N_LyST: a simple and rapid screening test for Lynch syndrome

Aims We sought to use PCR followed by high-resolution melting analysis to develop a single closed-tube screening panel to screen for Lynch syndrome. This comprises tests for microsatellite instability (MSI), MLH1 methylation promoter and BRAF mutation. Methods For MSI testing, five mononucleotide markers (BAT25, BAT26, BCAT25, MYB, EWSR1) were developed. In addition, primers were designed to interrogate Region C of the MLH1 promoter for methylation (using bisulphite-modified DNA) and to test for mutations in codon 600 of BRAF. Two separate cohorts from Nottingham (n=99, 46 with MSI, 53 being microsatellite stable (MSS)) and Edinburgh (n=88, 45 MSI, 43 MSS) were tested. Results All the cases (n=187) were blind tested for MSI and all were correctly characterised by our panel. The MLH1 promoter and BRAF were tested only in the Nottingham cohort. Successful blinded analysis was performed on the MLH1 promoter in 97 cases. All MSS cases showed a pattern of non-methylation while 41/44 cases with MSI showed full methylation. The three cases with MSI and a non-methylated pattern had aberrations in MSH2 and MSH6 expression. BRAF mutation was detected in 61% of MSI cases and 11% of MSS cases. Finally, 12 cases were blind screened by using the whole panel as a single test. Of these, five were identified as MSS, four as MSI/non-LS and three as MSI/possible LS. These results were concordant with the previous data. Conclusion We describe the Nottingham Lynch Syndrome Test (N_LyST). This is a quick, simple and cheap method for screening for Lynch syndrome.


AbsTrACT
Aims We sought to use PCR followed by high-resolution melting analysis to develop a single closed-tube screening panel to screen for Lynch syndrome. This comprises tests for microsatellite instability (MSI), MLH1 methylation promoter and BRAF mutation. Methods For MSI testing, five mononucleotide markers (BAT25, BAT26, BCAT25, MYB, EWSR1) were developed. In addition, primers were designed to interrogate Region C of the MLH1 promoter for methylation (using bisulphite-modified DNA) and to test for mutations in codon 600 of BRAF. Two separate cohorts from Nottingham (n=99, 46 with MSI, 53 being microsatellite stable (MSS)) and Edinburgh (n=88, 45 MSI, 43 MSS) were tested. results All the cases (n=187) were blind tested for MSI and all were correctly characterised by our panel. The MLH1 promoter and BRAF were tested only in the Nottingham cohort. Successful blinded analysis was performed on the MLH1 promoter in 97 cases. All MSS cases showed a pattern of non-methylation while 41/44 cases with MSI showed full methylation. The three cases with MSI and a non-methylated pattern had aberrations in MSH2 and MSH6 expression. BRAF mutation was detected in 61% of MSI cases and 11% of MSS cases. Finally, 12 cases were blind screened by using the whole panel as a single test. Of these, five were identified as MSS, four as MSI/non-LS and three as MSI/possible LS. These results were concordant with the previous data. Conclusion We describe the Nottingham Lynch Syndrome Test (N_LyST). This is a quick, simple and cheap method for screening for Lynch syndrome.

InTrOduCTIOn
Colorectal cancer (CRC) is a leading cause of cancer-related mortality. [1][2][3] Most CRCs arise sporadically without any antecedent family history. There are, however, several cancer syndromes in which development of CRC is part of the phenotypic spectrum. 1 2 The most common of these is Lynch syndrome (LS, also known as hereditary non-polyposis colorectal cancer), which is responsible for 2%-4% of all CRCs. 3 4 Patients with LS are susceptible to the development of CRCs and to the development of extra-colonic tumours-most notably endometrial, ovarian and small intestine adenocarcinomas. 5 6 LS arises as a consequence of germline mutation in one of four DNA mismatch repair (MMR) genes (ie, MLH1, PMS2, MSH2, MSH6). 7 8 Loss of any of the proteins results in loss of MMR function and an increase in the rate of gene mutation. One of the manifestations of this is an increase in insertion-deletion (indel) mutations especially at DNA microsatellites-known as microsatellite instability (MSI). 8 9 Tumours arising in LS therefore usually show both loss of expression of at least one of the MMR proteins (ie, deficient mismatch repair, dMMR) and MSI. 10 11 Thus, dMMR and MSIalthough they are distinct phenomena-are usually regarded as synonymous. Conversely, proficient mismatch repair (pMMR) is considered synonymous with a microsatellite stable (MSS) phenotype.
Numerous studies have shown that, due to the high risk of multiple cancers and its relatively high prevalence, there is a clinical and economic benefit to be gained by screening CRCs for LS. [12][13][14][15] While a definitive diagnosis of LS can only be made by demonstration of a germline mutation in an MMR gene, the possibility of LS can be inferred if a tumour is shown to be dMMR or shown to have MSI. However, approximately 10%-15% of sporadic CRCs will also show dMMR/MSI due to somatic loss of MMR function. 10 Epigenetic silencing of the MLH1 gene is the most common cause of dMMR in sporadic tumours and very rarely occurs in LS. 16 17 Thus, sporadic tumours with dMMR/MSI can be distinguished from tumours arising in LS by demonstrating methylation of the MLH1 promoter. Similarly, somatic mutation of BRAF is common in sporadic tumours with MSI but very rarely occurs in tumours arising in LS. [17][18][19] Guidance from the National Institute of Clinical and Healthcare Excellence (NICE) recommends that all CRCs should be screened for the possibility of LS. 12 The pathway suggested involves two steps: first, identify cases with dMMR/MSI and then filter out sporadic cases by testing for BRAF mutation and MLH1 promoter methylation. For the first step, testing for dMMR can be performed by immunohistochemistry (IHC) while testing for MSI involves PCR followed by capillary electrophoresis. For the second step, PCR followed by mutation screening or sequencing is required for the detection of BRAF mutation. Testing for MLH1 promoter methylation can be performed by PCR on modified DNA followed by sequencing or gel electrophoresis.
This strategy uses multiple tests and requires downstream analysis of the PCR products on different platforms. We believed that testing could Original article be simplified using high-resolution melting (HRM) analysis. HRM is an exquisitely sensitive method for detecting variations in DNA sequence. [20][21][22] It can be performed at the end of a PCR without needing to transfer PCR products to another tube (ie, a closed-tube test). We have shown previously that HRM can be used for testing for MSI, 23 24 for detection of BRAF mutation [23][24][25][26][27] and to identify promoter methylation. 25 Here, we sought to create a single-panel test in which a single PCR run followed by HRM can be used to screen for patients at risk of LS.

MATerIAls And MeThOds Cell lines
CRC cell lines were kindly donated by Professor Ian Tomlinson. The cell lines DLD1, HCT116, RKO, LoVo and LS1034 have previously been shown to have MSI while the cell lines SW480, SW620, HUTU80 and SW837 have been shown to be MSS. 28 DNA was extracted from cell lines using the Qiagen DNeasy kit (Qiagen, UK) as per manufacturer's instructions and adjusted to a concentration of 20 ng/µL. Identity of the cell lines was confirmed by mutation profiling as previously described. 25 Two diploid cell lines were chosen for spiking experiments in order to perform limit of detection experiment. DNA extracted from HCT116 (an MSI cell line) was spiked into DNA extracted from SW837 (MSS) to produce mixtures of DNA containing various proportions of HCT116 of ≈50%, ≈25%, ≈12.5%, ≈6%, ≈3% and ≈1.5%.

Primary colorectal cancers Nottingham cohort
Ninety-nine cases of CRC, which had previously been tested by IHC for expression of MMR proteins, were retrieved from the archives of Nottingham University Hospital Pathology Department. Of these, 46 cases were dMMR (and by inference had MSI). The remaining 53 cases were pMMR (and by inference were MSS).

Edinburgh cohort
Eighty-eight cases of CRC were retrieved form the archives of the Royal Infirmary of Edinburgh, Pathology Department, which had previously been tested for expression of MMR proteins or MSI. Of these, 45 cases were dMMR/MSI and 43 cases were pMMR/MSS.

dnA extraction for formalin-fixed tissue
DNA was also extracted from formalin-fixed, paraffin-embedded (FFPE) tumour samples. One or two 20 µm thick sections (depending on tissue surface area) were cut from each block. DNA was extracted using the QIAamp DNA FFPE tissue kit (Qiagen, UK) following the manufacturer's protocol. All DNA samples were adjusted to a concentration of 20 ng/µL.

Validation of IhC as a marker of MMr deficiency
The Nottingham cohort had been tested for expression of the MMR proteins by IHC. In order to confirm that the interpretation of the IHC was a correct reflection of the MMR function, a group of 33 cases (15 MSI/18 MSS) were tested by PCR followed by capillary electrophoresis (CE). PCR and CE testing was performed by the Molecular Genetics Laboratory at Nottingham University Hospitals NHS trust using the Promega MSI System V.1.2 in accordance with manufacturer's instructions. Five mononucleotide markers for MSI testing (BAT-25, BAT-26, NR-21, NR-24, MONO-27) 29 and two pentanucleotide markers (Penta-D, Penta-E) for sample identity checking were amplified using fluorescently labelled primers in a multiplex PCR. Products were analysed by CE on an Applied Biosystems 3130xl Genetic Analyzer (Life Technologies) using the kit internal lane standard. Data were analysed using GeneMapper software. Samples that showed MSI at ≥2 mononucleotide loci were interpreted as having MSI.
The optimum annealing temperatures of the primer pairs was ascertained as previously described. 38 Online supplementary table 1 lists all the mononucleotide repeat microsatellite markers, their genomic locations, amplicon sizes, the lengths of the mononucleotide repeats and the ranges of optimum annealing temperature. A range of metrics were used in order to define the best primers including reproducibility, PCR efficiency and range of functioning annealing temperature.

Testing for MsI using hrM analysis
In order to test for MSI using HRM, PCR was carried out each sample on the ABI 7500 FAST Real-Time PCR System (Applied Biosystems). Each reaction was carried out in a final volume of 10 µL and contained 5 µL of 2× Hot Shot Diamond PCR master mix and 0.5 µL of 20× (25 µM) EvaGreen dye; each primer final concentration was at 0.25 µM and 20 ng DNA template. The PCR was performed using a three-step procedure: 1 cycle of 95°C/5 min; 45 cycles of 95°C/10 s×1, 55°C/30 s×1, 72°C/30 s×1; and 1 cycle of 72°C/2 min. HRM was performed in-tube immediately after PCR and consisted of heating to 95°C for 15 s, rapid cooling to 60°C and maintenance at 60°C for 1 min. This was followed by slow ramping up at 0.03°C/s to 95°C during which fluorescent data were captured.
The melting data were analysed following normalisation but without temperature shifting using the ABI HRM software V.2.0. Samples were regarded as MSI if ≥2 markers (40%) showed instability; otherwise, they were regarded as MSS tumours.
The limit of detection for MSI by both CE and HRM was tested using spiked DNA samples (as described above).

novel primers for brAF testing
We have previously designed primers for screening for BRAF mutation using the nested quick-multiplex-consensus PCR protocol. 26 27 For the purposes of this protocol, which requires a single-stage PCR, novel primers were designed specifically for detection of mutation at codon 600.

Original article
Testing for methylation of the Mlh1 promoter Primer design Bisulphite modification of DNA causes a conversion of non-methylated cytosine residues to uracil while the methyl group of the methylated cytosines protects against this change (and cytosines are preserved). Following PCR on bisulphite-modified DNA, the methylated cytosines remain while non-methylated cytosines are converted to thymine residues. The sequence of methylated/ non-methylated DNA is therefore different and can be discriminated by HRM.
The promoter of MLH1 contains four CpG-rich regions (labelled A-D), which are the targets of epigenetic modification. It is generally considered that hypermethylation of the CpG island in region C is related to MLH1 silencing. 39 Furthermore, it is reported that region C exists in a dichotomous state, that is, all CpG residues being either methylated or non-methylated 39 40 without a state of partial methylation. However, the exact location of region C is not well defined and the number of reported CpG residues varies between 5 and 8. 39 40 Using the publicly available data, we identified a part of region C (located -46 to -111 from the transcription start site; NCBI sequence ID: NC_018914.2), which would contain all eight of the reported methylated CpG residues (online supplementary figure 1). Primers were designed to interrogate the whole CpG island of region C using the exactly the same cycling and HRM parameters as for the MSI markers. All tests (both sequencing and HRM) for MLH1 region C promoter methylation were performed on bisulphite-modified DNA.

Bisulphite conversion of DNA
In order to test for methylation of the MLH1 promoter, it was necessary to modify the DNA. Bisulphite conversion of 400 ng of genomic DNA from each sample was carried out using the EZ-DNA Methylation-Lightning Kit (Zymo Research, USA), according to the manufacturer's protocol. Optimisation of the methylation detection HRM-PCR assay was carried out using completely methylated or non-methylated human control DNA (Qiagen, UK).

Sequencing of region C of the MLH1 promoter
In order to confirm the dichotomous methylation state of region C, 20 CRCs (10 pMMR, 10 dMMR) were selected from the Nottingham cohort for Sanger sequencing of modified DNA. PCR prior to sequencing was performed using the reverse primer as described above. The forward primer, however, was modified to include a 'squirrel' tail to allow sequencing of short fragments as previously described. 38 PCR products were purified using the QIAquick kit (Qiagen) and the products sent to the DNA sequencing facilities (School of Life Sciences, University of Nottingham) and sequenced using Applied Biosystems BigDye Terminator V.3.1 Cycle Sequencing Kit and 3130xl ABI PRISM Genetic Analyzer (Data collection software V.3.0, Sequence analysis software V.5.2). The chromatograms were interpreted using Finch TV V.1.4.0 free software (from http://www. geospiza. com/ finchtv).

evaluation of the n_lysT panel
All the biomarkers were tested together as a single-panel test. Twelve cases were selected from the Nottingham cohort. This selection contained five cases designated as MSS, four cases designated as MSI with MLH1 deficiency and three cases designated as MSI with deficiency of MSH2/MSH6. They were assigned a new ID and were tested blind.

statistical analysis
GraphPad Prism software V.5.0 was used for statistical analysis. The χ 2 test was used to test for association between different factors. A value of P <0.05 was taken as being statistically significant.

Validation of IhC as a marker of MMr function
The Nottingham cohort had been selected using IHC expression of MMR proteins as a marker of MMR function. To confirm the association between IHC data and the presence or MSI, 33 cases from this cohort were tested by CE for the presence of MSI. Of these, 15 had been designated dMMR and 18 were pMMR. There was 100% concordance between the IHC analysis and MSI test results.

use of hrM for detection of MsI
From 11 different potential microsatellite loci, a panel of five markers comprising BAT25, BAT26, BCAT25, MYB and EWSR1 was chosen as the one showing the best performance. Our panel was compared with the commercial CE panel for their limit of detection for MSI calling. Using spiked samples containing varying proportions of DNA from MSI/MSS cell lines, the CE method and HRM were comparable with a limit of detection ≈6.25% (figure 1).
Our panel was used to test the Nottingham cohort of 99 cases of CRC (46 dMMR and 53 pMMR), and both observers correctly called every case. The Edinburgh cohort of 88 (45 dMMR/MSI and 43 pMMR/MSS) were tested separately. The HRM data were analysed by the same two observers and one observer correctly called all cases while the other observer miscalled two cases of MSI as MSS. Although we applied the generally used threshold of instability at ≥2 markers (40%) for a call of MSI, most cases usually showed instability at four to five markers and only one case, out of the total of 91 cases designated dMMR/MSI, was found to have instability at only two markers. Of the cases designated as MSS, 7% (7/96) had instability at one marker only while the remainder did not show any alteration in the microsatellite markers.

screening for brAF mutation
New primers to screen for BRAF codon 600 mutation were designed and optimised to work as a single-stage test using the cycling conditions for MSI testing. Primers were optimised and tested on cell lines with known BRAF mutation status (data not shown). All cases in the Nottingham cohort were tested and 28/46 (61%) of cases designated as MSI showed mutation while 6/53 (11%) of the MSS cases showed mutation. This frequency of mutation is consistent with published data and confirms the significant association of MSI with BRAF mutation (χ 2 test, P<0.0001).

Analysis of Mlh1 promoter methylation
Sequencing of region C Twenty cases of CRC from the Nottingham cohort (10 MSI, 10 MSS) were tested for MLH1 promoter methylation by direct sequencing. Our findings replicated published data with 10/10 cases of MSS CRC showing conversion of all 8 of the cytosines at the CpG sites to thymine without any cases suggesting partial methylation (ie, methylation at some residues but not others). In contrast, 10/10 cases of the MSI CRCs showed retention of the cytosines at the CpG sites (figure 2). The MSI samples did, however, show a double signal at the CpG sites, that is, a cytosine and a thymine. Since tumour samples contain both tumour epithelium and stroma, it is expected that the methylated signal   High-resolution melting analysis of region C of the MLH1 promoter. In order to define the melting patterns of methylated/non-methylated region C of the MLH1 promoter, PCR was performed on fully methylated or fully non-methylated control DNA following bisulphite modification. (A) is a derivative plot of the PCR products and shows that each condition (ie, methylated or non-methylated) had a distinct melting peak. The melting temperature (T m ) of the methylated peak (double arrow) was higher than that of the non-methylated peak (single arrow) reflecting the higher proportion of cytosine residues within the fully methylated samples. (B) shows the melting pattern of tumour samples, which are proficient mismatch repair (pMMR) (single arrow) and deficient mismatch repair (dMMR) due to loss of MLH1 expression (double arrow). The pMMR tumours gave a single non-methylated peak. The dMMR tumours gave a double peak representing a methylated peak (from the tumour cells) and non-methylated peak (from the stromal cells). All pMMR tumours tested gave a single peak and all dMMR tumours gave a double peak, reinforcing the data that this region does not have a state of partial methylation. comes from the tumour cells while the signal from the stroma would be non-methylated.

HRM analysis of region C
HRM was performed following PCR with primers targeted to amplify around the CpG island of region C of the MLH1 promoter. Amplification, following bisulphite modification, of both fully methylated and fully non-methylated DNA gave a single peak ( figure 3A). The melting temperature (T m ) of the PCR product from the non-methylated DNA (ie, the 'non-methylated peak') was lower than that of PCR product from the methylated DNA (the 'methylated peak') reflecting (G) is a derivative plot and tumours demonstrate two discrete melting forms: 'methylated' comprising two melting peaks that represent methylated DNA (from tumour epithelium) and non-methylated DNA (from tumour stroma) or 'non-methylated' comprising one melting peak that characterises a completely non-methylated tumour and stroma cell population. WT, wild type.

Original article
the enrichment of the latter with cytosine residues within the methylated sequence. All cases in the Nottingham cohort were tested for methylation of region C. Two cases (both designated as MSI) could not be tested due to failed PCR post-bisulphite modification of DNA. Of the 97 successfully tested cases, two distinct melting patterns were seen, that is, a single peak low T m peak (corresponding to the non-methylated peak) and a double peak with both low and high T m (corresponding to both the non-methylated peak and the methylated peak, figure 3B). All cases designated as pMMR/MSS showed only a single non-methylated peak, that is, there was no promoter methylation. We regard this as the 'non-methylated pattern'. Of the 44 cases with MSI, 41 showed a double peak indicating both methylated DNA and non-methylated DNA. The double peak was associated with loss of MLH1 expression (χ 2 test, P<0.0001), and we regard this as the 'methylated pattern'. The double peak is mostly likely due to methylated DNA being present in the tumour epithelium while the stromal cells are likely to contain non-methylated DNA. The three remaining MSI cases showed a single non-methylated peak. These cases were deficient in MSH2 and MSH6.

screening for lynch syndrome using n_lysT
In order to test the N_LyST panel, 12 cases were blind-tested in a single PCR run. The outcome of N_LyST is to categorise cases 'probable Lynch syndrome' if they show MSI, have wild-type BRAF and have a non-methylated pattern for region C of the MLH1 promoter. Any other pattern would be categorised as 'not Lynch syndrome'. All cases were correctly identified by the panel (table 1, figure 4).

dIsCussIOn
In this paper, we have described the Nottingham Lynch Syndrome Test (N_LyST) as a single-panel, closed-tube test for Lynch syndrome screening. The cases used to develop this test were selected on the basis of MMR protein expression and, to validate the use of these cohorts for our assay, we first confirmed that dMMR based on IHC was very strongly correlated with MSI.
N_LyST incorporates the three components of LS screening (ie, testing for MSI, MLH1 promoter methylation and BRAF mutation) into a single PCR run. First, we developed a panel of five microsatellite markers, which includes two established markers (BAT25, BAT26) and three novel markers (BCAT25, MYB and EWSR1). When tested in 187 CRCs (from two different institutions), there was near perfect concordance with the IHC/ CE designation. Analysis of the HRM data was undertaken by two observers, demonstrating that the analysis is easy and reproducible. The HRM method has a similar limit of detection as CE analysis (≈6.25% mutant DNA), but CE analysis can be complicated by stutter bands that can cause difficulty in allele sizing. 33 41 Next, we designed an assay to detect mutations in codon 600 of BRAF. Reassuringly, the detected mutation frequencies (61% in MSI tumours, 11% in MSS tumours) were in the expected range and the association of MSI with BRAF mutation (χ 2 test, P<0.0001) was seen.
The third step was the design of an assay to test for methylation of region C of the MLH1 promoter. Our sequencing and HRM data confirmed the dichotomous state of region C, that is, either non-methylated or fully methylated. The HRM assay clearly discriminated the two states and, when tested on the Nottingham cohort, all dMMR cases with loss of MLH1 expression by IHC had MLH1 promoter methylation (ie, the methylated pattern of two peaks) and were therefore sporadic tumours. None of the cases that were pMMR or dMMR due to MSH2/MSH6 loss had MLH1 promoter methylation. Finally, all components of N_LyST were put together and tested as a panel. Twelve cases of CRC were blind tested and perfectly categorised as 'non-LS' or 'probable LS'.
N_LyST involves a panel of seven PCRs that are performed in a single run using a single cycling programme. It could hugely improve work flow in a diagnostic laboratory since HRM is performed in-tube on completion of the PCR and transfer of PCR products to another platform for further analysis is not required. Since the test involves a panel applied to all tumours, it does mean that some tumours that are MSS will be unnecessarily tested for BRAF mutation and MLH1 promoter methylation. However, the cost of this is more than offset by savings made on manpower and consumable due to the removal of downstream analyses of PCR products. Furthermore, since it is a closed-tube test, the risk of laboratory contamination with PCR products is eliminated.
Most modern real-time PCR machines will have HRM capabilities and expensive specialist equipment is not required for N_LyST. The ease of the methodology and data interpretation means the N-LyST could probably be performed in non-specialist diagnostic pathology laboratories. This becomes pertinent when considering that MSI testing is likely to increase as it provides information, which extends beyond LS testing, for example, MSI can be used to stratify patients into groups eligible for treatment with 5-fluorouracil-based therapy 42 43 or immunotherapy. 44 45 Such high-throughput analysis will require a rapid and simple test such as N_LyST.
An important question is whether N_LyST-since it is a screening test-is relevant in the era of next-generation sequencing (NGS). The sheer sequencing power of NGS platforms would allow the MMR genes and multiple microsatellites to be sequenced in a single test. 46 However, microsatellite regions can be problematic from some NGS platforms and, where there is low tumour epithelium content, great sequencing depth may be required. In addition, MLH1 promoter methylation testing would require Methyl-Seq to be performed. The economic analyses performed as part of the NICE guidelines concluded that it was more cost-effective to screen the tumour samples prior to germline sequencing. 12 Since N_LyST can be performed in less time than that required for library preparation and sequencing with NGS, a case for including N_LyST in the testing pathway can be made.
In summary, N_LyST is based on PCR and HRM and uses a panel of seven markers to test for MSI, MLH1 promoter methylation and BRAF mutation in a single PCR run. It can be performed on most real-time PCR machines and, as a closed-tube test, it can improve laboratory workflow and reduce turnaround times for testing. It is a robust test that represents a quick, cheap and easy way to screen for LS.

Take home messages
► Nottingham Lynch Syndrome Test (N_LyST) is a closed-tube test for screening for Lynch syndrome in a single PCR run. ► It consists of a panel that simultaneously tests for microsatellite instability, BRAF mutation and MLH1 promoter methylation. ► A real-time PCR machine with ability to perform highresolution melting (HRM) at the end of the PCR is required. Apart from PCR/HRM, no other technique is used. ► N_LyST is as sensitive as the currently used techniques such as PCR followed by capillary electrophoresis, but it is quicker and simpler.