P46. Modelling individual amyotrophic lateral sclerosis (ALS) disease courses in the PRO-ACT data base using the D50 progression model

Abstract

Introduction Evidence shows that ALS progression is curvilinear and heterogenous. However, the routinely used progression rate ([PR = (48-ALSFRS-R/disease duration)] only reflects progression at a circumscribed time point rather than across the disease. It is crucial to develop indices that can describe individual progression over the disease course. Objectives Our objective was to develop a model that uses regularly collected ALSFRS-R scores to describe individual events within the ALS disease course, thereby reducing noise associated with the ALSFRS-R. Methods We used a sigmoidal decay function to describe the transition from full health to maximum disease for patients in the PRO-ACT database ( Atassi, 2014 ). The model yields three parameters: D50 = time taken for ALSFRS-R to reach 24, dx = slope of ALSFRS-R decrease, relative D50 (rD50) = calculated value describing individual disease covered in reference to D50, where 0 and 0.5 indicate onset and the time point of halved functionality, respectively. D50 was used to mathematically derive disease phases: Phase I (early semi-stable phase), Phase II (early progressive phase), and Phases III and IV (late progressive and late semi-stable phase). Results D50, dx, and rD50 for 4838 patients were determined using ALSFRS-R scores and disease duration from onset to ALSFRS-R date. The relationship between D50 and dx was highly linear (r = 0.794), indicating that the entire disease course can be described using D50 alone. A significant correlation between D50 and survival was observed (r = 0.870). Although significant correlations were observed between D50 and log transformed first (r = −0.931) and last (r = −0.794) recorded PRs, the difference in the correlations indicates that PR varies greatly over time and introduces a time dependent bias. When selecting for patients for whom the model obtained a high quality fit to actual data, riluzole treatment significantly prolonged mean D50 (45.7 vs. 41.5 months). Using rD50, we were able to show progressive worsening of forced and slow vital capacity for most patients. Statistical significance was set at p Conclusion D50, rD50 and dx enable comparisons between vastly different time scales of disease courses and also enable the staging of individual events. This may aid the discovery of early prognostic markers and bolster understanding of the pathomechanisms underlying the heterogeneity in ALS progression. rD50 offers an alternative reference point to survival, given that the latter is often a function of the quality of care the patient receives, rather than genuine disease progression. The model is a crucial first step towards integrating the abundant complexity at play in ALS and may help to reduce data noise that may be responsible for the lack of findings from previous clinical trials. Acknowledgment This work is supported by BMBF in the framework of the E-RARE programme (PYRAMID) and JPND (OnWebDUALS).