We appreciate the interest expressed in our article by Patel and associates. The fact that they have concerns about our study indicates a need for clarification. Our objectives were to describe outcomes associated with penetrating keratoplasty (PK) in eyes that had previously undergone placement of an Ahmed valve (AV); to determine the incidence of graft failure in such eyes; and, because not every patient experiences graft failure, to identify risk factors other than the presence of an AV that also contribute to graft failure in this population. Our study was not designed to demonstrate a direct causal relationship between the presence of an AV per se and graft failure.

With regard to the issue of cohort homogeneity, we were referring only to the type of glaucoma drainage device and to the temporal relationship between PK and placement of an AV. We specifically stated that we had limited our study to patients who underwent PK as a separate procedure after the eye already had an AV, as occurs typically at our institution. Clearly, the cohort was not homogeneous in terms of age, previous surgeries, or number of glaucoma medications, as we were looking to identify other risk factors for graft failure in this setting. We acknowledged the fact that our statistical analyses may have been affected by multiple comparisons, but we definitely did not state that the results were spurious; in fact, many of the relationships we identified are consistent with those found in previous studies, and all are biologically plausible.

Patel and associates asked about graft failure in the subgroup of eyes that had undergone a first PK after a single AV. They also suggested that we should have compared that subgroup to eyes without glaucoma drainage devices that had undergone single PK. We can provide the incidence data that they requested. Graft survival at 2 years for our patients with single AVs who underwent first grafts (n = 38) was 56.5%, which is substantially less than the 10-year survival data for primary grafts (78%–82%) reported in other studies and is also less than the reported 2-year survival for repeat grafts, in the absence of glaucoma drainage devices (63.9%–83%). We chose not to create our own retrospective control group of patients with grafts but no glaucoma drainage devices, because of the many factors that influence whether or not to place such devices; these factors could also influence graft survival, and thus comparison to such a control group would not have isolated the AV as an independent risk factor (confounding-by-indication). Only a prospective, randomized clinical trial would achieve that goal, and it is unlikely that such a trial can be undertaken. Because we did not limit our study to patients with single grafts, we were able to show that repeat PK is an additional risk factor for graft failure in this population.

Patel and associates point out that associations between multiple grafts, corneal neovascularization, and graft failure have been reported previously. As stated in our article, identification of these known associations in this cohort supports the quality of our data. Nowhere in our article did we claim that these associations were the “main” conclusions of our study, as suggested by Patel and associates.

Following are the main conclusions we drew from our study. Survival of corneal grafts in the presence of an AV located in the anterior chamber is poor. These graft failures are associated with progressive loss of endothelial cell function, without observed immunologic rejection episodes. Despite the presence of an AV at the time of PK, an escalation in glaucoma therapy typically follows PK. Careful control of intraocular pressure in these patients may reduce the risk of graft failure.