Operationally Seamless & Inferentially Seamless Adaptive Designs

Posted by Esha Senchaudhuri

Dec 4, 2014 11:30:00 AM

Croton_Lechleri_to_Crofelemer_Molecule

Fulyzaq® from Napo/Salix was the first drug ever to be approved using an adaptive two-stage "seamless" clinical trial design. However, when Napo Pharmaceuticals sought orphan drug status for Crofelemer (made from the croton lechleri plant, pictured above left), it received a short window from the FDA in which to complete a Phase 3 trial for safety and efficacy. Aiming to diminish both time and cost, Napo submitted an adaptive design for a Crofelemer trial to the FDA. Initially, the design was rejected on grounds that it would not demonstrate strong control of the type-1 error rate.

Napo then approached Cytel consultants which advised it to consider a seamless adaptive trial. A seamless adaptive trial combines 2 phases of a study into one trial, thereby allowing a trial to complete within a short window. Using appropriate techniques, it is possible to conduct such studies while maintaining strong control of type-1 error.  

A seamless adaptive design may be operationally seamless or inferentially seamless. An operationally seamless design is one in which a confirmatory trial proceeds after an exploratory one, but the data from the two are kept distinct. By contrast, an inferentially seamless trial combines data from both phases to make the final inference. Due to these two varieties of seamless adaptive designs, Napo/Salix had the following 3 trial designs from which to choose:

Trial 1: A Single Four Arm Trial
A single four arm trial which would use the Bonferroni-Holm Test and require enrolling 520 patients.  

Trial 2: An Operationally Seamless Adaptive Design

An operationally seamless design would involve a four arm learning trial with 50 patients in each arm, from which the sponsor could select the best dose during interim analysis. Moving forward with this dose, the second phase of the adaptive trial would involve a confirmatory trial with two arms, each of which would have 108 patients. As a result, the design would require enrolling 416 patients.

Trial 3: An Inferentially Seamless Adaptive Design
   
The inferentially seamless design (also called an integrated design) would consist of a two stage trial in which the first stage selected the dose that the second stage would then confirm. As in the operationally seamless trial, 50 patients would enroll into each of four arms during the first stage of the trial. The second stage would continue with 100 of these initial 200 patients (50 from the placebo group and 50 from the active arm), but then add an additional 75 patients to both arms of the trial. As a result,  the inferentially seamless trial would only require enrolling 350 patients for the entire trial.  

 

Napo Pharmaceuticals eventually chose to use the inferentially seamless design, which combined information from Phase 2 and Phase 3 using a method called the Posch Method  (Statistics in Medicine, 2005)The results of the ADVENT trial phase 3 were announced in Novemeber 2010, and an NDA submission was made in December 2011. The FDA granted Crofelemer Priority Review Status in February 2012 and full approval in December 2012. 

Read Case Study


 

Related Items of Interest
Scott Harris speaks on Fulyzaq Trial Design
 
 
 
Posch, Martin, et al. "Testing and estimation in flexible group sequential designs with adaptive treatment selection." Statistics in Medicine 24.24 (2005): 3697-3714.

 

Topics: Sample Size, Trial Design, Safety, type 1 error, Dose-Finding

The Cytel blog keeps you up to speed with the latest developments in biostatistics and clinical biometrics.  Sign up for updates direct to your inbox. You can unsubscribe at any time.

 

Posts by Topic

see all

Recent Posts