Curr Neurol Neurosci Rep (2014) 14:465 DOI 10.1007/s11910-014-0465-5
NEURO-OPHTHALMOLOGY (A KAWASAKI, SECTION EDITOR)
Obstructive Sleep Apnea and Optic Neuropathy: Is There a Link? Clare L. Fraser
# Springer Science+Business Media New York 2014
Abstract Over the last decade, there has been an emerging interest in the link between obstructive sleep apnea (OSA) and ocular health. Though the evidence for OSA playing a role in cerebrovascular disease risk seems clear, the same cannot be said for optic neuropathies. The association between OSA and glaucoma or non-arteritic anterior ischemic optic neuropathy (NAION) has been postulated to be secondary to direct hypoxia or mechanisms of optic nerve head vascular dysregulation. Papilledema and increased intracranial pressure have also been reported in OSA and are thought to be due to increased cerebral perfusion pressure and cerebral venous dilation secondary to hypoxia and hypercapnia. This article reviews the evidence for possible pathophysiological links between OSA and optic nerve pathology. The epidemiologic and clinical evidence for an association, direct or indirect, between OSA and glaucoma, non-arteritic anterior ischemic optic neuropathy (NAION), and papilledema or idiopathic intracranial hypertension is presented.
Keywords Obstructive sleep apnea . Glaucoma . Non-arteritic anterior ischemic optic neuropathy . Papilledema . Disc edema . Intracranial hypertension . Optic neuropathy
This article is part of the Topical Collection on Neuro-Ophthalmology C. L. Fraser Save Sight Institute, University of Sydney, 8 Macquarie Street, Sydney, NSW 2000, Australia C. L. Fraser (*) Department of Ophthalmology, Macquarie University, 3 Technology Place, Sydney, NSW 2109, Australia e-mail: [email protected]
Introduction Obstructive sleep apnea (OSA) is characterized by intermittent upper airway obstruction during sleep, resulting in apnea and hypoxia. The prevalence of OSA in community-screened patients varies between 2 %–6 % for moderate disease, and up to 14 % for mild disease [1]. Therefore, it is very likely that a similar proportion of our ophthalmology patients will have OSA. Those with OSA have a significantly higher body mass index (BMI) than those without, 31.4 vs 28.3, respectively [2]. Other risk factors to consider in patients include: male gender, older age, upper airway abnormalities, and enlarged neck girth [3]. In a clinical outpatient setting, the most useful observation for identifying patients with OSA is a history of nocturnal choking or gasping [1]. Though snoring is present in 94 % of those with OSA, it is not useful in establishing a higher risk for diagnosis as not all snorers have OSA [2]. The Berlin Questionnaire is a validated scale that can be easily administered in clinic to reliably identify those likely to have OSA [4]. A low score in response to this questionnaire identifies patients unlikely to have OSA, and has been shown to be of use as a screening tool in recognizing ophthalmic patients who do not need further investigation with sleep studies [5]. Overnight diagnostic polysomnography (dPSG) performed in a sleep laboratory is used as the reference standard for diagnosing OSA [2]. The dPSG measures respiratory indices, the electroencephalogram, and limb movements during sleep. The apnea-hypopnea index (AHI) is the sum of the apnea (cessation of oro-nasal flow for >10 seconds) and hypopnea (reduced oro-nasal flow of >50 % with a 4 % oxygen desaturation) events during an hour of sleep. In most papers OSA is defined as an AHI≥5, with an AHI of 5–15 considered mild, 16–30 moderate, and >30 being severe OSA [3]. The respiratory disturbance index (RDI) is defined as the sum of apneas, hypopneas, and abnormal respiratory events per hour
465, Page 2 of 9
of sleep. However, these terms are used interchangeably in some studies, and the cut-off levels for diagnosis of OSA vary between studies. Continuous positive airway pressure (CPAP) therapy is the treatment of choice, using airway pressure to effectively splint the throat open, thereby reducing apneas and hypopneas.
Basic Systemic Physiology of Obstructive Sleep Apnea The intermittent upper airway obstruction in OSA results in hypoxia (reduced arterial partial pressure oxygen, PaO2) and hypercapnia (increased PaCO2). This hypoxia results in sympathetic activation, oxidative stress, metabolic dysfunction, and systemic inflammation [6]. These are considered surrogate markers of cardiovascular risk, as they are all intermediary mechanisms that can lead to increased rates of hypertension, ventricular dysfunction, and early atherosclerosis. OSA is increasingly recognized as an independent risk factor for hypertension, stroke, and coronary heart disease [7]. Elevation in the odds of hypertension is observed in even those with mild OSA [8]. Epidemiologic studies suggest an association between OSA and increased atherosclerosis and large vessel endothelial disease [9]. However, the mechanistic link remains unclear. The Wisconsin Sleep Cohort Study showed that AHI is an independent predictor of hypertension, increased body mass index (BMI), insulin resistance, and cardiac dysfunction [6]. The metabolic dysfunction found in OSA exacerbates a tendency to obesity, with increased insulin resistance and nonalcoholic fatty liver disease [10]. Interestingly, obese patients without OSA do not show the same increase in markers of metabolic dysfunction. This suggests that OSA is an independent risk factor for these metabolic changes, rather than OSA being an epiphenomenon of obesity. OSA has also been found to be an independent risk factor for stroke in large epidemiology studies [11••]. The relationship between OSA and small-vessel cerebrovascular disease is complicated by the fact that the other underlying risk factors, namely hypertension, diabetes mellitus, and atherosclerosis, are also associated with OSA. Therefore, any relationship between OSA and small-vessel disease could be due to confounding factors [11••].
Physiology of Optic Nerve Pathology in OSA The interaction among various physiological parameters during sleep, including ocular perfusion pressure, aqueous outflow in the supine position, and circulating hormone levels, complicate the prediction of vascular changes and intraocular pressure (IOP) fluctuations overnight. The association between OSA and glaucoma or non-arteritic anterior ischemic
Curr Neurol Neurosci Rep (2014) 14:465
optic neuropathy (NAION) is postulated to be secondary to direct hypoxia or mechanisms of vascular dysregulation at the optic nerve head [12•]. 1. Direct Hypoxia Repetitive apneic episodes might lead to direct anoxic damage to the optic nerve [13, 14]. As the AHI increases, more hypoxia results in reduced mean oxygen saturation. However, Sergi et al. found no relation between ophthalmic measures of glaucoma and post-apneic falls of PaO2, change in oxygen saturation or the percentage of sleep time spent with oxygen saturation less than 90 % [15]. 2. Atherosclerosis and Arterial Blood Pressure Variation OSA has been shown to be associated with hypertension [16], endothelial cell dysfunction [17], and autonomic dysfunction [18] all of which may alter ocular vascular regulation. Both animal and clinical data shows that OSA promotes cellular changes in the vascular wall, in particular the carotid artery intima media thickness. The occurrence and progression of atherosclerotic plaques also appears related to the AHI, in the absence of significant comorbidity [17]. Endothelial function is also impacted by the oxidative stress of OSA, with resultant reduction in repair capacity and changes in vasomotor tone. However, these studies focus on larger peripheral arteries [19]. These changes could result in small vessel hyalinosis, thereby causing optic nerve head ischemia, though further histopathologic studies focusing on the optic nerve are required. Prolonged sympathetic overactivity is a characteristic feature of OSA, with bursts of sympathetic activity occurring before each nocturnal arousal and during the day in response to sleepiness. However, the role of the sympathetic system in regulating cerebral and ocular circulation remains a matter for ongoing research efforts. Physiological research in humans indicates that brain perfusion is significantly distorted in OSA due surges in blood pressure and blood velocity [20•]. Repetitive apneic episodes have been shown to have indirect effects on optic nerve head blood flow due to sympathetic surge-induced arterial hypertension [13, 14]. 3. Impaired Auto-Regulation at the Nerve Head If cerebral perfusion pressure decreases due to increased intracranial pressure (ICP) during apneic episodes, then any cerebral hypoxia will be worsened. Furthermore, dynamic cerebral blood flow auto-regulation is impaired in response to hypo-perfusion in OSA patients [21]. These fluctuations and impaired responses put the brain at risk of vascular damage and ischemia. Doppler studies on the ophthalmic and central retinal arteries found no difference in the systolic and enddiastolic blood velocities (resistivity index) between OSA patients and controls [22]. The posterior ciliary
Curr Neurol Neurosci Rep (2014) 14:465
arteries and the choroidal vessels supply the prelaminar and laminar portions of optic nerve head, as the nerve travels through the sclera. Only the surface layer of the optic disc has capillaries derived from retinal arterioles, but even there it is not uncommon for this region to be supplied by choroidal vessels [23]. Given that most optic disc pathology is thought to arise from changes in the prelaminar and laminar sections, the choroidal circulation is of most interest in pathophysiology. There is evidence from various human studies that retinal blood flow increases during hypercapnia [24–26]. A linear relationship exists between choroidal blood flow as measured on a laser Doppler flowmeter and PaCO2, however there is no effect from changes in PaO2 [27]. On examination of OSA patients compared with controls, there was no difference in choroidal blood flow reactivity to different inhaled concentrations O2 or CO2, and no change after CPAP treatment [28]. However, this study did not look at obese patients, or those who had a protracted course of OSA. Data suggest that cerebral blood flow is abnormal in OSA, with blood flow velocity surges [20•]. Under normal conditions, the pial arteries act to stabilize and counteract changes in cerebral blood flow. However, hypercapnia impairs the pial artery responses to changes in cerebral blood flow [29]. The orbital portion of the optic nerve and the intra-cranial part of the optic nerve are supplied by pial circulation. Therefore, if the pial system at the optic nerve head can no longer compensate for changes in cerebral blood flow, impaired auto-regulation at the nerve head may be the site of ocular pathology in OSA. 4. Imbalance Between Vasodilator and Vasoconstrictor Substances Intermittent hypoxia promotes production of reactive oxygen species, increasing oxidative stress, activating systemic inflammation, and leading to decreased bioavailability of endothelial nitric oxide. Vascular dysregulation due to nitric oxide/endothelin imbalance or abnormal platelet aggregation also has indirect effects on optic nerve head blood flow [30]. Recent research has focused on the association between retinal micro-vascular dysfunction, cardiovascular disease and OSA. Boland et al. found no association with OSA and the ratio of retinal arteriolar narrowing and venular widening [31]. However, retinal arteriolar changes are associated with hypertension, but retinal venule changes are associated with inflammatory and metabolic abnormalities [32]. Shankar et al. analyzed retinal vessels independently, and found that AHI was associated with retinal venular widening independent of confounders, but not associated with retinal arteriolar narrowing [9]. In a qualitative study of retinal photos, changes resembling hypertensive retinopathy (arteriolar narrowing, arterio-
Page 3 of 9, 465
venous-nipping and retinal hemorrhages) have been associated with AHI, independent of blood pressure [33]. These potential associations with microvascular changes in the retina are of the most interest when considering the link between OSA and optic nerve pathology, because the same changes may be occurring in the choroidal and posterior ciliary vessels.
Glaucoma The association between glaucoma and OSA has been widely reported, However, the results vary and a clear link is still uncertain. The prevalence of sleep-disordered breathing is between 20 %–50 % in patients with primary open angle glaucoma (POAG) [34, 35] and normal tension glaucoma (NTG) [36]. The inverse is also reported, with 7.2 % of sleep apnea patients having glaucoma, compared with the expected population rate of 2 % [37]. However, other studies have found no difference in the prevalence of glaucoma within an OSA population compared with the normal population. Geyer et al. found no relation between the RDI and intra-ocular pressure (IOP) or the presence of glaucoma [38]. While it is difficult to compare the various studies examining the association between glaucoma and OSA due to small numbers, different criteria of OSA, and differing means of glaucoma assessment, the following paragraphs provide a summary of the recent research. Retinal Nerve Fiber Layer (RNFL) Thickness The link between OSA and glaucoma has been investigated by examining the preclinical changes in RNFL thickness. The appearance of RNFL thinning suggests atrophy, presumably due to decreased ocular perfusion, hypoxia, and vasospasm. A trend for loss of RNFL to correlate with the severity of OSA was reported in a small group of patients by Kargi et al., using scanning laser polarimetry [39]. In a study of 105 patients and 22 controls examined with Stratus ocular coherence tomography (OCT), Lin et al. also found that RNFL was lower in patients with moderate/severe OSA (AHI >15) compared with those with mild OSA (515) than controls, 14.2+/−3.5 vs 12.2+/−3.6 mmHg [39]. Other studies have shown positive correlation between IOP and increasing AHI scores [22, 15]. However, the study inclusion criteria was of IOP 15), and diagnosed glaucoma in 27 %, a higher prevalence than expected [49]. However, there was no evidence for a relationship between AHI and IOP or the diagnosis of glaucoma (disc and field changes) - only increasing age correlated with glaucoma diagnosis. Examining a subset of the Beijing Eye Study, data on snoring (as reported by the subject’s life partner) and glaucoma was available for 3146 patients [50]. “Severe snoring” was not associated with open-angle glaucoma, angle-closure glaucoma, cup:disc ratio, or RNFL. Another interview based study found a higher prevalence of sleep-disordered breathingrelated symptoms in 212 POAG patients than controls [35]. A cross-sectional study that reviewed diagnoses given to patients in a managed care program, found that there was no difference in the hazard of being diagnosed with POAG or NTG in those with OSA (regardless of CPAP treatment) and those without OSA [51]. Using diagnostic codes to identify comorbities, Girkin et al. found that individuals who developed glaucoma were more likely to have a previous sleep apnea diagnosis relative to controls, However, once adjusting for other vascular and glaucoma risk factors (hypertension, migraine, diabetes) no significant difference was seen [52]. In a large retrospective population based study, Lin et al. compared the 5-year hazard ratio for developing open-angle glaucoma between 1012 subjects after a diagnosis of OSA (based of dPSG) and 6072 controls [53••]. All patients and controls were known to have attended an ophthalmologist
Curr Neurol Neurosci Rep (2014) 14:465
without being diagnosed with glaucoma in the 4 years prior to the “index year”. Records for OSA patients and the controls were then examined from the index year forwards over 5 years. After adjusting for multiple confounding factors, the hazard ratio for developing glaucoma after OSA was 1.67 that of controls. This study concludes that OSA was independently associated with an increased risk of POAG. The data for a link between OSA and glaucoma is variable. The pathophysiological association is plausible, and though there does seem to be consistent evidence for loss of RNFL, the clinical significance needs to be reviewed. The direct link to NTG in OSA appears stronger than the case for high IOPrelated POAG.
Non-arteritic Anterior Ischemic Optic Neuropathy NAION is an ischemic insult to the optic nerve head, and an annual incidence in persons 50 years and older is between 2.3 and 10.2 per 100,000 [54••]. Typically found in patients over the age of 50, a small cup to disc ratio “disc at risk” is probably one of the chief risk factors for NAION. NAION is also associated with acquired vascular risk factors such as systemic arterial hypertension, diabetes, and atherosclerosis. Circulatory insufficiency within the optic nerve head is thought to precipitate NAION, however, the exact location of the vasculopathy remains unknown [55•]. Histopathologic studies have implicated the retrolaminar region as being the area of infarction, with fluorescein studies finding delayed filling in the prelaminar optic disc [56]. Further research is needed to determine if and how OSA contributes to an isolated vascular insult or compartment syndrome within the optic nerve. In a review of 544 episodes of NAION at least 73 % were discovered upon first awakening, leading to the hypothesis that nocturnal hypotension might be a precipitating factor [57]. However, the hypoxia of OSA promotes hypertension, producing a nocturnal non-dipping pattern otherwise known as “masked hypertension” [58]. Therefore, other vascular changes in OSA must be related to the NAION process, if a link truly exists. As mentioned previously, vascular autoregulation may be hindered by sympathetic surges or imbalance of vasoactive substances such as endothelin-1 (vasoconstrictor) and nitric oxide (vasodilator) [53••], triggering NAION. In 2 studies of NAION, 71 %–89 % had OSA diagnosed on dPSG (RDI >10 or AHI >15) compared with 18 % of a control population [59, 60]. In addition, OSA was 1.5–2 times more frequent than other known risk factors such as hypertension and diabetes. However, in one of the two aforementioned studies, the 17 patients and controls were only matched by age and sex
Page 5 of 9, 465
[58]. In the other study, the prevalence of OSA in NAION was compared with published population studies, showing a 4.9 relative risk of OSA in NAION patients [59]. More recently Arda et al. performed polysomnography on 20 NAION patients and 20 controls matched for age, sex, body mass index (BMI), diabetes, and hypertension. Of those diagnosed with NAION 85 % had OSA, compared with 65 % in controls. By matching for hypertension and diabetes, which are found in association with OSA, these rates reported are much higher than expected for the general population. The high rates may also have been found because the study used an AHI >5 for OSA [61]. In a prospective study of 27 NAION patients within 1 month of symptom onset underwent dPSG. Controls who were statistically similar for systemic risk factors and BMI [54••] also underwent dPSG. A diagnosis of OSA was made, based on AHI >20, in 55.6 % of NAION patients and 22.2 % of controls. Using these stricter criteria for OSA than other studies, and appropriate matching for hypertension, diabetes, hypercholesterolemia, and coronary artery disease this study concluded that OSA should be considered a significant risk factor for NAION. These studies using dPSG are time consuming and have small sample sizes. By using the validated Sleep Apnea Scale of the Sleep Disorders Questionnaire (SA-SDQ), Li et al. were able to assess 73 NAION cases and 88 controls [62]. The patients included more women, and patients with a higher BMI than previous studies. They still found an increased prevalence of presumed OSA, 30 % vs 17.8 % in those with NAION, and an odds ratio of 2.62 (1.03–6.60) when controlling for some confounders. However, the temporal relationship could not be established. The prevalence of NAION in a population with diagnosed OSA has also been examined. In a large retrospective cohort study based on billing records, it was found that individuals diagnosed with OSA and not treated with CPAP had a 16 % increased hazard of developing NAION [50]. This increased risk was not found in those with treated OSA. However, reports of patients developing NAION while undergoing CPAP therapy have also been published [63]. There are no studies that look at whether CPAP can prevent NAION in patients with other risk factors. Given that there is no effective treatment for NAION, management of all potential risk factors including hypertension, diabetes and OSA is advisable, even though based on evidence to date, a direct causal relationship cannot be presumed.
Physiology of Cerebrospinal Fluid Pressure in OSA Several mechanisms have been proposed as the cause of increased cerebrospinal fluid (CSF) pressure in OSA [64].
465, Page 6 of 9
1. Increased Cerebral Perfusion Pressure Increased systemic blood pressure seen in OSA may result in a secondary increase in cerebral perfusion pressure, and therefore increased ICP. There is a strong correlation between the duration of apneic episodes and blood pressure fluctuations [64]. However, animal studies have shown that acute hypertension is tolerated with minimal perturbation of cerebral blood flow, whereas only if there was pre-existing intracranial hypertension was there a loss of autoregulation in response to arterial hypertension [65]. Human work has shown that cerebral blood flow velocities are lower in patients with OSA than normal controls [66], whereas other studies show a steady increase in cerebral blood flow during an episode of apnea [11••]. 2. Hypoxic and Hypercapnia Related Cerebral Vasodilation Vasodilation and increased brain water content from alteration in cerebral autoregulation may change intracranial volumes. A reduction in brain volume of 4 %, without focal changes, has been documented in OSA patients after treatment with overnight oxygen [67]. The increase in cerebral venous blood volume may be responsible or partially responsible for elevated ICP in OSA. Increases in ICP are seen in both rapid eye movement (REM) and non-REM stages of sleep. During overnight CSF monitoring of 3 OSA patients, ICP was episodically elevated in parallel to apneic episodes and hypoxia [64]. In subsequent study of patients with severe OSA, awake values of ICP were demonstrated to be elevated, with further increases in ICP related to apneic episodes while asleep [68]. Sustained elevation of ICP causes papilledema due to disruption of retrograde axonal transport at the optic disc. Two patients with intracranial hypertension but no disc edema had overnight ICP monitoring, and dPSG. Both were obese (BMI >30), had severe OSA with AHI >30 and had rapid sustained rises in ICP associated with hypoxemia. However, blood pressure did not change during these episodes, suggesting that cerebral vasodilation and increases in cerebral blood volume were the precipitating factors [69]. In another case series, a patient was shown to have lumbar puncture opening pressure of 16 cmH2O in the day, but overnight ICP monitoring revealed pressures of 48 cmH2O during episodes of oxygen desaturation [70]. The question remains as to whether these intermittent changes in ICP alone would be sufficient to cause papilledema. Apneic events are also associated with hypercapnia, which has been shown in some studies to lead to cerebral venous dilation. Hypercapnia has long been reported to decrease vascular resistance and increases cerebral blood flow. However, recent data has altered our understanding of the dynamic cerebral blood flow changes that occur in
Curr Neurol Neurosci Rep (2014) 14:465
OSA with changes in blood pressure [19]. Given the limited number of studies and the conflicting results, final conclusions cannot be drawn. 3. Mechanical The mechanical airway obstruction in OSA, leads to increased respiratory effort, increased intra-abdominal pressure and intra-pleural pressure resulting in an impediment to cerebral venous drainage [71]. Cerebral venous hypertension reduces CSF re-absorption through the arachnoid villi, thereby increasing ICP.
Papilledema and Idiopathic Intracranial Hypertension The first report suggesting a link between OSA and disc edema found resolution of disc edema after upper airway surgery in the setting of normal ICP [72]. A case series from Purvin et al. described 4 patients with bilateral optic disc edema and OSA [70]. Treatment with CPAP has been reported to precede resolution of the disc swelling, consistent with an association but does not prove a direct link [70, 73]. In a study using diagnostic codes from a large managed care organization, Stein et al. found that after adjusting confounding factors, untreated patients with OSA had a 29 % higher hazard and patients with CPAP-treated OSA had over twice the hazard of papilledema (105 % higher) when compared with patients without OSA [50]. However, in seeming contradiction to this, the same paper reports similar rates of idiopathic intracranial hypertension (IIH) between those without OSA and CPAP-treated OSA, but that the untreated OSA patients had over twice the hazard of IIH (103 %) compared with those without OSA. Marcus et al. found that of 53 patients with IIH, 70 % had evidence of sleep disturbance. Fourteen of these patients then went on to dPSG, with 13/14 being subsequently diagnosed with OSA [74]. In a larger review of 721 IIH patients, Bruce et al. found 4 % of women had OSA compared with 24 % of men [75]. Lee et al. found 6/18 men with IIH also had a diagnosis of OSA based on dPSG [73]. There has only been one small study regarding the prevalence of papilledema and symptoms of increased ICP in OSA patients. Of 95 recently diagnosed OSA patients contacted, only 35 had an eye examination, none of who were found to have papilledema [71]. However, 34 % gave a history of transient visual obscurations. A study of fundus photographs taken on 250 patients undergoing dPSG, found no patients with optic disc edema (95 % CI: 0 %–3 %) [32]. In a further study, patients with newly diagnosed IIH underwent dPSG, the AHI values were then compared with the population-based expected AHI for each patient based on
Curr Neurol Neurosci Rep (2014) 14:465
age, sex, race, BMI, and menopausal status. The prevalence and severity of OSA in IIH patients was not significantly different from predicted values [76]. A retrospective analysis of patients with IIH found that the prevalence of OSA might be higher in these individuals than in an obesity-matched population [77]. While OSA is strongly associated with obesity and is known to be associated with IIH on the basis of case reports and series, the actual prevalence of IIH in OSA patients is unknown. In a study of 41 Chinese OSA patients that was directed at investigating glaucoma, 2 were found to have disc swelling compared with no patients in the control population [45]. Increased macular thickness and increased disc area on OCT was seen in patients with mild-moderate OSA (5
NEURO-OPHTHALMOLOGY (A KAWASAKI, SECTION EDITOR)
Obstructive Sleep Apnea and Optic Neuropathy: Is There a Link? Clare L. Fraser
# Springer Science+Business Media New York 2014
Abstract Over the last decade, there has been an emerging interest in the link between obstructive sleep apnea (OSA) and ocular health. Though the evidence for OSA playing a role in cerebrovascular disease risk seems clear, the same cannot be said for optic neuropathies. The association between OSA and glaucoma or non-arteritic anterior ischemic optic neuropathy (NAION) has been postulated to be secondary to direct hypoxia or mechanisms of optic nerve head vascular dysregulation. Papilledema and increased intracranial pressure have also been reported in OSA and are thought to be due to increased cerebral perfusion pressure and cerebral venous dilation secondary to hypoxia and hypercapnia. This article reviews the evidence for possible pathophysiological links between OSA and optic nerve pathology. The epidemiologic and clinical evidence for an association, direct or indirect, between OSA and glaucoma, non-arteritic anterior ischemic optic neuropathy (NAION), and papilledema or idiopathic intracranial hypertension is presented.
Keywords Obstructive sleep apnea . Glaucoma . Non-arteritic anterior ischemic optic neuropathy . Papilledema . Disc edema . Intracranial hypertension . Optic neuropathy
This article is part of the Topical Collection on Neuro-Ophthalmology C. L. Fraser Save Sight Institute, University of Sydney, 8 Macquarie Street, Sydney, NSW 2000, Australia C. L. Fraser (*) Department of Ophthalmology, Macquarie University, 3 Technology Place, Sydney, NSW 2109, Australia e-mail: [email protected]
Introduction Obstructive sleep apnea (OSA) is characterized by intermittent upper airway obstruction during sleep, resulting in apnea and hypoxia. The prevalence of OSA in community-screened patients varies between 2 %–6 % for moderate disease, and up to 14 % for mild disease [1]. Therefore, it is very likely that a similar proportion of our ophthalmology patients will have OSA. Those with OSA have a significantly higher body mass index (BMI) than those without, 31.4 vs 28.3, respectively [2]. Other risk factors to consider in patients include: male gender, older age, upper airway abnormalities, and enlarged neck girth [3]. In a clinical outpatient setting, the most useful observation for identifying patients with OSA is a history of nocturnal choking or gasping [1]. Though snoring is present in 94 % of those with OSA, it is not useful in establishing a higher risk for diagnosis as not all snorers have OSA [2]. The Berlin Questionnaire is a validated scale that can be easily administered in clinic to reliably identify those likely to have OSA [4]. A low score in response to this questionnaire identifies patients unlikely to have OSA, and has been shown to be of use as a screening tool in recognizing ophthalmic patients who do not need further investigation with sleep studies [5]. Overnight diagnostic polysomnography (dPSG) performed in a sleep laboratory is used as the reference standard for diagnosing OSA [2]. The dPSG measures respiratory indices, the electroencephalogram, and limb movements during sleep. The apnea-hypopnea index (AHI) is the sum of the apnea (cessation of oro-nasal flow for >10 seconds) and hypopnea (reduced oro-nasal flow of >50 % with a 4 % oxygen desaturation) events during an hour of sleep. In most papers OSA is defined as an AHI≥5, with an AHI of 5–15 considered mild, 16–30 moderate, and >30 being severe OSA [3]. The respiratory disturbance index (RDI) is defined as the sum of apneas, hypopneas, and abnormal respiratory events per hour
465, Page 2 of 9
of sleep. However, these terms are used interchangeably in some studies, and the cut-off levels for diagnosis of OSA vary between studies. Continuous positive airway pressure (CPAP) therapy is the treatment of choice, using airway pressure to effectively splint the throat open, thereby reducing apneas and hypopneas.
Basic Systemic Physiology of Obstructive Sleep Apnea The intermittent upper airway obstruction in OSA results in hypoxia (reduced arterial partial pressure oxygen, PaO2) and hypercapnia (increased PaCO2). This hypoxia results in sympathetic activation, oxidative stress, metabolic dysfunction, and systemic inflammation [6]. These are considered surrogate markers of cardiovascular risk, as they are all intermediary mechanisms that can lead to increased rates of hypertension, ventricular dysfunction, and early atherosclerosis. OSA is increasingly recognized as an independent risk factor for hypertension, stroke, and coronary heart disease [7]. Elevation in the odds of hypertension is observed in even those with mild OSA [8]. Epidemiologic studies suggest an association between OSA and increased atherosclerosis and large vessel endothelial disease [9]. However, the mechanistic link remains unclear. The Wisconsin Sleep Cohort Study showed that AHI is an independent predictor of hypertension, increased body mass index (BMI), insulin resistance, and cardiac dysfunction [6]. The metabolic dysfunction found in OSA exacerbates a tendency to obesity, with increased insulin resistance and nonalcoholic fatty liver disease [10]. Interestingly, obese patients without OSA do not show the same increase in markers of metabolic dysfunction. This suggests that OSA is an independent risk factor for these metabolic changes, rather than OSA being an epiphenomenon of obesity. OSA has also been found to be an independent risk factor for stroke in large epidemiology studies [11••]. The relationship between OSA and small-vessel cerebrovascular disease is complicated by the fact that the other underlying risk factors, namely hypertension, diabetes mellitus, and atherosclerosis, are also associated with OSA. Therefore, any relationship between OSA and small-vessel disease could be due to confounding factors [11••].
Physiology of Optic Nerve Pathology in OSA The interaction among various physiological parameters during sleep, including ocular perfusion pressure, aqueous outflow in the supine position, and circulating hormone levels, complicate the prediction of vascular changes and intraocular pressure (IOP) fluctuations overnight. The association between OSA and glaucoma or non-arteritic anterior ischemic
Curr Neurol Neurosci Rep (2014) 14:465
optic neuropathy (NAION) is postulated to be secondary to direct hypoxia or mechanisms of vascular dysregulation at the optic nerve head [12•]. 1. Direct Hypoxia Repetitive apneic episodes might lead to direct anoxic damage to the optic nerve [13, 14]. As the AHI increases, more hypoxia results in reduced mean oxygen saturation. However, Sergi et al. found no relation between ophthalmic measures of glaucoma and post-apneic falls of PaO2, change in oxygen saturation or the percentage of sleep time spent with oxygen saturation less than 90 % [15]. 2. Atherosclerosis and Arterial Blood Pressure Variation OSA has been shown to be associated with hypertension [16], endothelial cell dysfunction [17], and autonomic dysfunction [18] all of which may alter ocular vascular regulation. Both animal and clinical data shows that OSA promotes cellular changes in the vascular wall, in particular the carotid artery intima media thickness. The occurrence and progression of atherosclerotic plaques also appears related to the AHI, in the absence of significant comorbidity [17]. Endothelial function is also impacted by the oxidative stress of OSA, with resultant reduction in repair capacity and changes in vasomotor tone. However, these studies focus on larger peripheral arteries [19]. These changes could result in small vessel hyalinosis, thereby causing optic nerve head ischemia, though further histopathologic studies focusing on the optic nerve are required. Prolonged sympathetic overactivity is a characteristic feature of OSA, with bursts of sympathetic activity occurring before each nocturnal arousal and during the day in response to sleepiness. However, the role of the sympathetic system in regulating cerebral and ocular circulation remains a matter for ongoing research efforts. Physiological research in humans indicates that brain perfusion is significantly distorted in OSA due surges in blood pressure and blood velocity [20•]. Repetitive apneic episodes have been shown to have indirect effects on optic nerve head blood flow due to sympathetic surge-induced arterial hypertension [13, 14]. 3. Impaired Auto-Regulation at the Nerve Head If cerebral perfusion pressure decreases due to increased intracranial pressure (ICP) during apneic episodes, then any cerebral hypoxia will be worsened. Furthermore, dynamic cerebral blood flow auto-regulation is impaired in response to hypo-perfusion in OSA patients [21]. These fluctuations and impaired responses put the brain at risk of vascular damage and ischemia. Doppler studies on the ophthalmic and central retinal arteries found no difference in the systolic and enddiastolic blood velocities (resistivity index) between OSA patients and controls [22]. The posterior ciliary
Curr Neurol Neurosci Rep (2014) 14:465
arteries and the choroidal vessels supply the prelaminar and laminar portions of optic nerve head, as the nerve travels through the sclera. Only the surface layer of the optic disc has capillaries derived from retinal arterioles, but even there it is not uncommon for this region to be supplied by choroidal vessels [23]. Given that most optic disc pathology is thought to arise from changes in the prelaminar and laminar sections, the choroidal circulation is of most interest in pathophysiology. There is evidence from various human studies that retinal blood flow increases during hypercapnia [24–26]. A linear relationship exists between choroidal blood flow as measured on a laser Doppler flowmeter and PaCO2, however there is no effect from changes in PaO2 [27]. On examination of OSA patients compared with controls, there was no difference in choroidal blood flow reactivity to different inhaled concentrations O2 or CO2, and no change after CPAP treatment [28]. However, this study did not look at obese patients, or those who had a protracted course of OSA. Data suggest that cerebral blood flow is abnormal in OSA, with blood flow velocity surges [20•]. Under normal conditions, the pial arteries act to stabilize and counteract changes in cerebral blood flow. However, hypercapnia impairs the pial artery responses to changes in cerebral blood flow [29]. The orbital portion of the optic nerve and the intra-cranial part of the optic nerve are supplied by pial circulation. Therefore, if the pial system at the optic nerve head can no longer compensate for changes in cerebral blood flow, impaired auto-regulation at the nerve head may be the site of ocular pathology in OSA. 4. Imbalance Between Vasodilator and Vasoconstrictor Substances Intermittent hypoxia promotes production of reactive oxygen species, increasing oxidative stress, activating systemic inflammation, and leading to decreased bioavailability of endothelial nitric oxide. Vascular dysregulation due to nitric oxide/endothelin imbalance or abnormal platelet aggregation also has indirect effects on optic nerve head blood flow [30]. Recent research has focused on the association between retinal micro-vascular dysfunction, cardiovascular disease and OSA. Boland et al. found no association with OSA and the ratio of retinal arteriolar narrowing and venular widening [31]. However, retinal arteriolar changes are associated with hypertension, but retinal venule changes are associated with inflammatory and metabolic abnormalities [32]. Shankar et al. analyzed retinal vessels independently, and found that AHI was associated with retinal venular widening independent of confounders, but not associated with retinal arteriolar narrowing [9]. In a qualitative study of retinal photos, changes resembling hypertensive retinopathy (arteriolar narrowing, arterio-
Page 3 of 9, 465
venous-nipping and retinal hemorrhages) have been associated with AHI, independent of blood pressure [33]. These potential associations with microvascular changes in the retina are of the most interest when considering the link between OSA and optic nerve pathology, because the same changes may be occurring in the choroidal and posterior ciliary vessels.
Glaucoma The association between glaucoma and OSA has been widely reported, However, the results vary and a clear link is still uncertain. The prevalence of sleep-disordered breathing is between 20 %–50 % in patients with primary open angle glaucoma (POAG) [34, 35] and normal tension glaucoma (NTG) [36]. The inverse is also reported, with 7.2 % of sleep apnea patients having glaucoma, compared with the expected population rate of 2 % [37]. However, other studies have found no difference in the prevalence of glaucoma within an OSA population compared with the normal population. Geyer et al. found no relation between the RDI and intra-ocular pressure (IOP) or the presence of glaucoma [38]. While it is difficult to compare the various studies examining the association between glaucoma and OSA due to small numbers, different criteria of OSA, and differing means of glaucoma assessment, the following paragraphs provide a summary of the recent research. Retinal Nerve Fiber Layer (RNFL) Thickness The link between OSA and glaucoma has been investigated by examining the preclinical changes in RNFL thickness. The appearance of RNFL thinning suggests atrophy, presumably due to decreased ocular perfusion, hypoxia, and vasospasm. A trend for loss of RNFL to correlate with the severity of OSA was reported in a small group of patients by Kargi et al., using scanning laser polarimetry [39]. In a study of 105 patients and 22 controls examined with Stratus ocular coherence tomography (OCT), Lin et al. also found that RNFL was lower in patients with moderate/severe OSA (AHI >15) compared with those with mild OSA (515) than controls, 14.2+/−3.5 vs 12.2+/−3.6 mmHg [39]. Other studies have shown positive correlation between IOP and increasing AHI scores [22, 15]. However, the study inclusion criteria was of IOP 15), and diagnosed glaucoma in 27 %, a higher prevalence than expected [49]. However, there was no evidence for a relationship between AHI and IOP or the diagnosis of glaucoma (disc and field changes) - only increasing age correlated with glaucoma diagnosis. Examining a subset of the Beijing Eye Study, data on snoring (as reported by the subject’s life partner) and glaucoma was available for 3146 patients [50]. “Severe snoring” was not associated with open-angle glaucoma, angle-closure glaucoma, cup:disc ratio, or RNFL. Another interview based study found a higher prevalence of sleep-disordered breathingrelated symptoms in 212 POAG patients than controls [35]. A cross-sectional study that reviewed diagnoses given to patients in a managed care program, found that there was no difference in the hazard of being diagnosed with POAG or NTG in those with OSA (regardless of CPAP treatment) and those without OSA [51]. Using diagnostic codes to identify comorbities, Girkin et al. found that individuals who developed glaucoma were more likely to have a previous sleep apnea diagnosis relative to controls, However, once adjusting for other vascular and glaucoma risk factors (hypertension, migraine, diabetes) no significant difference was seen [52]. In a large retrospective population based study, Lin et al. compared the 5-year hazard ratio for developing open-angle glaucoma between 1012 subjects after a diagnosis of OSA (based of dPSG) and 6072 controls [53••]. All patients and controls were known to have attended an ophthalmologist
Curr Neurol Neurosci Rep (2014) 14:465
without being diagnosed with glaucoma in the 4 years prior to the “index year”. Records for OSA patients and the controls were then examined from the index year forwards over 5 years. After adjusting for multiple confounding factors, the hazard ratio for developing glaucoma after OSA was 1.67 that of controls. This study concludes that OSA was independently associated with an increased risk of POAG. The data for a link between OSA and glaucoma is variable. The pathophysiological association is plausible, and though there does seem to be consistent evidence for loss of RNFL, the clinical significance needs to be reviewed. The direct link to NTG in OSA appears stronger than the case for high IOPrelated POAG.
Non-arteritic Anterior Ischemic Optic Neuropathy NAION is an ischemic insult to the optic nerve head, and an annual incidence in persons 50 years and older is between 2.3 and 10.2 per 100,000 [54••]. Typically found in patients over the age of 50, a small cup to disc ratio “disc at risk” is probably one of the chief risk factors for NAION. NAION is also associated with acquired vascular risk factors such as systemic arterial hypertension, diabetes, and atherosclerosis. Circulatory insufficiency within the optic nerve head is thought to precipitate NAION, however, the exact location of the vasculopathy remains unknown [55•]. Histopathologic studies have implicated the retrolaminar region as being the area of infarction, with fluorescein studies finding delayed filling in the prelaminar optic disc [56]. Further research is needed to determine if and how OSA contributes to an isolated vascular insult or compartment syndrome within the optic nerve. In a review of 544 episodes of NAION at least 73 % were discovered upon first awakening, leading to the hypothesis that nocturnal hypotension might be a precipitating factor [57]. However, the hypoxia of OSA promotes hypertension, producing a nocturnal non-dipping pattern otherwise known as “masked hypertension” [58]. Therefore, other vascular changes in OSA must be related to the NAION process, if a link truly exists. As mentioned previously, vascular autoregulation may be hindered by sympathetic surges or imbalance of vasoactive substances such as endothelin-1 (vasoconstrictor) and nitric oxide (vasodilator) [53••], triggering NAION. In 2 studies of NAION, 71 %–89 % had OSA diagnosed on dPSG (RDI >10 or AHI >15) compared with 18 % of a control population [59, 60]. In addition, OSA was 1.5–2 times more frequent than other known risk factors such as hypertension and diabetes. However, in one of the two aforementioned studies, the 17 patients and controls were only matched by age and sex
Page 5 of 9, 465
[58]. In the other study, the prevalence of OSA in NAION was compared with published population studies, showing a 4.9 relative risk of OSA in NAION patients [59]. More recently Arda et al. performed polysomnography on 20 NAION patients and 20 controls matched for age, sex, body mass index (BMI), diabetes, and hypertension. Of those diagnosed with NAION 85 % had OSA, compared with 65 % in controls. By matching for hypertension and diabetes, which are found in association with OSA, these rates reported are much higher than expected for the general population. The high rates may also have been found because the study used an AHI >5 for OSA [61]. In a prospective study of 27 NAION patients within 1 month of symptom onset underwent dPSG. Controls who were statistically similar for systemic risk factors and BMI [54••] also underwent dPSG. A diagnosis of OSA was made, based on AHI >20, in 55.6 % of NAION patients and 22.2 % of controls. Using these stricter criteria for OSA than other studies, and appropriate matching for hypertension, diabetes, hypercholesterolemia, and coronary artery disease this study concluded that OSA should be considered a significant risk factor for NAION. These studies using dPSG are time consuming and have small sample sizes. By using the validated Sleep Apnea Scale of the Sleep Disorders Questionnaire (SA-SDQ), Li et al. were able to assess 73 NAION cases and 88 controls [62]. The patients included more women, and patients with a higher BMI than previous studies. They still found an increased prevalence of presumed OSA, 30 % vs 17.8 % in those with NAION, and an odds ratio of 2.62 (1.03–6.60) when controlling for some confounders. However, the temporal relationship could not be established. The prevalence of NAION in a population with diagnosed OSA has also been examined. In a large retrospective cohort study based on billing records, it was found that individuals diagnosed with OSA and not treated with CPAP had a 16 % increased hazard of developing NAION [50]. This increased risk was not found in those with treated OSA. However, reports of patients developing NAION while undergoing CPAP therapy have also been published [63]. There are no studies that look at whether CPAP can prevent NAION in patients with other risk factors. Given that there is no effective treatment for NAION, management of all potential risk factors including hypertension, diabetes and OSA is advisable, even though based on evidence to date, a direct causal relationship cannot be presumed.
Physiology of Cerebrospinal Fluid Pressure in OSA Several mechanisms have been proposed as the cause of increased cerebrospinal fluid (CSF) pressure in OSA [64].
465, Page 6 of 9
1. Increased Cerebral Perfusion Pressure Increased systemic blood pressure seen in OSA may result in a secondary increase in cerebral perfusion pressure, and therefore increased ICP. There is a strong correlation between the duration of apneic episodes and blood pressure fluctuations [64]. However, animal studies have shown that acute hypertension is tolerated with minimal perturbation of cerebral blood flow, whereas only if there was pre-existing intracranial hypertension was there a loss of autoregulation in response to arterial hypertension [65]. Human work has shown that cerebral blood flow velocities are lower in patients with OSA than normal controls [66], whereas other studies show a steady increase in cerebral blood flow during an episode of apnea [11••]. 2. Hypoxic and Hypercapnia Related Cerebral Vasodilation Vasodilation and increased brain water content from alteration in cerebral autoregulation may change intracranial volumes. A reduction in brain volume of 4 %, without focal changes, has been documented in OSA patients after treatment with overnight oxygen [67]. The increase in cerebral venous blood volume may be responsible or partially responsible for elevated ICP in OSA. Increases in ICP are seen in both rapid eye movement (REM) and non-REM stages of sleep. During overnight CSF monitoring of 3 OSA patients, ICP was episodically elevated in parallel to apneic episodes and hypoxia [64]. In subsequent study of patients with severe OSA, awake values of ICP were demonstrated to be elevated, with further increases in ICP related to apneic episodes while asleep [68]. Sustained elevation of ICP causes papilledema due to disruption of retrograde axonal transport at the optic disc. Two patients with intracranial hypertension but no disc edema had overnight ICP monitoring, and dPSG. Both were obese (BMI >30), had severe OSA with AHI >30 and had rapid sustained rises in ICP associated with hypoxemia. However, blood pressure did not change during these episodes, suggesting that cerebral vasodilation and increases in cerebral blood volume were the precipitating factors [69]. In another case series, a patient was shown to have lumbar puncture opening pressure of 16 cmH2O in the day, but overnight ICP monitoring revealed pressures of 48 cmH2O during episodes of oxygen desaturation [70]. The question remains as to whether these intermittent changes in ICP alone would be sufficient to cause papilledema. Apneic events are also associated with hypercapnia, which has been shown in some studies to lead to cerebral venous dilation. Hypercapnia has long been reported to decrease vascular resistance and increases cerebral blood flow. However, recent data has altered our understanding of the dynamic cerebral blood flow changes that occur in
Curr Neurol Neurosci Rep (2014) 14:465
OSA with changes in blood pressure [19]. Given the limited number of studies and the conflicting results, final conclusions cannot be drawn. 3. Mechanical The mechanical airway obstruction in OSA, leads to increased respiratory effort, increased intra-abdominal pressure and intra-pleural pressure resulting in an impediment to cerebral venous drainage [71]. Cerebral venous hypertension reduces CSF re-absorption through the arachnoid villi, thereby increasing ICP.
Papilledema and Idiopathic Intracranial Hypertension The first report suggesting a link between OSA and disc edema found resolution of disc edema after upper airway surgery in the setting of normal ICP [72]. A case series from Purvin et al. described 4 patients with bilateral optic disc edema and OSA [70]. Treatment with CPAP has been reported to precede resolution of the disc swelling, consistent with an association but does not prove a direct link [70, 73]. In a study using diagnostic codes from a large managed care organization, Stein et al. found that after adjusting confounding factors, untreated patients with OSA had a 29 % higher hazard and patients with CPAP-treated OSA had over twice the hazard of papilledema (105 % higher) when compared with patients without OSA [50]. However, in seeming contradiction to this, the same paper reports similar rates of idiopathic intracranial hypertension (IIH) between those without OSA and CPAP-treated OSA, but that the untreated OSA patients had over twice the hazard of IIH (103 %) compared with those without OSA. Marcus et al. found that of 53 patients with IIH, 70 % had evidence of sleep disturbance. Fourteen of these patients then went on to dPSG, with 13/14 being subsequently diagnosed with OSA [74]. In a larger review of 721 IIH patients, Bruce et al. found 4 % of women had OSA compared with 24 % of men [75]. Lee et al. found 6/18 men with IIH also had a diagnosis of OSA based on dPSG [73]. There has only been one small study regarding the prevalence of papilledema and symptoms of increased ICP in OSA patients. Of 95 recently diagnosed OSA patients contacted, only 35 had an eye examination, none of who were found to have papilledema [71]. However, 34 % gave a history of transient visual obscurations. A study of fundus photographs taken on 250 patients undergoing dPSG, found no patients with optic disc edema (95 % CI: 0 %–3 %) [32]. In a further study, patients with newly diagnosed IIH underwent dPSG, the AHI values were then compared with the population-based expected AHI for each patient based on
Curr Neurol Neurosci Rep (2014) 14:465
age, sex, race, BMI, and menopausal status. The prevalence and severity of OSA in IIH patients was not significantly different from predicted values [76]. A retrospective analysis of patients with IIH found that the prevalence of OSA might be higher in these individuals than in an obesity-matched population [77]. While OSA is strongly associated with obesity and is known to be associated with IIH on the basis of case reports and series, the actual prevalence of IIH in OSA patients is unknown. In a study of 41 Chinese OSA patients that was directed at investigating glaucoma, 2 were found to have disc swelling compared with no patients in the control population [45]. Increased macular thickness and increased disc area on OCT was seen in patients with mild-moderate OSA (5
