Premium
A design of phase II cancer trials using total and complete response endpoints
Author(s) -
Lu Ying,
Jin Hua,
Lamborn Kathleen R.
Publication year - 2005
Publication title -
statistics in medicine
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.996
H-Index - 183
eISSN - 1097-0258
pISSN - 0277-6715
DOI - 10.1002/sim.2188
Subject(s) - complete response , clinical trial , medicine , sample size determination , flexibility (engineering) , stage (stratigraphy) , phases of clinical research , regimen , oncology , computer science , statistics , mathematics , chemotherapy , paleontology , biology
Phase II clinical trials in oncology are used for initial evaluation of the therapeutic efficacy of a new treatment regimen. Simon's two‐stage design based on total response (TR) rate is commonly used for such trials. Several authors have proposed alternative strategies to consider either response and toxicity or response and early progression. Because TR consists of both partial response (PR) and complete response (CR) and these two types of responses have different effects on subsequent patient outcome, Lin and Chen proposed a flexible design that is based on a weighted average of PR and CR rates as a way to recognize the differential significance of the two levels of response. Panageas and colleagues, on the other hand, used a trinomial model and direct search to consider a rejection region for PR and CR separately. In this paper, we reformat their hypotheses to assess efficacy based on TR and CR. A new two‐stage optimum phase II trial design based on TR and CR is developed. We provide guides on searching the stopping and rejecting regions and on determining sample size. An example of a phase II trial for glioblastomas treatment is presented. In this trial, physicians would be interested in either stable disease (SD), PR, or CR as an indication of efficacy. However, because PR and CR rarely occur, observation of any PR or CR will lean towards acceptance of the treatment. Our design has the advantage of being close to the traditional Simon two‐stage design while still having the flexibility to treat responses (PR and CR in this example) differently than SD. Copyright © 2005 John Wiley & Sons, Ltd.