Predicting prognosis in amyotrophic lateral sclerosis: a simple algorithm

Abstract

The objective of the study was to develop and validate a practical prognostic index for patients with amyotrophic lateral scleroses (ALS) using information available at the first clinical consultation. We interrogated datasets generated from two population-based projects (based in the Republic of Ireland and Italy). The Irish patient cohort was divided into Training and Test sub-cohorts. Kaplan–Meier methods and Cox proportional hazards regression were used to identify significant predictors of prognoses in the Training set. Using a weighted grading system, a prognostic index was derived that separated three risk groups. The validity of index was tested in the Irish Test sub-cohort and externally confirmed in the Italian replication cohort. In the Training sub-cohort (n = 117), significant predictors of prognoses were site of disease onset (HR = 1.7, p = 0.012); ALSFRS-R slope prior to first evaluation (HR = 2.8, p < 0.0001), and executive dysfunction (HR = 2.11, p = 0.001). The risk group system generated using these results predicted median survival time in the Training set, the Test set (n = 87) and the Italian cohort (n = 122) with no overlap of the 95 % CI (p < 0.0001). In the validation cohorts, a high-risk classification was associated with a positive predictive value for poor prognosis of 73.3–85.7 % and a negative predictive value (NPV) for good prognosis of 93.3–100 %. Classification into the low-risk group was associated with an NPV for bad prognosis of 100 %. A simple algorithm using variables that can be gathered at first patient encounter, validated in an independent patient series, reliably predicts prognoses in ALS patients.