Retinal microvascular changes in patients recovered from COVID-19 compared to healthy controls: A meta-analysis

Abstract

Purpose

To investigate changes in retinal microcirculation in patients recovered from COVID-19 infection compared to healthy controls, using optical coherence tomography-angiography.

Methods

Meta-analysis of eligible studies comparing retinal microcirculation between patients recovered from COVID-19 infection and healthy controls up to 7th of September 2022 was performed, according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses 2009 guidelines. The following search algorithm was used: (COVID-19 OR coronavirus) AND (retina OR optical coherence tomography OR optical coherence tomography angiography OR vessel density OR foveal avascular zone). Standardized Mean Difference (SMD) with 95% confidence interval (CI) was calculated to compare continuous variables. Revman 5.3 was used for the analysis.

Results

12 studies were included in our analysis. Foveal avascular zone (FAZ) area was larger in patients recovered from COVID-19 infection compared to healthy controls, while there was no statistically significant difference in FAZ perimeter between the two groups. The foveal, parafoveal and whole image vessel density in the superficial capillary plexus showed no significant difference between the two groups. The foveal, parafoveal and whole image vessel density in the deep capillary plexus was statistically lower in patients recovered from COVID-19 compared to healthy controls.

Conclusion

FAZ area was enlarged and foveal, parafoveal and whole image vessel density in deep capillary plexus were reduced in patients recovered from COVID-19 infection compared to healthy controls, suggesting that COVID-19 infection may induce long-term retinal microvascular changes in patients recovered from the virus infection.

1. Introduction

Coronavirus disease 2019 (COVID-19) is caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), a virus that was first identified in Wuhan, China in December 2019 and caused a global pandemic [1]. Most of the clinical manifestations of the disease involve the respiratory tract, however extrapulmonary symptoms have been also identified. According to a recent meta-analysis, ocular manifestations of COVID-19 infection have been found to involve 11% of the infected population [2], including complications of both the anterior and posterior segment of the eye [3], [4], [5].

It is worthy to note that SARS-CoV-2 can create endothelial dysfunction and cytokine overproduction, leading to multiple organ disease [6]. Specifically, it can affect the renin-angiotensin system (RAS) by recognizing its receptors, which are known to exist in many organs. As such, the eye contains RAS receptors, located in the anterior part of the eye, but also in the retina and choroid [7]. Interestingly enough, several studies have identified and reported retinal microvascular changes in COVID-19 infection [8], [9], [10].

Optical coherence tomography-angiography (OCTA) is a non-invasive method that allow us to visualize and obtain details of the retinal vasculature [11]. It provides information about the foveal avascular zone (FAZ) area, its perimeter, and the vessel density (VD) in the superficial (SCP) and deep capillary plexus (DCP). Recently, many studies utilized OCTA to examine retinal vascular indices in groups affected by COVID-19 infection compared to healthy control patients with controversial results. In light of the above, the purpose of this meta-analysis is to investigate the potential changes in OCTA parameters in patients recovered from COVID-19 infection compared to healthy control subjects.

2. Methods

2.1. Data sources and search strategy

A meta-analysis of observational studies was performed according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2009 guidelines [12]. Two reviewers (DK, GM) independently identified the relevant studies by an electronic search of the MEDLINE database from inception to 7th of September 2022. The following search algorithm was used: (COVID-19 OR coronavirus) AND (retina OR optical coherence tomography OR optical coherence tomography angiography OR vessel density OR foveal avascular zone). Articles and book chapters cited in the reference lists of initially identified articles by this query were reviewed, in order to identify any supplemental studies (“snowball procedure”). The final list of eligible articles was filtered manually to exclude duplicates.

2.2. Inclusion and exclusion criteria

Eligible studies have to fulfill the following inclusion criteria: (1) evaluated an adult population recovered from a previous confirmed diagnosis of COVID-19 (with positive PCR test) and a healthy control population; (2) evaluated at least one of the outcomes of interest, including superficial and deep VD, FAZ area and perimeter in recovered patients and controls. Studies were excluded if: (1) they were not published in English language; (2) the population of COVID-19 recovered patients and healthy controls were not comparable in terms of age; (3) they included case reports; (4) they evaluated patients not recovered from COVID-19 infection; (5) they did not employ a clear definition of COVID-19 infection diagnosis.

2.3. Data extraction

Data were independently extracted and reviewed from each study by two reviewers (DK, GM). Any discrepancy between data extractions was resolved by discussion or a third reviewer (IC). The following data were extracted: first author, year of publication, country, study design, number of patients and controls, patients’ age, OCTA device used and descriptive statistics of retinal microcirculation characteristics in patients and controls.

2.4. Quality assessment

Quality assessment of the included studies was conducted via the Newcastle-Ottawa Scale (NOS) [13], in which a study was judged on three categories: selection (four items, one point each), comparability (one item, up to two points), and exposure/outcome (three items, one point each). A nine-point scale of the NOS (range, 0–9 points) was eventually used for the evaluation. Two authors (DK, GM) discussed the implementation of this quality assessment tool and independently assessed the studies. Studies were defined as high quality if they had more than seven points, as medium quality if they had between four and six points, and as poor quality if they had fewer than four points.

2.5. Outcomes of interest

The pre-specified primary endpoint was the difference in FAZ area and perimeter and in VD in foveal, parafoveal and whole image in SCP and DCP.

2.6. Statistical analysis

The descriptive statistics were described as mean ± standard deviation (SD). For continuous outcomes the standardized mean difference (SMD) with 95% confidence interval (CI) was used as the summary statistic and trial-specific data were pooled with the inverse-variance random-effects method. When mean and SD were not available, they were derived from sample size, median and range based on a method previously described by Wan et al. [14] or from standard error and number of participants. The presence of heterogeneity among studies was evaluated with the Cochran Q chi-square test with p ≤ 0.1 considered to be of statistical significance, estimating the between-studies variance tau-square, and using the I² test to evaluate inconsistency. The I² statistic is derived from the Q statistic (100%×(Q - df)/Q), and describes the percentage of total variation across studies that is due to heterogeneity; a value of 0% indicates no observed heterogeneity, and larger values show increasing heterogeneity. I² values of 25%, 50%, 75% have been assigned adjectives of low, moderate, and high heterogeneity, respectively. A leave-one out sensitivity analysis was performed by iteratively removing one study at a time to confirm that our results were not driven by any single. In addition, a sensitivity analysis by calculating SMD using the inverse-variance fixed-effects method was performed for all outcomes of interest. Publication bias were assessed by visual inspection with funnel plots. All analyses were performed with Review Manager, version 5.3 (Copenhagen: The Nordic Cochrane centre, The Cochrane Collaboration, 2014). The guidelines summarized in the PRISMA and MOOSE statements were followed [12,15].

3. Results

The electronic database search identified 754 articles. After screening of all titles and abstracts of potentially relevant articles, a total of 12 studies met the inclusion criteria [16], [17], [18], [19], [20], [21], [22], [23], [24], [25], [26], [27], [28] (Fig. 1 ). The study characteristics and Newcastle Ottawa scores of the included studies are presented in Table 1 .

Fig. 1.

Flow-chart of the selection process of the included studies.

Table 1.

Characteristics of eligible studies, which were included in the meta-analysis.

Author	Year	Region	Study Design	No of patients	No of controls	Age of patients	Age of controls	COVID-19 recovery definition	OCT Device	Image quality
Zapata	2020	Spain	Case-control	27	21	39	44	Hospital discharge No negative PCR reported	DRI OCT Triton, Topcon	SSI ≥40
Abrishami	2021	Iran	Cross-sectional	31	23	40.4 ± 9.2	36.6 ± 7.1	Symptoms’ free for at least 2 weeks No negative PCR reported	AngioVue, RTVue XR Avanti, Optovue	Quality scan index>7
Bilbao-Malave	2021	Spain	Case-control	17	17	58.0 ± 2.7	57.2 ± 2.7	Review 6 months after hospitalization No negative PCR reported	DRI OCT Triton, Topcon	Poor quality scans excluded
Cennamo	2021	Italy	Observational,cohort	40	40	49.7 ± 12.6	48.6 ± 12.2	PCR negative test	AngioVue, RTVue XR, Optovue	SSI ≥80
Aydemir	2021	Turkey	Cross-sectional	39	40	45.23 ± 13.20	47.60 ± 12.14	PCR negative test	AngioVue, RTVue XR Avanti, Optovue	SSI ≥80, poor quality scans excluded
Hazar	2021	Turkey	Case control	50	55	37.00±5.93	35.14±6.95	PCR negative test	AngioVue, Optovue	SSI ≥50
Naderi Beni	2021	Iran	Cross-sectional	51	37	41.13± 12.6	45.64 ± 13.7	PCR negative test	AngioVue, RTVue XR Avanti, Optovue	SSI ≥45
Turker	2021	Turkey	Case-control	27	27	38.74 ± 10.70	37.44 ± 10.04	PCR negative test	AngioVue, RTVue XR Avanti, Optovue	SSI ≥70
Kal	2022	Poland	Case-control	63	45	51.33 ± 11.51	47.76 ± 9.26	No negative PCR reported, patients recovered after hospitalization	DRI OCT Triton, Topcon	Image quality ≥65%
Dipu	2022	India	Cross-sectional	35	12	27–60	27–60	PCR negative test	Nidek	NR
Erogul	2022	Turkey	Cross-sectional	32	33	49.60±13.50	48.40±12.40	PCR negative test	AngioVue, RTVue XR Avanti, Optovue	Quality scan index ≥7
Ozbas	2022	Turkey	Case-control	105	95	40.65±12.53	41.61± 12.21	Review at least 14 days after COVID-19 infection No negative PCR reported	AngioVue, RTVue XR Avanti, Optovue	NR

NR: not reported; OCT: optical coherence tomography; PCR: polymerase chain reaction; SSI: signal strength index.

3.1. Foveal avascular zone area

A total of 12 studies including 517 patients recovered from COVID-19 and 445 controls were included in the analysis. The FAZ area was significantly larger in patients recovered from COVID-19 compared to healthy controls (SMD=0.37, 95% CI=0.15–0.59, p = 0.001, I²=73%). In sub-analyses of each plexus, the FAZ area was significantly larger in the whole image analysis in patients recovered from COVID-19 (SMD=0.18, 95% CI=0.01–0.35, p = 0.04, I²=12%), while it was found to be larger in the SCP (SMD=0.45, 95% CI=−0.03–0.93, p = 0.06, I²=81%) and DCP (SMD=0.64, 95% CI=−0.16–1.44, p = 0.12, I²=88%), but the difference did not reach statistical significance (Fig. 2 ).

Fig. 2.

Forest plot of the foveal avascular zone (FAZ) area in COVID-19 recovered patients compared to healthy controls.

3.2. Foveal avascular zone perimeter

A total of 4 studies including 153 patients recovered from COVID-19 infection and 133 healthy controls were included in the analysis. There was no statistically significant difference in FAZ perimeter between the two groups (SMD=0.15, 95% CI=−0.09–0.38, p = 0.22, I²=0%), as it is shown in Fig. 3 . There was no heterogeneity across the studies.

Fig. 3.

Forest plot of the foveal avascular zone (FAZ) perimeter in COVID-19 recovered patients compared to healthy controls.

3.3. Superficial capillary plexus vessel density

10 studies, including 426 COVID-19 recovered patients and 384 healthy controls; 8 studies, including 365 COVID-19 recovered patients and 340 controls; 7 studies, including 348 COVID-19 recovered patients and 323 controls, were included in the analysis for the foveal, parafoveal and whole image VD in the SCP respectively.

There was no statistically significant difference in the foveal VD at SCP (SMD=−0.29, 95% CI=−0.59–0.02, p = 0.07, I²=77%), parafoveal VD at SCP (SMD=−0.13, 95% CI=−0.46–0.20, p = 0.45, I²=77%) and whole image VD at SCP (SMD=−0.22, 95% CI=−0.50–0.06, p = 0.13, I²=68%) in patients recovered from COVID-19 infection compared to healthy controls, as it is depicted in Fig. 4 . There was significant heterogeneity between the studies. We performed subgroup analysis based on the OCTA device used. No statistically significant difference in the foveal and parafoveal VD at SCP was found between COVID-19 recovered patients and healthy controls, using either Optovue or Topcon OCTA. All studies included in the whole image VD analysis at SCP used the Optovue device; therefore, no subgroup analysis was performed.

Fig. 4.

Forest plot of superficial capillary plexus vessel density in foveal (upper panel); parafoveal (middle panel) and whole image (bottom panel) in COVID-19 recovered patients compared to healthy controls.

3.4. Deep capillary plexus vessel density

9 studies, including 405 patients recovered from COVID-19 infection and 357 healthy controls, were included in the analysis for the foveal VD at DCP. There was a statistically significant difference in the VD between the two groups (SMD=−0.25, 95% CI=−0.49 to −0.01, p = 0.04, I²=60%), showing that the VD was decreased in the post infection group.

8 studies, including 365 patients and 340 controls, were included in our analysis for the parafoveal VD at the DCP analysis and revealed that the VD was decreased in the COVID-19 recovered patients (SMD=−0.29, 95% CI=−0.47 to −0.12, p = 0.001, I²=24%). We performed subgroup analysis for foveal and parafoveal VD based on the OCTA device used in the studies (Optovue or Topcon), showing the same results.

7 studies reported outcomes for the whole image VD at the DCP and showed that VD was decreased in patients recovered from COVID-19 infection compared to healthy controls (SMD=−0.27, 95% CI=−0.49 to −0.05, p = 0.01, I²=46%). All studies utilized Optovue for their analysis, so we did not perform any subgroup analysis. Fig. 5 illustrates the changes in VD at the DCP.

Fig. 5.

Forest plot of deep capillary plexus vessel density in foveal (upper panel); parafoveal (middle panel) and whole image (bottom panel) in COVID-19 recovered patients compared to healthy controls.

3.5. Risk of bias assessment

The quality assessment scores of the NOS are shown in Table 2 . 5 studies were of high quality and the remaining 7 were of moderate quality.

Table 2.

Newcastle-Ottawa scale scores of the studies which were included in the meta-analysis.

	Selection				Comparability		Exposure			Total
Study(year)	Case definition	Representativeness of the cases	Selection of controls	Definition of controls	On age	On other risk factors	Assessment of exposure	Same method of assessment in cases and controls	Non-response rate
Zapata (2020)	1	1	0	1	1	0	1	1	1	7
Abrishami (2021)	1	1	1	1	1	0	1	1	0	7
Bilbao-Malave (2021)	1	1	0	1	1	0	1	1	1	7
Cennamo (2021)	1	1	0	1	1	0	1	1	0	6
Aydemir (2021)	1	1	1	1	1	0	1	1	0	7
Hazar (2021)	1	1	0	1	1	0	1	1	0	6
Naderi Beni (2021)	1	1	0	1	1	0	1	1	0	6
Turker (2021)	1	1	0	1	1	0	1	1	0	6
Kal (2022)	1	1	0	1	1	0	1	1	0	6
Dipu (2022)	1	0	0	1	1	0	1	1	1	6
Erogul (2022)	1	1	0	1	1	0	1	1	0	6
Ozbas (2022)	1	1	1	1	1	0	1	1	0	7

3.6. Assessment of publication bias

Visual inspection of funnel plots showed no publication bias, as shown in Fig. 6 (supplemental material).

3.7. Sensitivity analysis

Leave one out sensitivity analysis showed that when the study of Naderi Beni et al. [22] was excluded from the analysis of the whole image VD at SCP, there was a statistically significant reduced VD in the COVID-19 recovered group compared to healthy controls (SMD=−0.37, 95% CI=−0.68 to −0.05, p = 0.02, I²=76%). In the DCP foveal VD, when the studies of Cennamo et al., Abrishami et al., Bilbao-Malave et al. and Erogul et al. were excluded from the analyses, the results were no longer statistically significant.

4. Discussion

In this meta-analysis, we focused on the microvascular changes in patients recovered from COVID-19 infection. We found that the FAZ area was larger and the foveal, parafoveal and whole image VD at the DCP was decreased in patients recovered from COVID-19 compared to healthy controls. These findings suggest that there is a long-term microvascular impairment in COVID-19 patients.

There are many parameters that can influence the VD in the macula. Age and sex have been previously identified as parameters that can affect the VD as measured by OCTA [29]. In our meta-analysis, we tried to include studies in which COVID-19 recovered patients and controls were comparable in age and sex, so as to minimize bias. In addition, severity of COVID-19 infection may influence the induced microvascular changes. Zapata et al. found decreased VD in patients with severe and moderate infection compared to healthy controls [16]. Abrishami et al. found that microvascular changes may be induced in a cohort consisted of hospitalized and non-hospitalized patients recovered from COVID-19 virus [17]. Kolkedi et al. studied patients with mild or asymptomatic patients recovered from COVID-19 and found no significant microvascular changes compared to healthy controls [24].

Several pathways may be involved in microvascular damage in COVID-19 patients, which can be regarded as a multi-systemic vascular disease. First of all, angiotensin‐converting enzyme 2 (ACE2) has been located in eyes in connection with Mueller cells, retinal pigment epithelium, and pericytes of endothelial cells and it is considered as the main receptor of SARS‐CoV‐2 cellular entry [30]. Microvascular endothelial injury and cytokine over-secretion are also thought to be key factors in the pathogenesis of microvascular changes, affecting the blood‐retinal barrier [31]. Moreover, there is probably a systemic activation of complement-associated microvascular injury with a related pro-coagulant state, leading to thrombosis especially in patients with severe COVID-19 disease [32]. This assumption may explain the fact that COVID-19 may affect several parts of the body, such as heart and brain, while patients may have persistent symptoms even after more than a year, accompanied by persistent capillary rarefaction [32].

It is worthy to note that COVID-19 infection creates a vicious cycle, as infection- and hypoxia-related inflammation cause capillary function to deteriorate, which in turn accelerates hypoxia-related inflammation and tissue damage [33]. Low oxygenation has been identified as a potential factor that affects retinal VD. Hypoxia is known to cause endothelial dysfunction and increased production of reactive oxygen species, affecting retinal tissues which are prone to oxidative stress, and leading to cellular damage and VD impairment [34]. Interestingly, Nakahara et al. found that the damage of the vascular network was more prominent in the DCP, based on an experimental model examining retinal ischemia in rats [35]. This was in line with our findings, showing that the DCP was more affected by COVID-19. A possible explanation for the vulnerability of the DCP in ischemia could be the distance from the bigger retinal arteries and the proximity to the outer retina which has bigger metabolic requirements due to the photoreceptors and retinal pigment epithelium's activity. In addition, patients that suffer from chronic obstructive pulmonary disease or sleep apnea syndrome have been found to have decreased retinal VD, implying that oxygenation plays an important role in macular microvascular damage [36,37].

There are some limitations of our meta-analysis. Firstly, there was significant heterogeneity in some observations of our analysis. This may be caused by the different devices and protocols used in the different studies. Specifically, most of studies included in the analysis used the Optovue OCTA device (8 out of 12), while Topcon and Nidek devices have been also used with variable signal strength index values. Moreover, there was not sufficient data to pursuit meta-regression analysis to identify if certain characteristics, such as age or sex, were responsible for the heterogeneity. Additionally, none of the included studies provided data on PCR cycles or severity of COVID-19 infection, which may affect the microvascular changes observed. Furthermore, the variability in the time from COVID-19 recovery until recruitment in the studies could influence the results. In fact, most of the studies included in our meta-analysis examined patients more than 3 months since their recovery from COVID-19 infection. However, Ozbas et al. and Abrishami et al. recruited patients at least 2 weeks after their recovery, while Dipu et al. examined patients 4–6 weeks after their discharge [17,26,28].

In conclusion, we found retinal vascular impairment in patients recovered from COVID-19 infection compared to healthy controls. Particularly, the FAZ area was enlarged, and the VD at DCP was decreased. Retinal microvascular changes in COVID-19 recovered patients indicates that routine ophthalmic examination in such patients may be necessary in the post-infection rehabilitation period. Moreover, long-term evaluation is needed to determine if these microvascular changes are transient or permanent. Future cohort studies are required to further investigate our findings in patients recovering from COVID-19 infection.

Funding

None.

Consent to publish

Not applicable, since this is a meta-analysis.

CRediT authorship contribution statement

Dimitrios Kazantzis: Writing – original draft, Methodology, Investigation, Conceptualization. Genovefa Machairoudia: Resources, Data curation. George Theodossiadis: Writing – review & editing, Resources. Panagiotis Theodossiadis: Writing – review & editing, Resources. Irini Chatziralli: Writing – review & editing, Visualization, Validation, Supervision, Methodology, Investigation, Conceptualization.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.