Detection of Mutations in SF3B1 in Chronic Lymphocytic Leukaemia Patients By Reverse Transcription (Rt) Polymerase Chain Reaction and High Resolution Melt Curve Analysis, an Alternative Approach to Next Generation Sequencing for Routine Molecular Diagnostic Laboratories?

Genetic makers such as TP53, ATM mutations and IgH somatic hypermutation status have been used to predict clinical behaviour and response to therapy in Chronic Lymphocytic Leukaemia (CLL) for the last decade. Recently identified genetic markers including splicing factor 3 B1 (SF3B1), NOTCH and BirC3 are also significant but their role cannot be fully evaluated until a robust assay is available for a routine molecular diagnostic laboratory

The cell spliceosome is involved in the precise excision of introns from pre-mRNA and is composed of five subunits of which SF3B1 is a core component. Patient samples with SF3B1 mutations display enhanced intron retention in a small subset of transcripts involved in cancer related regulation such as cell cycle control, apoptosis and angiogenesis. Studies of CLL patient samples has shown that those with mutations in SF3B1 have been associated with shorter treatment free and overall survival.

The screening methodology for SF3B1 mutations through deep sequencing, though very sensitive, is not available in routine diagnostic laboratories. Approximately 90% of CLL-associated SF3B1 mutations occur is 336bp region of the cDNA exome amenable to PCR amplification. We designed primers flanking this region and following reverse transcription (RT) PCR amplification, detected the products on standard agarose gels. Following bidirectional Sanger sequencing, we compared the sequences obtained with reference SF3B1 cDNA/amino acid code. We assessed the impact of the mutations by the polyphen database and compared this to the immunophenotype and cytogenetic data as well as response to treatment through minimal residual disease (MRD) tracking. We subsequently designed an assay for detection of sequence variants in the SF3B1 region of interest by High Resolution Melt (HRM) curve analysis. This has enabled us to identify genetic variants of the SF3B1 region prior to sequencing. The application of this HRM pre-sequencing screening assay has resulted in a significant reduction in the number of samples requiring sequencing.

The patient cohort consisted of 52 patients enrolled on a Phase II, multi centre study of fludarabine, cyclophosphamide and rituximab (FCR) and included patients < 65 yrs, WHO status of 0, 1 or 2 without a deletion 17p. All samples were collected and tested prior to treatment, at the end of treatment (after 4 or 6 cycles when MRD negative), 6 monthly post treatment for 5 years for minimal residual disease and then annually by 6 colour immunophenotyping. Patients (51/52, no RNA available for 1 patient) were screened prior to treatment for the presence of SF3B1 mutations. Five of 51 (9.8 %) patients had SF3B1 mutations detectable by this methodology shown in the table below.

Patients with mutations in SF3B1 all had favourable cytogenetics (normal or del 13q) and the majority (4/5) had unmutated IGHV genes. We detected two patients with a mutation at c.2146A>G, p.K700E and one at c.2267G>A, pG740E which have been described previously and two novel mutations at c.2179G>C, p.A711P and c.2032C>T, p.H662Y. All mutations detected were damaging as per the polyphen database. Two patients have weak or no expression of CD5. One patient has never achieved MRD negativity, and two patients who became MRD negative at 3 months reverted to MRD positivity 9 and 13 months later respectively. One patient's MRD levels could not be tracked due to negative CD5 expression and was followed by a less sensitive IgH clonality assay (sensitivity approx. 1%) and one patient was withdrawn from the study.

In summary, we have developed a simple PCR based technique to detect SF3B1 mutations in CLL patient samples with high disease burden which can be applied in routine molecular diagnostic laboratories without access to next generation sequencing. The use of HRM technology resulted in a > 90% reduction in the number of samples requiring sequencing.