Addressing challenges of molecular precision diagnostics for cancer patients in the genomics era

Genomic technologies have proven successful in research-led studies and their potential for characterising prognostic and predictive biomarkers in cancer patients has been recognised. However, such techniques and findings are yet to be applied in routine diagnostic environments. Indeed, many challenges remain before being able to establish a comprehensive genomics classification for all cancer patients, not least pre-analytical difficulties and variants interpretation. One of the most important pre-analytical challenges is the use of formalin-fixed paraffin-embedded (FFPE) tissues which are far more widely available in diagnostics compared with alternative fresh-frozen (FF) tissue samples. The key issue is that FFPE treatment of the tissue greatly impacts on the DNA quality and downstream analysis. This is important because genomic studies have discovered a significant number of mutations in multiple cancer driver genes. The presence of these mutations informs on patients' prognoses or responses to treatment and guides clinical management. With an increasing number of targeted therapies available it is critical to stratify patients using genomics technologies. However, current clinically available tests fail to identify known driver mutations in the majority of patients. The overall objective of the work presented in this thesis was to address current challenges and limitations of FFPE testing by improving knowledge in genomic technology and by providing a means to offer genomic-guided precision medicine to a greater number of cancer patients. The specific aims were (i) to explore the feasibility of using FFPE tissue from patients with solid tumour for clinical grade whole genome sequencing (WGS) and (ii) to use WGS and targeted sequencing data to search for novel genomic biomarkers of aggressive disease in CLL patients. In Chapter 3, a prospective WGS study of cancer patients using 52 matching FF and FFPE samples was performed. This highlighted 71% agreement in somatic single nucleotide variants and revealed sub-optimal copy number alteration (CNA) detection from FFPE samples (44% median correlation with FF) due to non-uniform coverage. In Chapter 4, after pre-analytical investigation and experimental optimisation, CNA detection was improved significantly using lower reverse crosslinking temperature in FFPE DNA extraction (80°C or 65°C depending the methods) or by controlling formalin fixation. The results showed that WGS data derived from FFPE tissue derived DNA is capable of detecting 96% of clinically actionable variants (including 97% of CNAs) and therefore the approach is considered suitable for implementation in the routine clinical diagnostic environment. Furthermore, the study of several clinical trial CLL cohorts presented in this thesis highlighted multiple novel genomic biomarkers of aggressive disease in CLL patients. In Chapter 5 relapse/refractory patients (R/R) were clustered according to the driver mutations identified using targeted sequencing; a multiple-hit profile defined as recurrent combinations of ≥ 2 mutations of TP53, SF3B1 and ATM was identified. This multiple-hit profile, found in 19% of patients was associated with a median progression-free survival of 12 months compared with 22.5 months in the remaining patients (p=0.003). Other potential biomarkers were identified by exploring the genome of 320 CLL patients, such as mutational signatures. Collectively, the work performed and the data generated provide new insights to overcome current challenges linked to the establishment of genomics technologies in the clinic in order to offer precision medicine to cancer patients.