Ocular ndings and retinal involvement in COVID-19 pneumonia patients: A cross-sectional study in an Italian referral centre

Maria Pia Pirraglia Sapienza University of Rome Giancarlo Ceccarelli Sapienza University of Rome Alberto Cerini Sapienza University of Rome Giacomo Visioli Sapienza University of Rome https://orcid.org/0000-0002-6152-433X Gabriella d'Ettorre Sapienza University of Rome Claudio Maria Mastroianni Sapienza University of Rome Francesco Pugliese Sapienza University of Rome Alessandro Lambiase Sapienza University of Rome Magda Gharbiya (  magda.gharbiya@uniroma1.it ) Sapienza University of Rome https://orcid.org/0000-0002-4991-9689


Introduction
As a result of its pandemic spread and the very limited therapeutic options, COronaVIrus Disease 19  is considered an unprecedented global health challenge. Italy was the rst European country affected by a severe outbreak of the SARS-CoV-2 epidemic that emerged from Wuhan region (China), and currently has 241.611 total cases and 34.861 deaths. [1][2][3] As a life-threatening condition, most of the research has been primarily focused on respiratory system for a survival rate improvement, leaving the effects on other systems or districts still unclear or unknown. [4][5][6][7] Regarding eye involvement in COVID- 19, only few data are currently available. 8 The subfamily of Orthocoronavirinae -in which SARS-CoVs belong to the Betacoronavirus genusis already known to occasionally affect ocular structures. [9][10][11][12] Several manifestations like uveitis, retinitis and optic neuritis have been described in animal models of murine and feline species, however these ndings have never been con rmed in humans. 13 First evidence on humans came from a few studies conducted during the early 2000s Severe Acute Respiratory Syndrome (SARS) outbreak caused by SARS-CoV-1. The SARS RNA was detected in the tears of little cohorts of patients, suggesting the eye as an entrance window for infection and/or as a hypothetical source of viral spread. [14][15][16] Eventually, the fast contention of the disease had rapidly turned off the scienti c interest about the ocular involvement. In more recent times, the SARS-CoV-2 epidemic has revived the interest in ocular manifestations especially since an associated conjunctivitis has been described and colonization of ocular surface has been reported, arising the same unsolved questions emerged in 2000s and suggesting the implementation of quick precautionary strategies to protect ophthalmologists and their patients. [17][18][19][20][21] Nevertheless, although recent data suggests that COVID-19 infection may be associated to changes in immune and coagulation systems and possible viral spread through blood-brain barrier, with clinical and anatomopathological ndings of disseminated intravascular coagulopathy (DIC), the effects of these alterations on the eye, speci cally regarding the posterior segment involvement, have never been studied. 13,[22][23][24] The main objective of the present cross-sectional study was to explore the possible retinal involvement in COVID-19 and to provide real-world data on ocular ndings from SARS-CoV-2 positive pneumonia patients.

Setting
We conducted a cross-sectional study at the Policlinico Umberto I, a large teaching hospital in Rome (Italy), that included a cohort of patients affected by COVID-19 who were hospitalized in one Intensive Care Units (ICU) and in two Infectious Diseases wards, from April 24 to May 24, 2020. The study was approved by the ethical board of the Sapienza University of Rome (Rif. 5965, Prot. 109/2020) and was conducted in accordance with the tenets of the Declaration of Helsinki. All patients gave written informed consent to the study.
During the study period, 68 patients were hospitalized at any stage of the disease in the above-mentioned units. Twenty-one patients were not able to be screened because in Continuous Positive Airway Pressure (CPAP) therapy and one patient denied the consent. Therefore, we nally screened a total of 92 eyes of 46 patients. Patients were treated with ad interim best available therapy (BAT), according to the Italian Society of Infectious and Tropical Diseases (SIMIT): Hydroxychloroquine 200 mg bid and azithromycin 500mg daily, plus Tocilizumab 8 mg/kg (up to a maximum of 800 mg per dose) twice with an interval of 12 hours. All patients were on weight-based low-molecular-weight heparin (LMVH) and systemic steroids treatment 0.5-1 mg/Kg. 25 Inclusion criteria were: (1) age between 18 to 90 years, (2) con rmed positive results for SARS-CoV-2 from nasopharyngeal or oropharyngeal swab testing at the time of the ophthalmological assessment, and (3) lung involvement related to COVID-19. We excluded patients who had: (1) active neoplasia; (2) history of any ocular diseases such as glaucoma, uveitis, retinal vascular occlusion or major eye surgery performed within the previous six months. Based on the clinical conditions, patients were strati ed following the COVID-19 phenotypic classi cation proposed by the Italian Society of Anesthesiology, Analgesia, Resuscitation and Intensive Care (SIAARTI): paucisymptomatic disease (Stage I), mild pneumonia (Stage II), moderate to severe pneumonia (Stage III), acute respiratory distress syndrome (ARDS, Stage IV), sepsis (Stage V), septic shock (Stage VI). 26 In order to classify anamnestic and prognostic comorbidity at baseline, we used the Charlson Comorbidity Index (CCI). 27 Finally, at time of ophthalmological screening, in order to assess the patient in ammatory status and thrombotic risk, we recorded the laboratory tests of the day. Erythrocyte sedimentation rate (ESR), C-reactive Protein (CRP), white blood cell (WBC) count and Lymphocytes (LYM) values were collected to evaluate the subject's in ammatory status, while the thrombotic risk was scored using the International Society on Thrombosis and Haemostasis (ISTH) criteria for DIC. 28 Ophthalmological evaluation Bedside ophthalmologic evaluation was performed in both eyes and included: ocular annexes and anterior segment examination using a direct lighting and a 20-dioptre lens for magni cation. In those cases, presenting mono or bilateral conjunctival hyperaemia, a conjunctival swab was performed in both eyes for RNA SARS-CoV-2 detection. Quantitative corneal sensitivity, scored from 0 to 6 (0 corresponding to the absence of sensitivity and 6 to the highest sensitivity), was further assessed following a previously described protocol, only in awake patients, using the Cochet-Bonnet aesthesiometer (COBO). 29 Ocular fundus examination was performed after pharmacological pupil dilation with 1% Tropicamide, using binocular indirect ophthalmoscopy and a 20-dioptre lens. Images of the posterior pole have been acquired by the same investigator (MPP), using a handheld fundus camera with a 40-degree eld of view, 9 internal xation targets for peripheral imaging and 5-megapixel resolution (Optomed Smartscope®).

Data source and collection
For every patient included in the study we collected demographic data, systemic and ocular history, laboratory test results, medical administration data, and ocular ndings. All data was recorded using an Electronic Case Report Form (eCRF) by one investigator (GV). Data is then securely transferred to a central database, where missing data or every discrepancy was corrected by a double-check analysis or after a collegial evaluation.

Statistical analysis
Normal distribution of data was analysed by the Shapiro-Wilk test. Continuous variables were reported as mean, median, maximum and minimum values, and interquartile ranges (IQR 25% and 75%). Categorical variables were reported as counts and percentages. All analyses were performed using SPSS v. 25.0 (SPSS, Inc., Chicago, IL).

Results
Starting from 46 screened patients, based on the inclusion and exclusion criteria, we excluded two patients with active neoplasia and one with chronic glaucoma. Finally, a total of 43 subjects (25 males and 18 females) with a median age of 70 [IQR 59-78] was included in the present study. The patients have been hospitalized after a median of 4 days (range, 0 to 11 days) from COVID-19 symptoms onset and the ophthalmological screening was performed after a median of 21.5 days (range, 1 to 47 days) from hospitalization. Comorbidity index was calculated at admission and ranged from 0 to 7, with a median of 1. Ten out of 43 patients (9 male and 1 female) were screened in an ICU setting. Baseline anamnestic and clinical characteristics of the study cohort are shown in Table 1. Patients' clinical status, as assessed by the SIAARTI COVID-19 classi cation, varied from stage II to stage V, with men globally in worse conditions than women. COVID-19 stages' distributions are shown in Table 2. Laboratory tests showed signs of systemic in ammation with general high CRP and low lymphocytes count. Overall, ddimer values were signi cantly elevated, however only one patient (2.3%) had suggestive criteria for DIC according to ISTH and 4 patients had a prior diagnosis of COVID-19 related pulmonary thromboembolism. Table 3 shows the principal laboratory tests results in our cohort.
Regarding ocular anterior segment ndings, we observed 3 cases (7%) of bilateral conjunctivitis: two patients were in stage II and one in stage III of SIAARTI COVID-19 classi cation. In all three patients the PCR assay from conjunctival swab for the detection of SARS-CoV-2 provided negative results. Corneal sensitivity score, in screened eyes, as assessed by esthesiometry, ranged from 4 to 6 with a median of 5. Regarding the ocular posterior segment ndings, apart from one patient with unilateral posterior chorioretinitis that is discussed below, no further retinal manifestation related to COVID-19 infection was found in our cohort. The patient with chorioretinitis, was a 67-year-old male, hospitalized in ICU in stage V disease, presented grade 1 vitreous haze, a wide area of deep chorioretinal whitening involving the posterior pole, associated with deep retinal haemorrhages (Figure 1a). According to standard protocol, an aqueous tap to rule out possible pathogens including SARS-CoV-2 was performed. A diagnosis of probable fungal retinitis was done, and the systemic antifungal therapy was changed accordingly, by replacing IV caspofungin with amphotericin B. Microbiological tests excluded the presence of SARS-CoV-2 in humour aqueous and blood culture subsequently con rmed the diagnosis of Candida parapsilosis infection. Chorioretinitis gradually improved ( Figure 1b) and blood culture became sterile, however the patient died for SARS-CoV-2 related pneumonia 4 weeks later. Table 4 shows the ocular ndings observed in our cohort.

Discussion
The primary outcome of the study was to evaluate the presence of posterior segment alterations through a cross-sectional sample at different stages of the COVID-19, and to the best of our knowledge this is the rst study to address this issue in a cohort of hospitalized people positive for SARS-CoV-2. All available evidence in humans is focused on the ocular surface -where almost only conjunctivitis has been described -and most data comes from case reports or ndings from human conjunctival samples. 17,30,31 In our patients, of whom we carefully considered baseline anamnestic, results of fundus examination seem to demonstrate that neither the retina nor retinal vessels are involved in the active phase of COVID-19 infection.
The rationale in focusing on potential ocular fundus alterations, relies on previous studies conducted in animals infected by viruses belonging to the large sub-family of Orthocoronavirinae. The occasional onset of uveitis or retinitis in feline and murine models has been described and linked to an underlying autoimmune process inducing vasculitis or to a viral-mediated in ammation. [9][10][11][12][13] Recent clinical and anatomopathological reports have described the endothelial damage as one of the most prominent causes of the systemic vascular thromboembolic and/or in ammatory manifestations of COVID-19. [32][33][34] In this setting, the retina as a privileged district for non-invasive and in vivo evaluation of systemic diseases, may reveal alterations such as vascular occlusion related to the thrombotic susceptibility and chorioretinitis or vasculitis directly mediated by the virus. Hence, as reported in the brain, we considered the possibility of a direct ocular spread of SARS-CoV-2 through the two blood-retinal barriers (BRBs). In the recent literature on COVID-19 there are only anecdotal reports of virus spread through the blood-brain barrier. [35][36][37] Although two proven cases of positive CSF testing for SARS-CoV-2 have been described and one post-mortem, there is no certain data proving that the virus is able to directly affect central nervous system. 38,39 Our ndings, on a cohort of subjects in different stages of the disease, including ICU patients, seem to demonstrate that the SARS-CoV-2 may not be able to cross the ocular BRBs. Furthermore, to the best of our knowledge, here we report the rst attempt to isolate SARS CoV-2 in human aqueous humour.
With respect to the known thrombotic susceptibility described in COVID-19, we did not nd any sign of retinal vascular involvement, such as venous or arterial occlusion. However, it should be considered that all patients in our cohort were treated with LMWH to prevent systemic vascular complications and only one patient addressed the criteria for DIC.
At the present time, there is only one report that described retinal lesions in a little cohort of asymptomatic SARS-CoV-2 positive patients. The authors found cotton-wool-like lesions and microhaemorrhages in 4 out of 12 patients and inner retinal OCT hyperre ective spots in the whole sample. However, apart from "normal blood parameters", the authors did not provide any speci c information enabling the clinical characterization of their patients. Indeed, no data regarding the presence of systemic comorbidities as well as no details regarding the patients' ongoing therapy were given. Hence, it cannot be excluded that their ndings may be ascribed to pre-existing non COVID-19-related systemic diseases affecting the retina, such as hypertensive or diabetic retinopathy or other infectious diseases. [40][41][42] In our cohort, largely composed of subjects with severe COVID-19, we merged retinal ndings with baseline anamnestic. Nevertheless, our negative results should also be attributed to the ongoing immune modulating treatment, including steroids or Tocilizumab, that could have concealed ophthalmoscopic ndings. Finally, with regard to the patient with chorioretinitis from fungal sepsis, it should be considered that severe COVID-19 patients, especially when hospitalized in an ICU setting, may be affected by super-infection due to several opportunistic pathogens with possible eye localization. 43 Regarding the anterior segment ndings, literature describes conjunctivitis as a part of the clinical manifestation of COVID-19 with a variable rate of presentation, going from 0 to 32%. Only in 4-7% of cases PCR revealed the presence of SARS-CoV-2 from conjunctival swab. 17,19,31,44 In all three patients presenting bilateral conjunctivitis that were observed in our study, the conjunctival swab was negative for SARS-CoV-2. Nevertheless, since other viruses (e.g. herpes and adenovirus) are known to induce and have been detected in conjunctivitis 45 , we cannot exclude a transient presence of SARS-CoV-2 on the ocular surface at any time before or after our swab. Alternatively, conjunctivitis may represent an epiphenomenon in hospitalized patients. Since it has been de nitely demonstrated that COVID-19 affects the peripheral nervous system, as shown by the reported 85-88% rate of olfactory and gustatory dysfunction 46 , we investigated the possible involvement of trigeminal sensory pathways by exploring the corneal sensitivity. However, results of aesthesiometry seem to demonstrate that, unlike herpes viruses, SARS-CoV-2 does not affect corneal sensitivity. 47 One strength of this cross-sectional study is to have explored posterior segment involvement in COVID-19 pneumonia patients at different stages of disease and in patients who had different comorbidities at the time of their hospitalization. Furthermore, considering our brief study period and the actual feasibility of an ophthalmological evaluation in an emergency setting, we obtained data from a relevant sample size even if a larger study population is probably necessary to con rm our ndings. Among the shortcomings of our research it should be disclosed potential bias resulting from systemic therapies put in place before the ophthalmological evaluation, and those deriving from the exclusion of patients in CPAP therapy.

Conclusion
In conclusion our study demonstrated the absence of retinal manifestations in SARS-CoV-2 pneumonia patients. Given the frequent drug-induced immunological dysfunction, ophthalmological evaluation may be useful to reveal systemic co-infections by opportunistic pathogens, especially in ICU patients.