Incidence, risk factors, and treatment of retroprosthetic membranes following Boston keratoprosthesis eyes and the impact of glaucoma surgery.

PURPOSE: To assess incidence, risk factors, and treatment of retroprosthetic membrane (RPM) formation in eyes following Boston keratoprosthesis (Kpro) implantation and their correlation with glaucoma drainage device placement (GDD).

METHODS: A retrospective review was performed on eyes that underwent Kpro type I or II implantation between 2005 and 2020 at a tertiary academic center. Multiple variables were collected including preoperative characteristics, presence of RPM, management of RPM, and outcomes including corrected visual acuity (VA). A Fischer's exact test was used to evaluate the significance of risk factors of RPM formation and an odds ratio was calculated for each possible risk factor. A Mann-Whitney U test was used to evaluate comparisons between outcomes and qualitative analyses.

RESULTS: Of the 87 eyes identified, 37 (43%) developed an RPM within an average of 1.5 years (range, 31 days-7.5 years) following Kpro implantation. Mean follow-up duration was 4.3 years. Eyes that developed RPM had significantly worse preoperative VA compared to those that did not (logMAR 2.55 vs. 2.28, p = 0.022). The mean number of prior penetrating keratoplasty procedures trended higher in eyes that developed RPM (2.46 vs. 2.18, p = 0.44) but was not significant. GDD placement after Kpro implantation was associated with an increased risk of RPM formation (RR = 1.69 p = 0.026). Of the 37 eyes that developed an RPM following Kpro, 17 (47%) were treated with Nd:YAG laser, and four of those 17 (21%) also underwent pars plana vitrectomy (PPV). Seven of 37 eyes (19%) underwent PPV without Nd:YAG. Comparisons between RPM occurrence and final VA were not significant.

CONCLUSIONS: The incidence of RPM formation following Kpro implantation was 43%. Eyes that developed RPM had significantly worse preoperative VA. GDD placement after Kpro implantation increased the risk of developing RPM. Final VA and occurrence of RPM were not significantly different between the Nd:YAG and PPV treatment groups.