GLA variant screening in patients with suspected demyelinating disease: Diagnostic yield and implications for the differential diagnosis of multiple sclerosis.

BACKGROUND

Fabry disease may resemble multiple sclerosis (MS) because of overlapping neurological manifestations and white matter lesions. The diagnostic yield of GLA testing for atypical demyelinating presentations remains uncertain and depends on variant classification.

METHODS

This retrospective, single-center cohort included patients screened for Fabry disease at a demyelinating diseases clinic from January 2016 to January 2026. GLA sequence analysis was performed on dried blood spot samples; sequence-negative cases underwent multiplex ligation-dependent probe amplification. Clinical, cerebrospinal fluid (CSF), and magnetic resonance imaging (MRI) findings were reviewed. Wilson 95% confidence intervals (CIs) were estimated; comparisons were exploratory.

RESULTS

Among 584 participants, 15 had reportable GLA sequence findings (2.57%; 95% CI, 1.56-4.19). Twelve carried p.Asp313Tyr (D313Y), a benign or likely benign pseudodeficiency allele. Pathogenic p.Ser148Asn and p.Arg227Gln variants were identified in two women (0.34%; 95% CI, 0.09-1.24). Neither fulfilled McDonald criteria, and final neurological assessment did not support MS. Four participants with reportable findings had confirmed MS; all carried variants not diagnostic of Fabry disease. Periventricular and corpus callosum lesions were nominally associated with reportable GLA findings (crude odds ratios, 3.61 and 3.09; P = 0.038 and 0.040, respectively), but estimates were imprecise and unadjusted for multiple comparisons.

CONCLUSIONS

Reportable GLA findings outnumbered pathogenic variants. The observed low frequency of pathogenic variants provides little support for testing based solely on suspected demyelinating disease. Targeted testing should be considered when atypical clinical, CSF, or MRI findings coexist with Fabry-compatible vascular, systemic, or familial clues. Detected variants require contemporary classification and phenotypic correlation.