False positives in retinal nerve fiber layer and minimum rim width in healthy subjects from spectral-domain optical coherence tomography

Abstract

Purpose : To evaluate the rate of false positive results of optical coherence tomography (OCT) retinal nerve fiber layer (RNFL) and minimum rim width (MRW) outside normal limits classification in healthy subjects.

Methods : A total of 100 eyes of 56 healthy subjects (European Descent (ED)=36, African Descent (AD)=20) with axial length ≤26.5 mm from the Diagnostic Innovations in Glaucoma Study, who had circumpapillary scanning by Spectralis SDOCT, were enrolled in this cross-sectional study. False positive of the RNFL and MRW color-code classifications were defined as one or more sectors outside normal limits compared to a reference database at the <5 % level (yellow or red). Demographic (age and ethnicity) and clinical factors (central corneal thickness (CCT), axial length, Bruch’s membrane opening area (BMO), presence of beta peripapillary atrophy (βPPA), and optic disc torsion) were also compared between eyes with normal findings and eyes with false positive results of RNFL and MRW outside normal limits classification. Multivariate logistic regression was used to estimate the strength of the association with these factors.

Results : The overall false positive rate was significantly higher for RNFL (23.0%) than for MRW classification (8.0 %) (p< 0.001). Being of African Descent was significantly associated with a higher likelihood of an outside normal limits classification of MRW (OR: 11.87 95% CI [1.18, 119.83], p= 0.036) and RNFL thickness (OR: 3.80 [1.03, 13.99], p= 0.044). Smaller BMO area was significantly associated with a lower likelihood of a MRW false positive occurrence (OR: 0.12 [0.02, 0.78], p= 0.027) but not with a RNFL thickness false positive occurrence (OR: 0.25 [0.05, 1.31], p= 0.102). Age, CCT, axial length, βPPA, and torsion were not significantly associated with RNFL or MRW false positive occurrence (p> 0.114 for all).

Conclusions : Spectralis SDOCT-measured MRW analysis has higher specificity than RNFL and its specificity is affected by race and BMO area. Sensitivity is another important aspect of diagnostic tools and should be considered when clinicians interpret color-coded SDOCT results.

This is a 2020 ARVO Annual Meeting abstract.