Investigation of changes in retinal vascular parameters and choroidal vascular index values during the early recovery period of COVID-19: The COVID-OCTA study

Abstract

Background

COVID-19 effects microvasculature in many tissues. This study investigated whether the choroidal structure is also affected.

Methods

This cross-sectional study included 80 patients with COVID-19 and the same number of age- and gender-matched healthy individuals. All participants' right eye measurements were examined. Optical coherence tomography angiography (OCTA) was used for imaging. Otherwise, two independent researchers used the Choroidal vascular index (CVI) for choroidal parameters calculation.

Results

Superior and deep flow values were lower in the COVID-19 group than in the control group, and vascular density (VD) values were lower in all regions in this group. Except for the superior mean VD, there was no statistically significant difference (p = 0.003). However, the COVID-19 group had significantly lower subfoveal choroidal thickness (SFChT) measurements than the control group (p = 0.001). In addition, no significant difference was observed between the groups in evaluating mean CVI values (p>0.05).

Conclusion

Noninvasive diagnostic tools such as OCTA and EDI-OCT can be used to monitor early changes in diseases affecting microvessels, such as from COVID-19.

1. Introduction

We have been experiencing the human coronavirus disease 2019 (COVID-19) pandemic since the end of 2019. The pandemic was caused by the acute respiratory syndrome coronavirus (SARS-CoV-2) and resulted in too many deaths [1]. There is currently no sign of an end to the epidemic, which has led to one of the deadliest events in the world. COVID-19 can affect the ocular vascular system in addition to many tissues [1,2]. Cell infection occurs due to the virus interacting with the angiotensin-converting enzyme-2 (ACE-2) receptor [1]. Immunohistochemical studies and single-cell RNA-sequencing datasets have revealed the extra- and intra-ocular localizations of SARS-CoV-2 receptors [3,4]. It is known that there are receptors for the virus in both anterior and posterior segment structures [4,5]. The choroid tissue forms the vascular layer of the eye and is one of the most vascularized tissues in the body [6]. The choroid exhibits critical physiological functions such as oxygenation and nutrition of the outer retina, regulation of retinal temperature, positional state of the retina, and secretion of growth factors [6]. SARS-CoV-2 causes vascular system damage through immune dysregulation, hyperinflammatory process, endothelial cell dysfunction, and hemodynamic instability [7]. Inflammation in the vascular system leads to diffuse microangiopathy and thrombosis. In COVID-19, in which the microvasculature in many tissues is affected, the easy accessibility of retinal and choroidal vessels in terms of observation and image acquisition creates a window to have an idea about the general vascular status.

Optical coherence tomography angiography (OCTA) is a new, noninvasive, reproducible imaging method for imaging choriocapillaris (CC) and retinal vascular structures. OCTA can detect local ischemia with vascular density (VD), foveal avascular zone (FAZ), superficial capiller plexus (SCP), deep capiller plexus (DCP) and CC flow measurements. In addition, by evaluating the choroidal anatomy and choroidal thickness (ChT) in different regions with the enhanced depth imaging (EDI) mode, structural features such as the percentage of vascularity of the choroid can be evaluated with the help of automated software [8]. Choroidal vascular index (CVI) is a newly defined marker and offers the opportunity to examine choroidal vascularity in different retinal pathologies. CVI shows the ratio of the vascular lumen area to the total choroidal area in the choroidal region between Bruch's membrane and the choroido-scleral space by decomposing the OCT image taken in EDI mode with the help of a computer program [9]. Since CVI is less affected by physiological parameters such as axial length, age, intraocular pressure, and systolic blood pressure than the subfoveal choroidal thickness (SFChT), it is thought to be an index that reveals choroidal changes more consistently [9,10].

Studies showing chorioretinal changes in COVID-19 have revealed different perspectives on the pathophysiological aspect of choroidal involvement. In this study, we aimed to observe the choroidal changes in COVID-19. This is the first study in which OCTA and CVI measurements were evaluated together.

2. Material and methods

The study was carried out as a cross-sectional retrospective, in accordance with the principles of the Declaration of Helsinki, with the ethical approval of the Clinical Studies Ethics Committee of Kahramanmaraş Sutçu Imam University. A total of 80 patients, 33 men and 47 women, whose diagnosis of COVID-19 was confirmed by real-time PCR test, and the data of patients who applied for eye symptoms at least two weeks after the end of their hospitalization were included in the study. Favipiravir and nasal oxygen were given to the included patient group during hospitalization. The study did not include the data of patients with intensive care and mechanical ventilator therapy. The ophthalmological and systemic histories of the cases were examined in detail. Cases with diseases or conditions that may affect the vascular system (such as diabetes, hypertension, cardiovascular disease, autoimmune disease, pregnancy, breastfeeding, migraine, or anemia) were excluded. In addition, cases with media opacity that may affect the quality of ophthalmologic imaging, cases with visual acuity less than 20/20, refractive errors of over spherical five diopters and cylindrical two diopters, previous ocular surgery, glaucoma, and other optic nerve diseases, uveitis, orbital trauma and cases with signs of retinal disease were not included in the study. The data of the same number of healthy individuals who applied for routine eye examinations and matched the COVID-19 group regarding age and gender were determined as the control group.

2.1. Optical coherence tomography angiography imaging

Before imaging, the subjects' pupils were dilated using a combination of 0.5% tropicamide and 2.5% phenylephrine HCL. Each session was performed at approximately the same time in the morning to avoid daily variability, minimizing the effects of day-to-day variability. Spectral-domain OCTA RTVue XR AngioVue (software version: 2015.1.0.90; Optovue, Inc., Fremont, CA, USA) was used to obtain high-quality images showing retinal vascular structures and vasculature. This device provides volumetric exposure of 304 × 304 A at 70,000 A-scans per second with an 840 nm light source with a 5 mm axial resolution. This technology also quantitatively analyzes data and calculates VD and flow area. The OCTA system is based on a split-spectrum amplitude-decorrelation angiography (SS ADA) algorithm that uses blood as intrinsic contrast. For this purpose, five repeated B-scans are performed at 216 scan locations, providing 3D scans per region of 3 × 3 mm. Each B-scan consists of 304 × 304 pixels, has 216 A transverse dimensions and has a frame rate of 270 frames per second. Each scan took approximately 3.0 s. OCTA images were created on the RTVue XR Avanti machine. This technology combines the integral SS ADA algorithm of the AngioVue system and the motion correlation technique. Two experienced ophthalmologists (AB, AM) performed OCTA scans and measurements during angiographic analysis on the RTVue XR Avanti equipped with AngioVue. The 3 × 3 mm frame was used to display the perifoveal region, the foveal avascular region (FAZ), and the parafoveal region. The term' flow area' indicated the percentage of vessels occupied in an area of 3 × 3 mm per square covered central region of FAZ. This software automatically calculates the flow area in the region of interest in both SCP and DCP (Fig. 1 A and B). The CC flow field was analyzed with the OptoVue software, in which we measured the flow function of a 3 × 3 mm area in the macular angiogram of the CC layer (Fig. 1C) [11,12]. A vascular region with a larger-than-normal gap between capillaries can be identified as a low flow signal on en-face angiography. As a fixation point on the retina, FAZ was used as an anatomical landmark. Measurements on SCP and DCP were made after the localization of FAZ (Fig. 1D and E). VD defines the percentage of vessels and microvessels in a selected area. The mean value of VD in the SCP and DCP layers was automatically measured (Fig. 1F and G).

Fig. 1.

Example figure showing OCTA measurements.

2.2. Choroidal imaging and choroidal vascularity index

Enhanced HD line scans were obtained on central macular thickness (CMT), SFChT, and CVI measurements. CMT and SFChT are made manually. For the CMT measurement, a built-in caliper tool is placed from the inner edge of the internal limiting membrane to the inner edge of the retinal pigment epithelium. For the SFChT measurement, a built-in caliper tool is placed from the outer edge of the retinal pigment epithelium to the choroid-sclera border perpendicularly at the level of the fovea. Measurements were done twice, and the mean of two discrete values was calculated for each participant.

The enhanced HD line images were binarised and segmented by the same examiners using the public domain software ImageJ 1.53 s (National Institutes of Health, Bethesda, MD), with a semiautomated method previously described [13]. Briefly, the OCT image was opened in ImageJ, and the polygon tool was used to select the region of interest across the entire length of the OCT scan. The upper boundary of the region of interest was traced along the choroidal–retinal pigment epithelium junction and the lower boundary along the choroidal–scleral junction to identify the choroidal area (TCA) (Fig. 2 a). After conversion to an 8-bit image Niblack's auto local threshold was applied to binarise the image and demarcate the lumen area (LA) and stromal area (SA) (Fig. 2b). The image was converted back to a red, green, and blue image, and the color threshold tool was used to select the dark pixels representing the LA (Fig. 2c). The TCA and LA were measured. The SA was calculated by subtracting LA from TCA. The CVI was computed as the LA divided by the TCA. Two investigators calculated the choroidal parameters separately, blinded to patients' characteristics (AB and AM). The mean value for each parameter calculated was used for the statistical analysis.

Fig. 2.

Image processing for CVI measurement.

2.3. Statistical analysis

Statistical analysis of all data was performed using SPSS software for Windows. (Version 22.0, IBM Co., Chicago, IL, USA). The conformity of the results of the patients with COVID-19 infection and the healthy control group to the parametric test analysis (conformity to normal distribution) was calculated with the Kolmogorov-Smirnov test. Statistical analysis of the results that met the parametric measurement conditions was calculated using the Independent Student T-test. The analysis of categorical variables was calculated with the 'Chi-Square test.'

3. Results

This retrospective study evaluated 80 COVID-19-positive patients and 80 age- and sex-matched control subjects. There were cases of equal gender in both groups (33 male, 47 female). The mean age was 37.99 ± 11.23 years in the patient group, while it was 39.64 ± 12.36 years in the control group (p = 0,689).

SCP, DCP, CC, FAZ (superior, deep), VD (mean/parafoveal; superior, deep), SFChT and CVI parameters are shown in Table 1 . SCP, DCP and CC values were measured lower in the COVID-19 group than in the control group, but the difference was not statistically significant (respectively; p = 0.239, p = 0.164, p = 0.158). sFAZ and dFAZ values were lower in all regions in the COVID-19 group but did not reach statistical significance (respectively; p = 0.488, p = 0.950). Superior mean VD was found to be significantly lower in the covid group. (p = 0,003) Other VD measurements (superior/deep parafovea and mean deep) were not found to be significant (respectively; p = 0.067, p = 0.116, p = 0.307). SFChT measurements were significantly lower in the COVID-19 group than in the control group (p = 0.001). Although TCA, LA, SA and CVI measurements were lower in the covid group than in the control group, they were not statistically significant (respectively; p = 0.083, p = 0.058, p = 0.662, p = 0.052).

Table 1.

Display of OCTA and choroidal measurement values of COVID-19 and control group^*.

	COVID-19 positive group	Control group	p value**
SCP (µm2)	1.41±0.10	1.43±0.12	0.239
DCP (µm2)	1.49±0.15	1.53±0.17	0.164
CC (µm2)	1.91±0.05	1.92±0.05	0.158
sFAZ (µm2)	0.32±0.11	0.33±0.09	0.488
dFAZ (µm2)	0.37±0.13	0.37±0.09	0.950
MEAN sVD (%)	49.42±3.14	51.05±3.76	0.003
PARAFOVEA sVD (%)	52.89±3.45	54.03±4.30	0.067
MEAN dVD (%)	57.47±3.86	58.48±4.18	0.116
PARAFOVEA dVD (%)	62.27±3.28	62.90±4.45	0.307
SFChT (µm)	276.82±18.25	288.90±27.33	0.001
TCA (µm2)	1598.65±24.05	1604.77±20.21	0.083
LA (µm2)	1062.60±12.66	1068.67±12.55	0.058
SA (µm2)	536.05±8.56	536.10±8.35	0.662
CVI (%)	0.6645±0.0091	0.6659±0.0087	0.052

^⁎Data expressed as mean±standard deviation.

^⁎⁎Independent Student's T test: significant values p<0.05, Show in bold.

SF; Superficial Flow, DF; Deep Flow, CC; Choriocapillaris sFAZ; Superficial Foveal avascular zone, dFAZ; Deep Foveal avascular zone, sVD; Superficial Vascular density, dVD; Deep Vascular density, SFChT; Subfoveal choroidal thickness, TCA; Total choroidal arae, LA; Lumen area, SA; Stromal area, CVI; Choroidal vascular index.

4. Discussion

The first step for SARS-CoV-2 cellular infection is the interaction of the virus S protein and the cell surface receptor ACE2. This is followed by hydrolysis of the S protein by the transmembrane protease serine 2 (TMPRSS2). ACE2 and TMPRSS2, as well as the newly identified CD-147 spike protein, are highly expressed in the human eye and are thought to play a role in ocular manifestations [14]. Confirming this information, the SARS-CoV-2 genome was detected by real-time reverse transcriptase-polymerase chain reaction (RT-PCR) in retinal biopsies in 3 of the 14 patients who lost their lives due to COVID-19 [15]. The clinical implications of this finding have been described in various studies as retinal lesions associated with COVID-19. These lesions include retinal hemorrhages, cotton wool retinal exudates, increased retinal vessel diameters, and increased vascular tortuosity [16,17]. In addition, OCT detected hyper-reflective lesions, micro hemorrhages, and nerve fiber infarcts in the ganglion cell and inner plexiform layers [18]. In histological studies, amorphous debris in the outer plexiform layer corresponding to hemorrhages and numerous cystoid lesions in the central retina were detected in eyes with COVID-19 [19]. These findings suggest that COVID-19 causes retinal vasculitis and ischemia. Vascular damage is thought to occur due to a disseminated intravascular coagulation-like hypercoagulability state and diffuse endothelial inflammation caused by a vasculitis-like process directly related to viral invasion [20].

In our study, images of patients with findings such as retinal hemorrhage and exudate due to ischemic causes were not taken, as this would increase the formation of artifacts in the images. However, we found reductions that we think are related to early ischemic findings due to vascular endothelial damage.

OCTA is a new-generation noninvasive imaging method that displays retinal and choroidal blood flow in high resolution and thus provides information about ischemia. Therefore, it is a beneficial method for quantitatively monitoring retinal and choroidal vascular changes that may develop due to covid. When studies on OCTA measurements in COVID-19 cases are examined, it is seen that different results are obtained. In a significant part of the literature, it is observed that VD significantly decreases in both the superficial and deep retinal and choroidal vascular layers in COVID-19 cases [21], [22], [23], [24], [25]. Therefore, changes in retinal and choroidal vascular density measurements are likely secondary to perfusion impairment and vasoconstriction due to COVID-19. Retinal and choroidal vascular involvement may be the first observable finding among the vascular component effects of COVID-19.

In our study, we could not find a significant difference between the groups, although there was a decrease in the data of patients who were positive for COVID-19 regarding flow and FAZ. However, there was a significant decrease in mean VD in the study group. We consider this as a secondary response to COVID-19 induced endothelial injury and vasoconstriction in the early stage of the disease.

On the other hand, some studies showed no significant difference between the COVID-19 and control groups in some studies [26], [27], [28], and even Naderi Beni et al. found no significant difference between the recovered COVID-19 group and the control group by pairwise comparisons. However, when parameters such as age, gender, axial length, and signal strength index (SSI) were adjusted, they observed that vascular density increased in parafoveal and perifoveal SCP and DCP, on the contrary [29]. These differences are likely due to the design of the studies. In addition, differences in factors such as the severity of the disease, possible undiagnosed vascular disorders, the cross-sectional or prospective nature of the studies, and the stage of the disease in which the measurements were taken, may affect the results achieved.

As a noninvasive diagnostic method, EDI-OCT allows for examining choroidal morphological features in high resolution by reducing the signal strength behind the retinal pigment epithelium in a cross-sectional manner [30]. However, the CVI measurements obtained by binarizing EDI-OCT images take this technique one step further. In a recent study, Agarwal et al. showed that CVI is a more stable index of choroidal changes than SFChT and is less affected by physiological variables such as axial length, age, intraocular pressure, and systolic pressure [9]. Also, CVI has less variability than SFChT. As with VD results, choroidal involvement has been observed to give different results in various studies: vascular and stromal depletion was reported on the one hand [21], while thickening was found in the macular region on the other hand [31]. In this study, it is seen that SFChT decreased significantly in the COVID-19 group compared to the control group in the early recovery period. In a prospective study, Erdem et al. compared patients who had recovered from covid for at least one month with the control group. ChT decreased in covid cases, especially in the subfoveal, temporal, and inferior areas [32]. It is known that hypoxia increases circulating proinflammatory cytokines and causes edema following vascular leakage. Inflammation and edema associated with hypoxia may cause a decrease in choroidal blood supply [32]. However, Cetinkaya et al. stated that there was no difference in CMT, Retinal Nerve Fiber Layer Thickness, Ganglion Cell Layer Thickness, and ChT (in 5 locations) between COVID-19 and the control group [21].

Similarly, other studies found that the central foveal and choroidal thicknesses decreased in the COVID-19 group, but this decrease did not reach statistical significance [33,34]. On the contrary, Abrishami et al. stated that the subfoveal nasal and temporal ChT was measured higher in the active period of the disease [31]. The authors also noted that in the binarization of EDI-OCT images, infection induces substantial edema in the choroidal stromal area. This finding shows the hemodynamic effects of choroidal vasculitis, systemic inflammation, or respiratory disease. The different results for ChT in the literature may be due to the severity of the disease, the number of patients, inclusion criteria, time differences and subjective reasons in the evaluation. As a result, although it seems certain that the choroidal vascular structures are affected by the disease, more studies are needed to understand the amount and nature of the effect and which effects are directly related to COVID-19.

CVI is thought to be reduced in inflammatory diseases [35]. Xin et al. reported the mean CVI value as 70.02% in healthy subjects. In this study, although the CVI value was statistically insignificant, it was decreased in the COVID-19 group compared to the control group. In their research, Kocamış et al. observed a statistically significant decrease in the mean total, choroidal, luminal, and stromal areas in patients with COVID-19 compared with healthy subjects. They stated that these results were associated with vascular damage, hypercoagulability, and hyperinflammation factors causing ischemia in the choriocapillaris, leading to a decrease in CVI. The authors state that the results at four months after remission are close to the measurements of the control group [34]. A study comparing the group with mild COVID-19 patients and the healthy group found that CVI, LA and TCA significantly decreased in the patient group. The authors attributed this to vascular damage that may result from systemic and local inflammation due to COVID-19. They also attribute the non-significant change in SA to the fact that the choroidal stroma is less affected by the choroidal vascular structure [36]. Similarly, our study found that TCA, LA, SA and CVI were decreased in the COVID-19 group compared to the control group, although it was not statistically significant. We think that this decrease may be due to vascular endothelial damage. The more limited decrease in SA can be explained by the edema caused by ischemia. Another prospective study reported that choroidal vascularity decreased, and the stromal/vascular area ratio increased significantly during the active period of COVID-19 cases. As a result, they observed that mean choroidal thickness and choroidal area increased, vascularity decreased, and stromal area increased during the active period of the disease [37].

Abrisami et al. showed that CVI increased significantly after one month but returned to baseline measurements three months after symptom onset. They stated that there was no significant change in choroidal thickness in the same period. Based on the time at which changes occur, the authors declare that serum extravasation and subsequent stromal edema develop in the active phase and resolve more rapidly than luminal enlargement [38]. Hepokur et al., on the other hand, showed a decrease in choroidal thickness in the early postinfectious period. They stated that this decrease occurred both in the choroidal stroma and the vessels, so the CVI did not change [39]. Similarly, this study found that SFChT was decreased in the COVID-19 group compared to the control group, but CVI did not differ between the groups. We attribute the decrease in SFChT to vascular damage. This decreasing trend in TCA, LA, SA, and CVI may be related to the fact that COVID-19 disease affects tissues differently. The variability of CVI was less than SFChT after COVID-19 disease. Therefore, we think CVI can be used to follow up on ocular side effects after COVID-19. Abrishami et al. also stated that the ChT was normalized in the late postinfectious period, and the CVI remained unchanged, thus possibly simultaneous healing of the stroma and vessels [38].

This study has several limitations. Firstly, this study was conducted in a single center. Secondly, the CVI measurement is partially subjective and choroidal measures do not cover the entire choroid. Additionally, since all patients received the same treatment and were evaluated at least two weeks after discharge, the possible effect of the treatment received during hospitalization on our findings should have been considered.

In conclusion, retinal and choroidal vascular structures are certainly affected due to vasculitic and inflammatory involvement in COVID-19 patients. In light of these results, we need more studies on the virus's direct effect, its relationship with systemic inflammation, and the persistence of the impact. This study demonstrates that OCTA and EDI-OCT can be used as convenient diagnostic tools to non-invasively monitor early changes in diseases affecting the microvasculature, such as COVID-19.

Funding

The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.

Ethics approval

The study was approwed by the Ethics Committee of Sutcu Imam University, Kahramanmaras, Turkey (Protocol No: 03/2021–149) and all procedures were applied in accordance with the Declaration of Helsinki.

Patient consent for publication

Not required.

CRediT authorship contribution statement

Abdullah Beyoğlu: Writing – review & editing, Writing – original draft, Visualization, Validation, Software, Methodology, Investigation, Formal analysis. Ali Küçüködük: Writing – original draft, Investigation, Data curation. Ali Meşen: Writing – original draft, Investigation, Data curation. Mustafa Aksoy: Writing – original draft, Resources, Project administration, Investigation, Data curation. Erhan Kaya: Writing – original draft, Investigation, Data curation. Betül Dağhan: Data curation.

Declaration of Competing Interest

None declared.

Data availability

• Data are available upon required.