Article Text
Abstract
Background/aims The authors estimated the association between corneal biomechanical properties and axial length in myopic children and verified the relationship between axial length and intraocular pressure (IOP).
Methods This cross-sectional study included 126 eyes of 63 children, aged 12.02±3.19 years. Corneal hysteresis (CH), corneal resistance factor (CRF), central corneal thickness (CCT) and IOP were recorded with the Reichert Ocular Response Analyzer. Axial length (AL) and anterior chamber depth (ACD) were measured by the IOLMaster, whereas corneal curvature was measured by an autorefractor.
Results In multivariable mixed model analysis, CH correlated significantly with AL (p<0.0001), ACD (p=0.044) and CCT (p<0.0001), but not with curvature; CRF correlated significantly with AL (p=0.004) and CCT (p<0.0001) but not ACD or curvature. The difference in CH between the two eyes of each patient correlated significantly with the difference in the AL between the two eyes (p=0.002). IOP did not correlate with age, gender or AL.
Conclusion Lower CH and CRF were associated with longer AL but not corneal curvature. The association between lower CH and deeper ACD demonstrated the importance of CH as a determinant of ocular biometry in both the anterior and posterior segments. Differences in corneal biomechanical properties may indicate more generalised structural differences between eyes.
- Childhood myopia
- corneal hysteresis
- corneal biomechanical properties
- cornea
- optics and refraction
- anatomy
- physiology
- imaging
Statistics from Altmetric.com
- Childhood myopia
- corneal hysteresis
- corneal biomechanical properties
- cornea
- optics and refraction
- anatomy
- physiology
- imaging
Introduction
The worldwide prevalence of myopia is quite high, reaching 80% in certain Asian populations.1 The well-documented myopic changes include elongated axial length (AL), deeper anterior chamber and vitreous depth,2 3 thinner retina and higher incidence of retinal detachment,4 and decreased scleral thickness and elasticity.5 Animal studies demonstrated that myopia development is associated with reduction in scleral thickness and extracellular matrix,6 increase in the collagen-degrading enzyme matrix metalloproteinase 25 and scleral extensibility.
The cornea and sclera together form the outer fibrous tunic of the eye and withstand both internal and external forces to maintain the eyeball shape. Although their optical properties differ, both cornea and sclera consist mainly of collagen fibres. Myopia-related corneal alterations include flatter corneal curvature, decreased corneal thickness, and decreased endothelial density.7 However, the biomechanical properties of myopic eyes have been less well studied.8 9 The biomechanical changes in collagens might precede changes in the extensibility of the sclera and ultimately lead to myopia development.5 Whether a change in the biomechanical properties leads to AL elongation and anterior chamber deepening, which subsequently result in the decreased scleral/corneal thickness and elasticity, remains controversial.5 7 8
Corneal hysteresis (CH) can be measured in vivo by the Ocular Response Analyzer (ORA; Reichert, New York) using a patented dynamic bidirectional applanation process. CH has been shown to reflect altered corneal characteristics in keratoconus, Fuchs dystrophy and following corneal surgery.10 Differences in CH have also been demonstrated to represent ocular properties not directly related to the cornea, such as different susceptibility to glaucomatous optic nerve damage.11–13
Lower CH has also been shown to be associated with longer AL in 2024 eyes of 1233 Chinese children.8 In contrast, Lim et al showed no correlation between CH and the AL in 271 right eyes in 271 children of a mixed ethnics population.9 As genetics could influence CH,14 and the mean CH among rural Chinese children was significantly lower than that previously reported for 42 normal Irish children,8 12 we thus proposed that comparing the difference in AL with the difference in CH in the same subjects could thus provide important information by excluding genetic influence, ethnic and/or other unknown factors that might affect CH.
Elevated intraocular pressure (IOP) is hypothesised to impose scleral stress and creep,15 resulting in axial elongation with scleral stretch. Myopia has been shown to be associated with increased IOP,16 but their relationship remains inconclusive.17 IOP readings from ORA correlate well with Goldmann applanation tonometry,18 whereas corneal-compensated IOP (IOPcc) seems to be independent of corneal thickness in non-glaucoma patients.18 19 It would thus be valuable to verify the relationship between the IOP and myopia using ORA.
Our primary purpose was to investigate the corneal biomechanical properties in children with myopia. The secondary aim was to investigate the association between AL and CH in different subjects and between the two eyes of individual patients. Our final purpose was to verify the relationship between AL and IOP.
Materials and methods
This study was approved by the institutional review board of the Far Eastern Memorial Hospital (FEMH No 97006). We included only Taiwanese (predominantly Han Chinese) children in our study. Patients younger than 18 years of age presenting for complete eye examination and myopic evaluation were invited to participate. In addition to obtaining the assent from the children themselves, informed consent was obtained from their parents. Thorough ocular examinations including visual-acuity assessment and anterior segment and fundus examinations were performed to exclude cases with ocular pathology other than refractive errors and myopia-associated chorioretinal changes. All subjects had a best-corrected visual acuity of 20/20 in both eyes. Exclusion criteria were (1) strabismus and/or amblyopia, (2) congenital or acquired glaucoma, (3) uveitis or retinal disease and (4) inability to cooperate for reliable ORA readings (as defined below).
After administration of one drop of 0.5% tropicamide every 5 min for three times, cycloplegic refraction was measured with an autorefractor (Topcon RK-8000). At least five readings had to be 0.25 dioptres (D) or less apart in both spherical and cylindrical components of each eye before they were accepted. Measurements of ocular biometric parameters including AL and anterior chamber depth (ACD) were performed with IOLMaster V4.8 (Zeiss, Oberkochen, Germany). Measurements were regarded as valid when the signal-to-noise ratio was greater than 2.0. Four acceptable measurements were considered for each eye, and the mean of the four readings was regarded as the representative measurement.
The ORA determines corneal biomechanical properties using an applied force–displacement relationship; details of its operation have been described.10 20 The ORA was used to obtain four measurements in each eye of every patient and the mean of these four readings was used in the analysis according to the manufacturer's instructions. We examined the raw signals of each reading to confirm reliable measurements. A reliable ORA reading was defined as (1) a clean signal at peak; (2) a fairly smooth signal at a descending curve, trough and ascending curve; and (3) a flat baseline signal and nearly the same amplitude on both sides, which means the cornea should return to its original state/pressure after the dynamic change. The ORA provides four variables including CH, corneal resistance factor (CRF), Goldmann-correlated IOP (IOPg) and IOPcc. ORA measurements were performed before pachymetry to eliminate the possible effect that applanation might have on the CH and IOP values. After the application of topical anaesthesia with 0.5% Proparacaine Hydrochloride (Alcaine, Alcon-Couvreur, Belgium), ultrasound pachymetry (20 MHz) was performed with the ORA integrated handheld pachymeter.
Data analysis
The spherical equivalent, in dioptres, was calculated by the spherical refractive error plus 0.5×cylindrical refractive error. We used the mean of five acceptable measurements, that is 0.25 D or less apart in spherical and in cylinder components as mentioned above, in each eye as our final spherical equivalent value.
We combined data for both eyes using the mixed model method,21 which adjusts for the correlation between the two eyes in a single person. A univariate mixed model analysis was used to determine the relationship of two continuous variables (eg, age and CH). The Pearson correlation was used to examine if the difference in CH between the two eyes of each patient correlated significantly with the difference in AL between the two eyes.
In order to retestify all variables in a univariate analysis, we constructed a mixed linear model, with CH or CRF as dependent variables, to evaluate the possible correlation of factors including age, gender, central corneal thickness (CCT), AL or ACD, and corneal curvature. To diminish the collinearity problem, spherical equivalent was not included in multivariate models because of its high correlation with AL. Because ACD is part of AL, the two independent variables were analysed in separate models (models 1 and 2). We also constructed a mixed linear model, with IOPcc or IOPg as the dependent variable, to evaluate the possible correlation of age, gender and AL. In the model with IOPcc as the dependent variable, CCT and CH were not included because of their high correlation with IOPcc. In all analyses, p values less than 0.05 were taken to indicate statistical significance. Statistical analyses were conducted using SPSS (version 13.0; SPSS, Chicago, Illinois).
Results
A total of 156 eyes of 78 children were originally included in the study. AL measurement by IOLMaster was successful in all children. ORA measurements failed in 20 eyes (12.8%) because of difficulty in steady fixation during measurement. Ultrasound pachymetry of CCT was not possible in 10 eyes (6.4%) because of a lack of patient cooperation. Complete data collection was eventually obtained for 126 eyes of 63 patients.
The mean age of the subjects was 12.02±3.19 years (range 7–18). Thirty-seven (58.7%) patients were boys. Table 1 summarises the patients' refractive status, biometric findings, CCT and ORA readings including CH, CRF, IOPcc and IOPg.
Results of univariate mixed model analysis of the association between ORA measurements and other clinical variables are shown in table 2. CH correlated with spherical equivalent, ACD, AL, CCT and IOPcc in a univariate analysis. In eyes with a lower CH, the cornea was thinner (b=0.015, p<0.001), the ACD was deeper (b=−1.429, p=0.012), and the AL was longer (b=−0.510, p<0.001), where b represented the estimated regression coefficient. AL correlated well with spherical equivalent (b=−0.35, p<0.001) and ACD (b=1.22, p<0.001) but not with CCT (b=0.005, p=0.175). The difference in CH between the two eyes of each patient correlated significantly with the difference in the AL between the two eyes (r=−0.410, p=0.002) (figure 1) but not with the difference in ACD (r=−0.135, p=0.341). Lower CRF was also associated with thinner cornea (p<0.0001; table 2). However, CRF did not correlate with AL and ACD in the univariate analysis, and the difference in CRF between the two eyes did not correlate with the difference in AL and ACD (p=0.307 and p=0.653, respectively).
In a multivariate mixed model analysis, a lower CH and CRF were significantly correlated with a thinner CCT (p<0.001 and p<0.001, respectively) and longer AL (p<0.001 and p=0.004, respectively). For example, a decrease of one unit of CH corresponded to the decrease of 0.019 (regression coefficient) unit of CCT and increase of 0.573 unit of AL (model 1, table 3). However, a lower CH but not CRF was significantly correlated with a deeper ACD (table 3) in a multivariate analysis (b=−1.168, p=0.044 and b=−0.343, p=0.627 for CH and CRF respectively). IOPcc was not significantly correlated with age, gender, or AL (table 4), whereas IOPg was significantly correlated with CCT (p=0.003) and CH (p=0.021) in a multivariate analysis.
Discussion
In our study, a multivariate mixed model analysis showed that a lower CH was associated with a longer AL and deeper ACD. However, multivariate analysis results alone could not exclude the confounding factor of genetic influence14 on CH value, which might either strengthen or weaken its correlation with AL. Our study further found that the difference in CH between the two eyes of each patient correlated significantly with the difference in AL between the two eyes (figure 1). This provided further evidence of the association between CH and AL by excluding the effect of genetic or other unknown factors that might have affected CH.8 14 However, this study could not answer whether the asymmetrical CH is the cause or the result of anisometropia.
Nineteen per cent subjects in our study had anisometropia of more than 1.5 D, and this hospital-based study may not stand for the normal population. This could be accounted for by the tendency that more parents of anisometropic children volunteered to join this study as designed in the research protocol, which intended to investigate the association between CH and AL by minimising the effect of individual difference from person to person.8 14
Flatter cornea curvature did not correlate with a lower CH in our series. This differed from Lim's report,8 which indicated that subjects with lower CH were more likely to have a flatter cornea. While the association between CH and AL was insignificant in Lim's study from a univariate analysis, it was significant in our study from a multivariate analysis. Meanwhile, we used a mixed model analysis. Because the parameters in the two eyes of an individual are considered highly correlated, classical statistical methods such as t tests and regression models are not valid for analysing these types of correlated data. To avoid the correlation problem, some studies were performed using only one eye for each patient.9 However, excluding some of the data may potentially lose information crucial for detecting associations between factors. In our study, linear mixed models were thus adopted to extract sufficient information from each patient. Moreover, the Lim study included a mixed ethnicity of Chinese, Malay, Indian and others, whereas only Taiwanese (predominantly Han Chinese) were included in our study. As such, our study probably circumvented any potential ethnic confounding factor,8 9 and analysing the difference between the two eyes in each subject minimised confounding effects related to heredity and environment.14 Additionally, utilising the non-contact IOLMaster to measure AL avoided the possible indentation-related inaccuracy caused by contact A-scan ultrasound biometry used in their study;8 9 22 this aspect could be particularly important for children, as we were unable to obtain pachymetry readings in 6.4% of our initial study population.
In our study, the reduced CH in myopic eyes might represent a primary alteration of the corneal and/or collagen properties of these eyes, suggesting that myopic eyes are more extensible and less resistant to the expansive force of the normal IOP, thus leading to increases in AL. This notion is supported by the demonstration of lower CH in keratoconus with disorganisation of stromal collagen lamellae.10 This reduced mechanical strength and manifested decreased CH in keratoconus is closely related to the upregulation of degradative enzymes and the associated tissue degradation.23 The increased sclera collagen-degrading enzyme, matrix metalloproteinase 2, is associated with thinner and structurally weakened sclera24 and increased scleral extensibility25 in myopia. We suggest that alteration in matrix metalloproteinase in the collagen-containing cornea and sclera might account for the decreased CH in myopia. In addition, a reduced CH has also been noted in both adults and children with glaucoma,12 13 suggesting that thinner lamina cribrosa could be associated with and detected by changes in corneal biomechanics. We currently are collecting additional longitudinal data to document the temporal relationship of the onset of the difference in CH and in AL to determine if the reduced CH is the result or the cause of axial elongation.
Both CH and CRF are influenced by viscoelastic properties, because they are both linear combinations of P1 (inward applanation pressure) and P2 (outward applanation pressure) signals in ORA measurement.10 20 CH describes the damping nature of the cornea (eg, collagen structure, hydration state) and represents the corneal dynamic resistance component; on the other hand, CRF emphasises inward applanation pressure and is weighted more heavily by elasticity.20 Nevertheless, the differences between CH and CRF are not fully understood currently. In the present study, CRF was significantly correlated with CCT but not with AL using univariate analysis (table 2). This suggested that CRF, as the parameter originally designed for maximal correlation with corneal thickness, is associated with corneal properties. Nevertheless, taking CCT as a covariate in a multivariate analysis (table 3), a lower CRF also correlated with a longer AL, as did CH. Whether the reduced CRF is the result or the cause of axial elongation still needs further longitudinal study to answer this question.
Our result of the significant association between lower CH and longer AL is similar to that reported by Song et al,8 but our study differed in four aspects. First, we noted that not only AL but also ACD was negatively associated with CH. This illustrates the importance of CH as a determinant of the ocular biometry in both the anterior and posterior segments. Second, we found a negative association between CRF and AL, demonstrating that not only the dynamic CH but also the static CRF plays a role in the biometric alteration during myopia development. Third, we noted that the difference in CH correlated significantly with the difference in AL between the two eyes of each patient. This suggests a possible role for CH/CRF measurement in the prediction of biometric alteration. Fourth, we used IOLMaster instead of ultrasound for better biometric accuracy.22
The mean IOPcc of 16.37±3.54 mm Hg and IOPg of 16.63±3.43 mm Hg in our study were similar to IOP measurements by ORA (IOPcc of 16.1±2.9 mm Hg and IOPg of 16.2±3.2 mm Hg) in normal Chinese adults.19 Our IOP measurement was similar to Song's results, 17.0±3.36 among 2024 eyes.8 The lack of a chin rest and the inability to monitor eye position during ORA measurement might have contributed to the relatively poor head and eye stability. These might contribute to the relatively large variation of IOP measurement, especially for measurements in children. In our multivariate analysis, IOPg was significantly associated with CCT, whereas IOPcc did not correlate with CCT. We did not find a significant correlation between the IOP and AL in the multivariate mixed model analysis, thus further reinforcing the notion that the differences in refractive error and AL in myopic eyes seem more likely attributable to genetically determined discrepancies in scleral structure and less related to differences in IOP.
Our cross-sectional study was limited to clinic-based patients. Nevertheless, we provide evidence that a lower CH in myopic children is associated with a longer AL within an individual. Our study has minimised the confounding effects of heredity and environment. It is possible that differences in corneal biomechanical properties may indicate more generalised structural differences between eyes. Whether the arrangement of myopic corneal lamellae or the weaker support of myopic sclera is related to the mechanical performance demonstrated in this study remains to be established. Further longitudinal studies on the change in hysteresis may provide further information on the progression of myopia.
References
Footnotes
Funding This project was supported by Far Eastern Memorial Hospital grant: FEMH-97-C-038.
Competing interests None.
Patient consent Obtained from the parents.
Ethics approval Ethics approval was provided by the institutional review board of Far Eastern Memorial Hospital.
Provenance and peer review Not commissioned; externally peer reviewed.