk factors ³⁵ have been examined with various results. 

Pressure-volume modeling of ocular rigidity in eyes undergoing cataract surgery ³⁶ has shown a positive correlation between ocular rigidity and age. Experimental studies have also shown age-related changes in the cornea, including reduced interfibrillary spacing and increased collagen fibril cross-linking. ³⁷ Intuitively, the lower mean CH values reported in studies on adult populations compared with those in children suggests that age has a significant impact on CH. Kotecha et al. ¹⁰ reported that CH declines with age, although caution was advised in the interpretation of this result due to the possible confounding effects of ocular hypertension in their study. Kirwan et al., ⁹ however, reported that CH and age did not correlate in children ranging in age from 4 to 18 years, and Shah also did not find any correlation between CH and age in both normal adult subjects and those with keratoconus. ²⁵  

The association between CH and sex has been assessed in only one study to date, in which Shah et al. ²⁵ reported that CH did not vary with sex in either normal or keratoconic adult eyes, and our study likewise did not find any association. 

Race has a well-documented effect on CCT. ^{22 38} Black patients have thinner CCT than white or Asian patients, with some investigators attempting to correlate the CCT with racial or iris pigmentation, ³⁹ and our study also found that Indian subjects had thinner CCT. The results of one-way ANOVA showed that race did not have an effect on CH or CRF values. CCT is related embryologically to the keratocyte population, ⁴⁰ and the common origin of keratocytes and melanocytes from the cranial neural crest population may in some way relate to the racial variations in CCT. 

Changes in axial length associated with myopia are known to be associated with a variety of changes in corneal structure. Axial globe elongation is associated with flatter corneal curvature and decreased corneal thickness. ^{14 15} van Alphen ⁴¹ has theorized that corneal flattening in myopia occurs due to equatorial stretching with axial elongation, whereas Siegwart and Norton ⁴² have proposed alterations in the force vectors generated by the extraocular muscles as a possible mechanism. The biomechanical properties of the cornea and sclera may thus be linked, and the ectatic changes of staphyloma formation in pathologic myopia suggest that the cornea in myopes may be in some way less rigid and resistant to deformation. Conversely, a finite-element model constructed ⁴³ from stress-strain analyses of sections from various regions of the globe showed that the biomechanical characteristics of the anterior segment approximated those of the whole globe. Finite element modeling of scleral stretch in tree shrews has also shown that the sclera behaves as a viscoelastic material, and changes in the biomechanical properties of the sclera have been demonstrated in myopic tree shrew and chick eyes. Siegwart and Norton ⁴² investigated scleral creep, a time-dependent elongation response of the sclera to stretching that is qualitatively similar to hysteresis and found that scleral creep rates were up to 200% higher in samples from the sclera of myopic eyes. In animal models of myopia, scleral remodeling is known to occur, with alterations in the content, composition and orientation of both the collagen fibrils and the extracellular matrix, ^{16 17 44} and similar changes may occur in the cornea. Several matrix metalloproteinases have been identified as potential mediators of scleral remodeling in myopia, ⁴⁵ In an avian model of myopia, globe enlargement after retinal image degradation by goggling was associated with the upregulation of a common 23.5-kDa serine proteinase in both the cornea and sclera, ¹⁸ demonstrating a direct biochemical link between corneal remodeling and scleral changes with axial elongation. Our study results, however, indicate that CH and CRF are not correlated with AL or refractive error. The slower time profile of scleral creep experiments and myopic deformation contrasts with the brief loading-unloading cycle of the ORA, and this may account for the lack of association. 

Among the corneal parameters measured, CH correlated significantly with CRF, IOPcc, IOPg, and CCT, whereas CCT did not correlate with IOPcc. CH, CRF, and CCT are known to be measurements that correlate highly, ^{6 7} a finding that was confirmed by our study. Both CH and CRF have moderately strong positive correlations with CCT (r = 0.52 and 0.55, respectively). Other studies have shown similar correlations in both normal (r = 0.42) and keratoconic eyes (r = 0.45). ²⁵  

The strengths of our study design include in vivo assessments of corneal biomechanical properties in Asian children and the availability of cycloplegic refraction and biometry. The availability of a normal healthy population allows the documentation of associations between CRF and CH with other correlates in a population without eye diseases such as glaucoma. Limitations include nonparticipant bias as the participants were subtly different from nonparticipants. As the study subjects were all enrolled from a single school, the results of our study may also not be applicable to all children of the same age in Singapore. 

In conclusion, our study reports the correlates of corneal biomechanical properties measured with the ORA in normal Asian children. Flatter corneas are associated with lower CH and CRF readings. CH and CRF measurements do not vary with age, sex, or race and are also not dependent on refractive error or AL. Our study provides baseline data that should be the foundation for further studies on corneal biomechanical variations in various disease states. Correlating the CH and CRF with in vitro models of corneal rigidity and viscoelasticity will also allow for a better understanding of these novel corneal measurements, and further research is needed to define their clinical role. 

 Supported by National Medical Research Council (NMRC) Grant NMRC/0975/2005.

 Submitted for publication December 30, 2007; revised February 11, and April 2 and 7, 2008; accepted July 23, 2008.

 Disclosure: L. Lim, None; G. Gazzard, None; Y.-H. Chan, None; A. Fong, None; A. Kotecha, None; E.-L. Sim, None; D. Tan, None; L. Tong, None; S.-M. Saw, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 1.

 

Pressure-applanation plot generated by ocular response analyser.

Figure 2.

 

Distribution of (A) corneal hysteresis and (B) corneal resistance factor.

Table 1.

 

Correlations between CH, CRF, IOPcc, and CCT

	CH	CRF	IOPcc	CCT
CH
r	1.00	0.86	−0.44	0.52
P	—	<0.001	<0.001	<0.001
CRF
r	—	1.00	0.09	0.55
P	—	—	0.15	<0.001
IOPcc
r	—	—	1.00	−0.04
P	—	—	—	0.53
CCT
r	—	—	—	1.00
P	—	—	—	—

Table 2.

 

Multiple Linear Regression Model of the Factors Associated with CH

	Unadjusted (95% CI)	Multivariable-Adjusted* (95% CI)	P
Age (y)	−0.13 (−0.34 to 0.08)	−0.17 (−0.35 to 0.01)	0.06
Sex (girls)	0.24 (−0.13 to 0.61)	0.23 (−0.09 to 0.55)	0.16
Ethnicity
Chinese	Reference	Reference
Malay	−0.49 (−0.98 to −0.01)	−0.17 (−0.59 to 0.25)	0.42
Indian	0.05 (−0.53 to 0.62)	0.57 (0.06 to 1.07)	0.03
Others	−0.34 (−2.50 to 1.80)	0.43 (−1.38 to 2.23)	0.64
IOPcc (mm Hg)	0.09 (0.03 to 0.15)	0.02 (−0.05 to 0.06)	0.931
CCT (mm)	0.02 (0.02 to 0.03)	0.03 (0.02 to 0.03)	<0.001
Refractive error (SE in diopters)	0.03 (−0.05 to 0.10)	0.03 (−0.04 to 0.10)	0.44
Corneal curvature (mm)	−0.65 (−1.37 to 0.07)	−1.28 (−1.92 to −0.64)	<0.001

*  Adjusted for all other factors in this table.

Table 3.

 

Multiple Linear Regression Model of the Factors Associated with CRF

	Unadjusted (95% CI)	Multivariable-Adjusted* (95% CI)	P
Age (y)	0.01 (−0.22 to 0.24)	−0.15 (−0.29 to 0.01)	0.06
Sex (girls)	0.36 (−0.05 to 0.77)	0.20 (−0.07 to 0.47)	0.15
Ethnicity
Chinese	Reference	Reference
Malay	−0.46 (−1.00 to 0.08)	−0.14 (−0.50 to 0.21)	0.43
Indian	−0.27 (−0.91 to 0.37)	0.47 (0.04 to 0.90)	0.03
Others	−0.39 (−2.80 to 2.00)	0.39 (−1.14 to 1.93)	0.61
IOPcc (mm Hg)	0.38 (0.33 to 0.43)	0.30 (0.26 to 0.35)	<0.001
CCT (mm)	0.03 (0.02 to 0.03)	0.023 (0.019 to 0.027)	<0.001
Refractive error (SE in diopters)	−0.02 (−0.10 to 0.07)	0.02 (−0.03 to 0.08)	0.43
Corneal curvature (mm)	−0.90 (−1.70 to −0.11)	−1.08 (−1.63 to −0.54)	<0.001

*  Adjusted for all other variables in this table.