The value of database controls in pilot or futility studies in ALS

Objective: To evaluate the use and reliability of database controls in place of a placebo group in pilot or "futility" ALS trials. Methods: We compared the rates of disease progression in the placebo arm of the clinical phase III US Insulin-like Growth Factor-I Trial (n = 90) with the rates of disease progression of 207 patients with ALS selected from 1,600 ALS database patients using the same inclusion criteria. Results: The mean rates of change in the Appel ALS (AALS) score were nearly identical in the placebo group (4.70 points/month) and in the database group (4.79 points/month). In addition, there was no significant difference in the median time to achieving a 20-point progression in AALS score: 143 days for database match vs 146 days for the placebo group (log rank p = 0.88). Furthermore, in the multivariate Cox analysis, both the rate at which the disease had progressed prior to first exam (preslope) (p < 0.001) and first exam AALS total score (p = 0.01) were shown to be covariates of subsequent rate of disease progression. Conclusion: The similarity in disease progression between placebo arm of clinical phase III trial and matched database group suggests the value of historical databases in futility trials. However, the proposed study design requires appropriate matching of study patients with database controls. Based on our results, we suggest matching by stage of the disease and rate of clinical decline in a contemporaneous ALS population.