Abstract:
|
Enrolling adequate patients in clinical trials for rare diseases can be challenging. Therefore, historical controls obtained through well conducted natural history studies may be considered. In recent years, FDA has approved therapies using historical controls, including asfotase alfa for perinatal/infantile- and juvenile-onset hypophosphatasia (HPP) in 2015 and cerliponase alfa for CLN2 disease in 2017. When considering historical controls, concerns include: the comparability of baseline characteristics and assessments in the proposed intervention and historical controls; and the ability to minimize bias due to the non-randomized setting. Potential confounders need to be considered and missing data in the natural history study may also be relevant. To overcome these challenges, we must carefully examine the data using different statistical methods and models. A robust well constructed sensitivity analysis plan is needed to evaluate the trial evidence. Several statistical and graphical methods will be used to illustrate potential solutions, including covariate-adjusted survival analyses; matching analyses using propensity scores; Fisher's exact and McNemar tests.
|