Infectious and Non-infectious Corneal Ulcers in Ocular Graft-Versus-Host Disease: Epidemiology, Clinical Characteristics and Outcomes.

PURPOSE: To evaluate the incidence, clinical characteristics, microbiological profile, and therapeutic outcomes of corneal ulcers in individuals with chronic ocular graft-vs-host disease (coGVHD).

DESIGN: Retrospective clinical cohort study METHODS: Review of individuals diagnosed with coGVHD following hematopoietic stem cell transplantation (HSCT) who were seen at the Bascom Palmer Eye Institute between May 2010 and November 2021. Baseline demographics, clinical characteristics, microbiological profile, risk factors for corneal ulceration, and treatment outcomes were collected. Etiology was deemed infectious in individuals with a positive culture or appropriate clinical scenario (presence of stromal infiltrate or hypopyon); otherwise, ulcers were presumed to be non-infectious. Treatment success was defined as re-epithelialization with infiltrate resolution, and treatment failure as progression to corneal perforation or keratoplasty. Kaplan-Meier survival analysis estimated the incidence of ulceration. Cox regression analyses examined demographic and risk factors. Infectious and non-infectious ulcer groups were compared using two-way independent t-tests, one-way analysis of variances (ANOVAs), and Chi-squared tests, as appropriate.

RESULTS: 173 individuals were included (53.7±14.4 years old; 59.0% male). Thirty-three individuals developed an ulcer 74.5±54.3 months after HSCT, with estimated 5- and 10-year incidences of 14% and 30%, respectively. Twenty-two (66.6%) ulcers were deemed infectious (15 microbiologically confirmed, 7 clinically) and 11 (33.3%) were deemed non-infectious. Risk factors for corneal ulceration included black race (hazards ratio [HR] 2.89, 95% CI 1.30-6.42, p<0.01), previous ocular surgery (HR 9.16, 95% CI 3.86-21.72, p<0.01), lid margin abnormalities (HR 3.44, 95% CI 1.69-6.99, p<0.01), and topical steroid use (HR 2.74, 95% CI 1.33-5.62, p<0.01). Conversely, contact lens use reduced the risk of corneal ulceration (HR 0.29, 95% CI 0.13-0.66, p<0.01). Infectious ulcers had a significantly higher frequency of treatment failure than non-infectious ulcers (57.1% vs 20.0%, p=0.04).

CONCLUSION: Corneal ulceration is a potential complication of coGVHD, with several clinical features identified as risk factors. Infectious ulcers had worse outcomes than non-infectious ulcers.