Development of a targeted next-generation sequencing gene panel to investigate recurrent mutations in chronic lymphocytic leukaemia

Chronic lymphocytic leukaemia (CLL) is a mono-clonal B-cell malignancy characterised by heterogeneous clinical course and response to treatment. Recent studies with whole genome or whole exome sequencing have identified novel recurrent genomic lesions, and associated them with adverse clinical outcome of this disease. Owing to their limited sensitivity, the true incidence of these mutations, subclonal complexity and evolution and their roles in disease progression and treatment resistance are still not quite clear. To gain insight into these issues, I developed a highly sensitive (with an average coverage depth being 2250x and the lowest limit of detection 1%) and robust next generation sequencing (NGS) test using HaloPlex and Ion Torrent PGM to target exons of 15 recurrently mutated genes including TP53, ATM, SF3B1, PCLO, NOTCH1, LRP1B, SAMHD1, FBXW7, BIRC3, HISTIH1E, XPO1, CHD2, MYD88, POT1 and ZFPM2 in CLL. In the initial study using this technique, samples from a cohort of 32 cases with progressive and/or therapy resistant CLL before (n = 10) or after chemotherapy (n = 22) were screened. 87.5% of the patients were found to carry at least one somatic non-synonymous mutation in the first 12 targeted genes (VAF: 2-98%). The most commonly mutated genes were SF3B1, ATM, TP53 and PCLO identified in 11, 10, 9 and 8 cases, respectively. Mutations in TP53 and its upstream regulator ATM appeared to be dominant over other concurrent gene mutations compared to other genes (P = 0.011). Combining NGS and global SNP array analyses revealed a significant association between genomic aberrations in the ATM-p53 pathway and genomic instability. Moreover, prior exposure to DNA-damaging chemotherapy was associated with the bigger numbers of mutation events and mutated genes, although the increases were borderline significant. These results suggest that ATM-p53 pathway defects contribute to the acquisition of additional genomic aberrations and that treatment with DNA-damaging chemotherapy may play a role in the induction or selection of mutations. In the subsequent longitudinal study of 33 additional samples of 23 mutated cases from the same cohort, I showed the existence of different mutational processes operative in CLL including mutations related to AID, ageing and other factors. I confirmed the significant contribution of AID-related mutations to CLL clonal evolution, implying on-going activity of this enzyme in off-target genes. In addition, I demonstrated the predominance of a linear path of clonal evolution in this cohort. Thus, 89.3% of the mutated clones/subclones identified at advanced disease stages were detectable at time of diagnosis or prior to treatment. Hence, the early detection of these mutations may potentially serve as predictive biomarkers to inform on therapeutic decisions. I also documented convergent clonal evolution with priori-selection of clones carrying deleterious mutations in 2 target genes including ATM and TP53. This observation suggests that not all the mutations in the same gene play an equal role in disease progression and/or treatment resistance. Analysis of mutation doubling time revealed that driver mutations, including those in TP53, BIRC3, NOTCH1 and SF3B1, were significantly correlated with faster evolution as indicated by the shorter doubling time of variant allele frequency. Importantly, I found that increased subclonal sizes were strongly associated with shorter treatment-free survival in patients with driver mutations. Taken together, the results from this thesis have provided strong evidence emphasising the importance and usefulness of applying the ultra-sensitive NGS test in the early detection and subsequent monitoring of these recurrent somatic mutations in CLL. A translational study based on findings in this thesis is now ongoing, with an aim to convert this test into a regional clinical service