A model incorporating ultrasound to predict the probability of fast disease progression in amyotrophic lateral sclerosis

Abstract

ObjectiveWe aimed to develop a model to predict amyotrophic lateral sclerosis (ALS) disease progression based on clinical and neuromuscular ultrasound (NMUS) parameters. MethodsALS patients were prospectively recruited. Muscle fasciculation (≥2 over 30-seconds, examined in biceps brachii-brachialis (BB), brachioradialis, tibialis anterior and vastus medialis) and nerve cross-sectional area (CSA) (median, ulnar, tibial, fibular nerve) were evaluated through NMUS. Ultrasound parameters were correlated with clinical data, including revised ALS Functional Rating Scale (ALSFRS-R) progression at one year. A predictive model was constructed to differentiate fast progressors (ALSFRS-R decline ≥ 1/month) from non-fast progressors. Results40 ALS patients were recruited. Three parameters emerged as strong predictors of fast progressors: (i) ALSFRS-R slope at time of NMUS (p = 0.041), (ii) BB fasciculation count (p = 0.027) and (iii) proximal to distal median nerve CSA ratio < 1.22 (p = 0.026). A predictive model (scores 0–5) was built with excellent discrimination (area under curve: 0.915). Using a score of ≥ 3, the model demonstrated good sensitivity (81.3%) and specificity (91.0%) in differentiating fast from non-fast progressors. ConclusionThe current model is simple and can predict the probability of fast disease progression. SignificanceThis model has potential as a surrogate biomarker of ALS disease progression.