Does Interrater Variation Influence the Classification According to the Diagnostic Criteria for Multiple Sclerosis?

Area of Science:

• Neurology
• Radiology
• Medical Imaging

Background:

• Limited and outdated studies exist on interrater agreement in MRI analysis for early multiple sclerosis (MS) diagnosis.
• This study addresses the need for current data on MRI lesion identification agreement.

Purpose of the Study:

• To evaluate interrater agreement for brain, optic nerve, and spinal cord lesion identification in early MS.
• To assess variability in applying the 2017 and 2024 McDonald criteria for MS diagnosis.

Main Methods:

• Seventy-eight patients with a first clinical episode suggestive of MS underwent standardized 3 Tesla MRI.
• Three readers with varied expertise analyzed lesions in the brain, optic nerves, and spinal cord.
• Interrater variability was quantified using Cohen's κ and Fleiss κ.

Main Results:

• Interrater agreement was slight to fair for total lesion counts but moderate to substantial for lesion presence.
• Highest agreement observed for spinal cord lesions (κ=0.84), periventricular T2 lesions (κ=0.70), and enhancing brain lesions (κ=0.75).
• Substantial agreement (κ=0.65) was found between the 2017 and 2024 McDonald criteria revisions.

Conclusions:

• Standardized 3T MRI protocols did not substantially improve interrater agreement for lesion counts in early MS.
• Agreement in classifying lesions based on diagnostic criteria remained consistent between the 2017 and 2024 McDonald criteria.