Objective Assessment of Corneal Backscattered Light in Myopic, Hyperopic, and Emmetropic Children

Purpose: To determine the corneal densitometry values by using Scheimpug imaging in myopic, and hyperopic children and to compare the results with emmetropic children. Methods: The corneal densitometry measurements of the subject were obtained with Pentacam Scheimpug tomography. The values were automatically measured in standardized grayscale units (GSU) over an area 12 mm in diameter, which was subdivided into 4 annular concentric zones (0-2 mm, 2-6 mm, 6-10 mm, 10-12 mm) and 3 corneal depths (anterior layer: anterior 120 µm; central layer: from 120 µm to the last 60 µm; posterior layer: last 60 µm). In addition, we evaluated the correlation between spherical equivalence and anterior corneal morphological parameters and the corneal densitometry values. Results: A total of 216 participants were included in this prospective cross-sectional study. The eyes were divided into three groups related to their spherical equivalent (SE) refractive error values as follows: 89 (41.2%) hyperopic eyes, 66 (30.6%) myopic eyes, and 61 (28.2%) emmetropic control eyes. The hyperopic eyes were found to have lower corneal densitometry values in 4 annular zones and the total diameter of all depths except the central layer. However, only the 6-10 mm annular zone of the central and posterior layers of the myopic eyes had lower corneal densitometry values than emmetropic eyes. There was also a signicant correlation between spherical equivalence and corneal densitometry values in some zones of the hyperopic eyes. Conclusion: Backward scattering of light was lower in hyperopic eyes and this could affect the visual quality.


Introduction
The cornea makes up the major part of the focusing power of the eye. The cornea and lens work together to focus the light entering the eye onto the retina. The cornea's refractive power depends on its smooth and stable curvature and its transparency [1].
Corneal transparency is related to the highly regular structure based on a uniform diameter of the collagen brils and the regular inter brillar distance [2,3]. Changes in this complex con guration results in an increased amount of light backscattering, reducing the amount of light reaching the retina and impairing the visual acuity [1]. Light scattering, diffraction, and wavefront aberrations are known to decrease the quality of the retinal image [4]. During a routine eye examination, the visual acuity measurements cannot re ect the effect of light backscattering on the visual quality [5]. However, a good level of vision may be dependent on the transparency and appropriate refractive shape of the cornea [6].
Corneal backscattering measurements are a re ection of corneal transparency. The Pentacam Scheimp ug system analysis can measure corneal densitometry values to enable quanti cation of corneal transparency. This non-invasive, rapid and reproducible optical system provides a map of the amount of backscattered light in various regions of the cornea. The system automatically gives quantitative values for corneal optical densitometry over an area 12 mm in diameter and analyzes the measurements in 4 annular zones and 3 layers.
Since corneal densitometry represents the health of the cornea, we therefore aimed to investigate whether these values were associated with refractive values and whether they could contribute to the optical quality of the cornea or enable balancing of refractive errors. To best of our knowledge, corneal densitometry values in children with various refractive errors have not previously been reported. Our aim in this study was to describe the corneal densitometry values in relation to the age and refractive errors in a group of children by using the Pentacam Scheimp ug system.

Materials And Methods
The 216 eyes of 216 children who had presented consecutively to our Ophthalmology Clinic for a routine eye examination at the Ulucanlar Eye and Training Research Hospital between January and June 2019 were included in this prospective cross-sectional study. All of the study procedures were conducted in accordance with the Declaration of Helsinki, and informed consent was obtained from the parents. The study was approved by the Local Ethics Committee.
All subjects underwent a comprehensive ophthalmologic examination including the best-corrected Snellen visual acuity, pupillary light re ex, slit-lamp biomicroscopy examination of the anterior and posterior segments without pupillary dilatation and using a 90-D lens, and intraocular pressure (IOP) measurement by non-contact tonometry.
Children aged 6 to 18 years were enrolled in the study. The groups were classi ed according to refractive error such as emmetropic, myopic, and hyperopic. Before enrolling the participants in the study, we performed a slit-lamp examination to make sure there was no corneal opacity or corneal epithelial problem. We excluded the cases with any ocular problems other than simple refractive errors resulting in the exclusion of those with ocular surface disease, corneal diseases, glaucoma, ocular surgery or trauma, cataract, uveitis, posterior segment abnormalities, and amblyopia. The patients who had any systemic diseases such as diabetes mellitus or autoimmune diseases were also not included in the present study.
The children who did not cooperate and were not able to xate with their eyes su ciently for good image quality were also excluded.
Corneal densitometry was evaluated by using the rotating Scheimp ug anterior segment analyzer (Pentacam HR, OCULUS, GmbH, Germany), which automatically provides quantitative values with the Pentacam software. The measurement consisted of 25 images (1003 X 520 pixels) over different meridians, with a uniform blue light source, taken under a uniform ambient light level of 4 lux in a windowless clinical room. The system automatically locates the corneal apex and analyzes an area with a diameter of 12 mm around it. The software algorithm shows the data in grayscale units (GSU). The units are graded as a minimum light scatter of 0 (maximum transparency) and a maximum light scatter of 100 (minimum transparency). An area 12 mm in diameter was divided into 4 annular concentric zones (0-2 mm, 2-6 mm, 6-10 mm, and 10-12 mm in diameter) centered at the apex of the cornea. In addition, the system provides the quantitative measurements of 3 different depth layers of the cornea that include the anterior layer (anterior 120 µm), central layer (from the rst 120 µm to the posterior 60 µm), and the posterior layer (posterior 60 µm). We also analyzed anterior and posterior corneal curvature, corneal thickness, corneal volume and white-to-white (WTW) diameter, and anterior chamber volume, depth and iridocorneal angle.
All examinations using the Scheimp ug system were performed by a single experienced technician, and in the same examination room. No contact ocular examination or pupil dilation was performed before this examination. In order to obtain a re ex-free image, all measurements were performed under standard dim-light conditions using the automatic-release mode to reduce examiner-dependent errors and during the same period of the day (between 9 and 12 am). We asked the participants to put their chin on the chin rest and touch their forehead to the forehead band. The eye was aligned with the visual axis using a central black xation target. A total of 25 single Scheimp ug images were obtained over approximately 2 seconds. Scans with a quality factor < 95% were excluded from the study. A single experienced grader (ES) who was blind to the groups reviewed all the Pentacam HR images. The cases with acceptable image quality were taken into account for statistical analysis.

Statistical analysis
Statistical analysis was performed using the SPSS software for Windows, version 22 (SPSS Inc., Chicago, IL, USA). The continuous variables were reported as mean ± standard deviation (SD) while the categorical variables were summarized with the use of frequencies. The normality of all data samples was checked with the Kolmogorov-Smirnov test. Only the right eye values were used for statistical purposes. The chisquare test was used in the analysis of categorical variables. Comparisons of the parametric values among the groups were performed with One-Way ANOVA. Comparisons of the nonparametric values among the groups were performed with the Kruskal-Wallis test. Tukey HSD and the Bonferroni-adjusted Mann-Whitney U-test were used as a post-hoc test for multiple comparisons between the groups. The Spearman rank test was used to investigate any relationship between both age and refractive errors and the anterior segment and corneal optical densitometry parameters. A two-tailed p value < 0.05 was considered signi cant.
Among the anterior segment parameters, the anterior, posterior and maximum mean corneal curvature values, the iridocorneal angle, and the anterior chamber volume and depth were statistically different between the groups (p < 0.001, p = 0.002, p = 0.007, p = 0.013, p < 0.001, and p < 0.001, respectively).
Conversely, the central and thinnest corneal thickness, the corneal volume, and the white-to-white corneal diameter were not statistically different (p = 0.290, p = 0.492, p = 0.258, p = 0.400, respectively). The differences between the groups are summarized in Table 2.
The densitometry of the anterior layer of the 0-2 mm concentric annular zone, and the posterior layer of the 0-2 mm and 2-6 mm concentric annular zones were statistically signi cantly lower in hyperopic group (the Bonferroni-adjusted Mann-Whitney U-test, p < 0.0166). Also, the anterior, posterior, and total layers of the total 0-12 mm diameter, and the total corneal layer of 10-12 mm diameter were statistically signi cantly lower in hyperopic group (the Bonferroni-adjusted Mann-Whitney U-test, p < 0.0166). The results of corneal densitometry values are summarized in Fig. 1 and Table 3.
Evaluation of the correlation between the anterior segment parameters and age revealed a signi cant positive correlation only between the age and anterior chamber depth (r = 0.218, p = 0.04) in hyperopic cases. When we evaluated the correlation between the corneal densitometry values and age, we only found a statistically signi cant positive correlation in emmetropic participants. The 0-2 mm and 2-6 mm zones for all depth layers were found to especially show a statistically signi cant positive correlation (p < 0.05). The correlation between corneal densitometry values and age was not statistically signi cant in myopic or hyperopic eyes. These parameters are summarized in Table 4.
We evaluated the correlation between the refractive errors and corneal morphological parameters in myopic and hyperopic patients. There was a signi cant negative correlation with maximal corneal curvature in myopic cases and with anterior chamber depth in hyperopic cases (p < 0.05) ( Table 5). In addition, there was a signi cant correlation between spherical equivalence and corneal densitometry, which was negative in the anterior and central 0-2 mm layer, and positive in the central 10-12 mm, and posterior 6-10 mm, 10-12 mm and total values in hyperopic eyes. The correlation values between spherical equivalence and corneal densitometry are summarized in table 6.
In addition, the white-to-white corneal diameter showed a statistically signi cant negative correlation in all annular zones except the 0-2 mm zone and the 2-6 mm zone in all three groups (data not shown, Spearman rank test, p < 0.05, r>-0.380 for emmetropic cases, r>-0.298 for myopic cases, and r >-0.297 for hyperopic cases).
We also investigated the correlation between the central corneal thickness and corneal optical densitometry values and found no correlation between these parameters in emmetropic cases. However, there was signi cant negative correlation in the 0-2 mm and 2-6 mm zones with different layers in myopic cases. Conversely, we found a signi cant positive correlation in all depth layers of the 6-10 mm annular zones in hyperopic patients. Table 7 shows the distribution of the correlation values between central corneal thickness and corneal optical densitometry values.

Discussion
Corneal densitometry is accepted to re ect both corneal clarity and health. Light scattering is minimal in the normal cornea thanks to its internal structure [7]. Increased corneal light backscattering has been reported in various corneal disorders such as bacterial keratitis, keratoconus, pseudoexfoliation syndrome, Fuchs endothelial dystrophy, and rheumatoid arthritis [7][8][9][10][11]. Light entering the cornea scatters either in a forward or backward direction. Backward light scattering results in a reduction in the amount of light reaching the retina [1]. Increased backward scattering decreases the quality of vision and degrades the retinal images [4]. We wanted to evaluate corneal light backscattering in healthy eyes with various refractive values by using corneal densitometry measurements. A literature survey did not reveal any other study that investigated the morphological structure of the cornea and optical densitometry values in the presence of various refractive errors in children. Our aim in the current study was to investigate whether a relationship was present between corneal densitometry values and refractive values as well as other ocular structure parameters and thus to identify whether these factors contributed to the optical status.
The main corneal scattering of light occurs at the interfaces between air and the tear lm, and between the tear lm and the cornea. The anterior super cial corneal epithelial cell layer and the posterior corneal endothelium are the major sources of light backscattering with the super cial epithelial layer making the highest contribution. The corneal stroma provides the cornea's transparency with the regular arrangement of collagen brils, the homogenous inter brillar distance, and the uniform diameter [1,12]. The posterior stroma with a more regular structural organization has more corneal transparency than the anterior stroma with a weaving structural pattern. In concordance with this, we found hyperopic cases had lower corneal optical density (COD) values in the 0-2 mm and 2-6 mm areas compared to both the myopic and emmetropic cases in the posterior layer, and lower densitometry throughout the total diameter was also seen in the anterior, posterior and total layers. This indicates a lower degree of scatter in the total diameter in all layers except the central corneal layer in hyperopia.
The zones where the hyperopic cases were not different than the myopic cases but had lower COD values than the emmetropic cases were the central 0-2 and 2-6 mm zones considering the total of these all corneas, this difference was signi cant in the 2-6 mm zone. These results indicated that the decisive corneal zones in terms of COD were the corneal apex (0-to 2-mm zone) and the pericentral cornea (2-6 mm zone) in eyes with refractive values. A small number of studies investigating corneal densitometry values in normal healthy corneas have reported a wide range of results in various ages without differentiating between refractive errors [5,[13][14][15]. Although comparison of the studies is di cult, Cankaya et al. found the lowest COD values to be in the total 0-12 mm total diameter of the posterior layer and the highest backscattering value to be in the anterior layer [5]. They reported the lowest COD value to be in the central 0-6 mm zone and to show a gradual signi cantly increase while advancing towards the limbus [5].
We observed higher COD values in the corneal apex, and pericentral cornea instead of the peripheral zones of 6-10 mm in all corneal layers for each group. However, the annular zone of 10-12 mm diameter in all layers for each group had the highest mean COD value. Furthermore, the total corneal layer of the 10-12 mm diameter was statistically signi cantly lower in the hyperopic group. Due to the nonhomogenous distribution of endothelial cell density in the cornea, the highest value is found in the 10-12 mm annulus [14], which means more corneal transparency because of the water pumping and barrier function of the endothelium. So, we can speculate that endothelial cell density in the 10-12 mm annulus has the highest value in hyperopic eyes. On the other hand, analysis of the 10-12 mm zone by using Scheimp ug devices had the lowest repeatability and reproducibility [15]. The assessment of COD values according to the position of the limbus and sclera during the measurement, especially in cases with corneas smaller than the 12-mm zones, can result in increased COD values in the peripheral zones. This could explain the negative statistically signi cant correlation between the WTW diameter and the COD values in all annular zones except the 0-2 mm and the 2-6 mm zones in all three groups in this study.
Evaluation of the correlation between the spherical equivalent values and COD values revealed a statistically signi cant correlation only in the hyperopic group. A signi cant but weak negative correlation was present between the spherical equivalent and corneal densitometry values in hyperopic cases, especially in the central 0-2 mm zone of the anterior, central and total layers. However, we saw that the COD value increased as the spherical equivalent value increased when the measurements were advanced towards the limbus and periphery (central layer of 10-12 mm) and towards deeper areas (posterior layer of 6-10 mm, 10-12 mm). This could result in decreased image quality on the retina in high hyperopic cases. In addition, the lower mean COD over the 12 mm diameter area in the anterior, posterior and total layers in only the hyperopic eyes suggested that this could have an effect on the balance of the formation of the optical qualities of the cornea in hyperopic children. In a study where high myopic values (> -6 dioptric (D) spherical refractive errors) were compared with age-and gender-matched < 5 D spherical cases, a high COD value was found in the 10-12 mm peripheral zone, while the 0-12 mm total diameter had lower corneal densitometry values in the other zones and in total in the high myopic group [13]. Although the groups we compared were different, we observed in our study that the myopic cases were not different than the hyperopic cases, but had lower COD values in the central and posterior 6-10 mm pericentral zone compared to emmetropic cases. However, high-diopter and pathologic cases were not included in our study.
We found a negative correlation between the central corneal thickness (CCT) and COD for the all depth layers of the central and pericentral annular zones in myopic cases in this study. Unlike our study, Dong et al. have found a positive correlation between the central corneal thickness and total corneal densitometry in highly myopic cases [13]. Cankaya et al. have found no correlation between CCT and corneal densitometry values [5]. We observed that the corneal densitometry values for all corneal depth layers increased only in the 6-10 mm zone as the CCT value increased, thus causing more corneal backscattering in hyperopic cases. This has raised the question of whether the COD values of the zones that are attened during refractive surgery have a possible effect on the resultant image quality as the mid-peripheral zone is targeted in hyperopic eyes while the central zone is targeted in myopic eyes. COD values increasing with age have been found in studies conducted with subjects at an advanced age where corneal densitometry values in various refractive errors were evaluated in healthy corneas [5,14,15]. Changes occurring in the structure of the cornea with age were presented as the reason for this result, due to the following changes: thickening of the epithelial basal membrane, reformation of the stromal collagen bers increasing the thickness of Descemet's membrane, and decreased endothelial cell function accompanied by increases in stromal hydration. A COD value increasing with age was only found in emmetropic cases in the age group of < 18 years in our study, but no correlation was found between the age and corneal densitometry values in myopic and hyperopic cases. The contrasting results in the literature may be due to the investigation of different age groups and different refractive errors. The emmetropization process seems to depend on the visual input being normal, and the lack of a normal input could disturb the process, resulting in refractive errors [16][17][18][19]. This made us consider that COD values may contribute to the development of refractive errors due to the lack of clear visual input.
The presence of COD values in various zones in myopic and hyperopic cases that are different than emmetropes may be balanced by various optical factors in eyes with different refractive errors. The main part of the emmetropization process is known to be completed within the rst 6 years of life [16]. Considering that the major part of ocular development had already been completed in our study that included pediatric patients above the age of 6 years, it would be appropriate to investigate with longitudinal studies how light backscattering in the 0-6 years age range can change during the emmetropization process.
The ocular biometric parameters including axial length (AL), anterior chamber depth (ACD), and lens thickness (LT) are known to be signi cantly correlated with refractive errors. These parameters have been found to be lower in hyperopic eyes than in myopic and emmetropic eyes [20]. Tomomatsu et al. have shown that the degree of hyperopia inversely correlates with the AL [21]. The increase in axial length during ocular growth in myopic eyes has been shown to be balanced by the decrease in the lens thickness [22]. Lee et al. have found the deepest ACDs in myopic eyes and the shallowest ACDs in hyperopic eyes in children [23]. Similarly, we observed ACD and anterior chamber volume values to be lowest in the hyperopic cases and highest in the myopic cases in our study, while both parameters were signi cantly different than in the emmetropic cases. Again similar to the same study [23], we encountered a shallower ACD in hyperopic eyes as the degree of hyperopia increased.
In conclusion, increased light backscattering results in the reduction of the amount of light reaching the retina and impairs the quality of vision. In the present study, the COD values were signi cantly lower in hyperopic eyes than in myopic and emmetropic eyes in children. The COD values also increased with age in emmetropic eyes. We believe that further and longitudinal studies are needed to support the signi cance of this result in clinical practice and its relevance for the optical system.

Declarations
Con icts of interest