Fluctuations in Central Subfield Thickness Associated With Worse Visual Outcomes in Patients With Diabetic Macular Edema in Clinical Trial Setting

•Central subfield thickness (CST) is classically a poor surrogate for visual acuity in patients with diabetic macular edema (DME), typically because of its inability to fully capture other optical coherence tomography metrics such as loss of the ellipsoid zone or disruption of the retinal inner layers.•This study of 1197 eyes from Diabetic Retinopathy Clinical Research Network protocols T and V found significant visual acuity differences based on fluctuations in CST in patients with DME. These differences remained significant, even while adjusting for mean baseline acuity, baseline CST, lens status, hemoglobin A1c, and treatment arm. Eyes with larger fluctuations in CST had worse visual acuity compared with eyes with the fewest CST fluctuations in both protocols.•Large fluctuations in CST may portend worse VA outcomes over time in patients with DME. This study examines the relationship between fluctuations in central subfield thickness (CST) and visual acuity (VA) in patients with diabetic macular edema (DME) using data from 2 large clinical trials. Clinical cohort study using post hoc analysis of clinical trial databases. Standard deviation (SD) of all recorded CSTs for each patient during the study period were used to quantify the fluctuations in CST. Patients from each protocol were grouped into quartiles based on the CST SD. Eyes with at least 3 CSTs and VA at 1 year were included. The main outcome measures were VA at 1 and 2 years for each protocol, stratified by SD quartile. A total of 1197 eyes were included in the analysis. There were significant VA differences based on CST SD quartile for both protocols while adjusting for mean baseline VA, baseline CST, lens status, hemoglobin A1c, and treatment arm. At week 52 in protocol T, the difference between the first and fourth quartiles was −1.61 Early Treatment Diabetic Retinopathy Study letters (95% confidence interval [CI] −3.52 to 0.30, P = .0986). At week 104, this difference was −3.59 letters (95% CI −6.17 to −1.00, P = .0066). In protocol V, at week 52, the difference between the first and fourth quartiles was −3.04 letters (95% CI −4.18 to −1.91, P < .0001). At week 104, this difference was −2.35 letters (95% CI −3.58 to −1.13, P = .0005). Large fluctuations in CST may portend worse VA outcomes at the 2-year end point in patients with DME.