Development of a Rasch-Built Amyotrophic Lateral Sclerosis Impairment Multidomain Scale to Measure Disease Progression in ALS

Abstract

Background and objectives: Current scales used in amyotrophic lateral sclerosis (ALS) attempt to summarize different functional domains or "dimensions" into 1 overall score, which may not accurately characterize the individual patient's disease severity or prognosis. The use of composite score risks declaring treatments ineffective if not all dimensions of ALS disease progression are affected equally. We aimed to develop the ALS Impairment Multidomain Scale (AIMS) to comprehensively characterize disease progression and increase the likelihood of identifying effective treatments.

Methods: The Revised ALS Functional Rating Scale (ALSFRS-R) and a preliminary questionnaire, based on literature review and patient input, were completed online by patients from the Netherlands ALS registry at bimonthly intervals over a period of 12 months. A 2-week test-retest, factor analysis, Rasch analysis, and a signal-to-noise optimization strategy were performed to create a multidomain scale. Reliability, longitudinal decline, and associations with survival were evaluated. The sample size required to detect a 35% reduction in progression rate over 6 or 12 months was assessed for a clinical trial that defines the ALSFRS-R or AIMS subscales as a primary endpoint family.

Results: The preliminary questionnaire, consisting of 110 questions, was completed by 367 patients. Three unidimensional subscales were identified, and a multidomain scale was constructed with 7 bulbar, 11 motor, and 5 respiratory questions. Subscales fulfilled Rasch model requirements, with excellent test-retest reliability of 0.91-0.94 and a strong relationship with survival (p < 0.001). Compared with the ALSFRS-R, signal-to-noise ratios were higher as patients declined more uniformly per subscale. Consequently, the estimated sample size reductions achieved with the AIMS compared with those achieved with the ALSFRS-R were 16.3% and 25.9% for 6-month and 12-month clinical trials, respectively.

Discussion: We developed the AIMS, consisting of unidimensional bulbar, motor, and respiratory subscales, which may characterize disease severity better than a total score. AIMS subscales have high test-retest reliability, are optimized to measure disease progression, and are strongly related to survival time. The AIMS can be easily administered and may increase the likelihood of identifying effective treatments in ALS clinical trials.

Figure 1. Flowchart of Study Design and Questionnaire Development Steps

The flowchart shows the study and questionnaire development steps used in this study. A preliminary questionnaire with 110 questions was sent to patients with ALS enrolled in the Netherlands ALS registry. Test-retest reliability, longitudinal follow-up, Rasch analysis, and optimization to measure disease progression using longitudinal data were performed to create the final unidimensional bulbar, motor, and respiratory subscales of the AIMS. AIMS = ALS Impairment Multidomain Scale; ALS = amyotrophic lateral sclerosis; ALSFRS-R = Revised ALS Functional Rating Scale.

Figure 2. Number of Questions vs Required Sample Size per Subscale

The required sample size for a 12-month clinical trial to detect a 35% reduction in progression rate with 80% power was used as an estimate of the sensitivity of candidate subscales to measure disease progression. For each number of questions, the combination of questions that resulted in the lowest required sample size is plotted. Solid rectangles indicate candidate subscales that result in the lowest (+5%) sample size. Initially, as questions are added, information increases, thus reducing the required sample size. However, as more similar questions are subsequently added, between-patient variability in the rate of decline increases, thus inflating the sample size required to detect treatment effects.

Figure 3. AIMS Question Locations and Thresholds

The figure shows the AIMS, question difficulty order, and targeting of each question. The most difficult question (e.g., chewing and swallowing) is on the far right, while the easiest question (e.g., respiratory insufficiency) is on the far left. The x-axis represents the logit measure for a patient's bulbar, motor, or respiratory disability. Worse question options sequentially become more probable as disability increases. AIMS = ALS Impairment Multidomain Scale; ALS = amyotrophic lateral sclerosis.

Figure 4. Association of AIMS Subscales With the ALSFRS-R and Survival Time

Bulbar, motor, and respiratory AIMS scores were strongly associated with the corresponding ALSFRS-R subscales (A–C) and survival time (D–F). There was a dose-response association between baseline scores and overall survival, lower scores resulting in lower survival probabilities after enrollment. AIMS = ALS Impairment Multidomain Scale; ALS = amyotrophic lateral sclerosis; ALSFRS-R = Revised ALS Functional Rating Scale.

Figure 5. Example Reporting of the AIMS in ALS Clinical Studies

In this example study, the AIMS subscales were used as primary endpoint. By weighing each of the domains, for example, according to patient, physician, or prognostic preferences, one can obtain 1 overall effect size with 1 overall p value. A similar approach is used by the Combined Assessment of Function and Survival, thereby prioritizing one outcome (survival time) over the other (ALSFRS-R). This approach results in 1 overall effect size; in this case, the probability or odds of having a better outcome when receiving the new therapy compared with that when receiving placebo. Other strategies for analyzing multiple domains are summarized in eTable 3 (links.lww.com/WNL/C898). AIMS = ALS Impairment Multidomain Scale; ALS = amyotrophic lateral sclerosis; ALSFRS-R = Revised ALS Functional Rating Scale.