gms | German Medical Science

64. Jahrestagung der Deutschen Gesellschaft für Medizinische Informatik, Biometrie und Epidemiologie e. V. (GMDS)

Deutsche Gesellschaft für Medizinische Informatik, Biometrie und Epidemiologie

08. - 11.09.2019, Dortmund

Designing pediatric phase I trials in oncology based on data from adult trials

Meeting Abstract

  • Dario Orlando Zocholl - Charité - Universitätsmedizin Berlin, Berlin, Germany
  • Geraldine Rauch - Charité - Universitätsmedizin Berlin, Berlin, Germany
  • Manuel Wiesenfarth - German Cancer Research Center (DKFZ), Heidelberg, Germany
  • Annette Kopp-Schneider - German Cancer Research Center (DKFZ), Heidelberg, Germany

Deutsche Gesellschaft für Medizinische Informatik, Biometrie und Epidemiologie. 64. Jahrestagung der Deutschen Gesellschaft für Medizinische Informatik, Biometrie und Epidemiologie e.V. (GMDS). Dortmund, 08.-11.09.2019. Düsseldorf: German Medical Science GMS Publishing House; 2019. DocAbstr. 256

doi: 10.3205/19gmds085, urn:nbn:de:0183-19gmds0853

Veröffentlicht: 6. September 2019

© 2019 Zocholl et al.
Dieser Artikel ist ein Open-Access-Artikel und steht unter den Lizenzbedingungen der Creative Commons Attribution 4.0 License (Namensnennung). Lizenz-Angaben siehe



In a phase I trial, a drug is administered to members of the target population for the first time. In oncology, the participants are patients instead of healthy volunteers, so there is a great need for highly efficient trial designs. The paradigm in oncological therapy is that efficacy and toxicity are related.

Hence, the aim of a phase I trial is to identify the maximum tolerable dose (MTD) from a set of doses to achieve a good trade-off between safety and efficacy and to build the basis for efficient drug dosing in subsequent phase II trials and potential off-label use, which is common practice in pediatric health care [1]. However, in oncology, pediatric phase I trials are rarely conducted, therefore pediatric drug dosing often relies on arbitrary rules of thumb based on the dose for adults. The lack of pediatric trials is due to very small sample sizes and ethical concerns. Although dose-finding methods like the Continual Reassessment Method (CRM) [2] and Bayesian regression models with one or two parameters [3] are designed for dose-finding in small samples of 20-30 patients, the requirements of pediatric trials are even stricter. In many potential applications in oncology, the desired sample size limit is distinctly below 20 pediatric patients, which limits the number of doses that can be explored and reduces the amount of information that can be collected.

In our study, we examine potential design adjustments of phase I trials that may allow drawing reliable conclusions from trials with approximately 10 patients. Promising approaches are particularly reducing the range of doses based on previous adult trials and borrowing information within the Bayesian framework. A pediatric dosing scheme may be based on the MTD identified by previous adult trials and a clinically reasonable range of doses around this MTD. Information from adult trials can be borrowed in form of an informative prior distribution for the pediatric trial, which expresses some degree of uncertainty about the maximum tolerable dose. The prior distribution's impact can be adjusted by applying a weighting parameter to take into account the degree of similarity between adult and pediatric patients. This weighting parameter can be specified beforehand based on clinical knowledge or it can be estimated using partial borrowing approaches [4]. The methods are compared in an extensive simulation by simulating adult and subsequent pediatric trials under various realistic settings. Real-world applicability of the design will be discussed as well as practical aspects of software implementation.

The authors declare that they have no competing interests.

The authors declare that an ethics committee vote is not required.


Gore R, Chugh PK, Tripathi CD, Lhamo Y, Gautam S. Pediatric off-label and unlicensed drug use and its implications. Current clinical pharmacology. 2017 Feb 1;12(1):18-25.
O’Quigley J, Pepe M, Fisher L. Continual reassessment method: a practical design for phase 1 clinical trials in cancer. Biometrics. 1990 March;46(1):33–48.
Neuenschwander B, Branson M, Gsponer T. Critical aspects of the Bayesian approach to phase I cancer trials. Statistics in medicine. 2008 Jun 15;27(13):2420-39.
Cunanan KM, Koopmeiners JS. Hierarchical models for sharing information across populations in phase I dose-escalation studies. Statistical methods in medical research. 2018 Nov;27(11):3447-59.