Erratum in: “Cellular-Level Visualization of Retinal Pathology in Multiple Sclerosis With Adaptive Optics” by Daniel X. Hammer, Katherine Kovalick, Zhuolin Liu, Chixiang Chen, Osamah J. Saeedi, and Daniel M. Harrison (Invest Ophthalmol Vis Sci. 2023;64(14):21), https://doi.org/10.1167/iovs.64.14.21.
Introduction
During the course of analysis for our pilot study to characterize the cellular features of the retina in multiple sclerosis (MS) using adaptive optics–optical coherence tomography (AO-OCT), the RGC soma density reported for some patients in the optic neuritis cohort (MS-ON) were incorrectly switched with respect to their values in the rest of the dataset, which included measures from the neurological exam, ophthalmic eye exam, and AO imaging session. The error affected the data presented in Figures 6A–C and Figures 7A, 7E, and 7F, as well as in Supplemental Tables S4, S5, and S6 (AO-OCT RGC density rows for each table). We have corrected the results, re-analyzed the statistical models, and present corrected figures below. Figures 63 and 7 and the Supplemental Section have been updated in the article online.
Figure 6.
Original.
Figure 7.
Original.
Figure 6.
Corrected.
Figure 7.
Corrected.
Results
Because the error only occurred within the MS-ON cohort, none of the group comparisons from the original article were affected. The overall trends in our results are unchanged, although a few new parameters reached statistical significance and the regression model coefficients increased. In no cases did the trends or significance become weaker or change the conclusions of the original article.
The corrected Figure 6 shows the updated associations between AO-OCT cellular and conventional OCT layer thickness metrics. The MS-ON cohort (solid blue symbols) follows clear trends with respect to the other groups. Accordingly, the correlation coefficients, which were analyzed in our model in the entire MS cohort (MS and MS-ON patients together), were more highly correlated with the corrected values (NFL: −0.122, P = 0.052; GCL: 0.115, P = 0.376; IPL: 0.945, P < 0.001) compared to the original erroneous values. Note that the HV values also changed, albeit only very slightly, because the statistical model is applied to the entire corrected dataset. For the corrected regional associations (nasal, temporal, all) between AO-OCT cellular and conventional OCT layer thickness metrics (Supplementary Table S4), statistical significance was found in the corrected data for the association between RGC density and GCC in the nasal region for MS subjects (P = 0.010) and between RGC density and GCL in the temporal region for healthy volunteers (P = 0.032).
The corrected Figure 7 shows the updated associations between AO-OCT cellular and MS visual and neurological disability metrics. The correlation coefficients were higher with the corrected values (2.5% LCLA: 0.675, P = 0.0012; EDSS: −3.282, P = 0.0143; PASAT: 0.506, P < 0.0001) compared to the original values. For the corrected regional associations between AO-OCT cellular and visual and neurological disability metrics, no different statistical significance was found for the visual metrics (Supplementary Table S5), whereas the association between RGC density and Symbol Digit Modalities Test (SDMT) in the nasal region was found to be significant (P = 0.0013) for the neurological disability metrics (Supplementary Table S6).
Discussion
The corrected results further reinforce the conclusions of the original article. Specifically, for all cohorts, and more so in the MS cohort with corrected values, lower GCL and IPL thicknesses were associated with reduced RGC soma density, and poorer performance on visual (lower 2.5% LCLA) and cognitive tests (higher EDSS and PASAT) were associated with reduced RGC soma density.
Supplementary Material
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.