Design and analysis of clinical trials

Appropriate statistical methods for design and analysis of clinical trials are required in order to obtain unbiased results in clinical research. Recent advances of new cancer treatments such as the matrix metelloproteinaise inhibitor, which is targeted to reduce the chance of metastases and prolong survival instead of tumour size shrinkage, require new methodology to assess its efficacy. The multinomial phase II design incorporating early progression in addition to tumour response as an endpoint was developed to meet this need. The performance of this new design was tested using real phase II data. In the drug development process for cancer such as hepatocellular carcinoma (HCC), incorporation of tumour marker information (e.g. AFP) may improve the likelihood of identifying an effective compound. The use of tumour markers as an endpoint in addition to response in a phase II study was considered. The research of using tumour markers to improve sensitivity and specificity for the diagnosis of HCC was one of the major focuses of the group. Other design issues such as sample size estimation for Phase III studies including Phase II portion has been investigated. Novel research on statistical methodology such as the development of a global test statistics for mixture endpoints of survival and dichotomous data for group sequential test in phase III cancer trials is in progress.

Quality of Life Research

The methodology of assessing whether treatment may improve quality of life of cancer patients is important so that appropriate clinical outcome information may be available at various levels of research and treatment assessment. Methods to assess both quantity and quality of life was applied in clinical trials data, but further study on these methodolgies are needed. The design and analysis issues related to quality of life outcomes in cancer clinical trials has been an on-going research interest of the group. Future studies and methodologies on whether patients predisposed to specific tumor markers may have a different level of susceptibility to pain, depression, and specific quality of life domains will be studied.