Diagnostic ability of a linear discriminant function for spectral-domain optical coherence tomography in patients with multiple sclerosis

Abstract

Purpose: To calculate and validate a linear discriminant function (LDF) for spectral-domain optical coherence tomography (OCT) to improve the diagnostic ability of retinal nerve fiber layer (RNFL) thickness parameters for the detection of multiple sclerosis (MS).

Design: Observational cross-sectional study.

Participants: Patients with multiple sclerosis (n = 115) and age-matched healthy subjects (n = 115) were enrolled in the study.

Methods: The Spectralis OCT system (Heidelberg Engineering, Heidelberg, Germany) was used to obtain the circumpapillary RNFL thickness in both eyes of each participant.

Main outcome measures: A validating set including 60% of the study subjects (69 healthy individuals and 69 patients with MS) was used to test the performance of the LDF in an independent population. Receiver operating characteristic (ROC) curves were plotted and compared with the RNFL parameters measured using OCT. Sensitivity and specificity were used to evaluate diagnostic performance.

Results: The optimized function was 4.965 - 0.40 × (mean thickness 15-30 degrees) - 0.17 × (mean thickness 300-315 degrees) + 2.743 - 0.032 × (mean thickness 105-120 degrees) - 0.031 × (mean thickness 120-135 degrees) - 0.018 × (mean thickness 225-240 degrees). The largest area under the ROC curve was 0.834 for our LDF in the validating population. At 95% fixed specificity, the LDF yielded the highest sensitivity values.

Conclusions: Measurements of RNFL thickness obtained with Spectralis OCT had good ability to differentiate between healthy individuals and individuals with MS. On the basis of the area under the ROC curve, the LDF performed better than any single parameter.