- Email: [email protected]
A HRM assay for identification of low level BRAF V600E and V600K mutations using the CADMA principle in FFPE specimens
Pathology (December 2017) 49(7), pp. 776–783
M O L E C U L A R PAT H O L O G Y
A HRM assay for identification of low level BRAF V600E and V600K mutations using the CADMA principle in FFPE specimens CLAUDIA HUEBNER, REMENY WEBER
AND
RICHARD LLOYDD
Department of Anatomical Pathology, Labtests Auckland Limited, New Zealand
Summary Melanoma patients with BRAF V600E and V600K mutations show complete or partial response to vemurafenib. Detection assays often scan for the common V600E mutation rather than the rare V600K variant, although this mutation can be found in a high proportion of melanoma patients in the South Pacific. Herein, we describe a BRAF high resolution melting (HRM) assay that can differentiate low level of V600E and V600K mutations using formalin fixed, paraffin embedded (FFPE) reference standards for assay validation. The assay is based on the competitive amplification of differentially melting amplicons (CADMA principle) and has a limit of detection of 0.8% mutant allele for V600K and 1.4% mutant allele for V600E. A differentiation between the two mutations based on the melting profile is possible even at low mutation level. Sixty FFPE specimens were scanned and mutations could be scored correctly as confirmed by castPCR. In summary, the developed HRM assay is suitable for detection of V600K and V600E mutations and proved to be reliable and cost effective in a diagnostic environment. Key words: BRAF mutational analysis; HRM; CADMA; reference standards. Received 28 March, revised 9 August, accepted 15 August 2017
INTRODUCTION BRAF codes for a non-receptor serine/threonine kinase component in the RAF-MEK-ERK signal transduction phosphorylation cascade. This pathway controls a variety of biological processes such as cell proliferation, differentiation and survival. The mutated version of the BRAF kinase results in constitutive activity, which is believed to be actively involved in oncogenic proliferation.1 Somatic mutations of BRAF have been documented in different kinds of tumours, predominantly in malignant melanoma, thyroid papillary cancer, sporadic colorectal tumours, low-grade ovarian serous carcinoma, and lung tumours.2,3 Hot spot mutations have been observed in exon 15 at codon 600 with the V600E (c.1799T>A, p.Val600Glu) mutation the most common. In general, 50–70% of melanoma tumours harbour a BRAF mutation, and of these, 80% are positive for BRAF V600E.4 Another mutation at this site involves the mutation of two adjacent nucleotides, substituting lysine for valine (c.1798_1799delinsAA, V600K). V600K mutations
are most frequently found in older patients (aged 65 years) and/or in those with evidence of chronic UV exposure.5 Recent studies have reported that the frequency of the V600K mutation may be as high as one-third of all BRAF mutations in melanoma.6,7 BRAF inhibitors have been designed which target the activated kinase domain. In 2011, vemurafenib (trade name Zelboraf, also known as PLX4032 and RO5185426) received approval from the US Food and Drug administration (FDA) for patients with metastatic or inoperable tumours that carry the BRAF activating V600E mutation.8,9 Patients carrying the V600K mutation have also shown response to vemurafenib.5,6 Based on the outstanding importance of BRAF as a molecular marker in cancer treatment, accurate identification of mutant cases is absolutely critical for appropriate therapy. Different methods are used to detect for BRAF mutations, which vary regarding specificity, sensitivity, turnaround time and costs.10 The traditional Sanger sequencing shows poor analytical sensitivity and often needs greater than 20% of tumour load in a specimen to render a reliable result. Pyrosequencing and next generation sequencing (NGS) are very sensitive with a reported limit of detection of 1% for NGS and 5% for pyrosequencing in a background of wildtype alleles.10 On the other hand, the drawbacks are the costs and the complex and challenging data analysis, which requires a suite of sophisticated bioinformatic tools to make sense of the data. Reports have shown that the CE-IVD marked Cobas 4800 BRAF V600 test (Roche Molecular Diagnostics, USA), widely used in the molecular analyses of melanoma in clinical settings, failed to detect 16.3% of the mutations eligible for therapy with vemurafenib.10 High resolution melting (HRM) analysis is a rapid, closedtube and comparatively inexpensive technique. It is based on the fluorescent monitoring of the dissociation behaviour of DNA when exposed to increasing temperatures. The HRM melting profile gives a specific sequence-related profile that differentiates wild-type sequences from homozygote or heterozygote variants.11 The sensitivity for HRM has been described as between 5–10% mutant allele, depending on the precise mutation change and amplicon length.12–14 However, identification of low-abundance mutations (below 5%) using this technique is only possible with modifications.15–18 Recently, Kristensen et al. described a modified HRM method (competitive amplification of differentially melting amplicons; CADMA) using a competitive oligonucleotide and very short amplicons, leading to sensitivities as low as 0.25% mutated V600E allele.15 The principle of the CADMA
Print ISSN 0031-3025/Online ISSN 1465-3931 © 2017 Royal College of Pathologists of Australasia. Published by Elsevier B.V. All rights reserved. DOI: https://doi.org/10.1016/j.pathol.2017.08.011
HRM ASSAY FOR LOW LEVEL BRAF MUTATIONS
777
method is the simultaneous amplification of wild-type and mutation containing sequences using a mutation specific primer, an overlapping primer that targets both sequences, and a common primer, which give rise to two different amplicons with different melting properties. The CADMA method is highly sensitive for the targeted mutation, however less specific for other significant mutations present in the amplicon. In order to get high sensitivities for different mutations a new assay has to be designed for each mutation of interest. Here we described a HRM assay based on the CADMA principle, which allows the simultaneous detection and accurate calling of V600E as well as V600K mutations in one assay. In order to evaluate the assay, commercial reference formalin fixed, paraffin embedded (FFPE) standards containing defined ratios of mutant alleles were used.
overlapping primer) from the above described, the assay was named CADMA assay V600E/K. In addition, one more experimental layout was performed incorporating the following four primers: BRAF for V600K, the two mutation specific primers and the overlapping primer BRAF rev. Each reaction mixture contained 2 mL DNA (10 ng/mL or 25 ng/mL), 300 nM mutation specific primer, 300 nM common primer, 150 nM overlapping primer and 5 mL MeltDoctor HRM Master Mix (Life Technologies, USA) in a total volume of 10 mL. Polymerase chain reaction (PCR) cycling and HRM analysis were performed on the Eco Real-Time PCR instrument (Illumina, USA), with an initial denaturation of 95 C for 10 min, followed by 48 cycles of 95 C for 10 s, 65 C for 10 s and 72 C for 20 s. HRM analysis was immediately performed from 55–95 C (0.01 C/s, 45 acquisitions/ ); and 40 C for 15 s. The derivatives of the raw data (melting curves analysis), normalised melting curves and difference graphs of test samples and wild-type controls were compared using the Eco software, and samples with an aberrant melting pattern were judged to carry a somatic mutation. Samples were run as replicates or triplicates in the test set. Mutation positive samples were repeated in the validation set.
METHODS
CastPCR
Samples
Samples were analysed by castPCR using the BRAF_476_mu and BRAF_473_mu allele assays paired with a gene reference assay for the detection of V600E and V600K mutations, respectively (Life Technologies). Mastermixes were prepared as recommended from the manufacturer in a total volume of 10 mL and 25 ng or 50 ng DNA were used as template. CastPCR was performed using Eco Real-Time PCR instrument (Illumina) with the following PCR conditions: an initial denaturation step of 10 min at 95 C, 5 cycles of 15 s denaturation at 95 C, 1 min extension at 58 C and 40 cycles of 15 s denaturation at 95 C and 1 min extension at 60 C. Real time data were collected during the last 40 cycles of amplification. Data were analysed by calculating the Ct value between amplification reactions for the mutant allele and the reference assay. Samples with a delta(d)Ct of less than 9.96 were considered positive for mutation.
FFPE cell line reference standards for BRAF (25% V600E, 1.4% V600E, 50% V600K, 0.8% V600K and 50% V600R) containing defined ratios of mutant alleles were purchased from Horizon Diagnostics (Cambridge, UK). Additional samples were gathered through the Quality Assurance program from the European Molecular Genetic Quality Network (EMQN). DNA from cell line WN-266-4, harbouring the c.1799_1800TG>AT mutation (p.V600D), was obtained from the American Type Culture Collection (ATCC). Reference human genomic DNA for spiking experiments was purchased from Roche Applied Science, Germany. Mutated DNA was diluted with reference DNA to produce 25%, 10%, 5%, 0.5% and 0.25% mutant allele in the DNA mix. Sixty routinely tested FFPE specimens were selected for validation. Paraffin embedded tissues were cut into 7 mm thick sections. The tumour area was marked on the haematoxylin and eosin (H&E) stained slides by an experienced pathologist and corresponding unstained slides were used for DNA isolation. Genomic DNA was isolated using the ReliaPrep FFPE gDNA Miniprep System kit (Promega, USA). The absolute yield of extracted DNA was quantified using a Nanodrop (Maestro, USA). The purity of the obtained DNA was assessed with the ratio of absorption at 260 nm versus 280 nm and at 260 nm versus 230 nm. Measurements were done in duplicate and the mean was calculated. High resolution melting All BRAF primers published by Kristensen et al., 2012 were checked for secondary structures and primer dimer formation (OligoAnalyser Tool; Integrated DNA Technologies, USA).15 Results revealed a high likelihood of primer dimer formation for the overlapping reverse primer 50 -TGATGGGACCCACTCCATCG-30 due to a stretch of three Gs followed by three Cs (DG = –8.09 kcal/mol). Based on this we decided to redesign the original overlapping primer and use the new design in combination with the already published mutation specific primer and common primer. In short, a mutation specific primer (BRAF mut rev: 50 -TGGGACCCACTCTATCGAGATTTCT-30 ) was combined with a overlapping primer (BRAF rev: 50 CCCACTCCATCGAGATTTC-30 ) and a common primer (BRAF for: 50 AGGTGATTTTGGTCTAGCTACAG-30 ), which amplifies both wild-type and mutation alleles. The common primer covers the first base of codon 600 and thus favours amplification of V600E and V600D mutations, in addition to the wild-type sequence. This assay set up is referred as CADMA assay V600E throughout the text. In order to increase sensitivity for V600K mutation detection, two additional primers were designed. A mutation specific primer (BRAF mut rev V600K: 50 -GGGACCCACTCTATCGAGATTTCTT-30 ) which has two 30 mismatches against wild-type and mutant alleles other than V600K, and a new common primer (BRAF for V600K: 50 -TAAGGTGATTTTGGTCTAGCTACA-30 ). The latter primer ends with the last base of codon 599 and does not cover any base of codon 600, thus enabling nonselective amplification of wild-type and mutant alleles. To distinguish the new set up (common primer V600K together with the BRAF mut rev and the
RESULTS Analytical sensitivity of CADMA assay V600E Primer concentrations and PCR conditions from Kristensen et al. were slightly modified to adapt the assay to the Illumina Eco instrument.15 The assay could detect 1.4% mutant allele in a wild-type background when using 20 ng DNA from commercial available FFPE cell mixtures. Analysis with diluted reference standards indicated the assay was sensitive to 0.5% mutant allele (Fig. 1A). However, 11 of 14 tested dilutions containing 0.25% mutant allele could be distinguished from wild-type replicas. Using a cell line blend the detection sensitivity was 0.25% mutant allele when 10 ng DNA was used. Cross reactivity of CADMA assay V600E with nonV600E mutations Cross reactivity of the assay was evaluated by testing BRAF non-V600E FFPE specimens (all from Horizon Diagnostics, UK) or the cell line WN-266-4, which is heterozygous for the V600D mutation, at various mutation levels. Using reference controls for 50% mutant allele in a wild-type background, V600K and V600R mutations could be detected in all replicas (10 replicas for V600K and 6 replicas for V600R). Decreasing the mutation load to 25% still led to a detection rate of 60% for V600R (3/5) and 40% for V600K (4/10). Twenty ng of DNA were used for these experiments. The assay also showed cross reactivity for V600D (25% mutation). Increasing the DNA input to 50 ng also led to detection of V600D with 10% mutant allele.
778
HUEBNER et al.
Pathology (2017), 49(7), December
Fig. 1 The analytical sensitivity of CADMA assays using 20 ng DNA of FFPE reference controls. (A) V600E assay, (B) V600E-K assay. Arrows highlight the low level mutations.
Viewing the melting plots (normalised and difference) revealed that samples containing V600K or V600R mutations were not distinguishable from samples with low level V600E mutations (3.4%). These samples would be misclassified as V600K/R or vice versa as V600E with low mutation load (Fig. 2). Testing the learning set of 12 samples obtained from the European molecular genetics quality network including specimens with V600E, V600K and V600R mutations confirmed this discrepancy. Specimens with V600E mutation were easily detected, but specimen with V600K or V600R mutations were misclassified as V600E variants with low mutation load. In addition two samples known to harbour BRAF V600K were also tested and would be detected as low level V600E mutation.
mutation specific V600K primer either as a three primer system or as a four primer system. Although a clear separation between V600K and V600E mutations at high mutation load could be obtained when running the assay with V600K mutation specific primer, low level mutations could not be distinguished from the wild-type. A clear differentiation for all V600E and V600K low level mutations were obtained with the other two assay designs. Surprisingly, the assay V600E/K with the redesigned common primer in combination with overlapping primer and the V600E mutation specific primer was as sensitive as the assay incorporating all four primers. Detection sensitivity for V600E mutation decreased to 1.4% mutant allele, however it increased to 0.8% for V600K mutation (Fig. 1B).
Analytical sensitivity of CADMA assay V600E/K
Cross reactivity of CADMA assay V600E/K
In order to provide a better discrimination between V600E and V600K mutations the assay was redesigned. Two new primers were introduced, with a forward primer that does not cover the V600 codon (BRAF for V600K) and a V600K mutation specific reverse primer. The assay was run with genetically defined reference standards using the redesigned forward primer and overlapping reverse primer in combination with the mutation specific V600E primer or the new
We analysed the cross reactivity for V600D and V600R using reference controls and cell line samples as described above. Cross reactivity was seen for V600D at 10% mutation (9 replicas) and V600R at 25% mutation (5 replicas). When diluting the V600R standard down to a fraction of 10% mutant allele, the melting curves were not distinguishable from the wild-type replicas. However, half of the replicas (3/ 6) containing 5% V600D mutant alleles could be distinguished from eight wild-type replicas. Experiments were
Melting curves for non-V600E mutations generated with V600E assay. The reference control 0.8% V600K was not distinguishable from the wild-type control. (A) Difference plot, (B) normalised plot.
Fig. 2
HRM ASSAY FOR LOW LEVEL BRAF MUTATIONS
repeated twice. Melting curves for samples containing 50% mutation were clearly distinct from samples harbouring V600E or V600K mutations. However, a mutation characteristic melting shape was not given any more for samples with less than 50% mutant allele (Fig. 3). Detection of BRAF mutation in clinical specimens For more specific evaluation of the V600E/K assay we used a total set of 60 FFPE carcinoma specimens. Twelve of these samples were pooled human FFPE specimens (2–3 specimens) without tumour enrichment by macrodissection. One of the specimens included in the pool proved to be challenging for DNA isolation as all isolations led to low DNA yield when isolated as one sample. This specimen was included in three of the pooled samples. Out of this sample set two specimens could not be successfully analysed. One sample yielded too little DNA and the other sample showed late amplification (Cq = 44). Noteworthy, this sample was spiked with a low quality FFPE sample DNA. Previous experiments regarding the quantification of nuclear DNA copies for a 96 bp fragment targeting the human EGFR gene using this particular sample revealed a reduced amount of DNA which is accessible for PCR (data not shown). Seventeen samples with V600 mutations were identified, with two of these at a mutation level less than 5% (Fig. 4A). One sample was called as V600K (~0.8% mutant sequence copies) and the other sample was assigned as V600E (~1.4% mutant sequence copies). Both of these samples were pooled human FFPE specimens. Of the remaining BRAF mutated samples, 10 samples were identified with V600E mutations, 4 samples with V600K mutations and one sample was harbouring the
Fig. 3
779
rare V600R variant. One of the samples with V600K mutation status would have been classified as V600E when using the V600E assay instead of the V600E/K assay (Fig. 4B). Furthermore, we identified one sample with an aberrant melting profile uncharacteristic for the tested V600 mutations. The melting curve for this sample contained an additional small interrelated peak. In comparison, a single second peak was specific for V600E, K, R and D mutations (Fig. 5). Sequencing revealed a mutation in the adjacent codon at position c.1801A>G. This change leads to a lysine to glutamic acid substation at codon 601 (K601E). To test the robustness of the technique, longer and shortterm precision (intra-assay and inter-assay variations) were tested using reference controls. For short-term precision, three V600E mutated samples were tested as 3 replicas and two wild-type samples as 5 or 3 replicas, respectively. In order to test for inter-assay variations, the same samples were tested on six different runs which led to a total number of up to 15 replicas for the mutated samples and up to 32 replicas for the wild-type samples. All replicas generated the same qualitative result. Mutation detection via castPCR In order to confirm the results obtained by HRM, predesigned TaqMan Mutation detection assays were purchased from Life Technologies. These assays are based on competitive allelespecific TaqMan PCR (castPCR) technology and allow detection of mutation loads as low as 0.1%.19,20 Assays were run in comparison to a reference gene assay, which detects a mutation-free region of the BRAF gene. Samples with a D Ct of less than 9.96 were considered positive for mutation. If
Cross reactivity of V600E-K assay with V600D and V600R. (A) Mutant allele 50%, (B) mutant allele 25%.
780
HUEBNER et al.
Pathology (2017), 49(7), December
Mutation screen of clinical specimens. (A) Two samples with low mutation load could be identified with HRM and were confirmed by castPCR. Red arrows indicate the mutant profile. In HRM graphs (difference plots), blue lines represent WT control, yellow line reference control 0.8% V600K, green line reference control 1.4% V600E, and pink line the sample. In castPCR, blue line indicates the reference assay, dark red line the V600E assay and green line the V600K assay. Slight cross reactivity of V600E positive sample with the V600K assay can be seen, but not reverse. CT value for this sample was above the threshold of 9.96 (DCt = 13.57) for V600K and therefore classified as mutation negative. DCt value for the V600E assay was 7.55 and sample was classified as mutation detected. (B) Genotype call for patient with V600K mutation when using assay V600E/K in comparison to assay V600E. Using the V600E/K assay melting curve for this sample identical overlaps with melting curve for reference control V600K in contrast to assay V600E. Here, melting curve for patient sample is located between reference controls 25% V600E and 1.4% V600E. Hence it was interpreted as V600E with less than 25% mutant sequence copies in wild-type background. CastPCR confirmed V600K mutation status (DCt = 0.1).
Fig. 4
amplification was only detected for the reference assay, samples were considered as mutation negative, or below the limit of detection for the TaqMan Mutation Detection Assay. Cross reactivity for BRAF_476 was observed with BRAF V600K mutations. Some of the samples which were positive for V600E mutations also gave low signal for V600K. The BRAF_473 assay showed cross reactivity for V600R mutations. The reference control 50% V600R was detected as V600K. All the samples containing V600K and V600E mutations identified by HRM were confirmed by castPCR (Fig. 4A). CastPCR failed to detect the K601E mutation because of the non-specificity of the probe.
DISCUSSION In the present study we designed a HRM assay that is able to differentiate between V600E and V600K mutations at low mutation levels (<2%) in one experimental set up. This level of sensitivity is essential for mutation detection in samples
with low tumour percentage or poor quality fragmented DNA.21 HRM is considered a screening tool to detect mutations without specifying the exact nature of mutation.20 Thus, a second method has to be implemented (mostly sequencing) in order to call the exact variation. To bypass this limitation of HRM we used the CADMA principle for our assay design as described.15 This method is based on a very short amplicon (55 bp in this case) and uses a three primer system. An overlapping primer to amplify both wild-type and mutated sequences and an additional third primer that binds specifically to the mutated allele. This mutation specific primer is designed to introduce melting temperature increasing mutations in the resulting amplicon so creating two products with different melting profiles in the case a mutation is present. One product represents the wild-type background, whereas the other product represents the mutation. In the absence of a mutation only one product is generated. We have taken advantage of the already published
HRM ASSAY FOR LOW LEVEL BRAF MUTATIONS
781
Melting peaks for V600 mutations detected by assay V600E/K using reference controls. (A) Reference controls with 50% mutant allele were used, except for V600E (25% mutant allele). * Reference controls 1.4% V600E and 0.8% V600K. (B) Patient with aberrant melting profile detected with V600E/K assay. This sample was classified as mutated with unknown mutation. Bidirectional Sanger sequencing identified mutation as K601E (c.1801A>G). The position of mutation (T to C change represented by IUB code of Y) is indicated by asterisk.
Fig. 5
CADMA assay for BRAF and modified it for better allelic discrimination.15,22 These modifications included the design of a new overlapping primer, since the original primer was prone to formation of secondary structures, and a new forward primer. The latter ended at codon Thr599 and allowed unbiased amplification of all potential BRAF alleles. As a general rule, mutation status was assigned by looking at the melting peaks in comparison to the controls. In case they did not conform a secondary assay was applied. At first we used the new designed overlapping primer together with the known common and mutation specific reverse primer.15 Using this technique we could detect samples down to 0.25% fraction of V600E mutated allele. Cross reactivity was shown for samples containing V600K and V600R mutations. Samples with 50% mutation level showed the same melting profile as samples with low abundance V600E mutations (3.4%) and therefore determining the true BRAF mutation status was not possible. Our sample set did not include any samples with the rare V600E2 (c.1799_1800delinsAT) mutation and hence we could not test for cross reactivity. However, we believe that there might be a minimised chance of cross reactivity compared to the original assay. There, the overlapping primer does not bind at the codon 600 and thus the inadvertent occurrence of amplifying other BRAF mutations is given. Yet the overlapping primer we used for our design covers the last base of codon 600, interfering with mutations occurring at this site such as V600E2. In particular the G to A transversion (both purine bases) at position c.1800 is of
importance, resulting in a mismatch at the 30 end of the overlapping primer. This matters, as the position of the mismatch between primers and template is crucial. There is a hierarchy of mismatch impacts described with the purine_purine (A-A, AG, G-A and G-G) and C-C mismatches result in the most detrimental effect.23 Despite the resulting mismatch in the overlapping primer, the assay still detected V600D mutation. Based on the employed overlapping primer, we assume that the original CADMA assay had even a higher likelihood of detecting this mutation, if it had been tested for. In a validation study using CADMA PCR as the detection method, Lade-Keller stated that the assay did not show any cross reactivity for V600K containing samples with 50% fraction of mutated allele.21 However, they included a second melting temperature decreasing mutation in addition to those already existing in the original mutation specific reverse primer; also, the utilised overlapping primer was different. Given these circumstances, together with different PCR cycling conditions, the uses of different master mixes and the use of a reference control with a defined allelic mutation ratio rather than a cell line mix might help to explain this discrepancy. In particular, the latter might be of importance as cell line mixtures lack the ability to give precise details regarding defined copy number and allelic burden. A cell line mixture experiment works on the assumption that the cell line is 100% monoclonal and that there is no pipetting error.21 Furthermore these are idealised samples which do not mimic real patient tumour samples, in which most of the DNA is compromised
782
HUEBNER et al.
due to the fixative formalin.24 In order to avoid these limitations of cell line mixtures we decided to use commercially available FFPE reference standards with precise allelic frequencies for our assay development. However, we could not use this strategy for the c.1799_1800TG>AT (V600D) mutation as no reference standard was available for this variant. Instead we used the cell line WN-266-4, which harbours a V600D mutation, and spiked them with a reference DNA for dilution experiments. In order to get a clear differentiation between V600K and V600E mutations we modified the common primer to be equally matched against the c.1798_1799GT>AA (V600K) mutation. After adjusting the common primer, a clear differentiation between V600E mutations and V600K mutations based on the different shape of the melting curve could be achieved. Assay sensitivity for V600K increased to 0.8% mutant allele in a wild-type background. Even at this mutation load the melting profile was different from samples with 1.4% mutated V600E alleles. All samples classified as V600K or V600E mutated by HRM could be confirmed with real time specific allelic PCR (castPCR). This method can identify the exact mutant sequence and allows the detection of mutation levels as low as 0.1%.20 One V600E sample and one V600K sample, both with less than 2% mutated allele, were confirmed with castPCR. As reported by others, we observed slight cross reactivity for eight V600E positive samples with the V600K assay when using castPCR (Cq value slightly below the threshold of 9.96).20,22 However all of these samples showed a low Cq value (1.5–2.5) when tested with the V600E specific TaqMan assay. In addition, the reference control 50% V600R also gave a signal in the V600K castPCR assay (Cq = 8.44) and would have been misclassified as V600K mutation. Using our HRM assay we also could assign V600D and V600R mutations with high fraction of mutated allele (50%) based on the different melting profiles. Samples with 25% mutated allele could also be detected but an exact calling of the mutant sequence based on the shape of the melting curve was no longer possible. In this case, a second method to give the exact mutation status is required. This also applies for low level V600K and V600E mutations as the melting curve resembles V600D/R mutation with 25% mutated allele, thus increasing the chance of misidentification when only relying on HRM. In these cases we recommend a second method be applied. We use castPCR in our routine diagnostic setting as method of choice for these contingencies. Furthermore, the assay detected a rare K601E (c.1801A>G) mutation that probably would have been missed using other detection platforms. In particular, several studies have shown that the FDA approved Cobas 4800 BRAF V600 test failed to detect mutations other than V600E occurring at codon 600.10,25,26 As the melting profile for K601E mutation was very different from the melting profile for the tested codon 600 mutation, the risk of miscalling a mutation based on the melting curve could be minimised. In conclusion, we developed an HRM based assay for the detection of low fraction mutated allele for V600E and V600K mutations in BRAF. Other mutations occurring in codon 600 also can be detected, albeit at lower analytical sensitivity. We could show that with HRM a precise and accurate calling of the mutation based on the different shapes of the melting curve is possible but is highly dependent on suitable reference controls to which the curves can be
Pathology (2017), 49(7), December
matched. The success of our assay relies heavily on commercial available reference controls with tightly defined allele burden and common genetic background which allowed us to distinguish V600K and V600E mutations even at low levels. Using this approach we could score all samples in our validation set correctly, as confirmed by castPCR. In our eyes, the strength of our assay design lies in the precise differentiating of the two most common mutations in melanoma in only one experimental set up using cost-effective HRM as the detection platform. Acknowledgements: We thank Bradford Tetlow and John Mackay for critical reading of the manuscript. Conflicts of interest and sources of funding: This work was supported by Labtests Limited, a subsidiary of Healthscope NZ. The authors state that there are no conflicts of interest to disclose. Address for correspondence: Claudia Huebner, Labtests Auckland, 37/43 Carbine Rd, Auckland 1642, New Zealand. E-mail: claudiahuebner17@ hotmail.com
References 1. Bogenrieder T, Herlyn M. The molecular pathology of cutaneous melanoma. Cancer Biomark 2011; 9: 267–86. 2. Davies H, Bignell GR, Cox C, et al. Mutations of the BRAF gene in human cancer. Nature 2002; 417: 949–54. 3. Ueda M, Toji E, Nunobiki O, et al. Mutational analysis of the BRAF gene in human tumor cells. Hum Cell 2008; 21: 13–7. 4. Vultur A, Villanueva J, Herlyn M. Targeting BRAF in advanced melanoma: a first step toward manageable disease. Clin Cancer Res 2011; 17: 1658–63. 5. Luke JJ, Flaherty KT, Ribas A, Long GV. Targeted agents and immunotherapies: optimizing outcomes in melanoma. Nat Rev Clin Oncol 2017; 14: 463–82. 6. Rubinstein JC, Sznol M, Pavlick AC, et al. Incidence of the V600K mutation among melanoma patients with BRAF mutations, and potential therapeutic response to the specific BRAF inhibitor PLX4032. J Transl Med 2010; 8: 67. 7. Amanuel B, Grieu F, Kular J, Millward M, Iacopetta B. Incidence of BRAF p.Val600Glu and p.Val600Lys mutations in a consecutive series of 183 metastatic melanoma patients from a high incidence region. Pathology 2012; 44: 357–9. 8. Chapman PB, Hauschild A, Robert C, et al. The BRIM-3 Study Group: improved survival with vemurafenib in melanoma with BRAF V600E mutation. N Engl J Med 2011; 364: 2507–16. 9. Sosman JA, Kim KB, Schuchter L, et al. Survival in BRAF V600mutant advanced melanoma treated with vemurafenib. N Engl J Med 2012; 366: 707–14. 10. Ihle MA, Fassunke J, König K, et al. Comparison of high resolution melting analysis, pyrosequencing, next generation sequencing and immunohistochemistry to conventional Sanger sequencing for the detection of p.V600E and non-p.V600E BRAF mutations. BMC Cancer 2014; 14: 13. 11. Wittwer CT, Reed GH, Gundry CN, Vandersteen JG, Pryor RJ. Highresolution genotyping by amplicon melting analysis using LCGreen. Clin Chem 2003; 49: 853–60. 12. Pichler M, Balic M, Stadelmeyer E, et al. Evaluation of high-resolution melting analysis as a diagnostic tool to detect the BRAF V600E mutation in colorectal tumors. J Mol Diagn 2009; 11: 140–7. 13. Boyd EM, Bench AJ, van ’t Veer MP, et al. High resolution melting analysis for detection of BRAF exon 15 mutations in hairy cell leukaemia and other lymphoid malignancies. Br J Haematol 2011; 155: 609–12. 14. Ney JT, Froehner S, Roesler A, Buettner R, Merkelbach-Bruse S. Highresolution melting analysis as a sensitive prescreening diagnostic tool to detect KRAS, BRAF, PIK3CA, and AKT1 mutations in formalin-fixed, paraffin-embedded tissues. Arch Pathol Lab Med 2012; 136: 983–92. 15. Kristensen LS, Andersen GB, Hager H, Hansen LL. Competitive amplification of differentially melting amplicons (CADMA) enables sensitive and direct detection of all mutation types by high-resolution melting analysis. Hum Mutat 2012; 33: 264–71.
HRM ASSAY FOR LOW LEVEL BRAF MUTATIONS
16. Fadhi W, Ibrahem S, Seth R, Ilyas M. Quick-multiplex-consensus (QMC)-PCR followed by high-resolution melting: a simple and robust method for mutation detection in formalin-fixed paraffin-embedded tissue. J Clin Pathol 2010; 63: 134–40. 17. Mancini I, Santucci C, Sestini R, et al. The use of COLD-PCR and highresolution melting analysis improves the limit of detection of KRAS and BRAF mutations in colorectal cancer. J Mol Diagn 2010; 12: 705–11. 18. Zhou L, Wang Y, Wittwer CT. Rare allele enrichment and detection by allele-specific PCR, competitive probe blocking, and melting analysis. BioTechniques 2011; 50: 311–8. 19. Didelot A, Le Corre D, Luscan A, et al. Competitive allele specific TaqMan PCR for KRAS, BRAF and EGFR mutation detection in clinical formalin fixed paraffin embedded samples. Exp Mol Pathol 2012; 3: 275–80. 20. Richter A, Grieu F, Carrello A, et al. A multisite blinded study for the detection of BRAF mutations in formalin-fixed, paraffin-embedded malignant melanoma. Sci Rep 2013; 3: 1659. 21. Mariani S, Bertero L, Osella-Abate S, et al. Extreme assay sensitivity in molecular diagnostics further unveils intratumour heterogeneity in
22. 23.
24. 25.
26.
783
metastatic colorectal cancer as well as artifactual low-frequency mutations in the KRAS gene. Br J Cancer 2017; 117: 358–66. Lade-Keller J, Rømer KM, Guldberg P, et al. Evaluation of BRAF mutation testing methodologies in formalin-fixed paraffin-embedded cutaneous melanomas. J Mol Diagn 2013; 15: 70–80. Stadhouders R, Pas SD, Anber J, Voermans J, Mes TH, Schutten M. The effect of primer-template mismatches on the detection and quantification of nucleic acids using the 50 nuclease assay. J Mol Diagn 2010; 12: 109–17. Gilbert MT, Haselkorn T, Bunce M, et al. The isolation of nucleic acids from fixed, paraffin-embedded tissues-which methods are useful when? PLoS One 2007; 2: e537. Halait H, Demartin K, Shah S, et al. Analytical performance of a realtime PCR-based assay for V600 mutations in the BRAF gene, used as the companion diagnostic test for the novel BRAF inhibitor vemurafenib in metastatic melanoma. Diag Mol Pathol 2012; 21: 1–8. Curry JL, Torres-Cabala CA, Tetzlaff MT, Bowman C, Prieto VG. Molecular platforms utilized to detect BRAF V600E mutation in melanoma. Semin Cutan Med Surg 2012; 31: 267–73.
M O L E C U L A R PAT H O L O G Y
A HRM assay for identification of low level BRAF V600E and V600K mutations using the CADMA principle in FFPE specimens CLAUDIA HUEBNER, REMENY WEBER
AND
RICHARD LLOYDD
Department of Anatomical Pathology, Labtests Auckland Limited, New Zealand
Summary Melanoma patients with BRAF V600E and V600K mutations show complete or partial response to vemurafenib. Detection assays often scan for the common V600E mutation rather than the rare V600K variant, although this mutation can be found in a high proportion of melanoma patients in the South Pacific. Herein, we describe a BRAF high resolution melting (HRM) assay that can differentiate low level of V600E and V600K mutations using formalin fixed, paraffin embedded (FFPE) reference standards for assay validation. The assay is based on the competitive amplification of differentially melting amplicons (CADMA principle) and has a limit of detection of 0.8% mutant allele for V600K and 1.4% mutant allele for V600E. A differentiation between the two mutations based on the melting profile is possible even at low mutation level. Sixty FFPE specimens were scanned and mutations could be scored correctly as confirmed by castPCR. In summary, the developed HRM assay is suitable for detection of V600K and V600E mutations and proved to be reliable and cost effective in a diagnostic environment. Key words: BRAF mutational analysis; HRM; CADMA; reference standards. Received 28 March, revised 9 August, accepted 15 August 2017
INTRODUCTION BRAF codes for a non-receptor serine/threonine kinase component in the RAF-MEK-ERK signal transduction phosphorylation cascade. This pathway controls a variety of biological processes such as cell proliferation, differentiation and survival. The mutated version of the BRAF kinase results in constitutive activity, which is believed to be actively involved in oncogenic proliferation.1 Somatic mutations of BRAF have been documented in different kinds of tumours, predominantly in malignant melanoma, thyroid papillary cancer, sporadic colorectal tumours, low-grade ovarian serous carcinoma, and lung tumours.2,3 Hot spot mutations have been observed in exon 15 at codon 600 with the V600E (c.1799T>A, p.Val600Glu) mutation the most common. In general, 50–70% of melanoma tumours harbour a BRAF mutation, and of these, 80% are positive for BRAF V600E.4 Another mutation at this site involves the mutation of two adjacent nucleotides, substituting lysine for valine (c.1798_1799delinsAA, V600K). V600K mutations
are most frequently found in older patients (aged 65 years) and/or in those with evidence of chronic UV exposure.5 Recent studies have reported that the frequency of the V600K mutation may be as high as one-third of all BRAF mutations in melanoma.6,7 BRAF inhibitors have been designed which target the activated kinase domain. In 2011, vemurafenib (trade name Zelboraf, also known as PLX4032 and RO5185426) received approval from the US Food and Drug administration (FDA) for patients with metastatic or inoperable tumours that carry the BRAF activating V600E mutation.8,9 Patients carrying the V600K mutation have also shown response to vemurafenib.5,6 Based on the outstanding importance of BRAF as a molecular marker in cancer treatment, accurate identification of mutant cases is absolutely critical for appropriate therapy. Different methods are used to detect for BRAF mutations, which vary regarding specificity, sensitivity, turnaround time and costs.10 The traditional Sanger sequencing shows poor analytical sensitivity and often needs greater than 20% of tumour load in a specimen to render a reliable result. Pyrosequencing and next generation sequencing (NGS) are very sensitive with a reported limit of detection of 1% for NGS and 5% for pyrosequencing in a background of wildtype alleles.10 On the other hand, the drawbacks are the costs and the complex and challenging data analysis, which requires a suite of sophisticated bioinformatic tools to make sense of the data. Reports have shown that the CE-IVD marked Cobas 4800 BRAF V600 test (Roche Molecular Diagnostics, USA), widely used in the molecular analyses of melanoma in clinical settings, failed to detect 16.3% of the mutations eligible for therapy with vemurafenib.10 High resolution melting (HRM) analysis is a rapid, closedtube and comparatively inexpensive technique. It is based on the fluorescent monitoring of the dissociation behaviour of DNA when exposed to increasing temperatures. The HRM melting profile gives a specific sequence-related profile that differentiates wild-type sequences from homozygote or heterozygote variants.11 The sensitivity for HRM has been described as between 5–10% mutant allele, depending on the precise mutation change and amplicon length.12–14 However, identification of low-abundance mutations (below 5%) using this technique is only possible with modifications.15–18 Recently, Kristensen et al. described a modified HRM method (competitive amplification of differentially melting amplicons; CADMA) using a competitive oligonucleotide and very short amplicons, leading to sensitivities as low as 0.25% mutated V600E allele.15 The principle of the CADMA
Print ISSN 0031-3025/Online ISSN 1465-3931 © 2017 Royal College of Pathologists of Australasia. Published by Elsevier B.V. All rights reserved. DOI: https://doi.org/10.1016/j.pathol.2017.08.011
HRM ASSAY FOR LOW LEVEL BRAF MUTATIONS
777
method is the simultaneous amplification of wild-type and mutation containing sequences using a mutation specific primer, an overlapping primer that targets both sequences, and a common primer, which give rise to two different amplicons with different melting properties. The CADMA method is highly sensitive for the targeted mutation, however less specific for other significant mutations present in the amplicon. In order to get high sensitivities for different mutations a new assay has to be designed for each mutation of interest. Here we described a HRM assay based on the CADMA principle, which allows the simultaneous detection and accurate calling of V600E as well as V600K mutations in one assay. In order to evaluate the assay, commercial reference formalin fixed, paraffin embedded (FFPE) standards containing defined ratios of mutant alleles were used.
overlapping primer) from the above described, the assay was named CADMA assay V600E/K. In addition, one more experimental layout was performed incorporating the following four primers: BRAF for V600K, the two mutation specific primers and the overlapping primer BRAF rev. Each reaction mixture contained 2 mL DNA (10 ng/mL or 25 ng/mL), 300 nM mutation specific primer, 300 nM common primer, 150 nM overlapping primer and 5 mL MeltDoctor HRM Master Mix (Life Technologies, USA) in a total volume of 10 mL. Polymerase chain reaction (PCR) cycling and HRM analysis were performed on the Eco Real-Time PCR instrument (Illumina, USA), with an initial denaturation of 95 C for 10 min, followed by 48 cycles of 95 C for 10 s, 65 C for 10 s and 72 C for 20 s. HRM analysis was immediately performed from 55–95 C (0.01 C/s, 45 acquisitions/ ); and 40 C for 15 s. The derivatives of the raw data (melting curves analysis), normalised melting curves and difference graphs of test samples and wild-type controls were compared using the Eco software, and samples with an aberrant melting pattern were judged to carry a somatic mutation. Samples were run as replicates or triplicates in the test set. Mutation positive samples were repeated in the validation set.
METHODS
CastPCR
Samples
Samples were analysed by castPCR using the BRAF_476_mu and BRAF_473_mu allele assays paired with a gene reference assay for the detection of V600E and V600K mutations, respectively (Life Technologies). Mastermixes were prepared as recommended from the manufacturer in a total volume of 10 mL and 25 ng or 50 ng DNA were used as template. CastPCR was performed using Eco Real-Time PCR instrument (Illumina) with the following PCR conditions: an initial denaturation step of 10 min at 95 C, 5 cycles of 15 s denaturation at 95 C, 1 min extension at 58 C and 40 cycles of 15 s denaturation at 95 C and 1 min extension at 60 C. Real time data were collected during the last 40 cycles of amplification. Data were analysed by calculating the Ct value between amplification reactions for the mutant allele and the reference assay. Samples with a delta(d)Ct of less than 9.96 were considered positive for mutation.
FFPE cell line reference standards for BRAF (25% V600E, 1.4% V600E, 50% V600K, 0.8% V600K and 50% V600R) containing defined ratios of mutant alleles were purchased from Horizon Diagnostics (Cambridge, UK). Additional samples were gathered through the Quality Assurance program from the European Molecular Genetic Quality Network (EMQN). DNA from cell line WN-266-4, harbouring the c.1799_1800TG>AT mutation (p.V600D), was obtained from the American Type Culture Collection (ATCC). Reference human genomic DNA for spiking experiments was purchased from Roche Applied Science, Germany. Mutated DNA was diluted with reference DNA to produce 25%, 10%, 5%, 0.5% and 0.25% mutant allele in the DNA mix. Sixty routinely tested FFPE specimens were selected for validation. Paraffin embedded tissues were cut into 7 mm thick sections. The tumour area was marked on the haematoxylin and eosin (H&E) stained slides by an experienced pathologist and corresponding unstained slides were used for DNA isolation. Genomic DNA was isolated using the ReliaPrep FFPE gDNA Miniprep System kit (Promega, USA). The absolute yield of extracted DNA was quantified using a Nanodrop (Maestro, USA). The purity of the obtained DNA was assessed with the ratio of absorption at 260 nm versus 280 nm and at 260 nm versus 230 nm. Measurements were done in duplicate and the mean was calculated. High resolution melting All BRAF primers published by Kristensen et al., 2012 were checked for secondary structures and primer dimer formation (OligoAnalyser Tool; Integrated DNA Technologies, USA).15 Results revealed a high likelihood of primer dimer formation for the overlapping reverse primer 50 -TGATGGGACCCACTCCATCG-30 due to a stretch of three Gs followed by three Cs (DG = –8.09 kcal/mol). Based on this we decided to redesign the original overlapping primer and use the new design in combination with the already published mutation specific primer and common primer. In short, a mutation specific primer (BRAF mut rev: 50 -TGGGACCCACTCTATCGAGATTTCT-30 ) was combined with a overlapping primer (BRAF rev: 50 CCCACTCCATCGAGATTTC-30 ) and a common primer (BRAF for: 50 AGGTGATTTTGGTCTAGCTACAG-30 ), which amplifies both wild-type and mutation alleles. The common primer covers the first base of codon 600 and thus favours amplification of V600E and V600D mutations, in addition to the wild-type sequence. This assay set up is referred as CADMA assay V600E throughout the text. In order to increase sensitivity for V600K mutation detection, two additional primers were designed. A mutation specific primer (BRAF mut rev V600K: 50 -GGGACCCACTCTATCGAGATTTCTT-30 ) which has two 30 mismatches against wild-type and mutant alleles other than V600K, and a new common primer (BRAF for V600K: 50 -TAAGGTGATTTTGGTCTAGCTACA-30 ). The latter primer ends with the last base of codon 599 and does not cover any base of codon 600, thus enabling nonselective amplification of wild-type and mutant alleles. To distinguish the new set up (common primer V600K together with the BRAF mut rev and the
RESULTS Analytical sensitivity of CADMA assay V600E Primer concentrations and PCR conditions from Kristensen et al. were slightly modified to adapt the assay to the Illumina Eco instrument.15 The assay could detect 1.4% mutant allele in a wild-type background when using 20 ng DNA from commercial available FFPE cell mixtures. Analysis with diluted reference standards indicated the assay was sensitive to 0.5% mutant allele (Fig. 1A). However, 11 of 14 tested dilutions containing 0.25% mutant allele could be distinguished from wild-type replicas. Using a cell line blend the detection sensitivity was 0.25% mutant allele when 10 ng DNA was used. Cross reactivity of CADMA assay V600E with nonV600E mutations Cross reactivity of the assay was evaluated by testing BRAF non-V600E FFPE specimens (all from Horizon Diagnostics, UK) or the cell line WN-266-4, which is heterozygous for the V600D mutation, at various mutation levels. Using reference controls for 50% mutant allele in a wild-type background, V600K and V600R mutations could be detected in all replicas (10 replicas for V600K and 6 replicas for V600R). Decreasing the mutation load to 25% still led to a detection rate of 60% for V600R (3/5) and 40% for V600K (4/10). Twenty ng of DNA were used for these experiments. The assay also showed cross reactivity for V600D (25% mutation). Increasing the DNA input to 50 ng also led to detection of V600D with 10% mutant allele.
778
HUEBNER et al.
Pathology (2017), 49(7), December
Fig. 1 The analytical sensitivity of CADMA assays using 20 ng DNA of FFPE reference controls. (A) V600E assay, (B) V600E-K assay. Arrows highlight the low level mutations.
Viewing the melting plots (normalised and difference) revealed that samples containing V600K or V600R mutations were not distinguishable from samples with low level V600E mutations (3.4%). These samples would be misclassified as V600K/R or vice versa as V600E with low mutation load (Fig. 2). Testing the learning set of 12 samples obtained from the European molecular genetics quality network including specimens with V600E, V600K and V600R mutations confirmed this discrepancy. Specimens with V600E mutation were easily detected, but specimen with V600K or V600R mutations were misclassified as V600E variants with low mutation load. In addition two samples known to harbour BRAF V600K were also tested and would be detected as low level V600E mutation.
mutation specific V600K primer either as a three primer system or as a four primer system. Although a clear separation between V600K and V600E mutations at high mutation load could be obtained when running the assay with V600K mutation specific primer, low level mutations could not be distinguished from the wild-type. A clear differentiation for all V600E and V600K low level mutations were obtained with the other two assay designs. Surprisingly, the assay V600E/K with the redesigned common primer in combination with overlapping primer and the V600E mutation specific primer was as sensitive as the assay incorporating all four primers. Detection sensitivity for V600E mutation decreased to 1.4% mutant allele, however it increased to 0.8% for V600K mutation (Fig. 1B).
Analytical sensitivity of CADMA assay V600E/K
Cross reactivity of CADMA assay V600E/K
In order to provide a better discrimination between V600E and V600K mutations the assay was redesigned. Two new primers were introduced, with a forward primer that does not cover the V600 codon (BRAF for V600K) and a V600K mutation specific reverse primer. The assay was run with genetically defined reference standards using the redesigned forward primer and overlapping reverse primer in combination with the mutation specific V600E primer or the new
We analysed the cross reactivity for V600D and V600R using reference controls and cell line samples as described above. Cross reactivity was seen for V600D at 10% mutation (9 replicas) and V600R at 25% mutation (5 replicas). When diluting the V600R standard down to a fraction of 10% mutant allele, the melting curves were not distinguishable from the wild-type replicas. However, half of the replicas (3/ 6) containing 5% V600D mutant alleles could be distinguished from eight wild-type replicas. Experiments were
Melting curves for non-V600E mutations generated with V600E assay. The reference control 0.8% V600K was not distinguishable from the wild-type control. (A) Difference plot, (B) normalised plot.
Fig. 2
HRM ASSAY FOR LOW LEVEL BRAF MUTATIONS
repeated twice. Melting curves for samples containing 50% mutation were clearly distinct from samples harbouring V600E or V600K mutations. However, a mutation characteristic melting shape was not given any more for samples with less than 50% mutant allele (Fig. 3). Detection of BRAF mutation in clinical specimens For more specific evaluation of the V600E/K assay we used a total set of 60 FFPE carcinoma specimens. Twelve of these samples were pooled human FFPE specimens (2–3 specimens) without tumour enrichment by macrodissection. One of the specimens included in the pool proved to be challenging for DNA isolation as all isolations led to low DNA yield when isolated as one sample. This specimen was included in three of the pooled samples. Out of this sample set two specimens could not be successfully analysed. One sample yielded too little DNA and the other sample showed late amplification (Cq = 44). Noteworthy, this sample was spiked with a low quality FFPE sample DNA. Previous experiments regarding the quantification of nuclear DNA copies for a 96 bp fragment targeting the human EGFR gene using this particular sample revealed a reduced amount of DNA which is accessible for PCR (data not shown). Seventeen samples with V600 mutations were identified, with two of these at a mutation level less than 5% (Fig. 4A). One sample was called as V600K (~0.8% mutant sequence copies) and the other sample was assigned as V600E (~1.4% mutant sequence copies). Both of these samples were pooled human FFPE specimens. Of the remaining BRAF mutated samples, 10 samples were identified with V600E mutations, 4 samples with V600K mutations and one sample was harbouring the
Fig. 3
779
rare V600R variant. One of the samples with V600K mutation status would have been classified as V600E when using the V600E assay instead of the V600E/K assay (Fig. 4B). Furthermore, we identified one sample with an aberrant melting profile uncharacteristic for the tested V600 mutations. The melting curve for this sample contained an additional small interrelated peak. In comparison, a single second peak was specific for V600E, K, R and D mutations (Fig. 5). Sequencing revealed a mutation in the adjacent codon at position c.1801A>G. This change leads to a lysine to glutamic acid substation at codon 601 (K601E). To test the robustness of the technique, longer and shortterm precision (intra-assay and inter-assay variations) were tested using reference controls. For short-term precision, three V600E mutated samples were tested as 3 replicas and two wild-type samples as 5 or 3 replicas, respectively. In order to test for inter-assay variations, the same samples were tested on six different runs which led to a total number of up to 15 replicas for the mutated samples and up to 32 replicas for the wild-type samples. All replicas generated the same qualitative result. Mutation detection via castPCR In order to confirm the results obtained by HRM, predesigned TaqMan Mutation detection assays were purchased from Life Technologies. These assays are based on competitive allelespecific TaqMan PCR (castPCR) technology and allow detection of mutation loads as low as 0.1%.19,20 Assays were run in comparison to a reference gene assay, which detects a mutation-free region of the BRAF gene. Samples with a D Ct of less than 9.96 were considered positive for mutation. If
Cross reactivity of V600E-K assay with V600D and V600R. (A) Mutant allele 50%, (B) mutant allele 25%.
780
HUEBNER et al.
Pathology (2017), 49(7), December
Mutation screen of clinical specimens. (A) Two samples with low mutation load could be identified with HRM and were confirmed by castPCR. Red arrows indicate the mutant profile. In HRM graphs (difference plots), blue lines represent WT control, yellow line reference control 0.8% V600K, green line reference control 1.4% V600E, and pink line the sample. In castPCR, blue line indicates the reference assay, dark red line the V600E assay and green line the V600K assay. Slight cross reactivity of V600E positive sample with the V600K assay can be seen, but not reverse. CT value for this sample was above the threshold of 9.96 (DCt = 13.57) for V600K and therefore classified as mutation negative. DCt value for the V600E assay was 7.55 and sample was classified as mutation detected. (B) Genotype call for patient with V600K mutation when using assay V600E/K in comparison to assay V600E. Using the V600E/K assay melting curve for this sample identical overlaps with melting curve for reference control V600K in contrast to assay V600E. Here, melting curve for patient sample is located between reference controls 25% V600E and 1.4% V600E. Hence it was interpreted as V600E with less than 25% mutant sequence copies in wild-type background. CastPCR confirmed V600K mutation status (DCt = 0.1).
Fig. 4
amplification was only detected for the reference assay, samples were considered as mutation negative, or below the limit of detection for the TaqMan Mutation Detection Assay. Cross reactivity for BRAF_476 was observed with BRAF V600K mutations. Some of the samples which were positive for V600E mutations also gave low signal for V600K. The BRAF_473 assay showed cross reactivity for V600R mutations. The reference control 50% V600R was detected as V600K. All the samples containing V600K and V600E mutations identified by HRM were confirmed by castPCR (Fig. 4A). CastPCR failed to detect the K601E mutation because of the non-specificity of the probe.
DISCUSSION In the present study we designed a HRM assay that is able to differentiate between V600E and V600K mutations at low mutation levels (<2%) in one experimental set up. This level of sensitivity is essential for mutation detection in samples
with low tumour percentage or poor quality fragmented DNA.21 HRM is considered a screening tool to detect mutations without specifying the exact nature of mutation.20 Thus, a second method has to be implemented (mostly sequencing) in order to call the exact variation. To bypass this limitation of HRM we used the CADMA principle for our assay design as described.15 This method is based on a very short amplicon (55 bp in this case) and uses a three primer system. An overlapping primer to amplify both wild-type and mutated sequences and an additional third primer that binds specifically to the mutated allele. This mutation specific primer is designed to introduce melting temperature increasing mutations in the resulting amplicon so creating two products with different melting profiles in the case a mutation is present. One product represents the wild-type background, whereas the other product represents the mutation. In the absence of a mutation only one product is generated. We have taken advantage of the already published
HRM ASSAY FOR LOW LEVEL BRAF MUTATIONS
781
Melting peaks for V600 mutations detected by assay V600E/K using reference controls. (A) Reference controls with 50% mutant allele were used, except for V600E (25% mutant allele). * Reference controls 1.4% V600E and 0.8% V600K. (B) Patient with aberrant melting profile detected with V600E/K assay. This sample was classified as mutated with unknown mutation. Bidirectional Sanger sequencing identified mutation as K601E (c.1801A>G). The position of mutation (T to C change represented by IUB code of Y) is indicated by asterisk.
Fig. 5
CADMA assay for BRAF and modified it for better allelic discrimination.15,22 These modifications included the design of a new overlapping primer, since the original primer was prone to formation of secondary structures, and a new forward primer. The latter ended at codon Thr599 and allowed unbiased amplification of all potential BRAF alleles. As a general rule, mutation status was assigned by looking at the melting peaks in comparison to the controls. In case they did not conform a secondary assay was applied. At first we used the new designed overlapping primer together with the known common and mutation specific reverse primer.15 Using this technique we could detect samples down to 0.25% fraction of V600E mutated allele. Cross reactivity was shown for samples containing V600K and V600R mutations. Samples with 50% mutation level showed the same melting profile as samples with low abundance V600E mutations (3.4%) and therefore determining the true BRAF mutation status was not possible. Our sample set did not include any samples with the rare V600E2 (c.1799_1800delinsAT) mutation and hence we could not test for cross reactivity. However, we believe that there might be a minimised chance of cross reactivity compared to the original assay. There, the overlapping primer does not bind at the codon 600 and thus the inadvertent occurrence of amplifying other BRAF mutations is given. Yet the overlapping primer we used for our design covers the last base of codon 600, interfering with mutations occurring at this site such as V600E2. In particular the G to A transversion (both purine bases) at position c.1800 is of
importance, resulting in a mismatch at the 30 end of the overlapping primer. This matters, as the position of the mismatch between primers and template is crucial. There is a hierarchy of mismatch impacts described with the purine_purine (A-A, AG, G-A and G-G) and C-C mismatches result in the most detrimental effect.23 Despite the resulting mismatch in the overlapping primer, the assay still detected V600D mutation. Based on the employed overlapping primer, we assume that the original CADMA assay had even a higher likelihood of detecting this mutation, if it had been tested for. In a validation study using CADMA PCR as the detection method, Lade-Keller stated that the assay did not show any cross reactivity for V600K containing samples with 50% fraction of mutated allele.21 However, they included a second melting temperature decreasing mutation in addition to those already existing in the original mutation specific reverse primer; also, the utilised overlapping primer was different. Given these circumstances, together with different PCR cycling conditions, the uses of different master mixes and the use of a reference control with a defined allelic mutation ratio rather than a cell line mix might help to explain this discrepancy. In particular, the latter might be of importance as cell line mixtures lack the ability to give precise details regarding defined copy number and allelic burden. A cell line mixture experiment works on the assumption that the cell line is 100% monoclonal and that there is no pipetting error.21 Furthermore these are idealised samples which do not mimic real patient tumour samples, in which most of the DNA is compromised
782
HUEBNER et al.
due to the fixative formalin.24 In order to avoid these limitations of cell line mixtures we decided to use commercially available FFPE reference standards with precise allelic frequencies for our assay development. However, we could not use this strategy for the c.1799_1800TG>AT (V600D) mutation as no reference standard was available for this variant. Instead we used the cell line WN-266-4, which harbours a V600D mutation, and spiked them with a reference DNA for dilution experiments. In order to get a clear differentiation between V600K and V600E mutations we modified the common primer to be equally matched against the c.1798_1799GT>AA (V600K) mutation. After adjusting the common primer, a clear differentiation between V600E mutations and V600K mutations based on the different shape of the melting curve could be achieved. Assay sensitivity for V600K increased to 0.8% mutant allele in a wild-type background. Even at this mutation load the melting profile was different from samples with 1.4% mutated V600E alleles. All samples classified as V600K or V600E mutated by HRM could be confirmed with real time specific allelic PCR (castPCR). This method can identify the exact mutant sequence and allows the detection of mutation levels as low as 0.1%.20 One V600E sample and one V600K sample, both with less than 2% mutated allele, were confirmed with castPCR. As reported by others, we observed slight cross reactivity for eight V600E positive samples with the V600K assay when using castPCR (Cq value slightly below the threshold of 9.96).20,22 However all of these samples showed a low Cq value (1.5–2.5) when tested with the V600E specific TaqMan assay. In addition, the reference control 50% V600R also gave a signal in the V600K castPCR assay (Cq = 8.44) and would have been misclassified as V600K mutation. Using our HRM assay we also could assign V600D and V600R mutations with high fraction of mutated allele (50%) based on the different melting profiles. Samples with 25% mutated allele could also be detected but an exact calling of the mutant sequence based on the shape of the melting curve was no longer possible. In this case, a second method to give the exact mutation status is required. This also applies for low level V600K and V600E mutations as the melting curve resembles V600D/R mutation with 25% mutated allele, thus increasing the chance of misidentification when only relying on HRM. In these cases we recommend a second method be applied. We use castPCR in our routine diagnostic setting as method of choice for these contingencies. Furthermore, the assay detected a rare K601E (c.1801A>G) mutation that probably would have been missed using other detection platforms. In particular, several studies have shown that the FDA approved Cobas 4800 BRAF V600 test failed to detect mutations other than V600E occurring at codon 600.10,25,26 As the melting profile for K601E mutation was very different from the melting profile for the tested codon 600 mutation, the risk of miscalling a mutation based on the melting curve could be minimised. In conclusion, we developed an HRM based assay for the detection of low fraction mutated allele for V600E and V600K mutations in BRAF. Other mutations occurring in codon 600 also can be detected, albeit at lower analytical sensitivity. We could show that with HRM a precise and accurate calling of the mutation based on the different shapes of the melting curve is possible but is highly dependent on suitable reference controls to which the curves can be
Pathology (2017), 49(7), December
matched. The success of our assay relies heavily on commercial available reference controls with tightly defined allele burden and common genetic background which allowed us to distinguish V600K and V600E mutations even at low levels. Using this approach we could score all samples in our validation set correctly, as confirmed by castPCR. In our eyes, the strength of our assay design lies in the precise differentiating of the two most common mutations in melanoma in only one experimental set up using cost-effective HRM as the detection platform. Acknowledgements: We thank Bradford Tetlow and John Mackay for critical reading of the manuscript. Conflicts of interest and sources of funding: This work was supported by Labtests Limited, a subsidiary of Healthscope NZ. The authors state that there are no conflicts of interest to disclose. Address for correspondence: Claudia Huebner, Labtests Auckland, 37/43 Carbine Rd, Auckland 1642, New Zealand. E-mail: claudiahuebner17@ hotmail.com
References 1. Bogenrieder T, Herlyn M. The molecular pathology of cutaneous melanoma. Cancer Biomark 2011; 9: 267–86. 2. Davies H, Bignell GR, Cox C, et al. Mutations of the BRAF gene in human cancer. Nature 2002; 417: 949–54. 3. Ueda M, Toji E, Nunobiki O, et al. Mutational analysis of the BRAF gene in human tumor cells. Hum Cell 2008; 21: 13–7. 4. Vultur A, Villanueva J, Herlyn M. Targeting BRAF in advanced melanoma: a first step toward manageable disease. Clin Cancer Res 2011; 17: 1658–63. 5. Luke JJ, Flaherty KT, Ribas A, Long GV. Targeted agents and immunotherapies: optimizing outcomes in melanoma. Nat Rev Clin Oncol 2017; 14: 463–82. 6. Rubinstein JC, Sznol M, Pavlick AC, et al. Incidence of the V600K mutation among melanoma patients with BRAF mutations, and potential therapeutic response to the specific BRAF inhibitor PLX4032. J Transl Med 2010; 8: 67. 7. Amanuel B, Grieu F, Kular J, Millward M, Iacopetta B. Incidence of BRAF p.Val600Glu and p.Val600Lys mutations in a consecutive series of 183 metastatic melanoma patients from a high incidence region. Pathology 2012; 44: 357–9. 8. Chapman PB, Hauschild A, Robert C, et al. The BRIM-3 Study Group: improved survival with vemurafenib in melanoma with BRAF V600E mutation. N Engl J Med 2011; 364: 2507–16. 9. Sosman JA, Kim KB, Schuchter L, et al. Survival in BRAF V600mutant advanced melanoma treated with vemurafenib. N Engl J Med 2012; 366: 707–14. 10. Ihle MA, Fassunke J, König K, et al. Comparison of high resolution melting analysis, pyrosequencing, next generation sequencing and immunohistochemistry to conventional Sanger sequencing for the detection of p.V600E and non-p.V600E BRAF mutations. BMC Cancer 2014; 14: 13. 11. Wittwer CT, Reed GH, Gundry CN, Vandersteen JG, Pryor RJ. Highresolution genotyping by amplicon melting analysis using LCGreen. Clin Chem 2003; 49: 853–60. 12. Pichler M, Balic M, Stadelmeyer E, et al. Evaluation of high-resolution melting analysis as a diagnostic tool to detect the BRAF V600E mutation in colorectal tumors. J Mol Diagn 2009; 11: 140–7. 13. Boyd EM, Bench AJ, van ’t Veer MP, et al. High resolution melting analysis for detection of BRAF exon 15 mutations in hairy cell leukaemia and other lymphoid malignancies. Br J Haematol 2011; 155: 609–12. 14. Ney JT, Froehner S, Roesler A, Buettner R, Merkelbach-Bruse S. Highresolution melting analysis as a sensitive prescreening diagnostic tool to detect KRAS, BRAF, PIK3CA, and AKT1 mutations in formalin-fixed, paraffin-embedded tissues. Arch Pathol Lab Med 2012; 136: 983–92. 15. Kristensen LS, Andersen GB, Hager H, Hansen LL. Competitive amplification of differentially melting amplicons (CADMA) enables sensitive and direct detection of all mutation types by high-resolution melting analysis. Hum Mutat 2012; 33: 264–71.
HRM ASSAY FOR LOW LEVEL BRAF MUTATIONS
16. Fadhi W, Ibrahem S, Seth R, Ilyas M. Quick-multiplex-consensus (QMC)-PCR followed by high-resolution melting: a simple and robust method for mutation detection in formalin-fixed paraffin-embedded tissue. J Clin Pathol 2010; 63: 134–40. 17. Mancini I, Santucci C, Sestini R, et al. The use of COLD-PCR and highresolution melting analysis improves the limit of detection of KRAS and BRAF mutations in colorectal cancer. J Mol Diagn 2010; 12: 705–11. 18. Zhou L, Wang Y, Wittwer CT. Rare allele enrichment and detection by allele-specific PCR, competitive probe blocking, and melting analysis. BioTechniques 2011; 50: 311–8. 19. Didelot A, Le Corre D, Luscan A, et al. Competitive allele specific TaqMan PCR for KRAS, BRAF and EGFR mutation detection in clinical formalin fixed paraffin embedded samples. Exp Mol Pathol 2012; 3: 275–80. 20. Richter A, Grieu F, Carrello A, et al. A multisite blinded study for the detection of BRAF mutations in formalin-fixed, paraffin-embedded malignant melanoma. Sci Rep 2013; 3: 1659. 21. Mariani S, Bertero L, Osella-Abate S, et al. Extreme assay sensitivity in molecular diagnostics further unveils intratumour heterogeneity in
22. 23.
24. 25.
26.
783
metastatic colorectal cancer as well as artifactual low-frequency mutations in the KRAS gene. Br J Cancer 2017; 117: 358–66. Lade-Keller J, Rømer KM, Guldberg P, et al. Evaluation of BRAF mutation testing methodologies in formalin-fixed paraffin-embedded cutaneous melanomas. J Mol Diagn 2013; 15: 70–80. Stadhouders R, Pas SD, Anber J, Voermans J, Mes TH, Schutten M. The effect of primer-template mismatches on the detection and quantification of nucleic acids using the 50 nuclease assay. J Mol Diagn 2010; 12: 109–17. Gilbert MT, Haselkorn T, Bunce M, et al. The isolation of nucleic acids from fixed, paraffin-embedded tissues-which methods are useful when? PLoS One 2007; 2: e537. Halait H, Demartin K, Shah S, et al. Analytical performance of a realtime PCR-based assay for V600 mutations in the BRAF gene, used as the companion diagnostic test for the novel BRAF inhibitor vemurafenib in metastatic melanoma. Diag Mol Pathol 2012; 21: 1–8. Curry JL, Torres-Cabala CA, Tetzlaff MT, Bowman C, Prieto VG. Molecular platforms utilized to detect BRAF V600E mutation in melanoma. Semin Cutan Med Surg 2012; 31: 267–73.