Screened selection design for randomised phase II oncology trials : an example in chronic lymphocytic leukaemia

Research output: Contribution to journalArticlepeer-review

Standard

Screened selection design for randomised phase II oncology trials : an example in chronic lymphocytic leukaemia. / Yap, Christina; Pettitt, Andrew; Billingham, Lucinda.

In: BMC Medical Research Methodology, Vol. 13, No. 1, 87, 03.07.2013.

Research output: Contribution to journalArticlepeer-review

Harvard

APA

Vancouver

Author

Bibtex

@article{086eb8a0db834d8c9dd05fe9b264f64a,
title = "Screened selection design for randomised phase II oncology trials : an example in chronic lymphocytic leukaemia",
abstract = "BackgroundAs there are limited patients for chronic lymphocytic leukaemia trials, it is important that statistical methodologies in Phase II efficiently select regimens for subsequent evaluation in larger-scale Phase III trials.MethodsWe propose the screened selection design (SSD), which is a practical multi-stage, randomised Phase II design for two experimental arms. Activity is first evaluated by applying Simon{\textquoteright}s two-stage design (1989) on each arm. If both are active, the play-the-winner selection strategy proposed by Simon, Wittes and Ellenberg (SWE) (1985) is applied to select the superior arm. A variant of the design, Modified SSD, also allows the arm with the higher response rates to be recommended only if its activity rate is greater by a clinically-relevant value. The operating characteristics are explored via a simulation study and compared to a Bayesian Selection approach.ResultsSimulations showed that with the proposed SSD, it is possible to retain the sample size as required in SWE and obtain similar probabilities of selecting the correct superior arm of at least 90%; with the additional attractive benefit of reducing the probability of selecting ineffective arms. This approach is comparable to a Bayesian Selection Strategy. The Modified SSD performs substantially better than the other designs in selecting neither arm if the underlying rates for both arms are desirable but equivalent, allowing for other factors to be considered in the decision making process. Though its probability of correctly selecting a superior arm might be reduced, it still performs reasonably well. It also reduces the probability of selecting an inferior arm.ConclusionsSSD provides an easy to implement randomised Phase II design that selects the most promising treatment that has shown sufficient evidence of activity, with available R codes to evaluate its operating characteristics.",
keywords = "Randomised Phase II, Selection Design, Screening Design, Play-the-Winner, Oncology, Moderate Sample Sizes",
author = "Christina Yap and Andrew Pettitt and Lucinda Billingham",
year = "2013",
month = jul,
day = "3",
doi = "10.1186/1471-2288-13-87",
language = "English",
volume = "13",
journal = "BMC Medical Research Methodology",
issn = "1471-2288",
publisher = "Springer",
number = "1",

}

RIS

TY - JOUR

T1 - Screened selection design for randomised phase II oncology trials : an example in chronic lymphocytic leukaemia

AU - Yap, Christina

AU - Pettitt, Andrew

AU - Billingham, Lucinda

PY - 2013/7/3

Y1 - 2013/7/3

N2 - BackgroundAs there are limited patients for chronic lymphocytic leukaemia trials, it is important that statistical methodologies in Phase II efficiently select regimens for subsequent evaluation in larger-scale Phase III trials.MethodsWe propose the screened selection design (SSD), which is a practical multi-stage, randomised Phase II design for two experimental arms. Activity is first evaluated by applying Simon’s two-stage design (1989) on each arm. If both are active, the play-the-winner selection strategy proposed by Simon, Wittes and Ellenberg (SWE) (1985) is applied to select the superior arm. A variant of the design, Modified SSD, also allows the arm with the higher response rates to be recommended only if its activity rate is greater by a clinically-relevant value. The operating characteristics are explored via a simulation study and compared to a Bayesian Selection approach.ResultsSimulations showed that with the proposed SSD, it is possible to retain the sample size as required in SWE and obtain similar probabilities of selecting the correct superior arm of at least 90%; with the additional attractive benefit of reducing the probability of selecting ineffective arms. This approach is comparable to a Bayesian Selection Strategy. The Modified SSD performs substantially better than the other designs in selecting neither arm if the underlying rates for both arms are desirable but equivalent, allowing for other factors to be considered in the decision making process. Though its probability of correctly selecting a superior arm might be reduced, it still performs reasonably well. It also reduces the probability of selecting an inferior arm.ConclusionsSSD provides an easy to implement randomised Phase II design that selects the most promising treatment that has shown sufficient evidence of activity, with available R codes to evaluate its operating characteristics.

AB - BackgroundAs there are limited patients for chronic lymphocytic leukaemia trials, it is important that statistical methodologies in Phase II efficiently select regimens for subsequent evaluation in larger-scale Phase III trials.MethodsWe propose the screened selection design (SSD), which is a practical multi-stage, randomised Phase II design for two experimental arms. Activity is first evaluated by applying Simon’s two-stage design (1989) on each arm. If both are active, the play-the-winner selection strategy proposed by Simon, Wittes and Ellenberg (SWE) (1985) is applied to select the superior arm. A variant of the design, Modified SSD, also allows the arm with the higher response rates to be recommended only if its activity rate is greater by a clinically-relevant value. The operating characteristics are explored via a simulation study and compared to a Bayesian Selection approach.ResultsSimulations showed that with the proposed SSD, it is possible to retain the sample size as required in SWE and obtain similar probabilities of selecting the correct superior arm of at least 90%; with the additional attractive benefit of reducing the probability of selecting ineffective arms. This approach is comparable to a Bayesian Selection Strategy. The Modified SSD performs substantially better than the other designs in selecting neither arm if the underlying rates for both arms are desirable but equivalent, allowing for other factors to be considered in the decision making process. Though its probability of correctly selecting a superior arm might be reduced, it still performs reasonably well. It also reduces the probability of selecting an inferior arm.ConclusionsSSD provides an easy to implement randomised Phase II design that selects the most promising treatment that has shown sufficient evidence of activity, with available R codes to evaluate its operating characteristics.

KW - Randomised Phase II

KW - Selection Design

KW - Screening Design

KW - Play-the-Winner

KW - Oncology

KW - Moderate Sample Sizes

U2 - 10.1186/1471-2288-13-87

DO - 10.1186/1471-2288-13-87

M3 - Article

C2 - 23819695

VL - 13

JO - BMC Medical Research Methodology

JF - BMC Medical Research Methodology

SN - 1471-2288

IS - 1

M1 - 87

ER -