Identifying High-Risk CLL to Predict Early Relapse after FCR Based Treatment Using Whole Genome Sequencing: First Results from the Genomics England CLL Pilot

Background

Chemo-immunotherapy (CIT) with fludarabine, cyclophosphamide, and rituximab (FCR) is the standard of care in frontline treatment of CLL. With this approach, 25% of patients relapse within 24 months, whereas approximately one third of patients with hypermutated immunoglobulin heavy chains (IgHV) achieve a functional cure (Hallek et al. Lancet. 2010; Tam et al. Blood, 2014, Fischer et al, Blood 2015; Philip A. Thompson et al. Blood 2016). So far, mutations and/or deletions of TP53 remain the only predictive marker screened for in routine clinical practice, accounting for only one third of patients relapsing early after CIT. Recent next-generation sequencing (NGS) studies have revealed novel candidate predictors of early relapse including somatic mutations in RPS15 (Landau et al. Nature, 2015) and SAMHD1 (Clifford et al., submitted). Taken together with TP53disruption, these only occur in a subset of high-risk patients. Here, we present a comprehensive analysis of high-risk patients using Whole Genome Sequencing (WGS).

Patients and Methods

Using WGS we investigated 149 CLL patients from 5 national UK clinical trials: CLEAR (n=8), RIAltO (n=45), CLL 210 (n=22), ARCTIC (n=32) and AdMIRe (n=42). The two first line FCR-based clinical trials (ARCTIC and AdMIRe) were studied in most detail: 56 patients relapsed within 24 months; this group of patients will be referred to as high risk patients. Leukemia samples (peripheral blood) and germline samples (saliva) were collected for each patient. We performed WGS on the HiSeqX (Illumina). After read alignment, we detected somatic variants using Strelka 2.4.7 for small variants detection (SNV and InDels), Manta 0.28.0 for Structural variant (SV) detection, and Canvas 1.3.1 for Copy number variant (CNV) detection (Illumina). Non-coding regions were annotated with information from primary CLL, CLL cell lines and B-cell ENCODE databases. We interrogated the data at a gene scale and global level in order to identify patterns of early relapsing patients. Operative mutational signatures were analysed according to Alexandrov et al. (Nature, 2013). Putative regions of kataegis were calculated based on Lawrence et al. (Nature, 2013) and Alexandrov et al. (Nature, 2013).

Results

The mean coverage for CLL tumour and germline samples was 105.2X and 33.7X, respectively. The analysis of the whole cohort highlighted 1,723,603 somatic SNVs (mean= 11,570/sample) and 555,179 InDels (mean= 3,726/sample). Somatic SNVs spectrum consisted mainly of C>T/G>A mutations (30% of total SNVs reported) as previously described. The analysis of 13,490 somatic functional SNVs and InDels revealed novel candidate genes as most commonly mutated in the cohort. In high-risk patients, we noticed an enrichment of mutations in known genes such as TP53, genes of the NF-κB pathway and novel candidate genes previously reported in other cancers. A specific analysis of the functional coding mutations of known CLL driver genes revealed ATM, SF3B1 and IGLL5 as most commonly mutated genes in FCR responders compared to TP53, RPS15 and EGR2 in high risk patients. In depth analysis of somatic non-coding regions also identified potential new candidate regions associated with early relapse. Next, we investigated 52,871 CNAs (mean= 380/sample) and 29,080 SVs (mean= 195/sample) and identified as expected del13q, del17p, del11q and tri12 as the most frequent aberrations. In addition, we identified SVs across genes of interest in CLL, for instance TP53, ATM and BIRC3. Finally, we performed global genome analyses with investigation of mutational signatures and kataegis analyses highlighting hypermutated candidate regions, including the previously described IGLL5gene.

Conclusion

Here we present initial analysis of WGS data on 149 CLL patients from 5 UK clinical trials. Different patterns of mutations between low and high risk clinical groups are suggested. More detailed analysis with greater numbers of samples is ongoing and will determine the true clinical significance of these preliminary findings. The possibility of using WGS to aid clinical decision-making is becoming a realistic goal.