Early corneal and optic nerve changes in a paediatric population affected by obstructive sleep apnea syndrome

The relation between OSAS and eye diseases is well known in adults, while very few and contradictory data can be found regarding paediatric ages. The aim of this study is to explore the early corneal, macular and optic nerve changes in paediatric patients with OSAS. Prospective study that enrolled children aged ≥ 4 years referred to the Paediatric Pneumology Clinic in Verona for suspected obstructive sleep apnoea syndrome (OSAS) and investigated with the overnight respiratory polygraphy. Patients with apnoea–hypopnea index (AHI) > 1 were classified as OSAS, while those with AHI < 1 were classified non-OSAS. All patients underwent comprehensive eye examination including slit lamp, refraction, intraocular pression (Goldman applanation tonometry), corneal tomography (corneal astigmatism, corneal keratometry at the apex, surface asymmetry index, central corneal thickness and thinnest corneal thickness) and optical coherence tomography (central macular thickness, macular volume and retinal nerve fibre layer). Seventy-two children were enrolled in the study. The overall prevalence of OSAS was 48.6%. Statistically significant differences were found between OSAS and non-OSAS group for corneal asymmetry (0.9 ± 0.5 and 0.6 ± 0.3, respectively; p = 0.02), thinnest corneal thickness (551.8 ± 33.9 and 563.7 ± 32.5; p = 0.04), average retinal nerve fibre layer (102.8 ± 10.5 µm and 98.1 ± 12.3 µm; p = 0.012) and in nasal quadrant (76.2 ± 15.4 µm and 66.5 ± 12.6 µm; p = 0.0002). A comprehensive eye examination with corneal and optic nerve imaging showed early corneal and optic nerve changes in children newly diagnosed with OSAS. These could be prelude of the known ocular manifestations associated with OSAS in adult patients.


Introduction
Obstructive sleep apnoea syndrome (OSAS) is a respiratory disorder characterized by repetitive collapse of the upper airway during sleep [1,2] reported in 1-3% of the general paediatric population [3]. The most frequent causes of OSAS are adenotonsillar hypertrophy and regional deposition of fat in the neck [3][4][5][6]. Anyway, the result of airway collapse is the reduction in the inspiratory flow leading to hypoxemia and hypercapnia which cause injury of blood vessels and organ involvement, including the eyes [6,7].
The association between OSAS and eye diseases is well known in adults. Several studies have linked OSAS with glaucoma, keratoconus, papilloedema, non-arteritic ischaemic optic neuropathy, age-related maculopathy and floppy eyelid syndrome [8,9], while very few and contradictory data can be found regarding paediatric ages [10,11].
The aim of this study is to explore the early corneal, macular and optic nerve changes in paediatric patients with OSAS.

Methods
This is a single-centre prospective study conducted in paediatric patients investigated for OSAS (ICD-9-CM 307.4x, 327.23, and 780.5x) by the overnight respiratory polygraphy (RP) between June and December 2020 at the Paediatric Pneumology Clinic in Verona.
The study was performed in accordance with the tenets of the Declaration of Helsinki. Approval of the local Institutional Ethics Committee was obtained.
Patients aged C 4 years, whose parents agreed to participate in the study by signing informed consent, were included. Preterm born patients and patients with comorbidities (pulmonary, cardiac, facial dysmorphism and other) were excluded from the study. Patients affected by congenital or acquired ocular abnormalities not related to OSAS (congenital cataract or glaucoma, megalocornea, sclerocornea, corneal or optic nerve lesions, previous ocular surgeries, amblyopia, trauma, etc.) were excluded.
Children underwent in-laboratory overnight RP in a dedicated room at the Paediatric Pneumology Clinic and performed a comprehensive ophthalmic examination with corneal and optic nerve imaging at the Eye Clinic of Verona University.
The collected medical clinical data included age, sex and body growth parameters. Trained personnel using standardized techniques measured height and weight. Body mass index [BMI, weight (kg)/height (m 2 )], BMI percentiles and BMI z-scores were calculated according to age and sex following the calculator tool for children aged 2 through 19 years old provided by the Centres for Disease Control and Prevention [12].
RP was performed using a portable ambulatory device (SOMNOscreen TM PSG, SOMNOmedics GmbH, Randersacker, Germany) following the indications of 2007 American Academy of Sleep Medicine (AASM) manual for the Scoring of Sleep and Associated Events [13].
The analysis of the entire recording session was performed automatically (DOMINO software, SOM-NOmedics v.2.6.0) and manually checked for artefactual or non-interpretable periods of nasal flow, thoracic effort, abdominal effort or oximetry channel. Periods with artificial data were withdrawn by the estimated total sleep time (TST), and movement periods were excluded. TST was calculated according to published criteria [13].
Respiratory events were scored accordingly to the AASM manual [13]. The number of apnoeas (apnoea for the duration of at least two breaths during baseline breathing associated with the presence of respiratory effort throughout the entire period of absent airflow) and hypopneas (reduction in airflow [ 50% for 10 s or greater, with a decrease in oxygen saturation of C 3%) was divided by the hours of TST and expressed as apnoea-hypopnea index (AHI).
Desaturation was defined a drop of C 3% of oxygen saturation. The oxygen desaturation index (ODI) was calculated as the total number of desaturations per hour divided by the TST.
Patients with AHI [ 1 were classified as OSAS, while those with AHI \ 1 were classified as non-OSAS.
Dilated fundus examination was performed with indirect ophthalmoscopy. The central macular thickness (CMT), macular volume (MV) and peripapillary retinal nerve fibre layer (RNFL) thickness were assessed by spectral domain optical coherence tomography (SD-OCT Spectralis, Heidelberg Engineering, Heidelberg, Germany; software version 6.0), adjusted for corneal curvature, fixation and patient positioning for patient age and level of cooperation.
The CMT, MV and RNFL values were generated automatically by the instrument, and scans were taken following the sequence published by Turk et al. [14]. All instrumental acquisitions have been performed by two trained examinators who did not know the overnight RP result. All accepted SD-OCT images showed Q-score C 20, and those of poor quality or with segmentation errors were excluded from this analysis. For corneal tomography, only measurements with quality rate higher than 70% were accepted and examinations were repeated until the achievement of this target.

Statistical analyses
Both eyes were examined and included in the analysis.
The prevalence for stratified clinical and demographic characteristics and the calculated confidence interval through the normal approximation interval were evaluated. For the quantitative variables, results of descriptive analyses were expressed as mean and standard deviation in the case of normal distribution (according to the Kolmogorov-Smirnov normality test), median and interquartile range (IQR) otherwise and as a count and percentage for categorical variables.
In dummy variables (sex, ODI), the chi-square test was used to assess differences between groups and the t test was used to assess differences between continuous variables (BMI, age). Statistical analyses were performed using the STATA 13.0 statistical package (Stata Corp. LP, College Station, TX). The significance level was set at 0.05.
The overall prevalence of OSAS was 48.61%, and in the 57.0% of the cases its severity was mild (Table 1). There were no differences in age, BMI and sex between OSAS and non-OSAS group (Table 1).
Family history of eye disease was positive in one child for glaucoma and in another one for keratoconus and high myopia. Both patients were classified in OSAS group. No differences were found between OSAS and non-OSAS groups regarding visual acuity (all children achieved 0.0 logMar) and SE ( -0.08 ± 2.76 D and -0.03 ± 2.50 D, respectively, p = 0.54). None required lenses modifications. No corneal or lens opacities and no ophthalmoscopic retina or optic nerve abnormalities were found.
Corneal tomography examination showed no difference for cyl (p = 0.89), AK (p = 0.42) and CCT (p = 0.19) between groups. However, significant differences were found for SAI (p = 0.02) and TCT (p = 0.04) ( Table 2). One child with moderate OSAS was diagnosed with early corneal ectasia and was referred to the cornea service for followup/management. All children showed IOP in normal range (12 ± 0.54 mmHg), without difference between groups (p = 0.73).
The analysis of the RNFL thickness in four segments (superior, nasal, temporal and inferior) showed higher values in the OSAS group in all quadrants; however, statistically significant difference was found in the nasal one (76.20 ± 15.40 lm and 66.50 ± 12.60 lm) (p = 0.00) and in the average retinal nerve fibre layer (102.80 ± 10.50 lm and 98.10 ± 12.30 lm, p = 0.012) ( Table 3).

Discussion
We examined children assessed for OSAS by overnight RP, to investigate for early signs of ocular changes related to airway obstruction during sleep, poorly documented and understood in the previous literature [10,11].
Retina and optic nerve have been examined in OSAS paediatric patients with adenotonsillar hypertrophy with controversial results. Cinici and Tatar [10] found no significant alterations of RNFL thickness, while Simsek et al. [11] showed that OSAS seems to influence RNFL and IOP parameters. Both studies highlighted the potential risk of long-term optic injury due to blood intermittent hypoxia and recommend prompt treatment of OSAS in young people. None evaluated corneal alterations in OSAS paediatric patients.
The major limitation of these studies was the assessment of OSAS diagnosis by questionnaire, contrasting with AAMS recommendations [15]. The questionnaires, indeed, are sensitive just for severe grades of apnoea, so they are not suitable for children due to low sensibility.
Although the gold standard for diagnosis of OSAS is the polysomnography, which is an expensive test, the set-up can be challenging, and the placement of numerous electrodes can be difficult in children.
In our study, we investigated the OSAS using RP in a dedicate room, in accordance with J. Corral et al. who found that RP management was similarly effective to polysomnography, with a substantially lower cost [16].
None of the children enrolled in the study had visual impairment, and all reached 0.00 logMar, with or without optical correction, without anomalies during ophthalmological examination.
However, although corneal parameters were within normal range in both OSAS and non-OSAS group, statistically significant differences were found for SAI and TCT values (p = 0.02 and p = 0.04, respectively).
SAI is the index of asymmetry of the corneal surface and is based on the specular correspondence between the two hemicorneas (superior and inferior), which normally does not exceed small physiological differences (\ 0.5). Values greater than 1 result in a notable corneal asymmetry which is suggestive for ectasia [17,18].
Our results suggest that OSAS children have a higher corneal asymmetry than non-OSAS. This finding, added to the detection of a lower TCT, leads to suspect a possible evolution towards ectasias. In fact, we found one OSAS patient with suspect corneal ectasia. This would be in keeping with the previous literature on the correlation between OSAS and keratoconus in adult patients [8,9]. The SAI values within the normal range, found in this study, could be explained by the shorter period of exposure to OSAS in children patients compared to adult ones. Indeed, the latter have generally several years of exposure to OSAS when a diagnosis of keratoconus is made. Therefore, a possible evolution towards ectasia in young patients should be suspected.
Similar behaviour has been found in RNFL parameters, whose values were found within the normal range in all patients [14]. However, the OSAS group showed higher RNFL thickness in all quadrants, reaching statistically significant difference in the average values and in the nasal quadrant values.
The relation between OSAS and optic injury is well known in middle-aged adults and the elderly, and most of the studies reported a reduction in RNFL thickness [8,9]. Recently, Lee et al. confirmed this relation also in young OSAS patients (aged 19-22 years) reporting RNFL thinning in inferotemporal and superotemporal segments [19].
This study shows that a correlation between OSAS and optic nerve fibres change is already detectable also in very young patients, but not in terms of a reduced RNFL thickness. On the opposite, our data showed increased RNFL in OSAS patients. These changes could reflect a mild swelling of the optic nerve fibres secondary to capillaries dilatation, due to hypoxic condition, or to a possibly increased intracranial pressure, which are common conditions in OSAS patients [20][21][22]. The peripapillary and retinal capillary tone, indeed, is under local vascular control, and it responds to O 2 and CO 2 concentration in the blood. Moreover, Jaki Mekjavic et al. [23] described in patients under experimental condition that hypoxia and acute hypercapnia induce dilation of retinal vessels and radial peripapillary capillary network, causing the increasing of RNFL thickness, as compensatory response.
As is known, in adults with moderate-severe OSAS, RNFL thickness is significantly reduced compared to healthy subjects and is associated with macular retinal thinning due to ganglion cells loss [24]. However, other authors have already noted that probably the optic nerve head changes through the stage of subtle oedema in the mild and moderate OSAS groups and moves to a secondary phase of reduction in RNFL thickness in patients with longstanding severe OSAS [20].
The fact that in our study a mild increase in RNFL thickness was not associated with any visual disturbances and CMT and RV appeared within normal limits could be a sign of very early and potentially reversible changes.
Most of our patients were newly diagnosed with a mild stage of OSAS, with a disease duration likely to be relatively short due to the young age of the patients. We cannot exclude that, in case OSAS is left untreated and progress with time to more severe disease stages, optic nerve head changes could evolve over the years to chronic nerve fibres distress with a reduction in RNFL thickness.
In this study, OSAS children show higher RNFL thickness in all quadrants, reaching statistically significant difference in the average and in the nasal one. Furthermore, the significant increase in RNFL only in the nasal sector could be explained by the swelling disc scheme proposed by Frisen [25], which showed that oedema begins precisely in the nasal part of the disc.
In our study, the correlation between OSAS and increased RNFL thickness could represent an early modification of the peripapillary nerve fibre layers, due to apnoeas and desaturation events. A limitation of this study is the lack of measurement of retinal ganglion cell layer values.
No significant differences were found in CMT and MV; however, higher values were found in the OSAS group. These findings could be due to the vascular response of retinal capillary network in OSAS group.
Further investigations with larger populations in paediatric patients with OSAS are needed to confirm our hypotheses. This is the first study that found a correlation between OSAS and corneal alterations in childhood, contributing to the knowledge of the ocular implications related to OSAS.
Ocular damages in children are completely asymptomatic, and these early signs could give insight into pathogenesis of corneal and optic disc changes that appear clinically evident in the adult population. So, based on our evidence, we suggest to plan earlier ophthalmological examination including evaluation of corneal asymmetry and RNFL thickness in paediatric patients affected by OSAS.