Identifying high-risk eyes for glaucoma surgery in cytomegalovirus anterior uveitis

Abstract

Cytomegalovirus (CMV) has recently emerged as a significant cause of anterior uveitis (AU) with intraocular pressure (IOP) elevation and corneal endothelial damage. The purpose of this study is to investigate the long-term prognostic factors for glaucoma surgery in patients with CMV-AU. In this retrospective cohort study, immunocompetent patients with CMV-AU were included from one academic institution and a uveitis-specialized private clinic between May 2015 and Mar 2024. Upon confirmation of polymerase-chain-reaction results, patients received anti-viral therapy (oral valganciclovir or 2% topical ganciclovir). When intraocular pressure was not controlled, glaucoma surgery (glaucoma filtering surgery or glaucoma drainage device) were performed. Indicators of glaucoma surgery were analyzed using Kaplan-Meir survival analyses and Cox-proportional hazard model. A total of 149 consecutive patients with CMV-AU included with a mean age of 44.5 ± 16.5 years; 65.1% being male. The baseline corneal endothelial cell (EC) density in affected eyes was 2088 ± 695 cells/mm². During follow-up, glaucoma surgery was performed in 24 patients (16.1%), of whom 75% underwent glaucoma filtering surgery and 25% underwent drainage device implantation. The median survival time from the first onset of the disease to glaucoma surgery was 73.0 ± 11.6 months. In the univariate Cox-regression analyses for receiving glaucoma surgery, baseline EC < 1000 cells/mm² (Exp(B) = 4.142, p = 0.013), baseline BCVA < 0.3 (Exp(B) = 2.733, p = 0.029), and number of anti-glaucoma medications at quiescent phase ≥ 2 (Exp(B) = 2.645, p = 0.024) were significantly associated with the future requirement for glaucoma surgery. In the multivariate analyses, baseline EC < 1000 cells/mm² was the only significant factor for predicting glaucoma surgery (Exp(B) = 3.960, p = 0.022). In conclusion, baseline corneal endothelial cell density can serve as a useful indicator for future requirement of glaucoma surgery in CMV-AU.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-026-52634-0.

Introduction

Anterior uveitis (AU) is the most common form of uveitis, and has both infectious and non-infectious causes¹. With the increasing use of polymerase chain reaction (PCR) in clinical practice, viral causes of AU have been increasingly recognized, particularly in immunocompetent individuals^2–4. Among these, cytomegalovirus (CMV) has emerged as a significant cause, especially in cases involving elevated intraocular pressure (IOP) and corneal endothelial damage^3,5.

In the long course of CMV-AU, two major complications— glaucoma and corneal endothelial damage, which may necessitate glaucoma surgery and keratoplasty—are significant and often irreversible^6,7. While corneal endothelial cell damage may not produce noticeable symptoms until corneal decompensation occurs, glaucoma can lead to progressive vision loss due to visual field (VF) constriction. This condition often requires long-term use of anti-glaucoma medications, and in many cases, finally glaucoma surgery when elevated IOP is not controlled with medications.

Posner-Schlossman syndrome (PSS), a key clinical manifestation of CMV-AU, was traditionally thought to be benign, characterized by transient elevations in IOP and anterior chamber inflammation that resolved without significant sequelae⁸. However, a long-term study has shown that patients with more than 10 years of PSS have a 2.8 times higher risk of developing glaucoma, with 17% eventually requiring glaucoma filtering surgery⁹. CMV infection has been recognized as a significant factor in PSS, with studies reporting CMV-positivity in 40% to 71% of PSS cases^10,11.

Several studies have compared CMV-positive and CMV-negative cases of AU, highlighting the role of CMV infection in the progression of the disease^6,8,12. Despite the potentially vision-threatening outcomes associated with long-term CMV-AU, there remains a lack of cohort studies investigating the disease’s progression and identifying prognostic factors for glaucoma surgery in PCR-proven CMV cases.

In the present study, we investigated clinical prognostic factors that predict need for future glaucoma surgery in long-term follow-up of patients with CMV-AU.

Results

Baseline patient characteristics

This study included 149 patients with anterior uveitis who had positive results on the PCR examination for CMV. Average follow-up period were 28.2 ± 22.7 months. Table 1 presents the demographic and clinical characteristics of included patients. The average age was 44.5 ± 16.5 years, and 65.1% of patients were male. The average BCVA was 0.74 ± 0.36, and 12.1% of patients had a value below 0.3. At the time of PCR examination, the average IOP was 35.21 ± 12.9 mmHg with 2.12 ± 0.81 anti-glaucoma medications. 135 patients received oral valganciclovir as the initial anti-viral treatment, while the remaining 14 patients received topical 2% ganciclovir.

Table 1.

Baseline demographic data and clinical findings in CMV anterior uveitis at initial visit (N = 149).

Characteristics		Mean ± SD
Demographics	Age, year	44.5 ± 16.5
	Gender
	Female	52 (34,9%)
	Male	97 (65.1%)
	Affected eye, n (%)
	Right	73 (48.9%)
	Left	76 (51.0%)
Visual acuity	BCVA, Snellen	0.74 ± 0.36
	BCVA < 0.3 (%)	18 (12.1%)
Glaucoma	Baseline MD, dB	− 5.6 ± 6.8
	Baseline PSD, dB	− 4.01 ± 3.62
	Baseline peripapillary RNFL thickness, μm	79.1 ± 17.5
	Baseline average GCIPL thickness, μm	71.2 ± 12.2
	Baseline average GCIPL thickness, μm	71.2 ± 12.2
Corneal endothelial cell	Baseline corneal EC, cells/mm2	2088 ± 695
	Baseline corneal EC < 1000 cells/mm2, %	12 (8.2%)
	Relative cell count compared with the other eye, %	79.7 (25.5)

CMV, cytomegalovirus; BCVA, best-corrected visual acuity; PCR, polymerase chain reaction; EC, corneal endothelial cell density.

Glaucoma surgery

Among the 149 patients with CMV-AU, glaucoma surgery was performed in 24 patients: 18 (75%) had glaucoma filtering surgery and 6 (25%) had glaucoma drainage device implantation as an initial surgery. Subjects who underwent glaucoma surgery had lower corneal EC than those who did not (1801 ± 630 cells/mm² vs. 2132 ± 697 cells/mm²; p = 0.039) and were older (50.0 ± 11.2 years vs. 43.4 ± 17.2 years; p = 0.020, Table 2). Furthermore, these patients were more likely to use steroid eye drops during quiescent periods (p = 0.022) and required a higher number of anti-glaucoma medications during the quiescent period (p = 0.045), with longer duration of elevated IOP (p = 0 0.019). Concerning glaucomatous change, patients with glaucoma surgery exhibited lower MD (− 10.26 ± 8.99 dB vs. − 4.48 ± 5.71; p = 0.014), higher pattern standard deviation (6.53 ± 4.49 dB vs. 3.39 ± 3.11 dB; p = 0.009), lower average RNFL thickness (67.0 ± 20.9 um vs. 81.6 ± 15.6 um; p < 0.001), and lower average GCIPL thickness (66.1 ± 9.4 um vs. 73.0 ± 12.4 um; p = 0.034). Figure 1 illustrates the survival curve for glaucoma surgery. The survival time was defined as the time interval from the disease onset to glaucoma surgery. The median survival time was 73.0 ± 11.6 months (Fig. 1A). Subjects with baseline corneal EC < 1000 cells/mm² had a significantly higher risk of receiving future glaucoma surgery than those with EC ≥ 1000 cells/mm² (p = 0.003; log-rank test, Fig. 1B).

Table 2.

Comparisons of clinical characteristics between CMV-AU patients who required glaucoma surgery and patients who did not (n = 149).

	Without glaucoma surgery (n = 125)	With glaucoma surgery (n = 24)	P value
Initial baseline characteristics
Age, year	43.4 ± 17.2	50.0 ± 11.2	0.020
Sex, male, n (%)	79 (63.2%)	18 (75.0%)	0.352
Baseline peak IOP, mmHg	34.90 ± 12.59	36.95 ± 16.95	0.516
Baseline BCVA	0.79 ± 0.26	0.60 ± 0.35	0.020
The initial number of anti-glaucoma medications, n	2.10 ± 0.77	2.23 ± 1.02	0.571
Baseline MD, dB	− 4.48 ± 5.71	− 10.26 ± 8.99	0.014
Baseline average RNFL thickness, μm	81.6 ± 15.6	67.0 ± 20.9	< 0.001
Baseline average GCIPL thickness, μm	73.0 ± 12.4	66.1 ± 9.4	0.034
Baseline corneal EC, cells/mm2	2132.5 ± 697.2	1801.6 ± 629.5	0.039
Follow-up characteristics
Number of anti-glaucoma medications at quiescent period, n	1.06 ± 0.67	1.59 ± 1.14	0.045
Trough IOP, mmHg	12.5 ± 3.2	14.6 ± 6.0	0.123
Steroid use at quiescent period, %	36 (35.2)	13 (61.9)	0.022
Duration of follow-up	28.3 ± 22.5	27.9 ± 24.3	0.942

CMV, cytomegalovirus; AU, anterior uveitis; IOP, intraocular pressure; BCVA, best-corrected visual acuity; MD, mean deviation; RNFL, retinal nerve fiber layer; GCIPL, ganglion cell-inner plexiform layer; PCR, polymerase chain reaction.

Fig. 1.

Survival curve for the probability of receiving glaucoma surgery in CMV anterior uveitis. The median survival time from the first onset of the disease to glaucoma surgery was 73.0 ± 11.6 months (A). Subjects with baseline corneal EC < 1000 cells/mm² had a significantly higher risk of receiving future glaucoma surgery than those with EC ≥ 1000 cells/mm² (p = 0.003; log-rank test) (B).

Prognostic factors for glaucoma surgery

Next, we analyzed the baseline clinical parameters associated with the future requirement for glaucoma surgery. In the univariate Cox-regression analyses for receiving glaucoma surgery, baseline EC < 1000 cells/mm² (Exp (B) = 4.142, p = 0.013), baseline BCVA < 0.3 (Exp (B) = 2.733, p = 0.029), and number of anti-glaucoma medications at quiescent phase ≥ 2 (Exp (B) = 2.645, p = 0.024) were significantly associated with the future requirement for glaucoma surgery. In the multiple analyses, baseline EC < 1000 cells/mm² was the only significant factor for predicting glaucoma surgery (Exp (B) = 3.960, p = 0.022) (Table 3).

Table 3.

Cox-regression analyses on the indicative factors for receiving glaucoma surgery in CMV Anterior uveitis.

	Simple analysis			Multiple analysis
	HR	CI	P value*	HR	CI	P value*
Age	1.008	0.981, 1.036	0.560
Sex	1.700	0.671, 4.309	0.263
Baseline EC < 1000 cells/mm2 (reference: ≥ 1000 cells/mm2)	4.142	1.357, 12.642	0.013	3.960	1.219, 12.872	0.022
Baseline BCVA < 0.3 (reference: ≥ 0.3)	2.733	1.109, 6.737	0.029	2.494	0.892, 6.975	0.081
Baseline IOP, mmHg	1.009	0.974, 1.046	0.606
First treatment of oral valganciclovir (reference: ganciclovir eyedrops)	1.653	0.531, 5.143	0.386
Steroid use at quiescent period (reference: no)	2.322	0.949, 5.682	0.065	2.286	0.857, 6.097	0.098
Number of anti-glaucoma medications at quiescent period > 1 (reference: ≤ 1)	2.645	1.135, 6.166	0.024	2.298	0.858, 6.155	0.098

CMV, cytomegalovirus; HR, hazard ratio; CI, confidence interval; EC, endothelial cell count; IOP, intraocular pressure; BCVA, best-corrected visual acuity.

Discussion

It is well known that CMV-AU is characterized by frequent recurrences, which results in long-term complications of the disease, such as advanced glaucoma or corneal decompensation, often lead to vision-threatening outcomes requiring glaucoma surgery or corneal keratoplasty. There are limited number of studies which addressed these outcomes: In a French series of CMV-AU, glaucoma with 4.13 years of follow-up, glaucoma surgery was necessary in 25.7%⁶. In a Taiwan series of CMV-AU with 40 months of follow-up, glaucoma surgery was performed in 9 (13.2%) of the CMV-AU, with more likelihood of receiving surgery in CMV-AU with a disease duration over 6 years¹².

For the present work, we specifically focused on defining indicative factors for the future need of glaucoma surgery in CMV-AU. It is well known that IOP is severely elevated in CMV-AU, and among cases of steroid-recalcitrant anterior uveitis, 75% were found to be CMV-AU¹⁸. A recent study by Terada et al.¹⁹ comparing the features of 3 types of viral anterior uveitis—herpes simplex virus, varicella-zoster virus, and CMV—found that CMV-AU was characterized by higher need for glaucoma surgery and a greater proportion of cases with elevated IOP. In the study of Lenglinger et al.²⁰ CMV-AU accompanied significantly reduced RNFL thickness and corneal EC compared with the unaffected eye, indicating glaucomatous damage.

The reason for high association of CMV anterior uveitis and glaucoma could be related to the fibrotic nature of CMV infection. The CMV virus can promote the epithelial to mesenchymal transition in tumor cells²¹, and has been associated with kidney graft failure though viral induction of TGF-β1, leading to fibrosis of graft²². In our previous study using a CMV-infected human trabecular meshwork cell model, CMV caused a progressive increase in TGF-β, a change that was reversed by application of steroid²³. Since TGF-β is a cardinal cytokine involved in the increased outflow facility and fibrotic changes seen in the outflow tract in primary open angle glaucoma, the higher prevalence of glaucoma in the opposite eye may suggest a possible pathophysiological link between CMV-AU and primary open angle glaucoma²⁴.

In CMV-AU, the corneal endothelium is affected, typically showing KPs or localized edema in the corresponding area⁷. Corneal EC count is reportedly significantly reduced in the affected eye compared to the unaffected eye,^20,25 with a strong correlation between EC loss and CMV viral load²⁶. CMV endotheliitis with ECD 1500 cells/mm² or less reportedly required a significantly longer course of anti-CMV treatment compared to those with an ECD above 1500 cells/mm²²⁷.

The most significant finding of our study is that baseline corneal EC is a prognostic indicator for future glaucoma surgery in patients with CMV-AU. In a previous CMV-AU cohort from Taiwan, Su et al.¹² reported that presence of CMV for longer than 5 years was significantly associated with higher likelihood of requiring glaucoma surgery. However, prognostic factors for glaucoma surgery within the CMV-AU context have not been reported yet. CMV is known to replicate in the human trabecular meshwork and corneal endothelial cells, both of which share a mesenchymal origin^23,28. Given the diverse clinical presentations of CMV-AU, including acute or chronic hypertensive anterior uveitis and corneal endotheliitis,²⁹ CMV infection in immunocompetent individuals has been understood as a spectrum of diseases. The finding that corneal endothelium indicates the future glaucoma surgery suggests that despite the diverse clinical spectrum of CMV-positive anterior chamber infection, CMV does not exclusively target the corneal endothelium or trabecular meshwork. Rather, the damage appears to be a result of both direct viral effect and responsive immunologic reaction, which parallel each other in contributing to the pathology.

In our cohort, glaucoma surgeries consisted of 18 filtering procedures and 6 drainage device implantations (Ahmed valve). Corneal endothelial cell density (ECD) differed significantly among eyes without glaucoma surgery, eyes that underwent filtering surgery, and those that received a drainage device (2132.5 ± 62.7, 1889.5 ± 668.0, and 1502.8 ± 388.8 cells/mm², respectively; p = 0.044). In subgroup analyses, baseline ECD < 1000 cells/mm² was the only significant factor associated with filtering surgery (HR, 4.687, 95% CI 1.422–15.446 p = 0.011, Supplementary Table 1), while statistical analyses could not be performed in the drainage device group because of the small sample size (n = 6). However, this group showed the lowest mean ECD, suggesting that more severe endothelial compromise may be associated with the need for drainage device implantation. Glaucoma surgery is one of the major complications of CMV-AU in the long course of the disease¹⁶. We found that patients requiring future glaucoma surgery were more likely exhibit baseline corneal endothelial cell count less than 1000 cells/mm², longer duration of high IOP, use of steroid eyedrop during quiescent period, and higher number of anti-glaucoma medications during quiescent periods. However, neither the number of anti-glaucoma medications at attack nor peak IOP were associated with either the future requirement for glaucoma surgery or glaucoma damage as represented by MD. While peak IOP or high number of anti-glaucoma medications at attack are important metrics for immediate disease control, it appears that chronic and pro-fibrotic change of outflow facility—reflected by the number of anti-glaucoma medications at the quiescent periods—may be more critical in determining insidious glaucomatous damage. In this regard, clinicians need to focus on IOP control in CMV-AU patients, particularly during the quiescent phase, to reduce the risk of long-term glaucomatous progression.

Our cohort study has several strengths, including a large number of CMV-PCR-proven cases. Nevertheless, several limitations should be noted. First, the retrospective design of the study may have introduced selection bias. Additionally, as the study was conducted in a uveitis-specialized or tertiary care setting, there is a possibility that the cohort includes a higher proportion of patients with more severe or uncontrolled disease. Second, our assessment of long-term outcomes was primarily limited to the need for glaucoma surgery, which may not capture the full spectrum of clinical progression. Finally, due to the relatively small number of patients who underwent glaucoma surgery, we were unable to perform detailed analyses stratified by the type of surgical intervention. It could be important to consider the differences between filtering surgery and drainage device surgery with regard to the clinical situation determining the surgery type or the effect of each surgery on corneal endothelium. However, subgroup analysis for drainage device was severely limited due to the small sample size (n = 6), and therefore, the results should be interpreted in light of this limitation. Also, for the survival curve, the number of patients at risk decreased substantially in the later follow-up period; therefore, survival estimates beyond this time point should be interpreted with caution. The small number of surgical events may also limit the robustness of our multivariate Cox-regression analysis. In terms of anti-viral treatment, the type of treatment was not associated with glaucoma surgery, as shown in Table 3. The lack of association between the type of anti-viral agent and glaucoma surgery may be partly explained by treatment pattern bias, in which most patients were treated with oral valganciclovir and only a minority received topical agents as the first anti-viral treatment.

In summary, within our cohort of CMV-AU patients, glaucoma surgery was required in 19.2% of total patients. A lower baseline corneal endothelial cell count, lower best-corrected visual acuity, and higher number of anti-glaucoma medications at the quiescent period were all associated with future requirement for glaucoma surgery. Notably, a baseline corneal endothelial cell density less than 1000 cells/mm² was the single most indicative factor for the need for glaucoma surgery in CMV-AU patients, suggesting its potential as an indicator of reduced trabecular meshwork cell function.

Method

Study protocol

This is a retrospective case series of consecutive CMV-AU patients at a tertiary reference center (St. Vincent’s Hospital, the Catholic University of Korea, Suwon, Korea) and the Uveitis Clinic of the Apgujung St. Mary’s Eye Center (Seoul, Korea) between May 2015 and Mar 2024. Only cases that tested positive for CMV on aqueous PCR were included based on the criteria suggested by the Standardization of Uveitis Nomenclature (SUN) Working Group¹³. Due to the retrospective nature of the study, the research Ethics Committee of the Catholic University of Korea waived the need of obtaining informed consent and approved the study protocol (IRB approval number: VC24RIDI0260). This study protocol followed the tenets of the Declaration of Helsinki. For eligibility purposes, the diagnostic criteria for CMV-AU were as follows: (1) proof of CMV infection in aqueous sample PCR; (2) unilateral, recurrent episodes of mild anterior chamber inflammation; (3) presence of endothelial keratic precipitates; (4) at least one transient episode of unilateral elevated IOP during attack in the presence of anterior chamber inflammation; and (5) absence of posterior synechiae or posterior inflammation^6,12,14. Aqueous tapping was performed in the presence of active inflammation during attack upon presentation. The quiescent period was defined as having returned to baseline IOP in the absence of anterior chamber inflammation.

At the first visit, subjects underwent a complete baseline ophthalmic examination, including measurement of best-corrected visual acuity (BCVA), Goldmann applanation tonometry, slit-amp examination, gonioscopy. A detailed examination on including standard automated perimetry using the 24–2 SITA program (Humphrey Visual Field Analyzer; Carl Zeiss Meditec, Inc., Dublin, CA) and spectral domain optical coherence tomography (OCT) (Cirrus High Definition-OCT; Carl Zeiss Meditec, Dublin, CA), was performed within one month following the resolution of inflammation. Demographic data and clinical information on baseline and each visit were recorded for age, sex, laterality, first onset of disease, best-corrected visual acuity, IOP, duration of high IOP ≥ 21 mmHg, use of steroid eyedrops and number of anti-glaucoma medications prescribed. Trough IOP is the lowest IOP recorded during follow-up period.

Baseline clinical assessment

Corneal endothelial cell density was measured using a non-contact specular microscope (Konan, ASKIN & CO GmbH, Wien, Austria). For structural assessment of the optic nerve head, OCT imaging was obtained after pupillary dilatation to a minimum diameter of 5 mm; only accurate images with a signal strength ≥ 6 (10 = maximum) were included. Detailed descriptions of the Cirrus HD-OCT, optic nerve head, and macular ganglion cell-inner plexiform layer (GCIPL) algorithm have been reported elsewhere¹⁵. As measurements of the retinal nerve fiber layer (RNFL), the average RNFL thickness and average GCIPL thickness were recorded for all patients. For assessment of optic nerve head deformation, vertical cup-to-disc ratio and cup volume were recorded. Glaucoma was defined as having a glaucomatous disc appearance associated with a typical reproducible VF defect. A glaucomatous VF defect was defined as having glaucoma hemifield test results outside normal limits and the presence of at least three contiguous points in the pattern deviation plot with p-values < 5%, with at least one point associated with a p-value < 1% on two consecutive reliable VF examinations.

Treatment protocol

After confirmation of CMV positivity in PCR examination, all CMV-AU patients received anti-viral treatment. The first choice of treatment was oral valganciclovir for a total of 6 weeks, starting with 450 mg twice a day^6,16. In the event a patient had contraindications to systemic treatment, or refused to take the medication, the alternative treatment was topical 2% ganciclovir, prepared by dissolving 500 mg of Cymevene lyophilized IV powder (Roche, Basel, Switzerland) in 25 mL of balanced salt solution, every 2–3 h as an induction therapy, and every 4 h as a long-term maintenance therapy^12,17.

Follow-up evaluation and glaucoma surgery

During follow-up, data on inflammation recurrence, intraocular pressure (IOP), and the number of anti-glaucoma medications were recorded both during episodes of active inflammation and during quiescent periods. Use of steroid eyedrops in the quiescent periods were also recorded. Trough IOP was defined as the lowest recorded IOP measurement during the follow-up period. When IOP was not controlled with anti-viral treatment and anti-glaucoma medication, glaucoma surgeries were performed by glaucoma specialists (PMH and CJA). In each eye, the choice of glaucoma filtering surgery (trabeculectomy or Ex-PRESS shunt (Alcon Laboratories, Fort Worth, TX) with adjuvant anti-fibrotic therapy or a glaucoma drainage device implant (Ahmed valve FP7, New World Medical, Rancho Cucamonga, CA) was based on the specialists’ clinical judgement. Follow-up duration was checked until the last visit for patients without glaucoma surgery and up to the time of surgery for those who underwent surgical intervention.

Statistical analyses

Statistical analyses were performed using IBM SPSS (Version 20, Chicago, IL). Numerical values were reported as mean and standard deviation, and categorical values as absolute number and percentage. Significant differences in numerical values were evaluated with Student’s t-test, and categorical values with the Chi-square test. Kaplan–Meier survival analysis was used to estimate the median time for glaucoma surgery after first visit. In the Kaplan–Meier analysis, censored observations were treated as surviving up to the time of censoring. Cox-regression analyses were performed to identify prognostic factors associated with the future requirement for glaucoma surgery. In multivariable Cox-regression analysis, variables significant at p < 0.150 in the simple analysis were included after adjusting for other potential clinical factors. P values less than 0.05 were considered to be significant.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

Soo Ji Jeon and Myoung Hee Park wrote the main manuscript text. Soo Ji Jeon, Myoung Hee Park, Ji Young Lee, Sung Eun Kim, Yong Sun Ahn, and Su Jin Lim performed the data review and analysis. The critical revision of the manuscript was done by Jin A Choi and Min Ho Kim. All authors reviewed the manuscript.

Funding

This study was supported by the 2025 Cheil-Nammyung Foundation Research Fund and by the St. Vincent’s Hospital Research Institute of Medical Science Foundation (SVHR-2025-06).

Data availability

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Declarations

Competing interests

The authors declare no competing interests.