The Library
The statistical design and analysis of clinical studies using personalised healthcare under biomarker uncertainty
Tools
Lanza, Ben (2023) The statistical design and analysis of clinical studies using personalised healthcare under biomarker uncertainty. PhD thesis, University of Warwick.
|
PDF
WRAP_Theses_Lanza_2023.pdf - Submitted Version - Requires a PDF viewer. Download (9Mb) | Preview |
Official URL: http://webcat.warwick.ac.uk/record=b3977890
Abstract
The improved availability and quality of molecular profiling data has driven the identification and use of predictive biomarkers within the drug development process. This has facilitated a shift towards personalised healthcare, where treatments are tailored to specific individuals at a genetic level using biomarker information. The motivation of this thesis is to develop methodology which utilises such biomarker information to optimally identify responding patient subgroups, helping to make clinical trial design and implementation safer and more efficient. The optimisation of biomarker defined patient subgroups is explored within a confirmatory clinical trial setting. Specifically, it is of interest to generalise work identifying an optimal dichotomising threshold for a continuous biomarker to the setting of dual biomarkers, where two biomarkers are simultaneously predictive of increased treatment effect and a dichotomising threshold value is sought for both.
Work in this thesis explores embedding dual biomarker threshold identification techniques into confirmatory clinical trial design. Feasibility is initially demonstrated by extending an existing trial design to the dual biomarker case. A variety of statisticalmethods are then implemented within a two-stage phase III adaptive trial design and their performance contrasted. It is shown that recursive partitioning displayed the best performance among the implemented methods, with respect to threshold identification accuracy and trial operating characteristics.
Novel research is also carried out to investigate how to optimally control the multiplicity arising from the optimisation of the patient population alongside the testing of multiple independent hypotheses. The use of resampling based techniques to control the family wise error rate (FWER) is investigated in the setting where efficacy assessments are carried out simultaneously within highly correlated subgroups. By implicitly accounting for the dependence structure between test statistics, it is shown that one can gain increased power over traditional methods of FWER control, whilst maintaining strong control of the FWER.
Item Type: | Thesis (PhD) | ||||
---|---|---|---|---|---|
Subjects: | R Medicine > R Medicine (General) R Medicine > RC Internal medicine > RC0254 Neoplasms. Tumors. Oncology (including Cancer) |
||||
Library of Congress Subject Headings (LCSH): | Biochemical markers, Biochemical markers -- Diagnostic use -- Data processing, Precision medicine, Clinical trials -- Design, Clinical trials -- Data processing, Tumor markers, Support vector machines | ||||
Official Date: | March 2023 | ||||
Dates: |
|
||||
Institution: | University of Warwick | ||||
Theses Department: | Warwick Medical School | ||||
Thesis Type: | PhD | ||||
Publication Status: | Unpublished | ||||
Supervisor(s)/Advisor: | Parashar, Deepak | ||||
Sponsors: | Medical Research Council ; Roche Products Limited ; Warwick Medical School | ||||
Format of File: | |||||
Extent: | xix, 406 pages : illustrations | ||||
Language: | eng |
Request changes or add full text files to a record
Repository staff actions (login required)
View Item |