Grading CMV Retinitis Progression Compared Between Two Reading Centers

Abstract

Abstract: : Purpose: Comparison of two different fundus photograph reading centers (FPRC's) in grading cytomegalovirus (CMV) retinitis features and progression. Methods: From an ophthalmology service, patients with Acquired Immune Deficiency Syndrome (AIDS) and CMV retinitis were enrolled in a prospective epidemiologic study of the prevalence and incidence of resistant CMV. Masked replicate gradings of monthly photographs were compared between two different FPRC's for retinitis progression (movement of an existing lesion border by > ½ disc diameter or occurrence of a new lesion) and for other features. Results: There were 77 patients with replicate gradings by the two FPRC's. Considering both eyes, there was 95% agreement on presence/absence of disease at baseline, and a concordance correlation coefficient of 0.75 for estimated retinitis extent. Kaplan-Meier plots of the proportion progressing over time were similar, with median times to progression of 65 vs. 69 days. Exact agreement between FPRC's on visit of progression was 51% (Κ = 0.37, "fair&rsquotdbl;), comprising 28 patients who progressed and 11 patients who did not. Including agreement on progression visit within 1 month increased the rate to 62%. Of 12 patients with progression graded as more than one month apart, 8 were only two months apart. Treating each visit as a choice point for presence/absence of progression, overall agreement was 78% (Κ = 0.55, "fair&rsquotdbl;). Agreement on progression was higher for lesions that had fully "active&rsquotdbl; (opacified) borders, or when the progression criterion was a new lesion. Average rates of retinal loss, calculated from monthly estimates of retinitis extent, were 2.8% per month vs. 2.0% per month (p=0.015). Rates of agreement within each FPRC, based on masked replicate grading of smaller subsets, were similar to or modestly higher than those observed between FPRC's. Conclusion: Different FPRC's agreed well on the presence and extent of CMV retinitis in eyes. Agreement on retinitis progression was fair regarding individual patients, but good on median time to progression for the population. Similar written protocols and procedures promoted reasonable consistency between centers, but ongoing comparison of gradings would probably be needed to further improve agreement.

Keywords: AIDS/HIV • clinical research methodology • cytomegalovirus