The Association Between Obstructive Sleep Apnea and Cotton-Wool Spots in Nonproliferative Diabetic Retinopathy

Abstract

Purpose:

This work reports the association of obstructive sleep apnea (OSA) and cotton-wool spots (CWS) seen in patients with nonproliferative diabetic retinopathy (DR).

Methods:

A random sample of patients diagnosed with DR between January 1, 2015 and December 31, 2018, were selected from medical-billing codes. Dilated funduscopic examination findings and medical history were analyzed by reviewing medical records.

Results:

CWS were present in 12 of 118 patients without OSA, compared with 11 of 32 patients with OSA (10.2% vs 34.4%, respectively; P = .002). OSA was more common in men (68.8%, P = .03) and associated with a higher body mass index (30.0 ± 5.0 without OSA vs 33.6 ± 5.5 with OSA, P < .001). When comparing those with and without OSA, there was no association with age; glycated hemoglobin A_1c; stage of DR; insulin dependence; presence of diabetic macular edema; smoking status; or a history of hypertension, hyperlipidemia, cardiovascular disease, or other breathing disorder.

Conclusions:

The presence of OSA is associated with CWS in patients with DR, as well as male sex and a higher body mass index. Further research is needed to determine the ophthalmologist’s role in the timely referral of patients with CWS for OSA evaluation.

Introduction

Obstructive sleep apnea (OSA) is the cessation of breathing during sleep that leads to hypoxia; it is known to increase a patient’s overall risk of death. ¹ The hypoxia induced by nightly cessation of breathing leads to oxidative stress, increased blood pressure, sympathetic nervous system activation, and systemic inflammation. ² This increases the risk of hypertension (HTN), glucose intolerance, cardiovascular disease (CVD), daytime somnolence, home- and work-related accidents, and worsened quality of life. ^{3 -5} In addition, patients with diabetes may have worse sleep-disordered breathing owing to autonomic neuropathy. ⁶ Nocturnal episodes of hypoxia are associated with inflammation, oxidative stress, and increased vascular endothelial growth factor, all of which may contribute to worsened microvascular and macrovascular complications. ^7,8

Furthermore, OSA has important implications in ophthalmology. OSA is associated with floppy eyelid syndrome, ⁹ central serous chorioretinopathy, ¹⁰ and nonarteritic anterior ischemic optic neuropathy, ¹¹ among other diseases. It is possibly associated with glaucoma and preclinical peripapillary nerve fiber layer thinning in young adults. ¹² It is believed that untreated OSA contributes to the development and progression of diabetic retinopathy (DR), ^13,14 the leading cause of blindness among working-aged individuals in the United States. ¹⁵ Multiple studies have found that OSA increases the risk and severity of DR, and even mild OSA seems to be independently associated with vision-threatening DR. ^5,14,16,17 Some researchers even recommend screening all patients with type 2 diabetes for OSA because the coexistence of these entities is associated with a high prevalence of microangiopathy. ¹⁸

DR is characterized by neurovascular and microvascular changes believed to be mediated by oxidative stress, ¹⁹ formation of advanced glycation end products, ²⁰ and inflammation, ²¹ all of which are worsened in patients with OSA. ²² These microvascular changes are clinically detected as microaneurysms, intraretinal hemorrhages, hard exudates, cotton-wool spots (CWS), and neovascularization. ²³ CWS are collections of axoplasmic material in the nerve fiber layer widely believed to be caused by retinal ischemia, ^24,25 and anecdotal evidence suggests that the presence of CWS may be associated with untreated OSA in patients with DR. ²⁶ However, no studies exist to prove this association. To our knowledge, we performed the first retrospective study aimed at identifying the association between CWS and OSA in patients with DR.

Methods

Study Participants

This was a multicenter, clinic-based, retrospective cross-sectional study. All electronic health record identification numbers of patients with an International Statistical Classification of Diseases and Related Health Problems, 10th revision code diagnosis of DR (including all 96 subcodes) who were treated within the San Antonio Military Health System between January 1, 2015 and December 31, 2018, were collected. Specifically, patients were treated at either the Wilford Hall Ambulatory Surgical Center or the Brooke Army Medical Center. A total of 1482 unique patient identification numbers were identified with a diagnosis of DR. Two hundred of these (13.5%) were blindly picked and investigated further. Fifty patients were excluded based on the criteria listed later.

A prior diagnosis of OSA was confirmed based on testing documented within the patient’s medical record, according to the American Academy of Sleep Medicine guidelines. ²⁷ Diagnostic criteria included: (1) clinical history of daytime somnolence with reports of apneic episodes at night; (2) polysomnography with more than 5 episodes of apnea/hypopnea per hour of sleep with evidence of respiratory effort; and (3) no other diagnoses that could better account for the patient’s symptoms.

Participants were excluded from the study if they had no confirmed diagnosis of DR, if they had any history of proliferative DR (PDR), or if they were lacking significant demographic or medical history data within their medical record. They were also excluded if they had any other confounding retinal conditions, such as retinal vein occlusions, macular degeneration, epiretinal membranes, ischemic optic neuropathy, or central serous chorioretinopathy.

Review of Medical Records

All data were manually screened by S.D.H. and another coauthor to verify accuracy of the data. Each participant was counted only once in the study, despite whether he or she had clinical findings of DR or CWS in 1 or both eyes. Only the first clinical examination within the time frame of the study was used to evaluate the stage of DR (Airlie House classification of DR), ²⁸ the presence of CWS, and all demographic and medical history data. Glycated hemoglobin A_1c (HbA_1c) level and body mass index (BMI) had to be recently measured within 6 months of the ophthalmological evaluation.

Statistical Analysis

The characteristics of patients with or without OSA and with or without CWS were compared using χ² tests for binomial variables and independent t tests for continuous variables. The Wilcoxon/Kruskal-Wallis tests (rank sums) were used when applicable, assuming nonnormal distribution, to verify the results. Factors significantly associated with CWS were included in a multifactorial logistic regression to determine whether significance was maintained after adjustment. All statistical analyses were performed using SPSS Statistics version 26 for Windows (IBM Corp). A P value of less than .05 was considered statistically significant.

Results

Study Participant Characteristics

A total of 150 patients (78 men [50.5%]) were included in the analysis. Participants were between ages 25 and 94 years (mean, 65.2 ± 11.0 years) at the time of funduscopic examination and medical-record review. Seventy-nine patients were using insulin (insulin dependent), whereas 71 were non–insulin dependent (52.7% vs 47.3%). Thirty-two (21.3%) had a known diagnosis of OSA. There were 83 (55.3%) patients diagnosed with mild nonproliferative DR (NPDR), 55 (36.7%) with moderate NPDR, and 12 (8.0%) with severe NPDR.

Main results are summarized in Table 1. Twenty-three patients were found to have CWS on funduscopic examination (15.3%). There was no association with sex (56.5% male with CWS vs 51.2% male without CWS, P = .64). Patients with CWS were significantly younger than patients without CWS (59.4 ± 10.9 vs 66.4 ± 10.7 years, respectively; P = .008) and had a higher average HbA_1c (9.0 ± 2.2 vs 7.9 ± 1.8, respectively; P = .01). Patients with CWS were more likely to have a worse stage of NPDR than those without CWS (26.1% vs 60.6% mild NPDR, 52.2% vs 33.9% moderate NPDR, 21.7% vs 5.5% severe NPDR; P = .008). Patients with CWS were no more likely to be insulin dependent (65.2% vs 50.4%, respectively; P = .19) or have diabetic macular edema (DME) (21.7% vs 13.4%, respectively; P = .32) than patients without CWS.

Table 1.

Comparison of Individuals With Obstructive Sleep Apnea and Those Without, as Well as Those With and Without Cotton-Wool Spots on Funduscopic Examination.

Characteristic	No OSA (n = 118)	OSA (n = 32)	P value	No CWS (n = 127)	CWS (n = 23)	P value
Age, y	65.7 ± 11.2	64.0 ± 10.2	P = .79	66.4 ± 10.7	59.4 ± 10.9	P = .008b
Sex			P = .03a			P = .64
Male	56 (47.5%)	22 (68.8%)		65 (51.2%)	13 (56.5%)
Female	62 (52.5%)	10 (31.2%)		62 (48.8%)	10 (43.5%)
CWS/OSA			P = .002b			P = .002b
(−) CWS	106 (89.8%)	21 (65.6%)	(−) OSA	106 (83.5%)	12 (52.2%)
(+) CWS	12 (10.2%)	11 (34.4%)	(+) OSA	21 (16.5%)	11 (47.8%)
Type of diabetes			P = .10			P = .19
IDDM	58 (49.2%)	21 (65.6%)		64 (50.4%)	15 (65.2%)
NIDDM	60 (50.8%)	11 (34.4%)		63 (49.6%)	8 (34.8%)
Stage of DR			P = .64			P = .008b
Mild NPDR	65 (55.1%)	18 (56.3%)		77 (60.6%)	6 (26.1%)
Mod NPDR	45 (38.1%)	10 (31.2%)		43 (33.9%)	12 (52.2%)
Severe NPDR	8 (6.8%)	4 (12.5%)		7 (5.5%)	5 (21.7%)
DME			P = .32			P = .32
(-) DME	99 (83.9%)	29 (90.6%)		110 (86.6%)	18 (78.3%)
(+) DME	19 (16.1%)	3 (9.4%)		17 (13.4%)	5 (21.7%)
HbA1c	8.0±1.9	8.2±1.9	P = .35	7.9±1.8	9.0±2.2	P = .01a
BMI	30.0±5.0	33.6±5.5	P < .001c	30.7±5.5	30.9±4.3	P = .44
Hx of HTN			P = .91			P = .61
(-) HTN	8 (6.8%)	2 (6.3%)		9 (7.1%)	1 (4.4%)
(+) HTN	110 (93.2%)	30 (93.7%)		118 (92.9%)	22 (95.6%)
Hx of HLD			P = .49			P = .87
(-) HLD	13 (11.0%)	5 (15.6%)		15 (11.8%)	3 (13.0%)
(+) HLD	105 (89.0%)	27 (84.4%)		112 (88.2%)	20 (87.0%)
Smoking			P = .18			P = .59
Nonsmoker	78 (66.1%)	19 (59.4%)		80 (63.0%)	17 (73.9%)
Prior smoker	24 (20.3%)	11 (34.4%)		31 (24.4%)	4 (17.4%)
Active smoker	16 (13.6%)	2 (6.2%)		16 (12.6%)	2 (8.7%)
Hx of CVD			P = .45			P = .19
(-) CVD	75 (63.6%)	18 (56.3%)		76 (59.8%)	17 (73.9%)
(+) CVD	43 (36.4%)	14 (43.7%)		51 (40.2%)	6 (26.1%)
BD			P = .87			P = .14
(-) BD	108 (91.5%)	29 (90.6%)		118 (92.9%)	19 (82.6%)
(+) BD	10 (8.5%)	3 (9.4%)		9 (7.1%)	4 (17.4%)

Abbreviations: BD, breathing disorder (including asthma, chronic obstructive pulmonary disease, pulmonary hypertension, dysphonia, or other breathing-related disorders); BMI, body mass index; CVD, cardiovascular disease; CWS, cotton-wool spots; DME, diabetic macular edema; DR, diabetic retinopathy; HbA_1c, glycated hemoglobin A_1c; HLD, hyperlipidemia; HTN, hypertension; Hx: history; IDDM, insulin dependent diabetes mellitus; Mod, moderate; NIDDM, non–insulin–dependent diabetes mellitus; NPDR, nonproliferative diabetic retinopathy; OSA, obstructive sleep apnea.

^a Significant at P < .05.

^b Significant at P < .01.

^c Significant at P < .001.

OSA was present in 11 of 32 (34.4%) patients with CWS on examination and only 21 of 127 (16.5%) patients without CWS (P = .002). In patients with and without CWS, respectively, there were no differences in BMI (30.9 ± 4.3 vs 30.7 ± 5.5, P = .44), history of HTN (95.6% vs 92.9%, P = .87) or hyperlipidemia (87.0% vs 88.2%, P = .87), smoking status (73.9% vs 63% nonsmoker, 17.4% vs 24.4% prior smoker, 8.7% vs 12.6% current smoker; P = .59), or history of CVD (eg, coronary artery disease, history of myocardial infarction or cerebrovascular accident, and peripheral arterial disease) (26.1% vs 40.2%, P = .14).

Similarly, CWS were more common in patients with OSA than those without (34.4% vs 10.2%, respectively; P = .002). OSA was more common in men (68.8% male, P = .03) and associated with a higher BMI (30.0 ± 5.0 without OSA vs 33.6 ± 5.5 with OSA, P < .001). When comparing those with and without OSA, respectively, there was no association with age (64.0 ± 10.2 vs 65.7 ± 11.2 years, P = .79); HbA_1c (8.2 ± 1.9 vs 8.0 ± 1.9, P = .35); stage of DR (56.3% vs 55.1% mild NPDR, 31.2% vs 38.1% moderate NPDR, 12.5% vs 6.8% severe NPDR; P = .64); insulin dependence (65.6% vs 49.2%, P = .10); presence of DME (9.4% vs 16.1%, P = .32); smoking status (59.4% vs 66.1% nonsmokers, 34.4% vs 20.3% prior smokers, 6.2% vs 13.6% active smokers; P = .18); or a history of HTN (93.7% vs 93.2%, P = .91), hyperlipidemia (84.4% vs 89.0%, P = .49), or CVD (43.7% vs 36.4%, P = .45).

Other breathing disorders were also considered, including any history of asthma, chronic obstructive pulmonary disease, pulmonary HTN, or dysphonia. These were not found to be associated with CWS on examination (17.4% with CWS vs 7.1% without, P = .14) or with a diagnosis of OSA (9.4% with OSA vs 8.5% without, P = .87).

Conclusions

This is the first study to our knowledge to evaluate CWS as part of the spectrum of DR and their association with OSA. Our data suggest a strong association between the coexistence of CWS in NPDR and a diagnosis of OSA, with patients 3.4 times more likely to exhibit CWS on examination if they had previously been diagnosed with OSA than if they had not (P = .002). Logistic regression to control for age and presence of HTN demonstrated that CWS are still more common in patients with OSA. We also found OSA to be more prevalent in men and in those with a higher BMI, findings that have been extensively described elsewhere.

In addition, this study shows that CWS are more common with more severe stages of NPDR. Although this association likely exists for patients with PDR as well, the decision was made to exclude these patients from the present study. Treatment with laser or antivascular endothelial growth factor medications affects disease activity and presentation, and many patients with a history of PDR have quiescent disease that is more likely to behave like nondiabetic eye disease or like a mild NPDR. In addition, CWS may be more challenging to identify in the presence of laser scars within the retina.

There are several reasons why CWS may be present more often in patients with OSA than those without. CWS are widely believed to be caused by ischemia, leading to impaired axoplasmic flow and accumulation of axoplasmic material. ²⁴ For this reason, CWS are often seen in other diseases such as ocular ischemic syndrome, malignant HTN, anemia, and hyperviscosity syndrome. Patients with diabetes are prone to retinal ischemia from hyperglycemia, which leads to endothelial dysfunction, leukostasis, and capillary occlusion. ²⁹ Frequent hypopneic or apneic episodes during sleep may worsen retinal ischemia in patient with diabetes. In addition, OSA is a common cause of secondary HTN, ³⁰ and patients with diabetes and OSA may be prone to develop retinal findings of both diabetic and hypertensive retinopathy, including CWS.

Several previous studies have explored OSA’s association to DR. Leong et al performed a meta-analysis of 1 cohort study and 15 cross-sectional studies, demonstrating that OSA was associated with a greater severity of DR in patients with type 2 diabetes. ³¹ However, they found insufficient evidence that OSA contributed to the development of DR or that OSA contributed to DME. Despite these findings, Zhu and colleagues conducted a separate meta-analysis of 6 studies (n = 608 for individuals with OSA vs n = 484 for individuals without OSA) and used a fixed-effects model that showed OSA to be a risk factor for DR in patients with either type 1 or type 2 diabetes. ¹⁴ These results are convincing, although only OSA was analyzed as a risk factor for DR, neglecting to account for the numerous risk factors shared between OSA and DR.

Chang et al performed a retrospective review of 317 patients with both DR and OSA. Although they did not evaluate for the presence of CWS, they determined OSA severity and performed overnight polysomnography, demonstrating an association between DR and severe OSA. ¹⁶ In addition, they found an association between severe OSA and both PDR and DME. Altaf et al performed a sizable longitudinal study that included 230 patients with no known respiratory disorders or OSA. They screened all patients for OSA and found a higher prevalence of vision-threatening DR in patients with OSA than those without. ¹⁷ After follow-up for longer than 3 years, the patients with OSA were more likely to develop severe NPDR or PDR as compared with those patients without OSA, and treatment with continuous positive airway pressure reduced that risk.

The present study differed from previous ones in that it attempted to determine the association of a single examination finding (CWS) with OSA. There are, however, several limitations to the present study. It was beyond the scope of this paper to determine the severity of a patient’s OSA and whether it was being adequately treated, or whether other patients in the study had unrecognized sleep-disordered breathing. In addition, there are other systemic medical conditions that may increase the occurrence of CWS, including collagen vascular diseases (such as systemic lupus erythematosus), hematologic abnormalities (such as anemia), clotting disorders, as well as cancers (such as leukemia). Future studies may aim to investigate the association of CWS with these diseases in patients with NPDR.

Each patient was examined by a single clinician, without confirming photographs or ocular coherence tomography imaging. The retrospective nature of the study and relatively small sample size limit the generalizability of our findings, especially with respect to subgroup analyses. As described in Methods, we analyzed 13.5% of patients identified with DR. Despite finding statistical significance within this group, future studies may investigate larger populations of patients.

Many questions remain. Would referring patients with diabetes and CWS help identify patients without obvious signs of OSA? Can instituting continuous positive airway pressure therapy change the funduscopic examination or improve the severity of DR? Further research is needed (including prospective and longitudinal data) to validate our findings and answer these important questions.

Several validated screening tools exist to identify patients with OSA, including the Berlin questionnaire, STOP-Bang Questionnaire, STOP Questionnaire, and the Epworth Sleepiness Scale. ³² However, these are not commonly used in the ophthalmologist’s clinic, and their diagnostic accuracy remains controversial. ³³ The ophthalmologist’s role in the early identification of patients with OSA has evolved over time, and more patients with floppy eyelids, central serous chorioretinopathy, and other known associations with OSA are now being referred for sleep studies. ³⁴ By identifying ocular associations with OSA, including CWS, ophthalmologists may help in the early diagnosis and treatment of OSA and its potentially fatal complications.