Evaluation of high-resolution melting curve analysis as a high-throughput mutation scanning tool

Taylor CF^*1, Harland M², Charlton R³ and Taylor GR^1,3.

¹Cancer Research UK Mutation Detection Facility, St James’s University Hospital, Leeds, UK, LS9 7TF; ²Cancer Research UK Genetic Epidemiology Division, St James’s University Hospital, Leeds, UK, LS9 7TF; ³Yorkshire Regional DNA Laboratory, St James’s University Hospital, Leeds, UK, LS9 7TF.

    ^*corresponding author : claire.taylor@cancer.org.uk

We have evaluated high-resolution melting curve analysis (MCA) as a high-throughput mutation scanning tool using the 96-well format LightScanner. Initial investigations were carried out using a panel of MLH1, MSH2 and TP53 mutants. 29 amplicons containing 52 nucleotide substitutions and 9 insertions/deletions were screened. Mutation detection sensitivity was 100%. We have carried out blinded analysis of panels of samples for mutations in CDKN2A (n=92), MSH2 (n=19), MLH1 (n=25) and TP53 (n=18). DNA was extracted from blood or cell lines; samples had previously been analysed by FSSCP, DHPLC and/or sequencing. All mutations were detected. 9/20 (45%) homozygous TP53 mutants were detected as homozygotes; 20/20 were detected after pre- or post-PCR mixing with wildtype DNA. Serial dilution of mutants with wildtype DNA demonstrates that the detection limit for minority species is variable and ranges between 25% to <10%. 25 blinded tumour samples were screened for variants in TP53. 22/23 variants were correctly identified; the missed mutation is currently being followed up.

In the blinded studies described above, specificity of MCA was initially determined to be approximately 95%. However, the majority of false-positive results are associated with a small minority (<5%) of DNA templates, which are objectively different to the remainder (e.g. degraded or low concentration DNA). Exclusion of these samples increases specificity to 97-99%. In total, 133 different variants in 58 fragments have been screened. Mutations at any position in the amplicon can be detected, including those immediately adjacent to the PCR primers. The fragment size range was 147-425 bp and the G+C content was 29-76% 
An estimate of the cost per analysis allowing for consumables, equipment depreciation and staff costs was made. We estimate that MCA using the LightScanner costs approximately 6 times less than DHPLC, 18 times less than FSSCP and more than 30 times less than DNA sequencing. 

In conclusion, MCA is a rapid, sensitive and cost-effective method for mutation scanning. No post-PCR sample processing is required, data for up to 96 samples can be collected in approximately 5 minutes; the data is simple in nature and its analysis is rapid. As a result of the successful evaluation of this technique, we are now making the transition to use of MCA as a routine mutation screening service tool.