Lead Investigator: Yasuhiro Hagiwara, The University of Tokyo
Title of Proposal Research: Adjusting for treatment switching in Oncology Trials: Sensitivity Analysis for Rank-Preserving Structural Failure Time Model
Vivli Data Request: 10052
Funding Source: Grant 22K17858 from the Japan Society for the Promotion of Science, KAKENHI
Potential Conflicts of Interest: None
Summary of the Proposed Research:
Health technology assessment (HTA) is used by the health regulatory bodies in the country in question to determine the pricing and reimbursement to the drug companies supplying / marketing the drugs. Generally, evidence from clinical trials is used as the gold standard for evaluating the efficacy of new drugs in HTA. In typical clinical trials, patients are randomly assigned to either the new treatment or the existing treatment, and the treatment effect is estimated by simply comparing outcomes. However, in many trials designed to develop anticancer drugs, the assessment of treatment effect could be complicated because patients assigned to an existing treatment often switch to a new treatment during the course of trials, which is knowns as “treatment switching”. To evaluate the treatment effects of new drugs, statistical adjustment for treatment switching is necessary.
To date, several statistical approaches are available to adjust for treatment switching. Among them, rank-preserving structural failure time model (RPSFTM) is the one of the most commonly used methods in HTA when analyzing overall survival outcomes (time to death from any cause). While RPSFTM can adjust for treatment switching with fewer variables and less modeling, this method assumes “constant treatment effect” assumption, which implies that treatment effect is the same no matter when patients switch to the new treatment. This assumption is often considered clinically invalid because treatment switching is usually allowed when patients experience disease progressions. The use of invalid models can lead to inappropriate reimbursement and pricing decisions, which in turn affects patients access to healthcare. Therefore, to support appropriate decision-makings, it is important to conduct sensitivity analyses to examine how treatment effects change when the model assumption is violated.
In our study, we will propose sensitivity analysis of RPSFTM to relax “constant treatment effect” assumption. Our proposed method is expected to reduce the potential bias that arises when estimating treatment effects with standard RPSFTM. We will apply the proposed method to several clinical trials with treatment switching and compare results with those obtained by standard RPSFTM. The ultimate goal of our research is to provide valid treatment effects and to help inform appropriate HTA decision-making when adjusting for treatment switching.
Requested Studies:
A Randomized, Open-label Study of the Effect of Herceptin Plus Arimidex Compared With Arimidex Alone on Progression-free Survival in Patients With HER2-positive and Hormone-receptor Positive Metastatic Breast Cancer
Data Contributor: Roche
Study ID: NCT00022672
Sponsor ID: BO16216
Phase 3, Randomized, Open-label Study Of The Efficacy And Safety Of Crizotinib Versus Pemetrexed/Cisplatin Or Pemetrexed/Carboplatin In Previously Untreated Patients With Non-squamous Carcinoma Of The Lung Harboring A Translocation Or Inversion Event Involving The Anaplastic Lymphoma Kinase (Alk) Gene Locus.
Data Contributor: Pfizer Inc.
Study ID: NCT01154140
Sponsor ID: A8081014
A Phase III Randomized, Double-Blind Study Of Sunitinib (SU011248, SUTENT) Versus Placebo In Patients With Progressive Advanced/Metastatic Well-Differentiated Pancreatic Islet Cell Tumors
Data Contributor: Pfizer Inc.
Study ID: NCT00428597
Sponsor ID: A6181111