Progress in biomedical science leads to highly effective targeted therapies in certain diseases, such as specific types of cancer or multiple sclerosis. These targeted therapies offer large benefits to often a small subpopulation of all patients affected by these specific diseases. Rare diseases and pediatric indications can also present a similar challenge inherent in their small population sizes.
For a drug development process, conducting a trial in a small patient population poses multiple challenges:
- The number of patients available for showing the effectiveness of the therapy in a clinical trial is low
- Due to the low numbers of patients available, the recruitment time in a clinical trial increase
- In order to include enough patients, it might be necessary to increase the number of participating clinical institutions, which increases the complexity and costs of a trial
- Very often there is large heterogeneity in these populations
Several statistical methods/designs can be used to minimise problems and the number of patients required for a trial:
- Designs where each patient serves as his/her own control, e.g., Cross-Over-Designs and N-of-1 trials
- Designs where the control group is not concurrently recruited but taken from data generated in the past, e.g., register data
- Designs that allow changes in the protocol of an ongoing trial, e.g., sequential designs or adaptive designs
For any study, careful consideration should be given to choosing the most suitable design in order to achieve the maximum statistical benefit.
The most important questions for choosing the best trial design in a specific situation are:
- Is the disease under investigation a chronic one or a progressive one?
- Are there any historical data available?
The nature of the disease under investigation – whether it is a chronic disease, which remains relatively stable over a certain time period and will return to a comparable situation once treatment is stopped, e.g., Diabetes Mellitus or Asthma, or whether it is a progressive disease, which will not return to a comparable situation once treatment is stopped, e.g., Cancer or Appendicitis with resection – is important to decide whether each patient can serve as his/her own control.
In this article, we will look at the pros and cons of designs that can be effectively used in order to overcome challenges posed by clinical trials in small patient populations with chronic and progressive diseases.
Cross-over design (chronic disease)
In a cross-over design, each patient serves as his /her own control and is treated in a pre-specified sequence on two different occasions with the compound under investigation and the comparator. Therefore, heterogeneity is addressed, and the sample size can be minimised. Possible carry-over effects, i.e. that treatment in period 1 still has an impact on period 2, may complicate the analysis.
Each patient serves as his/her own control in this design. The sequence of treatments is not pre-specified. More than one judgment of effect per patient is possible. This design is most effective if an endpoint that can be measured after a short time is appropriate.
Historical controls (chronic disease)
In this design the control group is taken from historical data, e.g. collected in a register and compared to prospective treated patients. This design is most appropriate if randomisation is unethical or not possible. We advise particular attention to the comparability of the groups, not only with respect to demographic and disease data but also to the evaluation method used for the endpoint.
Adaptive designs (progressive disease)
Adaptive designs minimise total development time and reduce the number of patients required in a trial as shown in the figure above. Sample size can be adapted to effect size at the interim analysis, which can also include a stop for futility. One or more interim analyses are necessary.
Sequential designs (progressive disease)
Sequential designs, either group sequential or full sequential, are most appropriate if a sponsor expects a large difference between treatments because they allow stopping for futility or success very early. A high administrative burden related to the multiple interim analyses compared to other designs has to be taken into account.
The Small Population Clinical Trials Task Force (IRDiRC) gives additional recommendations in order to minimise the number of patients in a trial:
- Collect natural history and patient registry data for rare diseases for the design of clinical trials.
- Trials should be long enough for complete follow up and patients should stay in trials for as long as possible
- Do not dichotomise continuous endpoints
- Use composites endpoints, by combining several outcomes into a single outcome measure, thereby increasing the number of events
- Use analysis of covariance (ANCOVA) instead of simple “change from baseline” analyses
- Make full use of longitudinal data: Analysis methods for repeated measurements lead to a potential reduction of sample size by 30% versus change score analysis
- EMA/CHMP: Guideline on clinical trials in small populations. Retrieved from: https://www.ema.europa.eu/documents/scientific-guideline/guideline-clinical-trials-smallpopulations_en.pdf (2006).
- FDA: Rare Diseases – Common Issues in Drug Development – Guidance for Industry. Retrieved from: https://www.fda.gov/downloads/Drugs/GuidanceComplianceRegulatoryInformation/Guidances/ucm458485.pdf
- Kravitz RL, Duan N, eds, and the DEcIDE Methods Center N-of-1 Guidance Panel (Duan N, Eslick I, Gabler NB, Kaplan HC, Kravitz RL, Larson EB, Pace WD, Schmid CH, Sim I, Vohra S). Design and Implementation of N-of-1 Trials: A User’s Guide. AHRQ Publication No. 13(14)-EHC122-EF. Rockville, MD: Agency for Healthcare Research and Quality; February 2014. Retrieved from: effectivehealthcare.ahrq.gov/N-1-Trials.cfm.
- Neuenschwander B, Capkun-Niggli G, Branson M, Spiegelhalter DJ. Summarizing historical information on controls in clinical trials. Clinical Trials. 2010 Feb;7(1):5-18
- Design of randomized controlled confirmatory trials using historical control data to augment sample size for concurrent controls. Yuan J, Liu J, Zhu R, Lu Y, Palm U. J; Biopharm Stat. 2019;29(3):558-573.
- Lim J, Walley R, Yuan J, Liu J, Dabral A, Best N, Grieve A, Hampson L, Wolfram J, Woodward P, Yong F, Zhang X, Bowen E. Minimizing Patient Burden Through the Use of Historical Subject-Level Data in Innovative Confirmatory Clinical Trials: Review of Methods and Opportunities. Ther Innov Regul Sci. 2018 Sep;52(5):546-559.
- Deepak L. Bhatt, M.D., M.P.H., and Cyrus Mehta, Ph.D.: Adaptive Designs for Clinical Trials N Engl J Med 2016;375:65-74.
- Cyrus Mehta, Ph.D.; Ping Gao, Ph.D.; Deepak L. Bhatt, MD, MPH; Robert A. Harrington, MD; Simona Skerjanec, PharmD; James H. Ware, Ph.D.: Optimizing Trial Design – Sequential, Adaptive, and Enrichment Strategies. Circulation. 2009;119:597-605. Retrieved from: http://circ.ahajournals.org/content/119/4/597