Discussion: View Thread

  • 1.  Recommendations for learning adaptive trial design?

    Posted 04-16-2014 10:16
    Hi Everyone, 

    I'm trying to come up to speed on adaptive trial design. Do any of you have textbooks or papers that you would recommend? I'm looking to get a good foundation in the theory so I can critically assess new methods and fit them to our needs. 

    Many thanks!

    Laila

    -------------------------------------------
    Laila Poisson
    Biostatistician
    Henry Ford Health System
    -------------------------------------------


  • 2.  RE:Recommendations for learning adaptive trial design?

    Posted 04-16-2014 10:49
    Dear Laila

    Good text books to start you off are:

    Bayesian Adaptive Methods for Clinical Trials by Berry, Carlin, Lee and Muller
    Adaptive Design Methods in Clinical Trials by Chow and Chang

    The DIA Adaptive Design Scientific Working Group published a series of papers in 2006 (when it was still under PhRMA auspices).  A good summary paper is:

    Adaptive Designs in Clinical Drug Development - an Executive Summary of the  PhRMA Working Group by Gallo et al, Journal of Biopharmaceutical Statistics, 16: 275-283, 2006

    But there were also a series of papers in the Drug Information Journal that year.

    I would also recommend reading:

    Good Practices for Adaptive Clinical Trials in Pharmaceutical Product Development, Gaydos et al, Drug Information Journal, Vol. 43, pp. 539-556, 200

    If you want to specifically focus on dose ranging then read the paper:

    Innovative Approaches for Designing and Analyzing Adaptive Dose-Ranging  Trials, Bornkamp et al, Journal of Biopharmaceutical Statistics, 17: 965-995, 2007

    Others may have suggestions but I think this is a good start.

    Kind regards

    Alun
    -------------------------------------------
    Alun Bedding
    Principal Statistical Scientist
    Roche Products Ltd
    -------------------------------------------








  • 3.  RE:Recommendations for learning adaptive trial design?

    Posted 04-16-2014 11:14
    Laila,

    If you Google that term you tend to get the phone book on things, you might have to be a little more specific in the area of adaptive designs you are looking at.  Most people that are looking for the new wave of stuff really want to look at trials with adaptive interventions.  I tend to think of this in contrast with other adaptive designs where the treatments don't adapt and the goal is to take iterative 'early' looks at the data to reduce total sample size or end early if results are really good or bad (i.e. Group Sequential Designs and such).  A great place to start with adaptive interventions is the Methodology Center at PSU (http://methodology.psu.edu/ra/adap-inter) which has info on SMART designs. 

    Hope this helps,

    Jason


    -------------------------------------------
    Jason Brinkley
    East Carolina University
    -------------------------------------------








  • 4.  RE:Recommendations for learning adaptive trial design?

    Posted 04-16-2014 12:40

    Alun, thanks for the reading list. I'm going to review these resources.  

    Jason, thanks for making that distinction! The concept of "adaptive trials" was brought up by a collaborator with respect to the I-SPY trials but adaptive interventions may be where we want to go. Thanks for this link. I see they have a reading list (current at least up to 2012). 



    -------------------------------------------
    Laila Poisson
    Biostatistician
    Henry Ford Health System
    -------------------------------------------






    Original Message


  • 5.  RE:Recommendations for learning adaptive trial design?

    Posted 04-16-2014 15:05
    Laila, An interesting set of responses and it highlights that the term "Adaptive Designs" means different things to different people.  Jason is referring to situations where the treatment regimen has moving parts.  So an example is using a treatment arm -- "start with A, if you see X use A+B, if you see Y go with A+C".  This means that the results of single patients changes how you treat them.  This is very cool stuff, and learning the best way to construct and learn adaptive therapies is quite interesting, but a very different topic from most of what Alun presented which was addressing a design that changes depending on the results of patients -- like changing randomization ratios, changing the patient population, seamlessly moving between trial designs, stopping early, extending longer, etc....  Both very interesting areas, but different topics.

    Given that the question arose in reference to iSpy -- it is likely pushing you in the direction of adaptive designs -- not adaptive treatments.  iSpy uses the former, not the latter.  Adaptive designs can use many different adaptations, like adaptive randomization, enrichment, early stopping for success and futility, using early biomarkers to forecast later events, etc... iSpy does all of that.  It's a complex adaptive trial.

    It is also the wave of the future -- it is in a general class of designs that are "platform trials".  It has multiple treatments in it at any given time -- and the set of treatments evolve as the trial evolves -- and heck, may never end!  As A and B leave, C and D enter, ...  There are many incredible advantages of platform trials, and I believe they are the inevitable future of clinical trials... we may all speak fondly of the days when pharma companies or academic centers started trials to study a drug.  Spent 2 years getting it running and then shut it down 2 years later...  In the future there will be platforms to study breast cancer, alzheimers, migraine, etc...

    Actually the future is kinda here...

    --Scott



    -------------------------------------------
    Scott Berry
    Berry Consultants
    -------------------------------------------






    Original Message