Retinal nerve fiber layer and optical coherence tomography angiography analysis in fibromyalgia using logistic regression

Abstract

AbstractPurposeTo evaluate the retinal nerve fiber layer thickness (RNFL) and the vascular density (VD) at the superficial vascular plexus (SVP) of the retina of patients with Fibromyalgia (FM) using Optical coherence tomography‐Angiography (OCT‐A) and to compare results with measurements of healthy eyes.MethodsFifty eyes of 50 patients with FM were included and compared with 232 healthy eyes. All subjects underwent peripapillary and macular RNFL thickness evaluation and macular SVP VD assessment using swept source OCT Triton and its OCTA module (Topcon, Japan). Both groups were compared, and the statistically significant different variables were used to create a logistic regression model. A Linear Discriminant Function (LDF) and its receiver operating characteristic curve (ROC) was obtained from the model.ResultsStatistically significant thinning was found at the peripapillary (temporal, p = 0.006; and supero‐temporal, p = 0.026) and macular (inferior, p = 0.029) RNFL thickness in the FM group compared to healthy controls. The best resultant LDF after logistic regression analysis was (Triton: −9.102 + (0.434 * RNFL macular inferior) + (0.088 * peripapillary RNFL nasal sector) – (0.220 * peripapillary RNFL temporal sector) showing a ROC curve with an area under curve (AUC) equal to 0.758. On the other hand, the SVP VD obtained with OCT‐A showed no statistically significant differences between groups in any of the analyzed areas (central p = 0.788; nasal p = 0.165; inferior p = 0.073; temporal p = 0.250; superior p = 0.538) so the data were not included in the regression analysisConclusionsFM patients present a reduction in the RNFL thickness compared to controls. The RNFL measurements might be a useful biomarker in the diagnosis of FM, complementing the current criteria based on clinical symptoms. However, it is necessary to explore the different imaging protocols to include them into a more robust statistical model.