Evaluating the Effect of Sodium Phenylbutyrate and Taurursodiol on Survival Outcomes in Amyotrophic Lateral Sclerosis Using a Propensity Score–Matched Cohort From the Pooled Resource Open-Access ALS Clinical Trials (PRO-ACT) Database
Sabrina Paganoni1, Melanie Quintana2, Yuehui Wu3, Jamie Timmons3, Merit Cudkowicz1
1Sean M. Healey and AMG Center for ALS & the Neurological Clinical Research Institute, Massachusetts General Hospital, Harvard Medical School, Boston, MA, USA, 2Berry Consultants, LLC, Austin, TX, USA, 3Amylyx Pharmaceuticals, Inc., Cambridge, MA, USA
Objective:

To assess the potential effect of placebo-to-active crossover on intent-to-treat (ITT) overall survival results from the CENTAUR trial evaluating an oral, fixed-dose sodium phenylbutyrate and taurursodiol combination (PB&TURSO) in amyotrophic lateral sclerosis (ALS), a post hoc analysis was performed utilizing an external, propensity score–matched (PSM), PB&TURSO-naïve cohort from the Pooled Resource Open-Access ALS Clinical Trials (PRO-ACT) database as a historical control comparator.

Background:

CENTAUR encompassed a 6-month randomized placebo-controlled phase and a long-term open-label extension (OLE) phase. ITT analysis at trial conclusion (maximum postrandomization follow-up, 44 months) showed a 4.8-month longer median survival time in those originally randomized to PB&TURSO (23.5 mo) versus placebo (18.7 mo; hazard ratio [HR]=0.64; 95% CI, 0.42–1.00; P=.048). Notably, 71% of placebo-randomized participants entered the OLE phase and crossed over to active treatment. Such placebo-to-active crossover may lead to underestimation of survival benefit of investigational therapies in trials incorporating this design. Comparison with an external control group may provide additional context beyond observed survival outcomes in this setting.  

Design/Methods:

PRO-ACT participants were control subjects from historical trials who met major enrollment criteria from CENTAUR and had known mortality information. PB&TURSO-randomized participants from CENTAUR (n=89) and historical control participants (n=85) were propensity score matched using covariates of prognostic significance. Survival probability through the date of final participant visit in CENTAUR was compared between groups using a Cox proportional hazards model.

Results:

PSM covariates were generally well balanced between groups. Median survival duration was 10.39 months longer in PB&TURSO-randomized participants from CENTAUR (23.54 months) versus historical controls (13.15 months; HR=0.48; 95% CI, 0.31–0.72; P=.00048).

Conclusions:

The results of this analysis suggest a greater survival benefit with PB&TURSO than seen on ITT analysis in CENTAUR. The survival benefit attributed to PB&TURSO in this analysis aligns with prior analyses using statistical models controlling for placebo-to-active crossover in CENTAUR.

10.1212/WNL.0000000000201898