Introduction

COVID-19, the most severe pandemic of the 21st century in terms of global mortality and morbidity, continues to cause significant illness in certain populations even five years after its onset1,2. Numerous studies have explored the ocular complications of this disease, particularly retinal vascular (or microvascular) occlusions and uveitis3,4,5. These complications are believed to arise from coagulopathy and vascular endothelial damage due to the cytokine storm, as well as virus-host antigen mimicry6.

Respiratory distress, a hallmark of SARS-CoV-2 infection, often leads to prolonged hypoxia in affected patients, with a subset requiring oxygen supplementation or even mechanical respiratory support1. In theory, such hypoxemia can induce retinal ischemia, potentially resulting in atrophy and thinning7,8. Given the unique anatomy of retinal circulation and perfusion, different retinal sublayers may exhibit varying levels of vulnerability to injury caused by hypoxemia.

Understanding these effects is crucial, as they may provide insights into the broader systemic impact of COVID-19 and its complications. Despite extensive research on COVID-19-associated retinal vascular changes, the effects of respiratory distress and prolonged hypoxemia on the retinal substructure remain underexplored.

This study aims to bridge that gap by assessing the OCT parameters of the fovea, peripapillary retinal nerve fiber layer (pRNFL), and macular sublayers in patients with systemic COVID-19 illness of varying severity. We hypothesize that prolonged hypoxemia from COVID-19-associated respiratory distress may result in measurable changes to retinal sublayer thickness, serving as a marker for previous ischemic damage. Highlighting these changes could advance our understanding of the interplay between systemic hypoxia and retinal health, emphasizing the importance of recognizing retinal alterations as a potential consequence of severe COVID-19 illness.

Materials and methods

Patients

In this prospective case-control study, consecutive patients were enrolled from a specialized COVID-19 screening and treatment center during a pandemic peak, between April and June 2021. This study design was selected due to its feasibility for implementation during the peak of SARS-CoV-2 infections and its effectiveness in detecting differences in outcome measures between cases with varying COVID-19 severity and controls.

Patients with a history of previous ocular trauma or surgery (other than uncomplicated cataract surgery), significant ocular diseases such as dense cataract, glaucoma, age-related macular degeneration (ARMD), presumed COVID-19-associated complications (such as retinal vein occlusion [RVO] or uveitis), refractive errors exceeding ± 3 diopters (D), or significant systemic conditions (e.g., diabetes, rheumatologic disorders, asthma, Alzheimer’s disease) were excluded. The inclusion and exclusion criteria were designed to eliminate cases with ocular or systemic conditions that could influence the magnitude or reliability of total retinal or retinal sublayer thickness measurements, as well as systemic diseases that could interfere with COVID-19 disease. Eligible patients were required to be able to attend the ophthalmology clinic six weeks after their initial referral to the COVID-19 emergency facility, cooperate with necessary imaging procedures, have no ophthalmic symptoms related to the recent SARS-CoV-2 infection, and exhibit a healthy-appearing macula as confirmed by fundoscopy and spectral-domain optical coherence tomography (SD-OCT) images.

There were two primary reasons for selecting 6-week interval. First, most SARS-CoV-2-infected patients were likely well enough to visit the ophthalmic imaging facility and cooperate with the imaging procedures. Second, retinal ischemia in the acute phase presents with hyperreflectivity and edema in the affected layer or sector. Typically by about six weeks, these changes progress to atrophy4,9. This interval was chosen to allow for the detection of chronic sequelae of the disease when it is likely to have stabilized.

All participants, or their guardians (in cases of severe illness), provided written informed consent, and the study adhered to the principles outlined in the Declaration of Helsinki. The study protocol was approved by the Ethics Committee at Shiraz University of Medical Sciences.

Based on SARS-CoV-2 PCR test results and clinical examinations, the patients were classified into four groups: Group 1 (controls; healthy individuals or those with a common cold, with negative PCR test results), Group 2 (outpatient SARS-CoV-2-positive cases), Group 3 (ward-admitted SARS-CoV-2-positive patients), and Group 4 (ICU-admitted SARS-CoV-2-positive patients). In this study, we did not collect data on the patients’ blood pressure or oxygen saturation.

Measurements

All patients underwent a comprehensive eye examination, which included measurement of best-corrected visual acuity (BCVA), slit-lamp biomicroscopy, and dilated fundus examination. OCT imaging was performed using the Spectralis OCT2 system (Heidelberg Engineering, Heidelberg, Germany). Imaging procedures were conducted by a trained examiner following the manufacturer’s recommendations. The device was calibrated prior to the study and at regular intervals throughout. For OCT imaging, a pRNFL scan of the optic nerve and a 30-degree macular volume scan were performed. The first successful scan with a quality score of at least 20 was recorded. Images with artifacts or uncorrectable segmentation errors were excluded. As cases with significant myopia or hyperopia were excluded from the study, the axial length scan depth setting on the OCT device was adjusted to “medium” (24.00 ± 2.50 mm).

The integrated HEYEX software was used to measure the total volume of each retinal sublayer within the 6-mm Early Treatment Diabetic Retinopathy Study (ETDRS) zone. The analyzed sublayers included the inner retinal layers (IRL; internal to the external limiting membrane), the outer retinal layers (ORL; external limiting membrane to the retinal pigment epithelium [RPE]), the retinal nerve fiber layer (RNFL), the ganglion cell layer (GCL), the inner plexiform layer (IPL), the inner nuclear layer (INL), the outer plexiform layer (OPL), the outer nuclear layer (ONL), and the RPE.

Statistical analysis

All statistical analyses were performed using IBM SPSS Statistics software (version 26; SPSS Inc., Chicago, Illinois, USA). Data are presented as adjusted means ± standard error of the mean (SEM). To account for the differing age distribution among groups, retinal sublayer values obtained from OCT imaging were compared among the four groups (controls and SARS-CoV-2-positive groups: outpatient, ward-admitted, and ICU-admitted) using analysis of covariance (ANCOVA), with age included as a covariate. Pairwise comparisons were conducted with Bonferroni correction to adjust for multiple comparisons. ANCOVA assumptions were tested using the Shapiro-Wilk test to assess the normality of residuals, Levene’s test to evaluate homogeneity of variance, and interaction tests to verify the homogeneity of regression slopes, and it was ensured that the assumptions were met before proceeding with the analysis. Our analysis was conducted on only one eye of the patients (right eyes) to eliminate bias arising from interocular correlations.

Results

Data from 78 cases with recent SARS-CoV-2 infection (29 outpatients, 32 ward-admitted, and 17 ICU-admitted) and 85 controls were collected and analyzed. Table 1 summarizes the baseline characteristics of the study participants. No significant differences were observed among the groups in terms of sex, spherical equivalent (SE) refraction, BCVA, or intraocular pressure (IOP). However, the age composition differed significantly between the groups. To address the potential influence of age on OCT variables, age was included as a covariate in all statistical analyses of OCT parameters.

Table 1 Baseline characteristics of the participants.
Full size table

The adjusted mean central subfield thickness (CSFT) ± SEM was 271 ± 3 μm in the control group, 251 ± 9 μm in the outpatient group, 260 ± 4 μm in ward-admitted patients, and 253 ± 5 μm in ICU-admitted cases (P = 0.093). Table 2 provides a detailed summary of the layer-specific retinal volume measurements across the four studied groups.

Table 2 Macular sublayer-specific volumetric data compared between patients with different severity of COVID-19 and normal controls.
Full size table

The differences in retinal sublayer volumes were statistically insignificant across the groups, except for the OPL, which demonstrated a declining trend with increasing disease severity (P = 0.006; pairwise comparison: p = 0.009 for ICU-admitted patients versus controls; Fig. 1).

Fig. 1
figure 1

Comparison of adjusted mean values of OPL volume within the 6-mm zone across the four studied groups.

Full size image

The ANCOVA test offers the advantage of providing adjusted mean values based on a controlled parameter (age in our study), allowing for a clearer comparison of the parameter values between groups. In addition, it facilitates comparisons among the four categorical groups while adjusting for covariates. However, recognizing the limitations of this method and to account for the possibility of type I statistical error, we also performed a linear regression analysis to identify independent predictors of OPL volume. Variables included in the analysis were age, sex, SE refraction, and SARS-CoV-2 infection status (control vs. positive). Through stepwise multiple regression analysis, the SARS-CoV-2 group emerged as the only independent predictor of OPL volume (unstandardized coefficient: −0.025; standard error: 0.011; P = 0.026), further supporting the ANCOVA findings.

No significant differences in pRNFL thickness were observed among the groups (Table 3).

Table 3 Peripapillary retinal nerve fiber layer thickness compared between patients with different severity of COVID-19 and normal controls.
Full size table

Discussion

Although COVID-19 illness has primarily presented as an acute respiratory syndrome, further reports reveal its complex pathophysiology, which involves an intricate immunological response, leading to coagulopathy and vasculopathy affecting multiple organ systems10. Ocular complications associated with SARS-CoV-2 infection are frequently reported. The posterior segment of the eye can be affected by conditions such as uveitis, central serous chorioretinopathy, RVO, and paracentral acute middle maculopathy (PAMM)3,4,5,11.

According to our findings, the OPL was the only retinal sublayer to show a significant difference in 6-mm volume among the groups. In the control group, the OPL volume was 0.832 μm3, while it measured 0.822, 0.814, and 0.785 μm³ in the outpatient, ward-admitted, and ICU-admitted SARS-CoV-2-positive patients, respectively (P= 0.006). This observation may be attributable to the unique anatomical location of the OPL, located in the watershed zone of the retina, between the retinal and choroidal circulation12,13,14 (Fig. 2). In addition, the deep capillary plexus of the retinal circulation, which contributes to OPL perfusion, has the lowest vascular density compared to the superficial and intermediate capillary plexuses14. In cases of systemic hypoxemia, oxygen tension easily drops in the watershed zone of the retina. The OPL, being one of the primary oxygen-consuming layers of the retina, is particularly vulnerable to such reductions in oxygen supply15,16.

Fig. 2
figure 2

An SD-OCT angiography B-scan of the macula shows the source of perfusion for different retinal sublayers. The outer plexiform layer (OPL) is located in the watershed zone between the retinal (central retinal artery, CRA) and choroidal circulation. This likely results in the lowest hydrostatic pressure (HP) and oxygen tension (O2) in the retina.

Full size image

The typical OCT presentation of retinal ischemia, as observed in conditions such as central retinal artery occlusion (CRAO), branch retinal artery occlusion (BRAO), PAMM, and acute macular neuroretinopathy (AMN), includes hyper-reflectivity and thickening of the inner retinal layers during the acute phase, followed by inner retinal atrophy in the subsequent weeks17,18. Bayram et al.19 reported OCT findings in severe cases of COVID-19 illness and found an increase in OPL thickness during the acute phase of the disease. Their results (indicating acute OPL ischemia) are consistent with our study, which showed OPL thinning six weeks after SARS-CoV-2 infection (chronic sequelae of OPL ischemia).

Although we proposed prolonged systemic hypoxemia as the most probable cause of OPL thinning in cases with SARS-CoV-2 infection, other COVID-19-associated immune-mediated mechanisms may also contribute. These mechanisms include microvasculopathy and coagulopathy6, which can affect various capillary beds and may have a detectable impact on the most vulnerable retinal vascular plexus (the deep capillary plexus) and sublayer (the OPL).

The OPL of the retina plays a crucial role in intraretinal visual processing, acting as the primary site for synaptic connections between photoreceptors (rods and cones), bipolar cells, and horizontal cells. This layer is essential for the transfer and integration of visual signals, contributing to key processes such as spatial resolution, contrast enhancement, and color discrimination. Damage to the OPL disrupts these critical synaptic interactions, impairing the transmission of visual information. As a result, fine visual tasks such as edge detection, color vision, contrast sensitivity, and pattern recognition may be adversely affected20,21,22. Such impairments may compromise daily living activities that require precise visual performance.

The OPL thinning associated with COVID-19 disease, as documented in our study, may have significant long-term implications for individuals with a history of severe SARS-CoV-2 infection. This thinning could potentially impact patients’ quality of life by reducing their ability to perform fine visual tasks. To better understand the functional consequences of this finding, further studies are required to explore its impact on visual performance and quality of life of such patients.

The differences in CSFT observed between the four groups in our study were marginally significant (P = 0.093). Although the difference between the control group and the overall COVID-19 patient cohort was 10–20 μm—potentially clinically relevant—there was no consistent trend within the SARS-CoV-2-infected subgroups (CSFT values in ICU-admitted patients were comparable to those in outpatients, but both were lower than in ward-admitted patients). Therefore, this study does not confirm a definitive impact of SARS-CoV-2 infection on CSFT, but the topic warrants further investigation. In addition, the lack of significant findings in this study regarding other retinal sublayers does not necessarily rule out their potential involvement in COVID-19. While subtle changes may exist beyond the detection capabilities of our study’s power, the adjusted mean values reported did not exhibit any consistent pattern. Therefore, the likelihood of a Type II error in this context is low.

The pRNFL thickness is a valuable biomarker for assessing the optic nerve head. Both thinning (as observed in conditions like glaucoma or chronic optic neuritis/ischemia) and thickening (as seen in acute ischemia or papilledema) can be indicative of early pathological changes in the optic disc. In our study, we included pRNFL analysis to evaluate the impact of COVID-19 disease and its associated respiratory distress on optic nerve head health, as could be detected through pRNFL measurements. This study did not find any clinically or statistically significant differences in pRNFL thickness between the groups (see Table 3). The existing literature on this topic presents contradictory findings. While some previous studies have reported a reduction in pRNFL thickness in specific sectors23,24,25, others have documented increased thickness19,26. These discrepancies may be attributed to statistical limitations, such as small sample sizes and methodological inconsistencies, as well as variations in the severity or duration of COVID illness among the populations evaluated in different studies. Further research is required to further elucidate this issue.

Powers and limitations

The present study involved the prospective enrollment of consecutive patients, which enhanced the reliability of its findings. The overall sample size was adequate to evaluate the primary outcomes; however, the modest sample size within the SARS-CoV-2 positive subgroups introduces the possibility of a type II statistical error. As such, the potential effects of COVID-19 disease on other retinal sublayers cannot be excluded, highlighting the need for future studies with larger sample size.

One important limitation of the study was that the groups were not age-matched, a factor inherent to the methodology of enrolling consecutive patients. Given the higher prevalence and severity of SARS-CoV-2 infection in older individuals, the average age of patients with severe disease was higher than that of those with mild disease or controls. However, age is a known biomarker affecting retinal OCT measurements27. To address this limitation, an ANCOVA approach was employed to control for age across all measurements. The adjusted mean values presented in this article were derived from the ANCOVA analysis. In addition, the observed positive finding of OPL volume changes between groups was reassessed using multiple linear regression. This further validation aimed to strengthen the reliability of the results and reduce the likelihood of a Type I statistical error.

In this study, our control group comprised individuals who either had exposure to COVID-19 patients or experienced upper respiratory tract symptoms that warranted a PCR test for SARS-CoV-2 infection (hence, the term “healthy controls” may be somewhat misleading). This inclusion strategy resulted from our prospective and consecutive sample collection approach from a COVID-19 screening facility. Nonetheless, we applied stringent criteria to ensure that participants with significant diseases affecting the outcome measures were excluded from the control group. Moreover, the likelihood of bias in control group selection is further diminished by the observed meaningful trend of OPL thinning in patients with more severe forms of COVID-19 illness.

Lastly, we did not collect data on mean blood pressure, oxygen saturation, or OCT angiography. As a result, our conclusion that the OPL may be the most vulnerable retinal sublayer to systemic hypoxemia remains speculative and warrants further investigation.

Conclusions

This study demonstrates that patients with severe SARS-CoV-2 infection, suffering from respiratory distress and prolonged hypoxia, may experience ischemia and atrophy of the OPL. These findings also suggest the potential vulnerability of the OPL to systemic hypoxemia, likely due to its anatomical location in the retina-choroid watershed zone and its high oxygen demand. However, this notion is hypothetical and need verification through future studies. The observed OPL atrophy may have clinical implications, including potential impacts on fine visual tasks, such as contrast sensitivity, color vision, and edge detection. These aspects were beyond the scope of the current study, and warrant further investigation to better understand the functional consequences of OPL changes in the context of COVID-19 illness.