Corneal endothelial cells are essential in maintaining stromal deturgescence,
14 but they are also the prime target of an immune-mediated attack during corneal allograft rejection. Endothelial cells have no mitotic capacity, but respond by spreading out in the process of wound healing. It is well known that even uncomplicated grafts lose endothelial cells more rapidly than the annual cell loss rate in normal adult eyes.
15 A decrease in ECD affects the ability of the endothelium to maintain its primary function, as is evidenced by an increase in corneal thickness as the ECD decreases. This eventually leads to endothelial decompensation and a hazy graft, typically when the cell density reaches 333–500 cells/mm
2.
14 This is corroborated by the significant negative association between graft failure and ECD.
15
There are several reports, mostly without control for the expected loss, claiming that allograft rejection leads to a strong decline in ECD.
6 7 8 9 10 In contrast, a large longitudinal cohort study of consecutive PKPs found no difference in endothelial cell loss at 10 years after surgery in grafts that had had rejection episodes compared with those that had not.
15 The purpose of our study was therefore to examine the decline in ECD in patients who had undergone PKP and experienced allograft rejection, by comparing this decline to the normal evolution in patients in whom the PKP follow-up was uneventful. The second purpose of our study was to identify possible factors predictive of this endothelial cell loss after corneal allograft rejection.
Approximately 15% of patients included in our study experienced an immunologic graft rejection. This rejection occurred on average at approximately 14 months after surgery. These results are in agreement with the 19% rejection rate reported by Naacke et al.,
16 the 13.6% reported by Jonas et al.,
17 and data reported in many other investigations.
We controlled for the expected decrease in ECD by quantifying the loss of endothelial cells over similar time intervals in a control population of patients with uneventful follow-up. Based on this comparison, three groups could be discriminated in the 45 cases experiencing allograft rejection. The observed percentage loss of ECD in almost half of the cases (21 corneas; 47%) was not significantly greater than expected (group I, normal decline). A second group of patients consisted of those (13 corneas, 29%) who showed a decline in ECD that was significantly greater than expected (group II, strong decline). Finally there were 11 corneas (24%) in which endothelial cells were no longer observable due to irreversible edema accompanying graft failure (group III, cells no longer observable). These results are in accordance with data reported by Musch et al.
9 who found only a difference of borderline statistical significance when comparing observed and expected endothelial cells in all 48 patients with keratoconus who experienced allograft rejection. Their result, however, was influenced greatly by the severity of the rejection episode. It was termed mild or severe based on the amount of keratic precipitates present and/or the increase in corneal thickness relative to the last visit at the time the allograft rejection was diagnosed. When they analyzed those with a severe rejection episode the difference between observed and expected endothelial cell loss became significant, whereas it remained statistically insignificant in cases of mild rejection. In our study we were not able to evaluate the severity of the rejection episode as a variable, because we did not have objective criteria at our disposal to define the severity (the number of keratic precipitates was not always noted and pachymetry was not routinely performed).
The results of our study are also in accordance with the findings of Ing et al.
15 They found no difference in endothelial cell loss in grafts that had had a rejection episode versus those that had not. However, their 10-year cohort excluded all grafts that had failed, often years after a rejection episode. Therefore, in fact, they probably excluded the patients who would have belonged in group III in our study (with a graft failure after irreversible rejection) as well as some patients in group II (strong decline), because some of those will probably end up with a graft decompensation because of insufficient ECD. Therefore in their setting, most patients with functional grafts 10 years after keratoplasty, would belong to group I and would show a decline in ECD similar to that in normal patients, as we have demonstrated.
Some patients experiencing allograft rejection did not show a greater than expected endothelial cell loss, whereas others showed a strong decline in ECD. This finding led us to examine those factors influencing the extent of the decline. We examined different donor, recipient, and surgical variables. We also evaluated clinical, as well as therapy data. Analysis was performed in two ways: first by comparing the patients with a normal decline in ECD with those with a strong decline (group I versus group II) and second by comparing the decline in group I with those in group II and group III taken together. We assumed this last group (group III) to be a continuum of the second group, because it is a well-known fact that a certain amount of endothelial cells is required to maintain corneal dehydration.
14 Therefore, even though specular microscopy photographs could no longer be obtained, these patients could be assumed to have an ECD of less than 400 to 500 cells/mm
2. This assumption was confirmed by the fact that both analyses resulted in the same significant risk factors after multiple logistic regression.
The only two risk factors that reached statistical significance after multiple logistic regression analysis turned out to be a delay in diagnosis and a recipient more than 60 years of age. None of the donor variables influenced the extent of endothelial cell loss. Grafts that are performed urgently (à chaud) have a risk of losing endothelial cells after allograft rejection similar to that in grafts performed for other indications, which is in line with our previous results.
18 A preoperative diagnosis of bullous keratopathy (KP), conversely, was associated with an odds ratio of 8.00; however, this did not reach statistical significance at the 0.05 level when group I was compared with group II, probably because of lack of power. It became significant at the univariate level in the analysis of group I versus group II and III taken together, but this was no longer the case in the multiple logistic regression analysis. There is evidence that the recipient endothelium is also reduced in cell density by allograft rejection. This may be explained by the cellular migration from the host peripheral cornea to the donor cornea
10 to compensate for the damage to the donor endothelial cells. Patients with pseudophakic bullous keratopathy have a poor recipient endothelium that cannot contribute to the wound healing process, and therefore may be at a higher risk of a strong decline in ECD after allograft rejection.
A risk factor that was significantly associated with a stronger decline in ECD after allograft rejection in both analyses was a recipient age of more than 60 years. There are several reports in which a young recipient age is associated with a higher chance of rejection: Alldredge and Krachmer
12 found a significantly higher incidence of rejection in patients younger than 50 years; Boisjoly et al.
19 reported significantly lower rejection-free transplant survival in recipients who were younger than 60 years; and, in the Collaborative Corneal Transplantation Studies Research Group,
13 recipients younger than 40 years had an elevated risk of graft reaction and rejection failure. The Corneal Transplant Follow-Up Study
20 reported an increased risk of failure associated with patients younger than 10 years. In contrast, the Australian Corneal Graft Registry found an opposite association. They reported that recipients older than 50 years at the time of graft had a poorer graft outcome
21 and in a study on PKP in recipients aged 80 year or older they found that graft survival declined significantly with increasing recipient age.
4 Whatever the effect of acceptor age on graft rejection, our study seems to suggest that patients older than 60 experiencing allograft rejection have a higher chance of losing more endothelial cells than do younger patients. This could be in part because these older patients often have additional ocular conditions (e.g., retinal problems or severe dry eyes) with chronic unrelated ocular symptoms that may mask the symptoms of allograft rejection or may lead to the fact that the symptoms of the rejection are not discernible to the patient,
22 thus leading to a longer delay and a more severe rejection episode before seeking medical advice. However, multivariate analysis clearly demonstrates that older recipient age is also an independent risk factor. Another possible explanation for our finding could be that older recipients have fewer endothelial cells in their remaining host cornea, thus reducing the possibility for cellular migration from the host peripheral cornea to the donor cornea to compensate for the damage to the donor endothelial cells.
The main risk factor for a strong decline in ECD in our study turned out to be a delay in diagnosis and treatment. This delay was defined as the time interval between the onset of symptoms, as noticed by the patient, and the moment of consultation at our department at which time a treatment for rejection was initiated. Patients treated in our department for corneal graft rejection episodes are closely questioned about the duration of their symptoms, and therefore data on the delay in diagnosis were available in 42 of 45 cases. In three cases (two in group I and one from group II) rejection occurred asymptomatically and was diagnosed during routine follow-up, making this time difficult to estimate. Three cases of 34 (groups I and II) patients (9%) without symptoms of rejection is low when compared with other figures reported in the literature and could be because our patients are carefully instructed of the possible alarm symptoms of rejection by handing out a list with the different alarm symptoms at the moment of discharge from the hospital and by repeating them at each follow-up visit. Naacke et al.
16 reported 30% of cases without symptoms in a mixed population of high- and low-risk cases. In their study on patient-reported symptoms associated with graft reactions in high-risk patients, Kamp et al.
22 stated that approximately one third of the patients experiencing a first or subsequent rejection reaction fail to report any symptoms.
There is only one other study in which researchers attempted to determine the relationship between loss of endothelial cell density and possible influencing factors.
9 The only factor analyzed by Musch et al.
9 was the severity of the allograft rejection. The authors found that only the severe rejection episodes showed a decrease in ECD that exceeded the expected loss significantly. They concluded that these must have been the cases in which detection was made late, because they assumed that most allograft rejection episodes are initially mild and become more severe over time; however, they did not record delay as a variable.
The results of our study confirm that there is a time window in which an allograft rejection can be treated and endothelial cell loss prevented. It is common knowledge to all cornea practitioners that a delay in diagnosis and treatment affects endothelial cell loss negatively. The longer one waits to treat rejection, the higher the loss of endothelial cells. The results of our study have supported this “common knowledge”—that delay affects outcome negatively. Even when the delay only exceeds 1 day, the outcome is already significantly worse when analyzed statistically. It is therefore imperative not only to educate patients and their relatives about the alarm symptoms of graft rejection, but also to instruct them to contact the hospital immediately through the emergency department at any time of the day or night (even during the weekend) whenever they suspect a graft reaction.
There are, of course, limitations to a study of this size. The relatively small number of corneas included results in a lack of statistical power, which is reflected by the large confidence intervals, and therefore also in low precision. It is possible that a larger number of cases would reveal other clinically relevant risk factors that did not reach statistical significance in our study because of this lack of power. However, those factors that were found to be of statistical significance despite the small numbers in our study, can be considered of high clinical relevance.
In conclusion, we can state that reversible immunologic rejection episodes do not necessarily lead to a greater than expected endothelial cell loss. Almost half of the patients in our study showed a decline in ECD that is comparable to the decline in patients who undergo PKP with an uneventful follow-up. The single most important variable influencing the extent of endothelial cells loss is the delay in diagnosis of rejection and therefore delay in institution of appropriate antirejection treatment. This is a clinically relevant finding because this means that there is room for decision-making or expert intervention to reduce the loss of valuable endothelial cells. It underscores the importance of patient education in recognizing the alarm symptoms of graft rejection and in seeking immediate medical assistance. Eye care providers should consider corneal graft rejection as one of the few real urgencies in ophthalmology, because vigorous and early treatment of graft rejections can mean the difference between keeping or losing endothelial cells.