4.1 Evidence for and prevalence of herpetic endothelial keratitis in patients with clinical AGR

Several case reports demonstrated an association between herpesviruses and AGR leading to graft failure.^{12-15, 22, 28} The role of herpesviruses in graft failure after keratoplasty has also been suggested in some larger studies.^{7, 8, 29} We had previously examined the prevalence of HSV in corneal buttons of patients with graft failure after AGR and found that it was positive in 7 of 95 patients.⁸ Van Gelderen et al.⁷ reached a similar conclusion by testing the corneal buttons and AH of 28 patients with allograft failure who underwent PKP for HSV and VZV using a combination of PCR tests of corneal tissue and of AH as well as testing for antibodies in AH. PCR testing for HSV was positive in 10 of 28 patients, two of whom also tested positive for VZV, suggesting a prevalence up to 35%. AH testing was simultaneously performed in 17 of the previous 28 patients. A test with a Goldmann-Witmer coefficient value >3 was considered positive.³⁰ The antibodies were positive in 7 of these patients (45%). They concluded that HSV could, therefore, be involved in corneal infection leading to allograft failure. Interestingly, PCR testing for HSV and VZV in corneal tissue was negative in 5 of the previous 7 patients with positive antibody production in AH. The simultaneous PCR test of AH of the previous 17 patients was also negative, probably due to the small sample size and the fact that the PKPs were performed in the quiescent phase of viral keratitis.⁷

While the previous findings confirm the role of herpesviruses in allograft failures, they also show that investigation of viral infection of corneal tissue versus AH directly via nucleic acid amplification techniques (NAT), such as PCR, may give rise to different results on the estimation of the prevalence of herpetic endothelial keratitis in corneal grafts compared to antibody testing of AH.

PCR testing of herpesviruses in AH has the advantage of good sensitivity in the early phase of intraocular infection with a very high specificity.³⁰ The test was truly negative in 140 of 140 in AH in patients who underwent cataract surgery,^{27, 30-32} although many of them had a positive serological test for these viruses. Thus, a positive PCR test result would reliably identify patients with active intraocular viral infection,^{27, 31-33} strongly suggesting a causative role of the detected viruses in the observed endothelial keratitis in our patients. However, due to the rapid clearance of viral DNA in AH compared with the slow clearance in avascular corneal tissue, it would likely not identify patients, in whom infection has resolved. AH could theoretically be contaminated with host blood, either during surgery or by spontaneous bleeding in the anterior chamber of the eye. This would potentially lead to a false-positive PCR test, especially in the case of CMV and EBV viremia without concomitant eye infection. However, such a contamination is very unlikely in practice. Circulating blood cells cannot access the anterior chamber of the eye under normal circumstances due to the immune privilege of the eye.³⁴ Additionally, the aspiration of the anterior chamber of the eye is done under sterile surgical conditions in the operation room, using a very fine (30 Gauge) needle that goes through the non-vascularized cornea. The aspiration is done before the corneal transplantation begins, hence minimizing the risk of contamination with blood through the following surgical manipulation. The negative PCR tests in the control group in our study as well as the high true negative values in the above-mentioned studies provide good evidence for the extremely low contamination rate of AH aspirates in clinical practice.

In contrast, testing for intraocular herpesvirus infection using antibodies in AH reaches good sensitivity with lower specificity no earlier than 5 days after onset of infection.³⁰ Antibodies also continue to be produced after the resolution of the infection, which makes it more suitable for identifying patients with a recently cleared or long-lasting herpesvirus infection.

Negative PCR in AH does not exclude a latent or recently resolved herpesvirus infection.⁷ Therefore, the true prevalence of herpetic keratitis masked as AGR is likely to be even higher than estimated in this study. On the other hand, antibodies in AH cannot reliably detect a herpetic infection in the early phase, and the DNA in corneal tissue persist long after active infection has resolved,^{25, 26, 30} limiting the usefulness of PCR test in corneal buttons.^{18, 35} An accurate estimation of prevalence would, therefore, likely require simultaneous testing with PCR and antibodies in AH, as well as PCR in corneal buttons, as these would complement each other and together provide a better understanding of the evolution of the disease over time.

4.2 Influence of a history of herpetic keratitis on the incidence of herpetic endothelial keratitis after keratoplasty

When comparing patients with AGR and HK+ to patients with AGR and HK−, there was no statistically significant difference between the two groups in our study (p = 0.8). The difference between patients with and without AGR was still statistically significant even after excluding patients with HK+ from the analysis. These results suggest that a history of herpetic keratitis does not increase the risk of herpetic endothelial keratitis after keratoplasty, when prolonged topical and systemic prophylactic therapy with antivirals is initiated.

In the study of Van Gelderen et al.⁷ A total of 8 of the 17 patients in whom AH was tested were HK+ and 9 were HK−. Antibodies were positive in 7 of these patients, 4 of whom were HK+ and 3 HK−. They concluded that HSV could, therefore, be involved in corneal infection during allograft failure, even in HK− corneas. These conclusions agree with those of the current study.

4.3 Association between a positive PCR test and repeat keratoplasty

In our study, 4 of the 9 patients with a positive test for herpesviruses underwent repeat keratoplasty, 3 underwent 2 repeat keratoplasties, and 2 underwent 3 repeat keratoplasties. After herpesvirus detection using PCR in AH and initiation of prophylactic treatment, 4 of the patients did not require further keratoplasties, 4 of them required another keratoplasty, and one required three keratoplasties and again had a positive PCR test for HSV in AH at the last surgery. This indicates that rapid initiation of therapy could prevent failure of the current graft and, if the current graft is already damaged, failure of subsequent grafts.

It was interesting to see that all patients with positive PCR tests had more than one keratoplasty, which could indicate that viral infection/reactivation is a major cause for graft failure and consequent repeat keratoplasty. Further studies with a larger cohort are required to explore this possibility.

4.4 Origin and risk factors for viral endotheliitis after keratoplasty

The relatively high estimated prevalence of viral activity of various herpesviruses after keratoplasty raises several questions about the origin and risk factors for these infections.

Infections with herpesviruses are very common in the general population. Such infections are usually mild and are controlled in most immunocompetent patients, but establish lifelong latency.³⁶ After primary infection, HSV and VZV remain latent in neural cells,³⁶ whereas CMV remains latent in polymorphonuclear leukocytes, monocytes, and hematopoietic progenitor cells.³⁷ EBV also establishes a lifelong latency in infected lymphocytes after an acute infection.³⁸ The prevalence of HSV-1 in the world population under 50 years of age is estimated at 67%.³⁹ More than 90% of the world's population is infected with EBV.⁴⁰ A total of 99% of the German population over the age of 40 are infected with VZV.³⁹ Furthermore, the seroprevalence of CMV infection was estimated to be 66%–90% worldwide.⁴¹

The high prevalence of positive serostatus in the population and the emergence of PCR tests have reduced the value of serologic tests in the diagnosis of herptic corneal infections. In the Japanese corneal endothelial study, 106 patients with corneal endotheliitis and positive PCR tests for CMV in AH samples were examined. The study showed that serum CMV immunoglobulin G (IgG) was positive in 100% of tested cases and that CMV IgM was negative in 100% of tested cases.⁴² Additionally, Serum CMV antigen is not typically elevated in patients with CMV endotheliitis.⁴³ However, recent studies on CMV infections after solid organ transplantation have shown that recipients with negative CMV serostatus from donors with positive CMV serostatus have a high risk of CMV infection after such transplantation.⁴⁴ To the best of our knowledge, there are no studies that have investigated the influence of donor/recipient CMV serostatus, nor of other herpesviruses, on the risk of viral keratitis after corneal transplantation. Therefore, exploring the influence of herpesvirus serostatus of recipients and donors as a risk factor for viral keratitis after corneal transplantation is warranted.⁴²

Herpesviruses could be reactivated under several circumstances leading to an infection of the corneal transplant. The high prevalence of latent herpesviruses makes reactivation of these viruses a likely explanation for the observed endothelial keratitis,^{8, 18} especially in patients with a previous history of viral keratitis. However, this may also occur in patients with no such history,^{8, 19-21} possibly through donor-to-recipient transmission.^{14, 22, 23}

Many of the herpesviruses’ reactivation risk factors are inherent to the corneal transplantation procedure. Risk factors for HSV and VZV reactivation include emotional stress, surgical trauma, postoperative tear dysfunction, use of corticosteroids and occasionally use of other immunospressants,⁴⁵ and neural stress response induced through the operation itself.³⁶ The Japanese corneal endotheliitis study showed that 96.3% of patients with CMV endotheliitis were using topical corticosteroids and 25.7% of patients had a history of corneal transplantation, underscoring the importance of these factors in CMV infection/reactivation.⁴² Immunosuppression is also a known risk factor for EBV reactivation.⁴⁶

Additionally, a history of herpetic keratitis is a known risk factor for infection, even after a successful corneal transplantation. This risk can be effectively reduced by the postoperative prophylactic use of antiviral therapy for at least 1 year.^{39, 47-50}

On the other hand, a positive PCR test for herpesvirus in the donor cornea does not appear to increase the risk of herpes keratitis after successful corneal transplantation. Morris et al.³⁵ tested 80 donor corneas for HSV PCR and with viral culture before transplantation. Although 3 out of 80 transplants were positive for HSV DNA, viral culture was negative in all three, and the follow-up after transplantation showed no sign of viral keratitis. Another study from Horstmann et al.³⁹ showed similar results. They tested donor corneas for HSV, VZV, and CMV PCR and found a prevalence of 8.2% (7/85) positive PCR tests. However, none of the patients developed viral keratitis after transplantation.

Various systematic factors are known to increase the risk of viral reactivation, including older age, diabetes mellitus, compromised immune system, autoimmune diseases, ocular rosacea, long-term immunosuppression, corticosteroid use and history of organ transplantation.

It is important to keep in mind that a viral infection could undermine the immune privilege of the anterior chamber and the cornea leading to exposure of transplant antigens to the host immune system and to a secondary AGR after corneal transplantation.³⁴ Additionally, AGR as well as other co-infections of the cornea could lead to differentiation of corneal dendritic cells and/or to activation of B-lymphocytes, thus triggering the reactivation of CMV⁵¹ and EBV,³⁸ respectively.

Endotheliitis following keratoplasty with accompanying viral replication is, therefore, more of an interplay between viral infection and immune response, whereby the two diagnoses are not mutually exclusive but can also occur simultaneously and act synergistically. Therefore, in the case of viral endotheliitis after keratoplasty, immunosuppressive therapy should always be administered in addition to appropriate antiviral therapy. In the case of significant or persistent AGR under corticosteroids, viral activity should also be excluded by means of AH aspiration.

4.5 Regular screening and prophylactic therapy for herpesviruses in patients with allograft rejection

Based on our findings, the number of patients needed to be screened (NNS) is estimated to be 12 when the AH of patients with current or recent clinical AGR is tested for herpesvirus DNA. Although the NNS is relatively high, the consequences of overlooking a viral infection, such as permanent graft failure and the need of repeat keratoplasty with a higher risk of viral keratitis recurrence after transplantation, warrant regular testing for herpesviruses in AH in patients with AGR and would possibly justify empirical use of antiviral therapy until test results are available. We hence recommend testing AH for herpesviruses in patients with AGR. Empirical antiviral therapy has also been proposed by van Gelderen et al.⁷ Further studies exploring other possible risk factors would help to further reduce NNS, and thus make screening more efficient. AH testing for herpesviruses with a combination of PCR testing and antibody testing with formation of the Goldmann-Witmer coefficient³⁰ could potentially increase the detection of herpes infections and overcome the shortcomings of both methods, leading to a more accurate estimate of the prevalence of actual as well as previous herpesvirus infection.³⁰ Further studies in this field are required to answer this question.

The present study also highlights the importance of testing not only for HSV but also for other herpesviruses including VZV, CMV, and EBV. Further studies with a larger number of positive cases would allow differentiation between the clinical manifestations and prognosis of infections with different herpesviruses.

This study is limited due to its retrospective nature and the fact that predominantly patients with moderate to severe acute graft rejection were included, as patients with mild AGR are usually treated as outpatients, selection bias cannot be excluded. However, the results strongly suggest that viral infection is a frequent cause of what is usually clinically perceived as endothelial allograft rejection.

Although a positive PCR test result in AH could also indicate an anterior uveitis, keratouveitis, or a deep stromal or endothelial viral keratitis,¹⁷ this distinction has no consequence for subsequent therapies with antiviral drugs and steroids and is, therefore, not of relevance in the clinical setting of acute graft rejection.

It is important to keep in mind that corneal transparency is extremely reduced after graft failure. Therefore, due to the limited visibility and the resulting difficulty in assessing the inflammatory status of the anterior chamber (cells/Tyndall), it is hard to say whether a new AGR has occurred after the first one. We, therefore, think that the actual time between AGR and keratoplasty might be overestimated in our study.

The antiviral preoperative treatment of patients with a history of herpes keratitis could reduce the sensitivity of a viral PCR test afterwards. Nevertheless, it is the only ethically acceptable approach and thus cannot be avoided in future studies.