To investigate whether sleep-disordered breathing is a risk factor for iris and/or angle neovascularization in patients with proliferative diabetic retinopathy (PDR).
Cross-sectional comparative case series.
Subjects and Methods
One hundred fifty-one consecutive patients with PDR who underwent surgery in our hospital were divided based on the presence of iris and/or angle neovascularization (NV group, 37 patients) or absence of NV (non-NV group, 114 patients). Pulse oximetry was conducted during the night and the mean SpO 2 , 4% oxygen desaturation index (4% ODI times/hour), the lowest SpO 2 % during sleep (lowest SpO 2 ), and the cumulative percentage of time at SpO 2 <90% in analysis times (CT90%) were calculated. When the 4% ODI exceeded 5 times/hour, sleep-disordered breathing was diagnosed. The results were compared between the 2 groups. Preoperative systemic parameters also were analyzed by logistic regression to clarify risk factors for the NV group.
A mean total of 50% (62% of the NV group and 46% of the non-NV group) was diagnosed with sleep-disordered breathing. The mean SpO 2 and lowest SpO 2 did not differ significantly between the 2 groups; the 4% ODI (12.3 vs 6.6) and CT90% (3.8 vs 1.7) were significantly higher in the NV group ( P =.02, for both comparisons). Logistic regression analysis identified insulin therapy (odds ratio [OR], 3.01; 95% confidence interval [CI], 1.26∼7.20; P = .01); and 4% ODI (OR, 1.09; CI, 1.01∼1.16; P = .02) as risk factors for the NV group.
In patients with PDR, nocturnal intermittent hypoxia/reoxygenation resulting from sleep-disordered breathing may be a risk factor for iris and/or angle neovascularization.
Iris neovascularization (NV) and angle NV in patients with proliferative diabetic retinopathy (PDR) were reported to be poor prognostic factors for vitreous surgery. Furthermore, progression of neovascular glaucoma in severe cases is a risk factor for severe visual loss in patients with diabetic retinopathy (DR).
Sleep-disordered breathing has been associated with arteriosclerotic diseases and identified as a possible risk factor for macroangiopathies, such as hypertension, coronary artery disease, and cerebrovascular disease, or microangiopathic disorders, such as renal disease.
We previously studied the relevance of DR and sleep-disordered breathing in screened patients with nonproliferative diabetic retinopathy (NPDR) and PDR for sleep-disordered breathing and reported that patients with PDR have a high incidence of sleep-disordered breathing; the 4% oxygen desaturation index (4% ODI) was an independent risk factor contributing to the diagnosis of PDR. In a reevaluation using more sleep-disordered breathing parameters and a larger number of patients, we reported that severe nocturnal intermittent hypoxia and reoxygenation occurred in patients with PDR and the lowest percutaneous oxygen saturation (SpO 2 %) during sleep, which is 1 of the sleep-disordered breathing parameters identified as a risk factor for PDR.
In the current study, we conducted sleep-disordered breathing screening in patients with PDR and divided the patients into groups based on the presence or absence of iris and/or angle NV (NV group and non-NV group). Our goal was to determine whether the sleep-disordered breathing parameters in this study were independent risk factors for the NV group.
A total of 151 consecutive Japanese inpatients with PDR (NV group, 37 patients; non-NV group, 114 patients) who had type 2 diabetes were included. Patients underwent filtering surgery for neovascular glaucoma or vitreous surgery for macula edema, vitreous hemorrhage, and tractional retinal detachment from April 1, 2006, to April 1, 2008, at the Department of Ophthalmology, Toho University Sakura Medical Center.
The same observer (Yu.S.) performed slit-lamp and gonioscopy examinations without mydriasis at the time when the need for surgery was determined. When iris and/or angle NV or neovascular glaucoma was present in 1 or both eyes of a patient, we diagnosed NV in this study.
The preoperative systemic parameters analyzed were gender; age; duration of diabetes treatment (years); preoperative HbA1c value (%); estimated glomerular filtration rate (eGFR for men, mL/min/1.73 m 2 ; for women, mL/min/1.73 m 2 × 0.742), which can be obtained by a simple formula from the Japanese Modification of Diet in Renal Disease; body mass index (BMI, kg/m 2 ); frequency of hypertension (%); and frequency of insulin therapy (%). The preoperative ophthalmic parameters analyzed were the frequencies of cataract surgery, panretinal photocoagulation (PRP), and retinal detachment. PRP was presumed to have been applied at the time of the decision to perform surgery. These items were confirmed by the medical records and compared between the NV and non-NV groups.
Screening for sleep-disordered breathing was performed after patients provided informed consent in accordance with the Declaration of Helsinki. A pulse oximeter (PMP-200G; Fuji Respironics Co, Ltd, Saitama, Japan) ( Figure 1 ) was placed on the patient’s wrist before bedtime to measure SpO 2 %, pulse rate, and body motions. The device was removed when the subject awoke. The data were stored in the memory of the instrument to calculate the average SpO 2 % during sleep (mean SpO 2 ), 4% ODI (times/hour), lowest SpO 2 (%) during sleep, and the cumulative percentage of time (CT90%) at which SpO 2 was below 90% during the analysis period, using analysis software SPO 2 Trend Chart G (Fuji Respironics Co, Ltd) ( Figure 1 ). The 4% ODI indicated the number of events per hour during the recording period in which the arterial oxygen saturation fell more than 4% below the baseline saturation but recovered thereafter to the baseline value. Decreases in oxygen saturation exceeding 4% during the interval of 90% to 100% saturation were regarded as 4% oxygen desaturation events. We previously reported that the 4% ODI with this device was correlated strongly with the apnea-hypopnea index times/hour, and when an ODI of 5 times/hour approximately corresponded to an apnea-hypopnea index of 15 times/hour. The American Academy of Sleep Medicine advocates that an apnea-hypopnea index of 15 times/hour be diagnosed as sleep apnea syndrome. For the current study, we diagnosed 4% ODI exceeding 5 times/hour as sleep-disordered breathing.
Pulse oximetry provided a 0% to 100% measurement range and 70% to 100% accuracy with ±2% error for SpO 2 and a range of 30 to 250 beats/minute with ±2% accuracy for the pulse rate, while the accuracy of the measurement of body motions was 0 ± 3 gravity. The oximeter was placed on the patient’s left wrist, and the probe was attached to the left first finger. Getting up at night (eg, to urinate) or displacement of the probe was verified by pulse changes or monitoring of body motions, and the corresponding data were excluded from the analysis.
The same experimenter (T.S.) administered the test preoperatively, because patient positions could be limited after surgery.
The sleep-disordered breathing parameters (ie, mean SpO 2 , 4% ODI, lowest SpO 2 , and CT90%) also were compared between the 2 groups. Logistic regression analysis was carried out to investigate whether the sleep-disordered breathing parameters were possible risk factors for NV.
Statistical analyses were conducted using the 2 × 2 χ 2 test, unpaired t test, and Mann-Whitney U test. A P value less than 5% was considered statistically significant. Stat View version 5.0 (SAS Institute Inc., Cary, North Carolina, USA) was used for statistical analysis.
The preoperative parameters for the NV and non-NV groups are shown in Table 1 . There were no significant differences in age, preoperative HbA1c value, eGFR, BMI, and the frequency of hypertension. The ratio of women was higher and the duration of diabetes treatment tended to be longer in the non-NV group; however, neither reached significance (NV group, men:women 17:20 vs non-NV group, 73:41, P = .05; duration of diabetes treatment [years], NV group, 9.6 ± 8.7 vs non-NV group, 12.7 ± 8.3, P = .06). The frequency of insulin therapy in the NV group was significantly higher than in the non-NV group (insulin therapy, NV group, 22/37 [59.5%] vs non-NV group, 37/114 [32.5%], P = .003).
|NV Group a||Non-NV Group b||P|
|Gender (male:female)||17:20||73:41||.05 c|
|Age (years), mean ± SD||61.5 ± 9.4||59.4 ± 10.0||.26 d|
|Period of diabetes treatment (years), mean ± SD||9.6 ± 8.7||12.7 ± 8.3||.06 d|
|HbA1c (%), mean ± SD||7.6 ± 1.4||7.4 ± 1.6||.53 d|
|eGFR (mL/min/1.73 m 2 ), mean ± SD||61.9 ± 25.5||68.8 ± 27.7||.18 d|
|BMI (kg/m 2 )||24.6 ± 4.4||24.5 ± 3.8||.94 d|
|Hypertension (%)||27/37 (73.0)||76/114 (66.7)||.47 c|
|Insulin therapy (%)||22/37 (59.5)||37/114 (32.5)||.003 c|
The preoperative ophthalmic parameters for the 2 groups are shown in Table 2 . The frequency of cataract surgery, PRP, and the effect of retinal detachment did not differ significantly between the 2 groups.
|NV Group a||Non-NV Group b||P|
|Cataract surgery (%)||8/37 (21.6)||17/114 (14.9)||.48 c|
|Panretinal photocoagulation (%)||26/37 (70.3)||93/114 (81.6)||.14 d|
|Retinal detachment (%)||8/37 (21.6)||17/114 (14.9)||.48 c|
The sleep-disordered breathing tests for the 2 groups are shown in Table 3 . The mean SpO 2 and lowest SpO 2 tended to be lower in the NV group than in the non-NV group but not significantly so (mean SpO 2 [%], 96.9% ± 1.9% vs 97.5% ± 1.4%, respectively, P = .07; lowest SpO 2 [%], 79.9% ± 9.9% vs 83.1% ± 9.2%, respectively, P = .07).
|NV Group a||Non-NV Group b||P|
|Mean SpO 2 (%), mean ± SD||96.9 ± 1.9||97.5 ± 1.4||.07 c|
|4% ODI mean ± SD||12.3 ± 11.6||6.6 ± 7.2||.02 d|
|Range||0 to 50||0.1 to 43.6||–|
|Lowest SpO 2 (%),mean ± SD||79.9 ± 9.9||83.1 ± 9.2||.07 c|
|CT90%, mean ± SD||3.8 ± 7.2||1.7 ± 3.9||.02 d|
|Range||0 to 35.7||0 to 32.6||–|
|SDB prevalence (%) e||23/37 (62.2)||52/114 (45.6)||.08 f|
|Analysis time (minutes), mean ± SD||418 ± 81.7||422 ± 57.5||.71 c|