Abstract
Precision medicine, also called personalized medicine, seeks to identify and classify individual patients according to their probable disease risk, prognosis and response to treatment, which relies on validated biomarkers. On the other hand, the molecular characterization of various malignancies has demonstrated that cancers can be highly heterogeneous even with the same origins. Although the biomarker development in oncology has been advanced by affordable high-throughput technologies and new biomarkers were identified by cohort studies, the traditional approach is largely limited by tumor heterogeneity, e.g., unbalanced distribution of rare subtypes and lack of independent validation to prevent potential false positive findings.This thesis presents a systematic approach to developing biomarkers in oncology for the purpose of diagnosis and prognosis prediction, with a focus on transcriptomic biomarkers.
Four main aspects are comprised in this framework: 1) Integrating multi-source data to obtain comparable cohorts with a sufficient sample size with a meaningful diversity of patients; 2) The validation of putative biomarkers in independent cohorts; 3) Transforming identified biomarkers into a feasible platform, e.g., biomarkers discovered using high-throughput measurements but PCR is preferred in clinical practice due to the cost; 4) Transforming biomarkers into less invasive methods, e.g., biomarkers discovered in tissue samples but pre-surgery biopsy is favorable.
Using this approach, we successfully identified and validated biomarkers for cancer diagnosis, lymph node metastasis and prognosis with robust predictive power. These new discoveries of transcriptomic signatures offer a promising aid in clinical decision making and improving the clinical management of such patients in the future.
| Date of Award | 9 Aug 2018 |
|---|---|
| Original language | English |
| Awarding Institution |
|
| Supervisor | Xin WANG (Supervisor) & Andrew Yen (External Co-Supervisor) |