Analysis of Risk Factors for Cytomegalovirus Corneal Infection

Objectives: The present study assessed the risk factors for cytomegalovirus (CMV) corneal infection and compared the odds ratios. Methods: This study was a retrospective case-control study (2014.12–2021.1) over a 6-year period. We reviewed the medical record data and compared the differences between the CMV-positive and virus-negative groups with regard to sex, age, rst diagnosis, and ocular complications to identify factors that may be associated with CMV corneal infections. The risk factors for CMV corneal infection were analysed using the chi-squared test and binary logistic regression models. Results: Ninety CMV-positive cases and 151 virus-negative controls were involved in this study. The risk factors for CMV corneal infection were Stevens-Johnson syndrome ocular complication (OR 12.851, 95%CI, 1.435–111.111; P=0.000), a penetrating keratoplasty (PKP) (OR 4.950, 95%CI, 2.632–9.346; P=0.024, and an anterior lamellar keratoplasty (DALK) (OR 5.290, 95%CI, 1.250–22.222; P=0.022). Conclusion: Severe corneal damage (PKP or DALK) and a severe corneal inammatory response (SJS ocular complication) may be risk factors for CMV corneal infections. syndrome: endothelial syndrome; IOL eyes: lens SJS ocular Stevens-Johnsons syndrome ocular complications; oGVHD: ocular graft-versus-host


Introduction
A Japanese ophthalmologist reported the rst case of cytomegalovirus (CMV) keratitis in 2006, and there has been a constant surge of cases since this period. The available evidence showed that CMV corneal infection primarily occurred in Asian populations 2 3 , and it primarily occurred as sporadic cases in European and American countries 4 . CMV corneal infections have gradually received widespread worldwide attention in recent years, and the pathology 5 , diagnosis 6 and treatment [7][8][9] of CMV keratitis have shown promising progress under the wisdom and work of scientists. However, the causes and mechanism of the disease remain poorly understood 10 11 . Whether the damage to the corneal tissue caused by CMV corneal infection is a primary or secondary cause of the primary disease is notclear 12 .
The current hypothesis is that immune cells carrying latent CMV reactivate after entering the anterior chamber 13 . The alternative hypothesis is that the corneal tissue itself is the latent site of CMV, and the transplantation procedure reactivates the latent CMV within the cornea of the donor or recipient 14 15 . CMV corneal infection occurs in different populations 2 with different anterior segment disorders 16 17 18 .
However, the risk factors for CMV corneal infection were not reported. The identi cation of risk factors would improve the prevention and treatment of CMV corneal infection.
CMV infection is closely related to the body's immune status 19 . Previous studies demonstrated that patients with immune system abnormalities often develop retinal lesions 20 , but patients with CMV infection in the anterior segment of the eye mostly have a normal immune status 21 . Local immune factors are the primary reason for viral relapse, but the identity of the factors that activate the virus and cause lesion recurrence are not clear. This may be related to the relative independence of the immunity of the anterior segment, and it requires further investigation. Clinical observations and previous studies reported that of CMV corneal infections occurred more often in patients after corneal transplantation and were particularly frequent after penetrating keratoplasty (PKP) 22 23 . Our study determined whether there was any correlation between corneal transplantation and CMV corneal infection. This study was a retrospective case-control study. Cases and controls were keratopathy patients who were hospitalized in a corneal ward during the same period, and there were controllable differences in the two groups in CMV DNA PCR test results from the aqueous humour/corneal tissue. We investigated risk factors for CMV corneal infections by analysing the differences between these two groups.

Hospital Cases and Controls
This study was a retrospective case-control study (2014.12-2021.1) performed over a 6-year period. It was performed according to the tenets of the Declaration of Helsinki, and the local ethics committee approved the study. All cases were collected from the cornea ward of the eye centre of Peking University Third Hospital. Viral DNA detection was routinely performed intraoperatively in patients scheduled for keratoplasty in our ward to understand ocular viral infections in that group population. The results of viral PCR were consistent with the results of intraoperative viral detection. Inpatients who were positive for aqueous humour/corneal CMV DNA were collected for the study group by reviewing the results of previous inpatient tests. The controls were selected from all inpatients admitted to our cornea ward during the same period (same month and year) and had tests within the same batch for virus detection as the cases but had negative aqueous humour/corneal viral test results. Hospitalized patients with negative virus tests must have no clinical symptoms or signs of virus infection. Cases with incomplete test results or poor case quality for various reasons unrelated to the study were excluded during the case selection process. A owchart of the inclusion and exclusion criteria is shown in Figure 1.
In-hospital patients were included in the case and control groups of this study, and outpatients were excluded due to the lack of more complete inpatient data. During the control group selection process, patients with negative results from the same period and the same virus detection batch as the CMVpositive patients who underwent viral testing were selected.

CMV DNA detection
Cornea tissues/aqueous humour samples were obtained during cornea transplantation procedures.
Su cient aqueous humour was drawn from the patient using a 1-ml syringe before removing the patients' diseased corneal tissues. The removed corneal tissues and aqueous humour were placed into 1.5-ml EP tubes. Fresh tissues were sent to the ophthalmic clinical testing laboratory of Peking University Third Hospital for viral detection. We extracted DNA using a Liferiver CMV Real Time PCR Kit (catalogue no. Z-OD-0022_02; Shanghai, China) according to the manufacturer's instructions. The limit of detection of viral DNA was 10 copies per sample size. Each sample was processed with the addition of an internal control for the assessment of the isolation and ampli cation e cacy. Positive, negative and internal controls were provided by the kit manufacturer.

CMV corneal infection 24
A CMV corneal infection was de ned as patients who had a lesion in the cornea (active or stationary phase) and positive CMV DNA PCR detection results from the corresponding corneal tissue or aqueous humour. All patients intended to undergo corneal keratoplasties, and the corneas all had varying degrees of pathological changes (diagnosis). Therefore, the de nition of CMV corneal infection was positive CMV DNA detection in corneal tissue or aqueous humour.

Observation procedures and data analysis
Information on all cases, including sex, age, diagnosis and ocular-related diseases, was collected for the study group and controls. The diagnosis included endothelial decompensation, graft rejection, congenital corneal leucoma, limbal stem cell de ciency, corneal ulcer and graft melting. Ocular-related conditions included history of glaucoma/glaucoma surgery, history of corneal transplantation, vitrectomy surgery, IOL eye/aphakia, eye trauma, Stevens-Johnson syndrome, ocular graft-versus-host disease (oGVHD) and uveitis history. Corneal endothelial decompensation was divided into bullous keratopathy, corneal endothelial dystrophy and iridocorneal endothelial syndrome (ICE syndrome) based on the aetiology. The diagnosis of graft rejection was classi ed based on the case data as graft rejection after Descemet's Stripping Automated Endothelial Keratoplasty (DSAEK), PKP and anterior lamellar keratoplasty (DALK). Corneal ulcers were classi ed as bacterial, fungal and immunological based on the preoperative or intraoperative smears of the lesion and the culture results. Systemic diseases were based on the diagnosis of hypertension, diabetes, heart disease, surgical history, neoplasm, hemopathy, allergic history, thyropathy, rheumatic immune disease and cerebral infarction in the patients. The medication history included whether the patient received topical immunosuppressive agents. The description of the history of corneal transplantation included the surgical approach and the number of procedures.

Statistical analysis
Continuous variables are expressed as the means±standard deviation. The independent samples t-test was used to compare the difference in age between the two groups. Chi-squared tests were used for comparisons between two groups. Odds ratios (ORs) and 95% con dence intervals (95% CIs) were initially ltered by the chi-squared test. A binary logistic regression model using a backward stepwise selection method was used to further study the possible risk factors associated with CMV corneal infection. P values less than 0.05 (two-tailed) were considered statistically signi cant. SPSS 23.0 (SPSS, Chicago, Illinois, USA) was used for statistical analysis.
The independent samples t-test was used to compare the difference in age between the two groups. The chi-squared test was used to compare the other factors in the study and control groups. P values less than 0.05 (two-tailed) were considered statistically signi cant. %: As a percentage of the total number of study cases or controls *P 0.05,**P 0.01 Further re nement of the analysis of differences in the history of corneal transplantation between study cases and controls is summarized in Table 2. The number of patients having graft rejection after penetrating keratoplasty between the study cases and controls was 27 (30%) and 22 (14.6%) χ 2 =11.511, P=0.001), respectively. There were signi cant differences between the study cases and controls in a history of keratoplasty after PKP, and there were 39 (43.3%) and 22 (14.6%) cases who had keratoplasty, respectively (χ 2 = 24.679, P = 0.000). The numbers of study cases and controls who were seen after a DALK were 4 (4.4%) and 2 (1.3%), respectively (χ 2 = 4.755, P = 0.029), and the number of study cases and controls who had DSAEK were 9 (10.0%) and 22 (14.6%), respectively (χ 2 = 1.051, P = 0.305). The number of study cases and controls with a limbal stem cell transplantation (LSCT) was 9 (10.0%) and 6 (4.0%), respectively (χ 2 = 3509, P = 0.061), which was not signi cantly different between the two groups. The numbers of study cases and controls who had corneal keratoplasty were 37 (41.1%) and 39 (25.8%), respectively (χ 2 = 6.101, P = 0.014), and 14 (15.6%) and 4 (2.6%), respectively (χ 2 = 13.592, P = 0.000).
The chi-squared test was used to analyse factors that were different between the study cases and controls, and the results were shown in Table 3 15 . The nal results of this study revealed that the risk factors for CMV corneal infections were PKP, DALK and SJS ocular complications. The results were most striking for PKP, which had the smallest range of 95% con dence intervals. PKP and DALK caused greater damage to the corneal tissue due to the 360° annular incision involving the corneal stromal layers. The complexity of the preoperative primary disease 28 , the intense in ammatory response and corneal neovascularisation 28 exacerbated the immune damage secondary to PKP and DALK in the anterior segment of the eye. DSAEK surgery was less traumatic, had a signi cantly better postoperative recovery than PKP, and had a milder in ammatory response 29 . CMV may be latent in the stromal layer of the cornea 24, and the DSAEK procedure did not contain latent CMV from a donor because there were few antigen-presenting cells (APCs) in the grafts 30 .
Topical immunosuppression and glucocorticoids after keratoplasties may render the anterior ocular segment immunocompromised and contribute to CMV reactivation 31 32 .Local or systemic immunosuppressive agents with stronger effects tended to be applied, especially after keratoplasty, after considering the higher odds of rejection 30 . However, novel conclusions were made in studies of other organ transplants, and early tacrolimus application after transplantation plays a protective role against CMV infection 33 , which may involve the relationship between CMV latent cells and killer cells 34 35 . However, more evidence is needed to make an accurate conclusion. Immunosuppressants in this study were excluded as confounders in the binary logistic model analysis and were not analysed in detail.
SJS ocular complication is a typically intense in ammatory disease of the ocular surface that manifests as a defect in the corneal epithelium in the acute phase and nonspeci c keratitis and corneal lysis in the chronic phase due to the presence of ongoing in ammation. The present study revealed SJS ocular complications as a risk factor for CMV corneal infections, but this factor had a large con dence interval range. The reason for this result may be that the disease caused persistent in ammatory reactions on the ocular surface 36 , but the severities and treatments varied greatly between individuals. There have been no studies of ocular SJS manifestations associated with CMV corneal infections. Most SJS ocular complication patients in this study presented with a CMV corneal infection that did not have a history of corneal transplantation, which suggests that a CMV corneal infection may also result from reactivation of a recipient CMV infection under corneal in ammation and lesions 26 . oGVHD with ocular surface damage was not a risk factor in this study for several possible reasons. 1) The sample size of hospitalised oGVHD patients included in this study was small. 2) The corneal in ammation caused by oGVHD was super cial, had timely follow-up and treatment, and few involved the stromal layer 37 .
In conclusion, severe corneal damage (PKP or DALK) and a severe corneal in ammatory response (SJS ocular complication) may be risk factors for CMV corneal infections. This study preliminarily identi ed the possible risk factors for CMV cornea infection in a limited retrospective cohort study. The results provide a method to analyse the mechanism of CMV corneal infection.

Declarations Funding
The study was supported by a National Natural Science Foundation (81970768). Dr. Hong had full access to all of the data and takes responsibility for the data integrity and accuracy of data analyses.

Con icts of interest
None of the authors have any con icts of interest related to this study.

Availability of data and materials
Some or all data and materials generated or used during the study are available from the corresponding author upon request. (List items) Code availability Some or all codes generated or used during the study are available from the corresponding author by request. (List items)

Authors' contributions
Yunxiao Zang performed the studies and drafted the manuscript. Rongmei Peng and Jinghao Qu participated in the design of the study and performed the statistical analyses. Gege Xiao and Lixue Shuai conceived the study, participated in its design and coordination and helped draft the manuscript. Linhui He and Xuanjun Zhang collected aqueous humour/cornea tissue viral DNA PCR detection results. The corresponding author Jing Hong is responsible for ensuring that the descriptions are accurate and agreed upon by all authors. All authors read and approved the nal manuscript.

Ethics approval
The present study was performed in accordance with the Declaration of Helsinki and approved by the Medical Ethics Committee of Peking University Third Hospital (201729901).

Consent to participate
These patients were contacted by telephone to obtain verbal informed consent.
Disclosure statement.
The authors report no con icts of interest.