Transcriptional gene signatures : passing the restriction point for routine clinical implementation

Abstract: Uncontrolled cell growth and cell division are central to the process of tumorigenesis and a number of gene expression signatures have been developed based on genes that are involved in the cell cycle. Notably, gene expression signatures are used extensively in breast cancer research to examine the disease at a molecular level to describe tumour progression, treatment response and patients’ survival. The subject of this thesis is to explore the potential prognostic capacity of gene expression signatures in breast cancer and additionally, determine the prognostic capacity of a transcriptomic cell cycle activity (CCS) signature within variety of cancer types. Several breast cancer gene expression signatures have emerged and been validated over the past two decades in large retrospective clinical trials. Although the clinical impact of these signatures has been clearly demonstrated, breast cancer therapeutic guidelines are still established on the basis of immunohistochemical markers (IHC) such as estrogen (ER), progesterone (PR), human epidermal growth factor 2 (HER2) and the proliferation marker Ki67. In Study I, the additional prognostic information derived from the combination of gene expression signatures and IHC/Ki67 was investigated in two Swedish breast cancer cohorts. Cohort I is comprised of 621 individuals with primary breast cancer tumours diagnosed between 1997 and 2005 in Stockholm region of Sweden. Cohort II consists of 484 individuals with primary breast tumours who diagnosed and received primary therapy in the Uppsala region of Sweden between 1987 and 1989. In Cohort I, Recurrence score (RS) and PAM50 gene expression signatures added prognostic information beyond Ki67 and IHC subtypes while only IHC subtypes provided additional prognostic information to all gene expression signatures with the exception of PAM50 gene signature in this cohort. Similar results were observed in Cohort II. The ability of gene expression signatures to provide prognostic and treatment predictive information has been tested in primary breast tumours; however, their capability to provide similar information in the metastatic breast cancer (MBC) patients has not been investigated. In Study II, the prognostic capacity of gene expression signatures in breast cancer was evaluated in the metastatic setting in a Swedish multicenter randomized clinical trial known as “TEX” with 304 patients diagnosed with advanced locoregional or distant breast cancer relapse. A large number of tumours were classified into intermediate or high4 risk groups by all gene expression signatures. PAM50 was the only gene expression signature that provided prognostic information from lymph node (LN) metastases. In Study III, the prognostic and treatment-specific potential of CCND1 amplification was assessed in two breast cancer cohorts with 1965 and 340 patients, respectively. In the combined cohort, patients with CCND1-amplified tumours show worse survival in ER+/HER2-/LN-, luminal A and luminal B subtypes. Moreover, luminal A subtype with CCND1-amplified tumours shared similar gene expression changes with and luminal B subtype. In Study IV, the DNA mutations and chromosome arm-level aneuploidy within tumours with different cell cycle activity (CCS) were explored. We showed that cell cycle activity varied broadly among and within different cancer types. Two well-known oncogenes (TP53 and PIK3CA) exhibit the highest rate of mutations within different CCS groups. Furthermore, chromosomal arm level aberrations present in all CCS groups with a higher number of gains in 7p, 20q whereas deletions were more frequent within 17p and 8p arms. In the survival analysis, patients with higher CCS tumours show worse Progression-free interval relative to low and intermediate CCS groups. In conclusions, we have shown that PAM50 and RS gene expression signatures can add prognostic information to Ki67 and IHC subtypes; however, IHC subtypes did not add any prognostic information to PAM50 signature. Moreover, PAM50 gene expression signature can provide prognostic information from LN metastases in MBC patients. Additionally, CCND1 gene amplification has the potential to stratify patients with worse survival outcome within good-prognosis luminal A subtype tumours. Finally, we have demonstrated that CCS can provide independent prognostic information across cancer types.