Introduction

The prevalence of obstructive sleep apnea (OSA) in children is approximately 2–3%, primarily affecting those aged 2 to 8 years1,2. The pathophysiology of OSA in this population is multifactorial; in addition to obesity, the two most significant risk factors are tonsil hypertrophy and adenoid hypertrophy3,4,5. Consequently, adenotonsillectomy (AT) is considered the first-line therapy for children with OSA. The diagnosis of pediatric OSA involves a comprehensive clinical history, a physical examination focusing on the Otolaryngology region, an assessment of nocturnal and diurnal symptoms, and the identification of comorbidities, supplemented by specific questionnaires (such as the OSA-18) administered to parents. Most importantly, polysomnography (PSG) is essential for an accurate diagnosis6.

OSA can cause damage to multiple organs and systems throughout the body, leading to various complications such as cardiovascular disease, diabetes, and neurological disorders7,8,9,10. The complications associated with OSA in the adult population are well documented11. Although complex diseases related to OSA are less common in children than in adults, the underlying physiological changes are objectively present12,13. Early diagnosis and treatment of OSA in children may reduce the occurrence of systemic complications over time13,14. In recent years, there has been growing evidence linking OSA to eyelid syndrome, dry eye, central plasma choroidal retinopathy, optic neuropathy, and optic papillary edema15,16,17,18,19. The pathogenesis of ocular complications associated with OSA has not been fully elucidated; however, it may be multifactorial, involving ischemia-reperfusion injury, hypercoagulability, the release of vasoconstrictors, increased oxidative stress, and endothelial dysfunction18,20. Continuous positive airway pressure (CPAP) therapy can mitigate ocular complications in adult patients with OSA, including glaucoma and stromal edema in the choroid18,21. Optical coherence tomography angiography (OCTA), a non-invasive and painless examination of fundus blood flow that images and quantifies retinal microvascular perfusion throughout the macula, has been widely used in diagnosing fundus diseases22. Using OCTA, researchers have found that adults with OSA exhibit smaller vascular densities in both the superficial capillary plexus (SCP) and deep capillary plexus (DCP) compared to normal controls, particularly in those with moderate to severe OSA23,24,25. However, the findings regarding changes in the foveal avascular zone (FAZ) remain controversial23,24. Additionally, OCTA measurements of choroidal thickness showed a statistically significant decrease when grouped by OSA severity, although they did not correlate with the apnea-hypopnea index (AHI)25. In summary, adult patients with OSA exhibit a significant association with choroidal vascular microcirculation. However, it remains to be confirmed whether analogous changes occur in pediatric OSA. Therefore, the aim of this study was to investigate the alterations in macular microcirculation among pediatric OSA patients and to determine whether tonsillectomy and/or adenoidectomy have a beneficial effect.

Subjects and methods

General information

Children diagnosed with OSA who will undergo tonsillectomy and/or adenoidectomy at the Department of Otolaryngology of the Affiliated Jiangning Hospital with Nanjing Medical University between July 2019 and December 2023 were selected as the OSA group (n = 100). Additionally, children with normal physical examinations from ophthalmic outpatient clinics, who exhibited no signs of retinal disease, were chosen as the control group (n = 50). The patient screening process is illustrated in Fig. 1. This study received approval from the Ethics Committee of the Affiliated Jiangning Hospital with Nanjing Medical University, and parents (or guardians) of all study subjects provided informed consent.

Fig. 1
figure 1

The patient screening process.

Full size image

Inclusion and exclusion criteria

Inclusion criteria

For the OSA group: (i) Children aged 3 to 14 years who are admitted to the Otolaryngology Ward for tonsillectomy and/or adenoidectomy. (ii) Body mass index (BMI) between 15.5 and 22.9 kg/m². (iii) Regular snoring (≥ 3 nights per week for a duration of at least 3 months), observable mouth breathing during sleep, and episodic apnea. (iv) Willingness to participate in this study, which includes receiving free portable PSG monitoring, with results confirming a diagnosis of OSA (AHI ≥ 5 events/hour). (v) Participation includes receiving free ophthalmic examinations, OCTA assessments, and OSA-18 score evaluations before and after surgery.

For the control group: (i) Children aged 3 to 14 years undergo physical examinations in the ophthalmology clinic. (ii) BMI between 15.5 and 22.9 kg/m². (iii) Absence of frequent sleep disturbances and snoring (≤ 3 nights per month) and no snoring reported in the past month. (iv) Willingness to participate in this study and attend the otolaryngology outpatient clinic to rule out OSA via free portable PSG monitoring (AHI < 5 events/hour). (v) Participation includes receiving free OCTA examinations and OSA-18 score evaluations.

Exclusion criteria

Exclusion criteria included limited visual fields in both eyes, refractive abnormalities such as myopia or hyperopia, axial length abnormalities, intraocular pressure abnormalities, and various fundus diseases. Additionally, children with a history of chronic diseases such as hypertension, those who have experienced an acute infectious disease or those with a history of trauma within the past month were excluded from the study.

Methods

Routine examination

All study subjects underwent routine portable PSG monitoring in the otolaryngology department, as well as a comprehensive eye examination in the ophthalmology department, which includes axial length, visual field, refractive status, intraocular pressure and fundus. In the OSA group, tonsillectomy was indicated for tonsils classified as II degree or above. Additionally, adenoidectomy was performed if the posterior nostril was obstructed by more than 50% of the adenoid tissue during nasopharyngoscopy, or if the A/N value from nasopharyngeal lateral radiographs was ≥ 0.70. If both the tonsils and adenoids met these criteria, AT was performed.

OCTA examination

Both the control group and the OSA group (preoperative and 6 months postoperative) attended the ophthalmic outpatient clinic for OCTA examinations, with data from the left eye being utilized for analysis. All the children were accompanied by a guardian during the day in a quiet dark room with no natural light or lighting. Blood flow images were acquired using OCTA (Cirrus HD-OCT5000, Zeiss, Germany). All subjects underwent testing following left eye pupil optometry. The internal gaze method was employed to ensure the green fixed cursor remained within the lens of the gaze device for more than 3 s. All scans utilized the FastTrac™ image tracking function, focusing on a 6 mm × 6 mm area of superficial retinal capillaries around the macula and optic disc. Only OCTA blood flow signal intensity readings of ≥ 8 were included for quantitative analysis via the built-in software. The macular area was segmented into grids according to the Early Treatment Diabetic Retinopathy Study (ETDRS) guidelines: the central 1 mm diameter circle constituted Area 1, the inner ring formed Area 2, the outer ring represented Area 3, and the complete 6 mm diameter ring was designated as Area 4. This study primarily assessed vascular length density (VLD) and vascular perfusion density (VPD) across Areas 1, 2, 3, and 4, while also manually reviewing the FAZ within these areas. All OCTA examinations in the subject must be independently completed by the same ophthalmologist and the other physician shall be responsible for the verification and testing.

OSA-18 scale scores

OSA-18 scale scores were performed in the control group and OSA group before and after surgery (6th postoperative month).

Observation indicators

OSA-18 scores, VLD1-VLD4, VPD1-VPD4 and FAZ.

Statistical methods

All statistical analyses were performed using SPSS 26 (SPSS, Inc., Chicago, IL, USA). Normality of the data was verified using the Shapiro–Wilk test. Chi- squared tests were used to determine the differences in patient sex ratios between the groups, and t test was used to compare the two groups (Mean ± SD). In addition, we used covariance analysis to adjust the potential bias caused by age, BMI, and sex on the observational indicators, and used the corrected P-value as the comparison standard. A p-value of less than 0.05 was considered statistically significant.

Results

General data

Six OSA children fell off at follow-up 6 months after surgery. Therefore, this study was finally included in the OSA group 94 children (male 46, female 48) and 50 healthy control group (male 25, female 25). There was no statistically significant difference in the age (6.95 ± 2.10 vs. 6.82 ± 2.12, P = 0.731), gender ratio (chi-square = 0.123a, P = 0.861) between the OSA group and the control group.

Comparison of OSA-18 (Table 1; Fig. 2A)

Table 1 Comparison of indicators between the OSA group and the control group.
Full size table
Fig. 2
figure 2

Comparison of observational indicators.

Full size image

Through t-tests, we found that the OSA-18 scores of children in the OSA group before surgery were higher than those in the control group, and the difference was statistically significant (72.532 ± 12.865 vs. 29.220 ± 6.911, P < 0.0001). However, at the six-month follow-up after surgery, the OSA-18 scores of children in the OSA group were lower than before surgery (31.096 ± 9.340 vs. 72.532 ± 12.865, P < 0.0001), and had fallen to the same level as those in the control group (31.096 ± 9.340 vs. 29.220 ± 6.911, P > 0.05).

Comparison of VLD (Table 1; Fig. 2B and E)

Through t-tests, we found that the VLD1 in the OSA group before surgery was lower than those in the control group, and the difference was statistically significant (9.541 ± 0.183 vs. 9.615 ± 0.098, corrected P = 0.008). However, at the six-month follow-up after surgery, the VLD1 of children in the OSA group were higher than before surgery (9.625 ± 0.142 vs. 9.541 ± 0.183, P = 0.001), and had increased to the same level as those in the control group (9.625 ± 0.142 vs. 9.615 ± 0.098, corrected P = 0.671).

VLD2 showed a consistent trend with VLD1, with preoperative VLD2 values in the OSA group being lower than those in the control group, a difference that was statistically significant (17.951 ± 0.131 vs. 18.821 ± 0.146, corrected P < 0.0001). However, at 6 months postoperatively, VLD2 values in the OSA group were significantly higher than preoperative levels (18.852 ± 0.131 vs. 17.951 ± 0.131, P < 0.0001), returning to the same level as those in the control group (18.852 ± 0.131 vs. 18.821 ± 0.146, corrected P = 0.213).

Through t-tests, no statistical difference was found between the preoperative VLD3 of OSA patients and the control group (18.399 ± 0.087 vs. 18.383 ± 0.093, corrected P = 0.282). Six months after surgery, the VLD3 values of OSA children did not change significantly compared with those before surgery (18.381 ± 0.088 vs. 18.399 ± 0.087, P = 0.167), and there was no significant difference compared with the control group (18.381 ± 0.088 vs. 18.383 ± 0.093, corrected P = 0.945).

VLD4 showed a consistent trend with VLD3, which no statistical difference was found between the preoperative VLD4 of OSA patients and the control group (18.162 ± 0.076 vs. 18.173 ± 0.070, corrected P = 0.410). Six months after surgery, the VLD4 values of OSA children did not change significantly compared with those before surgery (18.162 ± 0.071 vs. 18.162 ± 0.076, P = 0.953), and there was no significant difference compared with the control group (18.162 ± 0.071 vs. 18.173 ± 0.070, corrected P = 0.384).

Comparison of VPD (Table 1; Fig. 2F and I)

Through t-tests, we found that the VPD1 in the OSA group before surgery was lower than those in the control group, and the difference was statistically significant (0.247 ± 0.044 vs. 0.409 ± 0.027, corrected P < 0.0001). However, at the six-month follow-up after surgery, the VPD1 of children in the OSA group were higher than before surgery (0.399 ± 0.029 vs. 0.247 ± 0.044, P < 0.0001), and had increased to the same level as those in the control group (0.399 ± 0.029 vs. 0.409 ± 0.027, corrected P = 0.054).

VPD2 showed a consistent trend with VPD1, with preoperative VPD2 values in the OSA group being lower than those in the control group, a difference that was statistically significant (0.406 ± 0.028 vs. 0.429 ± 0.027, corrected P < 0.0001). However, at 6 months postoperatively, VPD2 values in the OSA group were significantly higher than preoperative levels (0.437 ± 0.030 vs. 0.406 ± 0.028, P < 0.0001), returning to the same level as those in the control group (0.437 ± 0.030 vs. 0.429 ± 0.027, corrected P = 0.100).

Through t-tests, no statistical difference was found between the preoperative VPD3 of OSA patients and the control group (0.453 ± 0.027 vs. 0.449 ± 0.013, corrected P = 0.291). Six months after surgery, the VPD3 values of OSA children did not change significantly compared with those before surgery (0.450 ± 0.015 vs. 0.453 ± 0.027, P = 0.364), and there was no significant difference compared with the control group (0.450 ± 0.015 vs. 0.449 ± 0.013, corrected P = 0.547).

VPD4 showed a consistent trend with VPD3, which no statistical difference was found between the preoperative VPD4 of OSA patients and the control group (0.372 ± 0.065 vs. 0.389 ± 0.067, corrected P = 0.131). Six months after surgery, the VPD4 values of OSA children did not change significantly compared with those before surgery (0.383 ± 0.066 vs. 0.372 ± 0.065, P = 0.256), and there was no significant difference compared with the control group (0.383 ± 0.066 vs. 0.389 ± 0.067, corrected P = 0.590).

Comparison of FAZ (Table 1; Fig. 2J)

Through t-tests, no statistical difference was found between the preoperative FAZ of OSA patients and the control group (0.277 ± 0.018 vs. 0.274 ± 0.013, corrected P = 0.363). Six months after surgery, the FAZ values of OSA children did not change significantly compared with those before surgery (0.273 ± 0.018 vs. 0.277 ± 0.018, P = 0.125), and there was no significant difference compared with the control group (0.273 ± 0.018 vs. 0.274 ± 0.013, corrected P = 0.639).

Discussion

OSA is characterized by recurrent episodes of partial or complete collapse of the upper airway during sleep, leading to oxidative stress, systemic inflammation, heightened sympathetic activation, and hemodynamic changes. Hypoxia induces endothelial dysfunction, which is marked by a reduction in the synthesis of nitric oxide that facilitates vasodilation in blood vessels, alongside an increase in endothelin-1 that promotes vasoconstriction. In OSA, airway obstruction elevates intrathoracic pressure, resulting in hypoperfusion due to decreased blood pressure, which causes vasospasm in blood vessels and increases resistance in smaller vessels26. The microcirculation within the retina is distinctive; there is minimal to no innervation in the retinal vessels, and it is believed that the regulation of blood flow in the macular area is achieved through active mechanisms. The retina can be adversely affected by episodes of hypoxia associated with OSA, which can lead to vascular hemodynamic changes, oxidative stress, mitochondrial dysregulation, and inflammation. These factors may initially cause tissue edema, followed by capillary dropout and subsequent tissue atrophy. Relative to its size, the oxygen consumption of the retina is higher than that of the brain and other tissues. The cells in the retina require substantial amounts of oxygen and glucose, necessitating continuous blood flow to meet these demands27,28. Given that retinal microcirculation may serve as a window for estimating systemic vascular status, it is hypothesized that alterations in micro vessels could reflect systemic hypoxia in OSA. OCTA is a non-invasive imaging method that facilitates this investigation. Currently, these imaging modalities are widely utilized in the diagnosis and management of ophthalmic diseases associated with OSA22,23,24,25,26,27,28.

A review of the literature reveals that retinal microcirculatory alterations were observed in adult patients with OSA, primarily characterized by decreased vessel density (VD) in both the SCP and the DCP and retinal nerve fiber layer (RNFL) thickness, along with an increased foveal avascular zone (FAZ) 23,24,29,30,31,32,33,34,35. However, the relationship between the severity of OSA and the extent of retinal microcirculatory changes remains contentious, as various studies report differing conclusions19,23,24,34,35. These discrepancies may stem from variations in study populations, methodologies for assessing OSA severity, and indicators used for evaluating retinal microcirculation, all of which could have contributed to biased outcomes. Although controversial on some issues, there is no doubt that OSA causes changes in retinal microcirculation.

Unfortunately, studies on retinal microcirculatory changes in children with pediatric OSA are more limited. In children, all OCTA parafoveal vessel indices—including VD, vascular area density (VAD), vascular skeleton density (VSD), and vessel perimeter index (VPI) in the macular SCP and DCP—as well as the FAZ in the DCP, were significantly lower in the OSA group compared to the control group. The indices of VD and VAD in the four parafoveal quadrants (temporal, superior, nasal, and inferior) for both layers were also significantly lower in the OSA group than in the control group36. Following adenoidectomy, the values of VD, VAD, VSD, and VPI in the SCP and DCP of the parafovea showed significant increases. Additionally, the FAZ was significantly reduced in the SCP and increased in the DCP, indicating a notable redistribution of oxygen supplies in cases of OSA children37. It has been reported that childhood undernutrition and overnutrition may influence the density of retinal capillaries38. Therefore, only children within the normal range of BMI were included in our study while screening the study population. Nonetheless, the potential bias of BMI on the results cannot be ignored. In this group of cases, we utilized analysis of covariance to control for the bias of BMI, gender, and age on the observed indicators, and the final results obtained were consistent with those before controlling for confounders. In our study, we conducted a preliminary investigation of the SCP in the macula, which revealed a statistically significant reduction in VPL and VPD in zones 1 and 2 among children with OSA. Notably, these parameters could revert to levels comparable to those of the control group six months following tonsillectomy and/or adenoidectomy. However, no changes were observed in VPL, VPD in zones 3 and 4, or in the FAZ. This may be attributed to the relatively dense vascularization of zones 1 and 2, which allows for an early and sensitive response to alterations in the ischemic-hemic-oxygenated internal environment. The lack of statistical changes in VPL and VPD in zone 3 may be since the vascular distribution in the outer ring zone is inherently sparser than in the inner ring zone, and the relatively large area of the outer ring zone means that zone 4, which reflects the overall situation, also shows no significant changes. Previous studies have not reached a consensus on the difference in FAZ area between the eyes of OSA patients and control eyes. As mentioned earlier, elevated FAZ values were observed in adult OSA patients23,30. In pediatric OSA patients, however, an opposite trend in FAZ values between the SCP and DCP had been observed37. The specific reasons for this phenomenon remain unclear. However, based on the presence of retinal microcirculatory disorders in zones 1 and 2, we are more inclined to suggest that FAZ values exhibit an increasing trend, and the data from our study indeed reflect this trend, though unfortunately it lacks statistical significance. Although the children with OSA in this study did not present with significant retinal disease, chronic sleep snoring and hypoxia have impacted the SCP. Therefore, we concluded that early surgical intervention to alleviate upper airway obstruction may reverse these physiological changes and potentially prevent further damage.

There are some shortcomings in our study, such as a small sample size and fewer observational indicators. In addition, this study did not fully utilize the PSG data and used it only as a diagnostic criterion. Therefore, the OSA group was not stratified by AHI severity, and the correlation between oxygen saturation and observational indicators was not analyzed. Due to the small sample size of children with severe OSA in this cohort, further analysis is needed to expand the sample size in the future. It is important to note that OCTA itself has several limitations in clinical trials. Currently, there is no standardized method for layer segmentation and definitions19. Furthermore, OCTA imaging has a limited field of view, which can result in the omission of peripheral pathology. While OCTA is a novel tool for evaluating retinal vasculature, these limitations may affect interstudy agreement39,40,41. Additionally, OCTA cannot be performed on sleeping patients, leaving unanswered questions regarding whether relevant outcome parameters differ between the sleep and awake states.

Although the precise mechanism remains unclear, it is hypothesized that oxidative stress and the inflammatory response may be the primary pathological mechanisms, which require further confirmation through animal experiments. The results of this study suggest that OCTA serves as a supplementary parameter for the early assessment of tonsillectomy and/or adenoidectomy in pediatric cases of OSA. Furthermore, OCTA may be utilized to evaluate the pathophysiology of retinal vascular changes associated with OSA, thereby providing a surrogate biomarker for systemic hypoxia.