Longitudinal motor function and biomarker correlates in treated adult spinal muscular atrophy: a single-center cohort study.

BACKGROUND

Long-term clinical trajectories and accessible biomarkers in adult spinal muscular atrophy (SMA) remain insufficiently characterized under real-world treatment conditions. We conducted a single-center longitudinal study to evaluate motor function, genotype-phenotype relationships, and clinically accessible biomarkers in treated adults with SMA.

METHODS

Twenty-three Japanese adults with genetically confirmed 5q-SMA treated with nusinersen or risdiplam (2018-2025) underwent repeated assessments of Revised Upper Limb Module (RULM), Hammersmith Functional Motor Scale Expanded (HFMSE), and laboratory, physiological, and electrophysiological measures. Associations were explored using correlation analyses, linear mixed-effects models, and exploratory machine-learning approaches.

RESULTS

Higher SMN2 copy number was associated with later onset and milder severity. Hybrid SMN alleles showed heterogeneous patterns depending on copy-number context. Longitudinally (mean follow-up 57 months), motor scores often improved or stabilized during the first year after treatment initiation, followed by plateau or gradual decline in some patients. In mixed-effects models, vital capacity (%VC) was independently associated with RULM, while other biomarkers showed consistent associations with motor scores. Exploratory machine-learning suggested that ulnar compound muscle action potential (CMAP) amplitude and creatine kinase (CK) contributed to model predictions, although baseline features did not reliably predict short-term motor changes.

CONCLUSION

In treated adults with SMA, simple clinical measures such as %VC, CK, ulnar CMAP amplitude are associated with motor status and may support routine monitoring, while prediction of treatment responsiveness remains challenging.