Association of Optical Coherence Tomography With Longitudinal Neurodegeneration in Veterans With Chronic Mild Traumatic Brain Injury

Importance: Mild traumatic brain injury (TBI) may predispose individuals to progressive neurodegeneration. Objective: To identify evidence of neurodegeneration through longitudinal evaluation of changes in retinal layer thickness using optical coherence tomography in veterans with a history of mild TBI. Design, Setting, and Participants: This longitudinal cohort study evaluated veterans who were receiving services at the Minneapolis Veterans Affairs Health Care System. Symptomatic or mild TBI was diagnosed according to the Mayo TBI Severity Classification System. Participants in the age-matched control group had no history of TBI. Participants with any history or evidence of retinal or optic nerve disease that could affect retinal thickness were excluded. Data analysis was performed from July 2019 to February 2020. Exposures: The presence and severity of mild TBI were determined through consensus review of self-report responses during the Minnesota Blast Exposure Screening Tool semistructured interview. Main Outcomes and Measures: Change over time of retinal nerve fiber layer (RNFL) thickness. Results: A total of 139 veterans (117 men [84%]; mean [SD] age, 49.9 [11.1] years) were included in the study, 69 in the TBI group and 70 in the control group. Veterans with mild TBI showed significantly greater RNFL thinning compared with controls (mean [SE] RNFL slope, -1.47 [0.24] μm/y vs -0.31 [0.32] μm/y; F_1,122 = 8.42; P =.004; Cohen d = 0.52). Functionally, veterans with mild TBI showed greater declines in visual field mean deviation (mean [SE] slope, -0.09 [0.14] dB/y vs 0.46 [0.23] dB/y; F_1,122 = 4.08; P =.046; Cohen d = 0.36) and pattern standard deviation (mean [SE] slope, 0.09 [0.06] dB/y vs -0.10 [0.07] dB/y; F_1,122 = 4.78; P =.03; Cohen d = 0.39) and high spatial frequency (12 cycles/degree) contrast sensitivity compared with controls. Cognitively, there was a significantly greater decrease in the number of errors over time during the Groton Maze Learning Test (GMLT) in controls compared with veterans with mild TBI (mean [SE] slope, -9.30 [1.48] errors/y vs -5.23 [1.24] errors/y; F_1,127 = 4.43; P =.04; Cohen d = 0.37). RNFL tissue loss was significantly correlated with both worsening performance on the GMLT over time (Spearman ρ = -0.20; P =.03) and mild TBI severity (Spearman ρ = -0.25; P =.006). The more severe the mild TBI (larger Minnesota Blast Exposure Screening Tool severity score), the faster the reduction in RNFL thickness (ie, the more negative the slope) across time. Conclusions and Relevance: This cohort study found longitudinal evidence for significant, progressive neural degeneration over time in veterans with mild TBI, as indicated by greater RNFL tissue loss in patients with mild TBI vs controls, as well as measures of function. These results suggest that these longitudinal measures may be useful biomarkers of neurodegeneration. Changes in this biomarker may provide early detection of subsequent cognitive and functional deficits that may impact veterans' independence and need for care.