Cytomegalovirus-positive Posner-Schlossman syndrome: to compare differences in retinal vessel area density between the affected and non-affected eye using optical coherence tomography angiography

Cytomegalovirus (CMV) is a ubiquitous herpes virus that might affect all ocular structures with long-term persistence, leading to recurrent infections. A rare form of anterior uveitis (AU) that has been linked to a CMV infection is the Posner-Schlossman syndrome (PSS). It is also termed glaucomatocyclitic crisis and was first described by Posner and Schlossman in 1948 [1]. It mostly occurs unilaterally and is characterised by recurrent episodes of non-granulomatous AU with remarkably increased intraocular pressure (IOP) [1, 2]. Other characteristics of PSS include few keratic precipitates, the absence of iris synechia, and only mild anterior chamber reaction [3]. The aetiology of PSS is still unclear, but recent studies have shown a causal relation between PSS and CMV infection, although CMV-negative PSS has been reported [4‐6]. An aqueous humour analysis is recommended to exclude other viral aetiologies, such as rubella virus (RV), herpes simplex virus (HSV), or varicella zoster virus (VZV) and to initiate adequate antiviral therapy [3, 8]. Earlier studies demonstrated a significant reduction in the recurrence rate after systemic treatment with the nucleoside analogue valganciclovir [7, 9, 10].

In the past, PSS was thought to be a more benign uveitis entity, but newer findings have shown a loss of corneal endothelial cells (CEC) and a significant thinning of the retinal nerve fibre layer (RNFL) in affected eyes [11, 12]. Furthermore, frequent recurrences of severely elevated IOP may be a risk factor for developing secondary optic neuropathy and chronic secondary glaucoma [2, 13, 14]. Interestingly, studies also suggest that CMV can cause vascular dysfunction and systemic hypertensive complications [15, 16]. Although there is an increased interest in morphologic changes in CMV-positive PSS, it remains unclear how it affects the retinal vasculature.

Optical coherence tomography angiography (OCTA) is a new non-invasive imaging method that visualises retinal blood vessels using multiple OCT scans [18]. It distinguishes blood flow from static tissue using an algorithm [19, 20]. It is a common procedure and is comparable to structural OCT measurements of retinal nerve fibre and ganglion cell thickness for differentiating glaucomatous from healthy eyes [21]. The use of OCTA to detect morphological changes in the retinal microvasculature, such as peripheral vascular dysfunction and reduced peripapillary and macular vessel area density (VAD), has been demonstrated in patients with various forms of glaucoma [19, 20, 22]. To date, three studies have investigated microvascular changes assessed by OCTA in PSS patients without evidence of CMV [23, 24].

We aimed to investigate the retinal microvasculature in CMV-positive PSS patients. Therefore, we analysed alterations of the VAD in the different retinal vascular plexuses on OCTA in PSS affected eyes in the subacute or remission phase compared to the non-affected contralateral eyes.

In total, 50 eyes of 25 patients with a mean age of 47 ± 13 years were included. Six patients were excluded. Four patients did not meet the IC or exclusion criteria (EC) for different reasons: one because the aqueous humour sample was negative for CMV and positive for VZV (DC 1), one because of iris synechiae and no evidence of IOP elevation (DC 2 and 4), one because of refractive errors greater than 3.50dpt in the sphere (EC 2), and one because of amaurosis in the contralateral eye (EC 4). Two patients had to be excluded because OCTA-imaging could not be performed or was of insufficient quality (image quality < 20). Baseline characteristics are presented in Table 1. Nineteen patients (76%) were receiving SAT at the time of the examination or had received SAT in the past. Six patients (24%) did not receive SAT due to side effects or family planning. Disease duration in our cohort ranged from 2 weeks to 22 years.

The ophthalmological findings are presented in Table 2. We found differences in the highest IOP, the VF MD, the RNFL thickness in three quadrants, the CDR, the number of anti-glaucomatous drugs, and the corneal endothelial cell density between the PSS affected and the contralateral eye. There was no difference in the current IOP and macular retinal thickness between two eyes, except in the nasal quadrant.

Table 3 summarises the results of the perfusion parameters assessed by the OCTA. There were no significant differences in the macular and optic disc vascular parameters between both groups. Only the superficial vascular complex on the peripapillary regions showed a significant decrease in the VAD in the PSS eye (46.1 ± 9.3% versus 50.1 ± 6.3%, p = 0.008, adjusted p = 0.048). All the deeper layers (DVC, CCL, CL) did not differ between the two groups.

In Table 4, we analysed the correlation between the clinical findings (MD of the VF, RNFL) and the vessel perfusion parameters assessed by OCTA. We found a significant correlation between the MD and the following parameters: the VAD of the macular, the optic disc, the peripapillary SVC, the peripapillary DVC, and the FAZ. Correlation analyses of the RNFL showed similar results with a significant correlation of the RNFL with the VAD of the macula, the optic disc, the peripapillary SVC, and the FAZ. The deeper peripapillary layers (CCL, CL) showed no correlation.

Through analysing the association between the results of the affected eye and the disease duration, we detected a dependence of the mean RNFL on the disease duration (p = 0.037; duration < 1 year: 87.1 ± 20.0 μm, duration ≥ 1 year: 78.0 ± 24.0 μm). The subanalysis of patients with disease duration < 1 year is shown in Table 5. There were no significant differences in VAD of all layers between the affected and non-affected eye. Table 6 shows the subanalysis of patients with disease duration ≥ 1year. The superficial vascular complex in the peripapillary regions showed a significant decrease in VAD in the PSS eye (44.1 ± 10.3% versus 49.6 ± 7.4%, p = 0.006, adjusted p = 0.036). All other layers (DVC, CCL, CL, macula, and ONH) did not differ between affected and non-affected eyes.

Additionally, patients receiving SAT showed a significantly better VAD of peripapillary CCL (p = 0.001, no therapy: 31.4 ± 1.9%, SAT: 35.0 ± 1.6%) and peripapillary CL (p = 0.009, no therapy: 34.2 ± 0.3%, SAT: 36.3 ± 1.8%), while all other parameters did not differ (supplementary data).

In this prospective observational case series, the VAD in the peripapillary SVC is significantly reduced in CMV-positive PSS-affected eyes, when compared to the non-affected eyes in the subacute or remission phase. This is consistent with a decrease in RNFL thickness and VF MD in PSS-affected eyes. Additionally, the reduction of RNFL seems to be dependent on the duration of the disease. The deeper layers showed no significant differences between CMV-positive PSS-affected and non-affected eyes. However, there might be an impact of the antiviral therapy on the VAD of the peripapillary CCL and CL layer.

The VAD in the peripapillary SVC is significantly reduced in CMV-positive PSS-affected eyes compared to non-affected eyes. Chen et al. reported comparable results in PSS-patients [23], while the study by Liu et al. in newly diagnosed PSS-patients without proof of CMV, where OCTA images were taken during the first acute attack and 1 week after treatment, presented opposite results [24]. However, this study is not comparable to ours as we investigated patients in the subacute or remission phase. Similar observations to our results have been reported for secondary IOP elevation in patients with unilateral FUS [42]. Our results are consistent with earlier research in patients with different forms of glaucoma, such as primary open angle glaucoma (POAG), acute primary angle closure (APAC), pseudoexfoliation glaucoma (PXG), and normal-tension glaucoma (NTG) [43‐47]. Liu et al. demonstrated that the peripapillary SVC is particularly affected in patients with perimetric glaucoma, and the authors hypothesised that this change is a sign of glaucomatous optic disc damage [49]. Additionally, Wang et. al. found reduced peripapillary VAD in patients with APAC even without measurable structural differences (RNFL thickness and ganglion cell complex thickness) compared to the non-affected eye [50]. Another study in patients with APAC indicated that a longer duration of an acute attack was correlated with reduced peripapillary VAD [51], and Moghimi et al. described a progressive decrease in circumpapillary VAD over the first 6 weeks after the initial insult [52].

The current research demonstrated no significant differences between the VAD of the optic disc and the deeper peripapillary layers. This is consistent with another study that found no differences in the VAD of the DVC in glaucomatous eyes compared to healthy controls [49].

No significant differences were found in macular thickness and the VAD of the macular SVC between the two groups. This is in line with the study of Liu et al., which also reported no changes in the VAD of the macular SVC [24]. However, in PSS eyes without confirmed CMV, two studies found a reduced superficial macular VAD compared to the fellow eye and a healthy control group [23, 25].

In contrast to our study, which captured OCTA images during the remission phase, in the study of Chen et al., OCTA images were taken closer to the acute phase (within one to seven days afterwards) [23]. This may explain the lower macular superficial VAD in PSS eyes, which probably recovers during remission. Trabeculectomy showed similar reversal effects on the VAD in POAG patients, with postoperative increase in VAD correlating to higher preoperative IOP and greater IOP reduction [53]. In contrast, Liu et al. found no changes in the VAD of the macular and peripapillary SVC despite high IOP reduction [24]. In this study, selected PSS patients displayed no changes in VF and RNFL thickness. These patients without glaucomatous damage might have strong retinal autoregulation to maintain the blood flow in the peripapillary and macula region under elevated IOP, while patients with mild and advanced glaucomatous damage might not.

Another possible explanation of the missing differences in the macular parameters in our study could be the beneficial effect of therapy with valganciclovir on the overall retinal and choroidal microvasculature. We found a better VAD of the CCL and CL in patients receiving antiviral therapy compared to those refusing antiviral treatment. Contrary to our study, in which we treated more than three-quarters of the patients with valganciclovir, none of the patients in the studies on PSS and microvascular changes on OCTA were treated with antiviral medication [23, 24]. Additionally, lowering IOP and improving retinal blood flow by an anti-glaucomatous therapy of the affected eye may also be a reason for a lack of differences in this layer [54, 55].

Additionally, our results might be explained by the chronological progression of glaucoma, as seen in other glaucoma types (POAG, PXG, APAC), where disease severity correlates with macular VAD reduction [45, 56, 57]. The SVC seems to be more affected than the DVC, like the peripapillary region [59]. Initial and mild glaucomatous damage is indicated by peripapillary VAD and RNFL changes, as supported by our data [60, 61]. Advanced stages may result in VF defects and macular VAD reduction. In our cohort, the glaucomatous damage was relatively mild, with minor VF alterations and small CDR differences. We cannot rule out the possibility that macular damage might be detectable in more advanced stages. Our thesis is supported by significant correlations between peripapillary, papillary, and macular VAD and reductions in RNFL thickness and VF MD [62]. Furthermore, the RNFL appears to be dependent on the duration of disease. Research on glaucoma patients with POAG showed similar results, particularly towards the macular VAD, correlating significantly with a reduction in RNFL thickness and an increase in VF MD [43, 58, 63, 64]. The results of Liu et al. confirm this assumption. In affected, newly diagnosed eyes, no glaucomatous damage, and no changes in VAD were present. Our study included patients with different disease duration. Some of them had no glaucomatous changes. Other patients, some with partially longer disease courses, were more likely to have mild or advanced glaucomatous damage, such as reduced RNFL thickness and consecutive first changes in VAD, as shown in our subgroup analysis.

In line with other studies on PSS, we found a higher proportion of males, an increased maximum IOP, an average age of 47 ± 13 years, differences in MD, CDR, and the amount of anti-glaucomatous agents [66]. As reported previously, a loss of CEC and RNFL thickness was also detectable in our patients [11].

It is unclear whether the glaucomatous damage in PSS is due to elevated IOP or due to CMV infection and vascular dysfunction. Experimental trials have shown that CMV is associated with immune activation in T-cells. This can cause atherosclerotic progression and chronic inflammation, leading to endothelial damage and vascular dysfunction [16, 17]. Infection of vascular endothelial cells and macrophages appears to be especially important for CMV latency and possible reactivation in the host. Furthermore, their finding of persistent ocular inflammation following CMV infection in a mouse model of an immunocompetent host underscores the need to consider CMV as a pathogen capable of inducing long-term inflammatory sequelae in the eye, including the neural retina [67]. In another experimental murine model, CMV infection in the eye is first detected in endothelial cells of the iris and causes inflammation and a latent infection of the retinal pigment epithelium [68]. These results may support our findings and hypothesis of vascular impairment in ocular tissue due to CMV infection. Our study allows only speculation about vascular dysfunction, as we did not compare our results to healthy controls, and the previously mentioned studies were not conducted in humans. However, we could detect differences in microcirculation between affected and non-affected eyes that might play a role in the development of glaucomatous damage.

In the Asian population, approximately 50% of PSS patients were identified to be CMV-positive [6]. To date, there are no comparable studies in a European cohort, but we suspect that the incidence of CMV-infection is lower in Germany. To analyse the data in a more homogeneous group, we based our study on CMV-positive PSS only. There are several reasons for this, as the clinical characteristics of CMV-positive and CMV-negative PSS appear to be similar. Nonetheless, CMV-positive PSS was associated with greater CEC loss and a larger CDR [69]. Furthermore, the number of eyes treated surgically was higher in the virus-proven group [70].

Currently, there is no standardised guideline for the optimal treatment of CMV-associated PSS, but patients might benefit from antiviral therapy [9, 10]. Previous studies reported a good response to valganciclovir, a lower recurrence rate, a reduction in glaucoma severity, and reduced need for glaucoma surgery [7, 10, 71]. Unfortunately, additional glaucoma surgery was required in some cases, especially when the disease had been present for more than 5 years [7, 70, 71]. Our results support these findings, as trabeculectomy was performed in two of our patients with more than 5 years of disease.

We are aware of certain limitations of this study. Firstly, the sample size was relatively small due to the rare occurrence and the often-asymptomatic clinical characteristics. Secondly, OCTA imaging was performed during the subacute or remission phase. This was related to the often-limited image quality due to acute IOP elevation and corneal oedema, which did not meet our quality requirements. Additionally, the RNFL examination may be misleading in the first days or weeks after the attack due to oedema [72]. Therefore, we recommend OCT and OCTA examinations during the remission phase. In addition, some of our patients received topical anti-glaucomatous therapy, which may affect retinal blood flow. In particular carbonic anhydrase inhibitors such as dorzolamide or brinzolamide were found to improve retinal blood flow [54, 55]. Accordingly, anti-glaucomatous medication should not lead to a reduction of superficial peripapillary VAD in the affected eye but are possible the reason for the lack of differences in the other layers and regions as discussed before.

Furthermore, we only included patients with confirmed CMV-positive PSS, and the antibody analysis took about 1 to 2 weeks. Additionally, the results covered different disease durations, which limits generalisation. However, longer disease duration correlated with a reduced VAD in the peripapillary SVC. Data collection relied on patients’ recall of their first attack, which may not be accurate as it remains unclear whether patients had previous undetected attacks. Thirdly, we only examined the retinal vasculature at a single time point. Longitudinal follow-up with OCTA would provide more reliable information on disease progression. The investigation of individuals with different disease progression partially addressed this limitation. However, larger longitudinal studies with a control group are needed to investigate long-term changes in the retinal microvasculature. Fourth, our study focused only on CMV-positive PSS. This limits the generalizability to all PSS patients.

This study is the first to investigate the peripapillary microvasculature in CMV-positive PSS in Europe. Challenges for future management include high IOP, glaucomatous damage, recurrence rate, and optimal treatment. More than three-quarters of our patients received SAT, which may benefit the retinal microvasculature. A more standardised VAD analysis with defined assessment of retinal layers is needed in the future. Nonetheless, lower peripapillary VAD in the SVC may indicate vascular dysfunction and impaired blood flow to the ONH. Therefore, we suggest OCTA as an additional tool for the follow-up of patients with PSS to detect changes in VAD that might lead to glaucomatous damage.