Development of an Ultra-Deep Sequencing Method on Ion Torrent PGM to Detect Mutations in a Panel of Genes Relevant to Disease Progression and Drug Resistance in Chronic Lymphocytic Leukaemia

Chronic lymphocytic leukaemia (CLL) is the most common adult leukaemia in the Western world. The disease is characterised by a great variability in clinical course and response to treatment. Acquired somatic mutations in genes related to multiple signal pathways have recently emerged as important factors contributing to this heterogeneity. These mutations accumulate with disease progression or development of therapy resistance. However, with conventional methods they are very difficult to be detected in a single test, particularly when the size of the affected cell population is small. Therefore, there is a clinical need for a rapid, reliable, affordable and highly sensitive sequencing method to monitor these mutations from an early stage of this disease.

Herein we developed an ultra-deep next generation sequencing approach based on Ion Torrent PGM to sequence a total of 246 exons of 15 genes including TP53, ATM, BIRC3, NOTCH1, SF3B1, MYD88, LRP1B, SAMHD1, FBXW7, POT1, HISTIH1E, XPO1, CHD2, PCLO and ZFPM2. These genes were selected because mutations in them have been reported to involve at least 5% of CLL patients and associate with poor outcome of CLL. Serial CLL samples were collected at multiple time points from diagnosis to disease progression/drug resistance from a cohort of 33 patients. Initially, target DNA in samples taken at advanced disease stages, without or with mutations in TP53 as detected and quantified by FASAY assay and Sanger sequencing, was enriched and amplified using Agilent HaloPlex probes, with 4604 amplicons for a sequenceable size of 135.91 Kbp of each sample. On each Ion 318 Chip (1 GB output), barcoded DNA enrichment preparations from 4 patients were sequenced by the PGM. For confirmation, the same patient samples were tested repeatedly in an independent experiment starting from DNA preparation. The sequence data were aligned to a human reference genome (Hg19); variants were called by the Torrent Variant Caller (v4.0-6) and visualised with the IGV. Results showed that the target coverage was 99.999% and an average coverage depth 3941x. Importantly, all of the known TP53 mutations were repeatedly detected at expected frequencies. In addition, extra mutations at lower levels (5% - 20% alleles) in multiple genes, including TP53 and SF3B1, were detected.

In a sensitivity test of this method, DNA samples from 5 cases, each with a known % of mutation in TP53measured by FASAY assay and Sanger sequencing, were pooled together and then mixed with a wild-type DNA sample to serially dilute these mutations to 20% - 0.2% alleles before target enrichment. With an average coverage depth of 1843x (range: 1610 - 2187) for these 5 target locations, each mutation was readily detected at 20%, 5% and 1%, with the average quality scores being 1497.3, 61.2 and 29.4, respectively. However, only 3 out the 5 mutations could be detected at the level of 0.2%.

Taken together, we have successfully developed a sensitive next generation sequencing method for detecting mutations in a CLL gene panel. This allows us to monitor multiple mutations affecting as low as 1% of alleles in CLL samples throughout the disease course and possibly discover clinically useful biomarker(s) for CLL progression and resistance to therapy.