Statistical Considerations for Rare Disease Clinical Trials

Organizers: Yeh-Fong Chen (FDA), Freda Cooner (Amgen), Jean Pan (Amgen)
Chair: Yeh-Fong Chen (FDA)
Vice Chair: Miaomiao Ge (Boehringer-Ingelheim)

Speakers:
Anita Zaidi (FDA)
Xiang Yin (Medidata)
Qing Liu (QRmedSci)
Ben Saville (Berry Consultants)
Ming-Dauh Wang (Bayer)

Drug development in rare diseases presents many challenges as well as opportunities for innovation and adoption of novel approaches. The substantial evidence of effectiveness has generally been interpreted as two adequate and well-controlled trials. This requirement could also be met by a single trial plus confirmatory evidence which may be more feasible in the rare disease population. However, the limited population of patients and limited availability of natural history information often hinders the ability to design an adequate and well controlled trial. Some rare diseases primarily (or exclusively) affect children and, thus, additional ethical considerations of enrolling children in clinical investigations come into play and need to be considered carefully. Lastly, selecting and validating clinical endpoints for demonstration of drug efficacy is often exceedingly difficult for various reasons and the use of biomarkers can become important but also includes complex considerations as a substitute to clinical evidence of efficacy.

Abstracts:

Title: Drug Development in Rare Diseases
Speaker: Anita Zaidi (FDA)

Drug development in rare diseases presents many challenges as well as opportunities for innovation and adoption of novel approaches. The substantial evidence of effectiveness has generally been interpreted as two adequate and well-controlled trials. This requirement could also be met by a single trial plus confirmatory evidence which may be more feasible in the rare disease population. However, the limited population of patients and limited availability of natural history information often hinders the ability to design an adequate and well controlled trial. Some rare diseases primarily (or exclusively) affect children and, thus, additional ethical considerations of enrolling children in clinical investigations come into play and need to be considered carefully. Lastly, selecting and validating clinical endpoints for demonstration of drug efficacy is often exceedingly difficult for various reasons and the use of biomarkers can become important but also includes complex considerations as a substitute to clinical evidence of efficacy.

Title: Synthetic control arm (SCA) as external controls in drug development for rare diseases
Speaker: Xiang Yin (Medidata)

In the drug development for treatments of rare disease, often it is impractical or unethical to conduct a randomized clinical trial with a concurrent control. External data sources such as historical clinical trials and real-world data can be considered to replace or augment a concurrent control. Synthetic control arm (SCA), created from patient-level data of historical clinical trials in the same disease indication with demographics and disease characteristics balanced to the patients participating the current trial investigating a new treatment, can be considered as an alternative to randomized control in rare, serious or life-threatening diseases with unmet medical needs. In this presentation, I will include some case studies where we used propensity score methodology to create SCA as external controls in drug development.

Title:  On Randomized Delayed-Start Design with Integrated Analysis of Efficacy for Drug Development in Rare Diseases
Speaker:  Qing Liu (QRmedSci)

A randomized delayed-start design consists of two-stages where patients in the first stage are randomized to receive a new treatment or a control and in the second stage patients randomized to receive the control in the first stage switch to receive the new treatment.  We develop an integrated analysis of efficacy (IAE) where efficacy from stage one from a parallel group comparison is combined with efficacy from second stage from intra-patient comparison. As a result of IAE, the trial is 55% to 80% more efficient than a traditional parallel group design, which is often not feasible for rare disease drug development.

Title: Drug development for Rare Diseases: A Bayesian disease progression model for Amyotrophic lateral sclerosis
Speaker: Ben Saville (Berry Consultants)

Amyotrophic lateral sclerosis (ALS) is a rare progressive neurodegenerative disease that affects the neurons controlling voluntary muscles, and typically leads to a patient’s death within 2-5 years from time of diagnosis.  To date, clinical trials have been largely unsuccessful in finding effective therapies to slow or stop ALS disease progression.  Given the large number of candidate therapies awaiting testing, there is a great need for innovative clinical trial designs that can accelerate the drug development process.  In collaboration with the Healey Center (Boston, MA), we have designed an innovative adaptive platform trial in which multiple drugs are both simultaneously and serially evaluated under a single master protocol.   The design includes sharing of placebo participants across treatments and frequent interim analyses to allow early stopping for success or futility of individual treatments.  A Bayesian primary analysis model quantifies a treatment benefit as a proportional slowing of disease progression using an ALS functional rating score (ALSFRS-R) over 6 months.  The model accounts for patient-level heterogeneity of disease progression, potential differences in placebo between treatment regimens, staggered entry of treatment arms, and baseline prognostic factors.  Simulation studies are used to assess the performance of the design, including statistical power, average sample sizes, probability of early stopping, and Type I error.  Compared to traditional drug development, we estimate that the ALS Platform Trial will find an effective therapy more quickly (average 3.4 vs. 8.5 years), with fewer total participants (average 880 vs. 1400), and fewer participants on placebo (average 220 vs. 700).

Title: Alternative statistical considerations for rare disease trials
Speaker: Ming-Dauh Wang (Bayer)

Clinical trials for rare diseases are typically smaller in sizes due to limitations inherent from small patient populations. In many a case, conventional large-sample statistical methods for judging whether a trial has significantly demonstrated intended efficacy and safety cannot be practically applied. Thus, reasonable modifications of conventional approaches or completely different statistical considerations are needed for making such trials implementable. Such alternatives are in no means to lower statistical standard, but to show them more appropriate and justifiable for the situations concerned. I will review a spectrum of rare disease programs at Regeneron and their distinct features for alternative statistical considerations. Interactions with regulatory agencies, particularly with FDA, will be shared to illustrate how non-conventional statistical perspectives can be reached between the sponsor and a regulatory agency for rare disease trials.