122IN - The evolution of clinical trials in oncology: Randomised controlled trials to real world studies

Date 29 September 2014
Event ESMO 2014
Session ESMO-ASCO: The evolution of the clinical trial landscape
Topics Bioethics, Legal, and Economic Issues
Translational Research
Basic Principles in the Management and Treatment (of cancer)
Presenter Gary Lyman
Citation Annals of Oncology (2014) 25 (suppl_4): iv41-iv42. 10.1093/annonc/mdu317
Authors G.H. Lyman
  • Public Health Sciences, Fred Hutchinson Cancer Research Center, 98109 - Seattle/US




Randomized controlled trials (RCTs) continue to represent the gold standard for the evaluation of comparative diagnostic, therapeutic and preventative strategies. However, generalizable data from large, well-designed RCTs are often unavailable or conflicting on major clinical questions in oncology. In addition, a number of important limitations to RCTs are recognized including high costs, time to completion, narrow eligibility criteria, and limited attention to treatment-related toxicities and quality of life. Therefore, RCTs often do not adequately address effectiveness and safety in the broader cancer population and important safety issues may not emerge until years later. A variety of approaches to making RCTs more relevant, generalizable, and rapid and to reduce their complexity and costs have been proposed. Nevertheless, the limitations of RCTs necessitate consideration of a broad range of research approaches in an effort to gather reliable evidence on critical questions related to the comparative benefit, harm, and value of available interventions. When properly applied, such approaches may provide reasonable, valid, and more generalizable estimates of comparative effectiveness, safety, and cost and may also generate hypotheses that form the basis of future confirmatory RCTs. However, non-randomized approaches such as prospective and retrospective observational studies and clinical models come with their own set of important limitations. Such studies can readily be biased by the inappropriate section of patients as well as controls and, as the intervention is not randomly assigned, the results may be influenced by confounding factors impacting on both treatment selection as well as clinical outcome. Clearly, there is a need to design better, more efficient, relevant, and generalizable RCTs and to complement them with other comparative effectiveness approaches utilizing creative but equally rigorous methods while understanding and disclosing the limitations of those methods and the conclusions derived. Such results may provide patients, providers, and policy makers with the most reliable information upon which to base valid recommendations on optimal management of patients with cancer.


The author has declared no conflicts of interest.