For the first time, we conducted a case-control study to investigate the associations of slit-lamp and confocal features and the CMV status in Chinese eyes with chronic or recurrent AU. Our data showed that classical features such as owl’s eye cells could be observed in both eyes with positive and negative CMV PCR. This supports a possible false negative PCR result and exemplifies the limitation of aqueous PCR. In contrast, eyes with absence of pseudoguttata and lower corneal ECD, and more severe iris depigmentation were independently associated with a higher predictive risk of CMV, either alone or in combination. In particular, the combination of absence of pseudoguttata and severe iris depigmentation raised the PPV to 100% and NPV to 62.5%. The combination of absent pseudoguttata and low corneal ECD, both identified on IVCM, also achieved a comparable PPV of 100% and NPV of 60%. Although we had shown that in carefully selected eyes, iris depigmentation could be a potential biomarker for CMV AU, it was imperative to consider alternative causes for pigment loss in iris, such as prior intraocular surgery, iatrogenic iris injury, acutely elevated intraocular pressure (for example, acute angle closure), pseudoexfoliation, or underlying genetic predisposition. The merit of utilizing IVCM in predicting CMV was the independence from a time-sensitive slit-lamp examination, and the subjectivity in grading iris depigmentation.

In our cohort, identification of pseudoguttata was more frequent in eyes without CMV. Pseudoguttata was first described by Krachmer and coworkers in 1981, referring to hyporeflective elevated spots seen among regularly arranged endothelial cells in eyes with corneal inflammation, that disappear when inflammation resolves [14]. Scanning and transmission electron microscopy of these lesions suggested transient endothelial cell edema to be the culprit, which could be secondary to infection. Histologically, unlike true guttata, pseudoguttata lacks excrescences around the endothelial cells. Owing to its transient nature, it is a scarcely reported clinical feature, and scarce reports on histology. The largest series of pseudoguttata reported its occurrence in 44 eyes suffering from contact lens-related keratitis, keratoconjunctivitis, corneal epithelial defect, corneal foreign body and keratitis [15]. In comparison, true guttata appears as hyporeflective elevated spots with a hyperreflective white dot in the center, and forms due to focal thickening of Descemet’s membrane [14, 15]. These lesions do not resolve with resolution of inflammation.

Other than the pseudoguttata, there was no other significant difference in confocal signs of endotheliitis between the two groups. Interestingly, while none of our patients had detectable anterior chamber cells at the time of assessment, pseudoguttata was identified in 13.3% of CMV positive patients and 53.3% of CMV negative patients. This was in contrast to a previous report which described pseudoguttata to resolve when the inciting episode of anterior segment inflammation subsided [15]. This could be due to a difference in the disease duration, or the stage of infection between their study and ours. It is unclear how long pseudoguttata would remain present after clinically detectable anterior chamber inflammation resolves. Another possible explanation was that IVCM performed more superiorly than slit lamp examination alone in detecting subclinical inflammation in the corneal endothelium, hence the results reflecting a higher proportion of subclinical inflammation present in our CMV negative patients.

Our current cohort replicated results from our earlier study, showing that severe or diffuse iris depigmentation, even when assessed at a much later timepoint after AC tapping, could still effectively differentiate CMV AU from eyes with pan-negative PCR [6]. Nonetheless, there was always an element of subjectiveness and operator-dependence in identifying and quantifying iris depigmentation, despite a standard description. Hence, we hope that IVCM could provide a more objective assessment, that was less sensitive to time. The combination of at least one confocal feature with either iris depigmentation or absence of pseudoguttata was shown to improve the predictive accuracy for CMV.

The majority of eyes in our cohort experienced hypertensive episodes during flares. However, there was a significantly larger proportion of eyes in the CMV positive group suffering from glaucoma (80%, vs. 40% in CMV negative group). This was in line with the findings by Shirahama and coworkers, who identified significantly higher prevalence and faster progression of secondary glaucoma in patients with CMV AU, when compared to HSV AU and VZV AU [16]. This reiterates the importance of earlier identification of CMV in eyes with AU, for more aggressive monitoring and management of glaucoma.

Our analysis of IVCM and clinical features of CMV AU carries important clinical implications. We showed that a combination of any two of the following factors, absence of pseudoguttata on IVCM, low corneal ECD and severe iris depigmentation effectively increased the prediction of CMV as the cause of AU. Apart from documenting the severity of iris depigmentation during clinical exam, which was subjective and multifactorial in nature, clinicians may utilize IVCM to evaluate for pseudoguttata and corneal ECD in the central cornea for patients with recurrent or chronic AU before AC tapping. IVCM is a non-invasive procedure which poses minimal risk to patients, and provides important data so that clinicians can selectively offer AC tapping to patients with at least two of the following three features: severe iris depigmentation, absence of pseudoguttata on IVCM and low corneal ECD.

Nevertheless, our study had several limitations. First, our patients had a relatively long duration of AU. IVCM was only performed once but at a variable time point after an AC tap. These subjects had received at least 3 months of treatment (topical corticosteroid with or without antiviral) before the start of this study. As stated in the results, all eyes had zero inflammatory cells and clear cornea on the day of IVCM. A prospective study which recruits fresh eyes before AC tapping and offers IVCM on the day of uveitic recurrence might reveal more information especially on any acute endothelial differences, however the presence of severe corneal oedema may affect the image quality of the endothelium during IVCM, or impede localization of keratic precipitates, if any, thus making analyses difficult. The impaired vision during recurrence may also reduce ability to fixate during image capture. Alternatively, a uniform, pre-determined time-point for IVCM may theoretically reduce bias but we could not entirely control for the variable disease course following AC tap and response to treatment. Second, our study was limited by a small sample size, which was reflected in the wide confidence intervals and the predictive values analyses. Nonetheless, the directions of effect were uniform and a study of larger sample size may help refine the precision of data in addition to a proof of concept. Moreover, both the identification of pseudoguttata on IVCM and assessment of iris depigmentation are subject to the experience of the observer. We attempted to mitigate this bias by using one single operator on IVCM with a standard protocol, and another independent, masked observer in grading the IVCM images.

In Chinese eyes with chronic or recurrent CMV AU, it is prudent to note for the extent and severity of iris depigmentation on slit lamp exam, and to monitor for development of glaucoma. Further imaging such as the non-invasive IVCM plays a role in eyes with high clinical suspicion for CMV. Detection of low ECD and absence of pseudoguttata increases the PPV for detection of CMV, which are both important factors to consider when counselling patients for AC tapping in hypertensive, acute or current AU.