The Associations of Obstructive Sleep Apnea and Eye Disorders: Potential Insights into Pathogenesis and Treatment

Obstructive sleep apnea (OSA) patients are at significantly increased risks for cardiovascular and cerebrovascular morbidities. Recently, there has been heightened interest in the association of OSA with numerous ocular diseases and possible improvement of these conditions with the initiation of OSA treatment. We reviewed the current evidence with an emphasis on the overlapping pathogeneses of both diseases. Currently available literature points to a substantial association of OSA with ocular diseases, ranging from those involving the eyelid to optic neuropathies and retinal vascular diseases. Since the retina is one of the highest oxygen-consuming tissues in the body, the intermittent hypoxia and hypercapnia ensuing in OSA can have deleterious effects on ocular function and health. Tissue hypoxia, autonomic dysfunction, microvascular dysfunction, and inflammation all play important roles in the pathogenesis of both OSA and ocular diseases. Whether OSA treatment is capable of reversing the course of associated ocular diseases remains to be determined. It is anticipated that future therapeutic approaches will target the common underlying pathophysiologic mechanisms and promote favorable effects on the treatment of known associated ocular diseases. Emerging evidence supports the association of ocular diseases with untreated OSA. Future studies focusing on whether therapeutic approaches targeting the common pathophysiologic mechanisms will be beneficial for the course of both diseases are warranted.


Introduction
Within the past 5 years, there have been several excellent published reviews addressing the relationship of obstructive sleep apnea (OSA) and ocular disorders [1, 2••, 3]. During this period, there has also been an intensification of the relationship of OSA with ocular disorders that has been inspired by new imaging technologies that can identify structural changes in the eye that may shed light on the pathogenesis of ocular disorders in the context of OSA. As we better understand the metabolic, immune, and biological attributes of OSA, it is worthwhile to revisit these associations in order to better understand potential models for the pathogeneses of these ocular conditions and to potentially identify therapeutic interventions that might impact their management. This review will attempt to summarize the evidence for the associations of OSA and ocular disorders, the newer structural and functional parameters of eye disorders and vision that may be linked to OSA, and attempt to consider the underlying metabolic, genetic, and structural mechanisms that may inform our understanding of both OSA and ocular conditions ( Table 1). In some instances, we will consider shared etiologies and risk factors, while for others, we will consider how the alterations caused by OSA may directly influence ocular function and disease. This third segment of the paper, based on peer-reviewed publications, is ultimately speculative but gives us some hints for future research directions and potential therapies.

Definition of OSA, Age-Dependent Differences, and Implications for Eye Disorders
Obstructive sleep apnea (OSA) is a subset of sleep-disordered breathing that is characterized by episodic sleep statedependent upper airway collapse, resulting in periodic reductions or cessations in ventilation, with subsequent hypoxia, hypercapnia, or arousals from sleep [103]. Risk factors for OSA are conditions that reduce the size of the resting pharynx or increase airway collapsibility such as obesity, male sex, persons with hypothyroidism or acromegaly, increased tonsillar and adenoid tissue, and certain craniofacial abnormalities [104].
In terms of OSA prevalence and age, although there is a gradual increase, prevalence tends to level off after 65 years [105]. Geriatric patients exhibit more severe and deeper nocturnal intermittent hypoxia compared to young adults, independent of OSA severity which could be reflective of the already present chronic hypoxemic conditions, due to the physiologic aging process [106]. When we consider the association of OSA with ocular conditions, we have to consider if these relationships are due to pleiotropic effects of shared risk factors (such as anatomic features and common physiologic pathways) and/or if the metabolic and systemic effects of the OSA itself contributes as a risk factor for these conditions, as illustrated in Fig. 1.

Association of OSA with Specific Ocular Conditions
The association of OSA with microvascular eye diseases, including non-arteritic ischemic neuropathy, retinal vein occlusion, and diabetic retinopathy, seems to be strengthened by a common disturbance in tissue perfusion and oxygenation as well as potential metabolic derangements that promote coagulation disorders. One can make a case that similar processes could underlie the association of OSA with both primary open-angle glaucoma and normal-tension glaucoma. However, it is certainly possible that structural changes in connective tissue that contribute to the intermittent loss of airway integrity could act as a shared risk factor for floppy eyelid syndrome (FES), and also result in increased optic nerve vulnerability to damage from elevated intraocular pressures or compression in high myopes [107,108]. These mechanisms do not seem to offer a clear explanation for the associations of OSA with central serous chorioretinopathy (CSR). While CSR is also intimately related to abnormalities of the choroidal circulation and closely associated with choroidal thickening (known as pachychoroid), the distinct clinical features of CSR, the focal or multifocal character with considerable ocular asymmetry, its intermittent acute flares and spontaneous remissions, and associations with endogenous cortisol or exogenous steroid exposure seem to be outside of known OSA risk factors. Yet, the association of OSA with CSR is so clearly established that some clinicians have even suggested that every patient with CSR undergo testing for OSA regardless of clinical symptoms [94].
There is one study that indicates that the clinical response to anti-vascular endothelial growth factor (VEGF) injections for treating exudative age-related macular degeneration (AMD) is impacted by whether or not an OSA patient is treated with continuous positive airway pressure (CPAP) [101]. If such a finding were to be replicated, it would suggest a very dynamic relationship between the physiologic changes during sleep and potential exacerbations of retinal hypoxia and secondary elevations of VEGF in the retina/choroid. Given our lack of therapies to prevent or slow the progression of nonexudative AMD and the relatively high percentage of patients with exudative AMD who demonstrate only a partial response to anti-VEGF therapies, it would be invaluable to explore the potential impact of OSA treatment on this condition. Figure 2 summarizes the effects of OSA on various ocular structures and associated ocular diseases.

Eyelid and Ocular Surface Morphology
Ever since Gonnering and Sonneland first reported a patient with both obstructive sleep apnea (OSA) and floppy eyelid syndrome (FES) in 1987 [7], there have been numerous papers looking at the association between the two diseases [4, 8-12, 14, 16] (Fig. 3). Lid laxity quantification is important in determining the presence or absence of structural eyelid change and there have been several methods suggested, such as measuring the "vertical lid pull," "vertical hyperlaxity of the lid," or simply "horizontal eyelid distraction distance" [10,11,13]. McNab measured the excursion of the upper lid margin from traction applied to the pretarsal skin in a vertical direction and found a vertical lid pull of 15 to 25 mm for FES patients, 7 to 16 mm for OSA, and 5 to 10 mm for age-and gender-matched controls [10]. Robert et al. found increased eyelid hyperlaxity in OSA patients, and Mojon et al. found positive correlation between respiratory disturbance index (RDI) and eyelid distraction distance [11,13]. However, Fox et al. recently performed a crosssectional observational study with individuals referred for overnight polysomnography and found no association between the presence of OSA and eyelid laxity [16]. The authors enrolled a large number of patients (201 individuals, 402 eyes) and attempted to employ validated quantitative measurements to objectively determine the presence of eyelid laxity. The lack of a gold standard for assessing and the relatively subjective methods with which previous studies investigated structural eyelid change are all possible reasons for the discrepancy in these study results. Age and body mass index (BMI) are both important factors associated with increased eyelid laxity and OSA and may possibly act as a confounding factor, but previous studies controlling for these factors nonetheless found an association between FES and OSA [8,10,11].
The number of elastin fibers has been found to be markedly decreased in FES patients with increased expression of matrix metalloproteinases implicated as a possible cause [109,110].  Interestingly, elastin fiber network disorganization in the distal uvula was found to be associated with the apnea-hypopnea index (AHI) [111] and such pathologic tissue changes may be the common pathophysiology underlying both FES and OSA. Therapeutic approaches that could mitigate these structural changes to the tissues would potentially benefit patients with either FES or OSA. However, a more functional interrelationship between these two conditions has been suggested by several studies that reported that FES improved in a group of OSA patients with successful use of CPAP [18,19]. The not-soclear-cut effects of CPAP therapy and OSA-related surgery on FES suggest that both shared factors related to the structural integrity of the palate and lid tissues, as well as dynamic features of OSA, can impact FES directly. This may also be due to inflammatory factors that appear to improve with OSA therapy and may contribute to the OSA severity, as well as to the severity of the lid swelling and laxity. Ocular surface changes have also been studied in conjunction with eyelid laxity in OSA patients. One study found that although 52% of OSA patients had abnormal eye findings and RDI correlated negatively with tear film break-up time (TBUT), corneal abnormalities were found in only 4.5%, with symptoms of ocular irritation being rare [11]. Another study conducted a more comprehensive study focusing specifically on ocular surface changes occurring in OSA patients and found that moderate and severe OSA is associated with lower Schirmer and TBUT, high scores on the ocular surface disease index questionnaire, and corneal staining pattern stage [112].

Corneal Hysteresis
Normal human corneal thickness is about 500 μm and diurnal variation is present with overnight swelling and resolution by early afternoon, possibly arising from hypoxia created by lid closure [113]. Both hypoxic and hypercapnic environments are known to affect corneal thickness with 7% swelling per hour observed in the normal human cornea [114]. One study analyzed changes in corneal thickness with/without CPAP application in OSA patients using an ultrasonic pachymeter and found a significant corneal thickness increase in only the without-CPAP group [115]. Another study looked at central corneal thickness (CCT), TBUT, and Schirmer's test in OSA patients according to severity defined by AHI scores [116]. CCT was significantly decreased in OSA patients compared to that in the control group, and as OSA severity increased, CCT decreased in a stepwise manner (mean CCT 570 mm, 561 mm, and 534 mm in mild, moderate, and severe OSA, respectively, p < 0.05). There were no significant differences in TBUT or Schirmer's test results among different OSA severity groups. Dikkaya  study corneal biochemical properties in OSA patients and showed significantly lower corneal hysteresis and resistance in the severe OSA group, which implies possible corneal biochemical changes in OSA, especially in the severe type [117].

Optic Disc
Retinal ganglion cells (RGCs) are responsible for conveying synaptic input initiated by a visual stimulus to the visual cortex within the brain. Ganglion cell axons converge at the optic disc and exit through the lamina cribrosa to form the optic nerve. Optic nerve examination, as well as examination of the peripapillary retinal nerve fiber layers (RNFL), consisting of ganglion cell axons, is useful for investigating the presence of neuroaxonal degeneration, either anterograde or retrograde. OSA has been reported to be associated with various optic nerve diseases, such as non-arteritic anterior ischemic optic neuropathy (NAION), primary open-angle glaucoma, normal-tension glaucoma, and papilledema [118, 119•, 120•, 121-123]. Huseyinoglu et al. looked specifically at changes in optic disc parameters in OSA patients, such as optic disc area, cup area, cup volume, cup/disc area ratio, and nerve head volume, using OCT to obtain 12 radial scans and 13 concentric rings, centered on the disc, but were unable to detect any significant differences [118].
Studies related to optic neuropathy/glaucoma and OSA have suggested that the negative effects of OSA on RNFL and nerve oxygenation may be reversed by CPAP. However, to date, IOP control is the only proven treatment for glaucoma and there is yet no evidence that IOP can be altered by CPAP therapy. Oxidative stress and lower oxygenation could also be contributing factors for ischemic optic neuropathy, which also has implications for diabetic retinopathy (which also seems to respond to CPAP therapy but not always consistently) [83], as well as complications of retinal vein occlusion (RVO).
Idiopathic intracranial hypertension (IIH), a syndrome of increased intracranial pressure (ICP) with unknown etiology, has also been reported to be associated with OSA [124][125][126]. Intermittent ICP elevation is not infrequently observed in OSA, and Sugita et al. hypothesized this to be due to a combination of factors such as increased central venous pressure and subsequent cerebrovascular volume increase, systemic arterial hypertension with secondary increased cerebral perfusion pressure and cerebral vasodilation due to hypoxia and hypercapnia, and resulting intracranial blood volume increase [127,128]. Since IIH and OSA share a common risk factor of obesity, it remains to be determined whether OSA itself is a comorbidity or an independent risk factor for IIH. There was a report of IIH symptom resolution after surgery for OSA in a pediatric patient and optic disc swelling resolution was observed after CPAP, but as with the association between OSA and IIH, there is no consensus at this time as to whether OSA treatment affects IIH [126,128,129].

Optic Nerve Vasculature: Non-arteritic Ischemic Optic Neuropathy
The perfusion of the retina by the central retinal artery and its branches is crucial for the maintenance of the inner retinal neurons. The outer retinal layers, specifically the photoreceptor layer, derive their vascular and metabolic support from the choroidal circulation. Effective blood flow and oxygenation are achieved by having sufficient perfusion pressure of the retinal vasculature to overcome the resistance caused by the intraocular pressure. Reductions in nocturnal blood pressure, in conjunction with the impairment of central retinal artery blood flow and/or elevated intraocular pressures, can lead to acute ischemic events such as NAION (Fig. 4) and, more rarely, central retinal artery occlusions (CRAO) [130,131]. Among 925 episodes of NAION, 73% of patients first reported visual symptoms after waking from sleep, and nocturnal arterial hypotension showed a significant correlation with progressive visual field deterioration in NAION, altogether implying that nocturnal blood pressure depression may precipitate NAION development in patients having additional risk factors [132,133]. In OSA, the imbalance between nitric oxide and endothelin results in reduced autoregulatory ability of the optic nerve microcirculation to compensate for fluctuations in blood pressure and decreased blood oxygen saturation during apnea may cause direct injury to the optic nerve head [134,135]. Such structural and functional changes occurring in OSA could act as the additional risk factor tipping the fine balance of perfusion towards ischemia for patients at risk for NAION. Prospective studies also reported the prevalence of OSA in NAION patients to be 55.6~89% in comparison to that of 18~22% in the general population or a control group, while a 12-year nationwide population-based retrospective cohort study found an increased risk of developing NAION in the OSA group compared to the non-OSA group (HR 3.80; 95% CI 1.46 to 9.90) even after adjusting for demographics, comorbidities, and comedications [136,57,53,105,119]. Hence, individuals who have experienced an episode of NAION or a CRAO should be carefully questioned regarding a history of OSA-related symptoms and a formal sleep study should be considered, even if other OSA comorbidities are not present. There is no evidence at this time that suggests that treatment of OSA will alter the recovery of the eye from one of these acute ischemic events [59] although nonadherence to CPAP treatment in patients with both OSA and unilateral NAION was found to increase the risk of fellow eye involvement [51].

Choroidal Layer
Choroidal circulation has autonomic regulation, sharing regulatory properties as that of cerebral blood flow [137]. Alternating hypoxia and arousal occurring in OSA is thought to stimulate the sympathetic nervous system leading to an increase in choroidal blood flow following hypercapnia and expression of hypoxia-inducible factor and vascular growth factors, resulting in vascular endothelium damage, changes in vessel permeability, and choroidal thinning. One study used choroidal laser Doppler flowmetry to study choroidal vascular reactivity in OSA men with a mean age of 50.9 years, having Fig. 4 A 66-year-old male non-arteritic ischemic optic neuropathy patient presenting with vision blurring in the right eye. Disc photograph of the right eye (a) showed blurring (white arrowheads) of the superotemporal optic disc margin compared to a normal-looking optic disc in the left eye (b). Humphrey visual field testing showed inferior visual field defect (black arrows) in the right eye (c) corresponding to the area of optic disc swelling, while the left eye showed normal visual field (d) no cardiovascular comorbidities, and found no impairments, i.e., hypercapnia-induced blood flow increase and stable choroidal blood flow in hyperoxia, suggesting long-term adaptive mechanisms coming into play in the ocular microcirculation [138]. As with RNFL, numerous studies have found thinner choroidal thickness while others have found no difference in OSA patients compared to controls, although a meta-analysis was able to confirm significant choroidal thickness reduction, especially in severe OSA [94,[139][140][141][142][143]. In terms of reversal of choroidal thickness changes after treatment, one study reported significant increases in choroidal thickness after 12 months of CPAP institution with possible improvements in choroidal function [144]. Different modalities and analysis techniques used to assess choroidal blood flow may be one of the underlying reasons for varied results in previous studies.
The biological basis between CSR, in which a thickened choroid is the most distinct characteristic, and OSA is perhaps the most complicated and unclear (Fig. 5). CSR does not appear to be triggered by topical, periocular, or intravitreal exposures to steroids, even though there are multiple reports of activation associated with systemic steroids, as well as from intra-articular, intranasal, and inhalant exposures. This paradox highlights the fact that the underlying mechanism of CSR is poorly understood. The relationship of CSR and OSA becomes an even more intriguing target for study. There is no evidence that the choriocapillaris is thicker in OSA patients, suggesting that this "pachychoroid" feature is relatively independent of OSA. It is possible that the association of OSA and CSR is not with the underlying pathology that is required to develop CSR but may be related to factors that cause acute activation and/or persistence of active CSR [94,139,141,145].
The initial activation and recurrences of CSR have suggested the possibility of an infectious and/or inflammatory etiology for this condition. There is an association of OSA with activation of herpes zoster in patients which has been suggested to be due to modulation of the immune system [146,147]. An infectious etiology for CSR has been proposed with an association of Helicobacter pylori [148], though the natural history of recurrent episodes would require either reactivation of infection or sensitization of the choroid with inflammatory cells that continue to reside in the choroid, and which are vulnerable to sporadic activation. Other studies have shown that H. pylori can both promote the accumulation and degranulation of mast cells in different tissues [149][150][151][152][153][154]. There are even several studies that have focused on the association of H. pylori itself with OSA. Whether such infectious etiologies play an important role in the pathophysiology of OSA and its comorbidities is a topic that requires further investigations [155][156][157][158][159][160][161][162].
Some investigators have suggested that CSR activation may be due to the degranulation of mast cells that are resident in the choriocapillaris, and this could certainly be triggered by low-dose steroid exposure. Whether or not H. pylori plays a role in CSR, the activation and degranulation of mast cells that are resident within the choriocapillaris may be a reasonable connection with other known associations of CSR such as stress or low-dose(not high-dose) steroid exposure and OSA, which can also modulate the immune system and trigger mast cell degranulation. This hypothesis that CSR may be driven and triggered by mast cell accumulation and degranulation, as well as local inflammation, would also be consistent with the purported role of mast cell accumulation and activation in floppy eyelid syndrome [163].
Recently, numerous meta-analyses have been published, in which all found a significant reduction of average RNFL thickness in OSA patients [122,123,166,167,174]. Recurrent obstruction of airflow occurring in OSA, leading to repeated hypoxemia, hypercapnia, and reduced perfusion in the optic nerve, would most likely result in retinal ganglion cell apoptosis and subsequent retinal nerve fiber layer reduction [118,121].
Sleep disturbance is one of the most prevalent non-motor symptoms in Parkinson's disease (PD) [175]. Rapid eye movement sleep behavior disorder (RBD), characterized by the loss of normal atonia during rapid eye movement (REM) sleep, has been implicated as a precursor to α-synucleinopathies such as dementia with Lewy body (DLB) and PD [176]. Peripapillary RNFL thickness, as well as inner retinal thickness in the macular area, has been reported to be reduced in both PD and RBD [177][178][179][180][181]. Melanopsin-containing retinal ganglion cells (mRGCs) are a subset of RGCs that also participate in non-image-forming functions such as circadian rhythm regulation or pupillary light reflex activation [182,183]. Their degeneration and impairment are one of the pathways in which sleep disturbance is hypothesized to occur in PD [184]. Neither PD nor RBD has any overlapping pathogenic mechanisms with OSA; however, both neurodegenerative diseases result in sleep disturbance and have ocular manifestations both functionally and anatomically, as in OSA, and are examples of a sleep-associated disease, such as OSA, in which ophthalmic examination may reveal more than meets the eye.

Hypoxia
In vitro model of intermittent hypoxia (IH)/reoxygenation shows activation of the pro-inflammatory transcription factor NFκB. Circulating tumor necrosis factor-α levels which were significantly higher in OSA patients normalized after CPAP therapy, demonstrating that selective activation of inflammatory pathways as a result of intermittent hypoxia in OSA may be one of the molecular pathways underlying macro/ microvascular diseases associated with OSA [185]. The proinflammatory nature of OSA could potentially play important roles in the pathophysiology of floppy eyelids, dry eyes, diabetic retinopathy, and age-related macular degeneration.

Autonomic Dysfunction
High sympathetic tone, increase in baseline heart rate, and elevated muscle sympathetic nerve activity are all commonly observed in OSA patients [186]. The following phenomena all come into play in OSA patients, and such changes may affect the choroidal vascular supply, which is mainly controlled by the autonomic nervous system, pathologic changes in which could have an impact on diabetic retinopathy, central serous chorioretinopathy, and age-related macular degeneration. However, treatments that would alter these autonomic states have not been explored as potential treatments for these ocular conditions as of yet.
1)Elevated catecholamines and muscle sympathetic nerve activity during apnea as while awake 2)Activation of the renin-angiotensin-aldosterone system 3)Differing vascular reactivity between OSA and controls 4)Changes in sympathetic and vascular reactivity with OSA treatment 5)Animals exposed to intermittent hypoxia or apnea show the above observations.

Microvascular Changes
One study used OSA patient skin biopsies to identify molecular biomarkers involved in the vascular dysfunction of OSA and found endothelial nitric oxide synthase (eNOS), tumor necrosis factor-α-induced protein 3, hypoxia-inducible factor 1 α, and vascular cell adhesion molecule 1 (VCAM-1) expressions to be significantly upregulated [187]. All of these factors have also been implicated in the pathogenesis of diabetic retinopathy, retinal vascular occlusive disease, and exudative forms of age-related macular degeneration [188][189][190]. However, there is yet no study that has considered the role of these factors in the combined context of OSA patients with any of these retinal diseases.

Oxidative Stress
When rats were exposed to 14 days of IH, superoxide ion expression in the ophthalmic artery (OA) wall and OA contractile response to endothelin-1 both increased, and nitric oxide-mediated relaxation was significantly delayed [191]. This shows the induction of oxidative stress in rat OA by chronic IH, combined with endothelial cell and nitric oxide synthase dysfunction. Given the known vulnerabilities of the optic nerve, as well as the retina, to damage from oxidative stress for many of the conditions that we know are associated with OSA, we need to consider if agents that reduce oxidative stress might be potentially useful in a subset of OSA patients and have a clinical impact on the ocular conditions in these individuals.

Conclusions and Future Directions
We can see that the ocular associations with OSA likely involve both structural comorbidities such as collagen mutations that affect the structural integrity of the airways and the eyelids, as well as the activation of mast cells that place a role in the atopic aspects of many cases of floppy eyelid syndrome as well as dry eye manifestations. At the same time, this modulation of mast cells and/or cortisol in patients with OSA may contribute to the risk of activation of CSR. Elevated inflammatory factors that have been reported in OSA may contribute to both AMD and diabetic retinopathy. Microvascular changes may also play a role in the severity and progression of diabetic retinopathy. The chronic, intermittent hypoxia of untreated OSA may play a role in patients' visual dysfunction and may offer a potential means of functionally monitoring these patients both before and in response to treatment. Central serous chorioretinopathy appears to be ameliorated by the treatment of OSA and may be related to either cortisol modulation and/or mast cell activation. These hypothetical connections offer new opportunities to better understand both OSA and the ocular conditions that are influenced by this disorder. They offer potential, testable interventions to limit the damage from a number of these eye conditions by either treating the underlying OSA or the shared pathways that may be contributory (even in individuals without clinical evidence of OSA). Using ocular imaging and visual function to assess therapeutic interventions for OSA and the impact of those treatments on ocular anatomy and metabolic states can help us better understand how the amelioration of OSA in patients with these ocular conditions may offer an adjunct to our current therapies.

Compliance with Ethical Standards
Conflict of Interest The authors report no conflicts of interest.
Human and Animal Rights and Informed Consent This article does not contain any studies with human or animal subjects performed by any of the authors.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.