Article Text

PDF

Validating a fully automated real-time PCR-based system for use in the molecular diagnostic analysis of colorectal carcinoma: a comparison with NGS and IHC
  1. Richard Colling1,2,3,
  2. Lai Mun Wang1,4,5,
  3. Elizabeth Soilleux1,6
  1. 1Department of Cellular Pathology, Oxford University Hospitals NHS Trust, John Radcliffe Hospital, Oxford, UK
  2. 2Oxford Molecular Diagnostics Centre, Molecular Haematology, Oxford University Hospitals NHS Trust, John Radcliffe Hospital, Oxford, UK
  3. 3Department of Oncology, University of Oxford, John Radcliffe Hospital, Oxford, UK
  4. 4Department of Laboratory Medicine, Changi General Hospital, Singapore, Singapore
  5. 5Ludwig Institute for Cancer Research, University of Oxford, Old Road Campus Research Building, Oxford, UK
  6. 6Nuffield Division of Clinical Laboratory Sciences, University of Oxford, John Radcliffe Hospital, Oxford, UK
  1. Correspondence to Dr Richard Colling, Department of Cellular Pathology, Level 1, John Radcliffe Hospital, Headley Way, Headington, Oxford OX3 9DU, UK; rtcolling{at}gmail.com

Abstract

Background Molecular testing is increasingly needed in colorectal carcinoma (CRC) and the current clinically relevant mutations are in BRAF, KRAS and NRAS. This study aimed to further validate a new alternative polymerase chain reaction (PCR) platform (Idylla, Biocartis) against existing next-generation sequencing (NGS) and immunohistochemistry (IHC) assays.

Methods 56 Idylla tests were performed on 43 CRC cases, in a total of 74 comparisons against an NGS panel (Ion Torrent) and the VE1 (anti-BRAF) antibody IHC. Discrepant cases were also compared with either conventional (Cobas) or droplet digital PCR (Bio-Rad).

Results Idylla showed an overall concordance of 100% (95% CI 93% to 100%) with comparator molecular testing and indications were that Idylla is likely to be more sensitive than routine NGS. BRAF IHC showed 90% concordance with NGS (95% CI 70% to 97%).

Conclusions This study validates Idylla in formalin-fixed, paraffin-embedded CRC tissue. BRAF IHC, however, is an unreliable substitute for molecular testing in CRC.

  • COLORECTAL CANCER
  • MOLECULAR PATHOLOGY
  • IMMUNOHISTOCHEMISTRY

Statistics from Altmetric.com

Introduction

Colorectal carcinoma (CRC) accounts for around one in 10 new malignancies diagnosed each year.1 Molecular testing is useful in guiding oncological management in CRC due to increased availability of newer targeted therapies and therefore such tests are becoming routinely incorporated into the reporting of CRC resection specimens.2

Routine molecular testing focuses on mutations in the mitogen-activated protein kinase pathway.2 This includes KRAS and (increasingly) NRAS testing, with the aim of stratifying patients with metastatic CRC for antiepidermal growth factor monoclonal antibody (anti-EGFR mAb) therapies, and BRAF testing in the context of Lynch syndrome (LS) screening. In LS screening, MLH1 protein-deficient tumours are tested for BRAF mutations and patients with BRAF wild-type tumours are followed up with clinical genetics consultations.3–10 Most centres use polymerase chain reaction (PCR)-based tests which can be expensive, time consuming, require specialist staff and facilities and have long turnaround times. Increasingly, larger centres are moving to next-generation sequencing-based (NGS) technologies. These include Ion Torrent, a chip-based hydrogen ion semiconductor detection method and Illumina, a flow cell-based, dye-reporter sequencing. NGS generally allows greater coverage than traditional sequencing or PCR and potentially has greater sensitivity. These technologies, however, require highly specialist facilities and staff and incur a very high cost with a greatly extended turnaround time.11 ,12 An immunohistochemical (IHC) test is also available for BRAF testing, although reported sensitivity and specificity vary.13

The real-time PCR-based Idylla platform offers a fully automated, fast and potentially cost-effective technology for molecular testing in CRC. The technology is based on a disposable cartridge format and carries out testing with all reagents on board. The test requires the user to insert rolled sections of fixed tissue into an assay cartridge with results available within a few hours, as described previously.14 ,15 The Idylla assays are based on one of two methodologies as detailed in table 1. The first of these is a conventional TaqMan reporter system (the Idylla BRAF Mutation Test). The second is a novel chemistry known as PlexPCR and is used for both the Idlla KRAS Mutation Test and the Idylla multiplexed NRAS-BRAF-EGFRS492R Mutation Assay, the focus of this study. This system uses primers (PlexPrime) that create amplicons containing a small region with a sequence different from that of target DNA. This works together with a specific amplicon sequence-matched reporter probe (PlexZyme) and this significantly increases the overall detection specificity and allows multiplexing of numerous gene targets in one assay.16 ,17 The Idylla system then can detect a broader range of mutations within KRAS, NRAS and BRAF than many currently available commercial PCR platforms.18–20 We previously reported the validation of the Idylla BRAF Mutation Test15 and the Idylla KRAS Mutation Test21 and the aim of this study was to extend the evaluation of Idylla to NRAS testing and validate all three targets together with NGS and available IHC.

Table 1

Details of the available BRAF, KRAS and NRAS assays on the Idylla system compared with common PCR (Cobas) and NGS (Ion Torrent) platforms

Materials and methods

CRC cases, which had previously undergone targeted NGS with the Ion AmpliSeq Cancer Hotspot Panel version 2 (v2) (Thermo Fisher Scientific), were taken from the departmental archive at the John Radcliffe Hospital. Where possible, sample sizes (20 cases; equal numbers of positive and negative cases by NGS) for each validation were selected in line with recommendations.22 In some cases, limited material was available for testing, for example, only 5 mutated NRAS cases were identified in the archive and therefore 5 positive and 13 negative (only 18 cartridges were available) were used.

The Idylla KRAS Mutation Test and the Idylla NRAS-BRAF-EGFRS492R Mutation Assay were validated in this study, details of which are given in table 1 along with a comparison of the alternative tests used locally. The general principles and methods for Idylla testing have been described previously.15 Briefly, formalin-fixed, paraffin-embedded (FFPE) tissue was macrodissected from a single 5 μm unstained section for each assay. At least 50 mm2 of tissue was used for each test. All cases were reviewed by a consultant histopathologist and an estimate of the tumour content was made. All cases met the Idylla minimum tumour cellularity for KRAS (10%), NRAS/BRAF (25%) and BRAF (50%) assays. Tissue was placed between blotting paper inside an Idylla test cartridge, which was loaded onto the Idylla system for processing. The Idylla console software autoanalysed the fluorescent amplification signal to report the presence or absence of a mutation. The presence of a mutation was considered a positive test and wild type was considered negative.15

Both Idylla assays were validated against the Ion AmpliSeq Cancer Hotspot Panel v2, which includes BRAF, KRAS and NRAS coverage. The manufacturer quotes an assay detection rate of 98% for mutations with at least a 5% variant frequency where the coverage is 1000×–4000×; details are given in table 1. This was run on the Ion Torrent Personal Genome Machine (PMG) platform (Thermo Fisher Scientific); for Ion Torrent testing, a tissue area of up to 250 mm2 on up to eight sections at a maximum thickness of 10 µm was used. A minimum of 30% tumour was required for the assay.23 ,24 In all cases, the same FFPE tissue block was used for Idylla and Ion Torrent testing and the same tumour area was macrodissected for processing.

The NRAS-BRAF-EGFRS492R Mutation Assay was run on a cohort of 18 cases and compared with NGS. Thirteen cases from this cohort and a five further separate cases (18 in total) also underwent testing with the KRAS Mutation Test. Discrepant cases were further tested where needed with droplet digital PCR (QX100 Droplet Digital PCR System Droplet Generator, Bio-Rad Laboratories), as described previously.15 Idylla EGFR results were not evaluated.

The Idylla BRAF Mutation Test was further compared with IHC carried out using the VE1 antibody (Ventana Medical Systems), specific for V600E-mutated BRAF on a Ventana Benchmark Immunostainer, according to the manufacturer's instructions25; 10 positive and 10 negative cases, as determined by the Cobas BRAF V600 Mutation Test (Roche Molecular Systems),26 were selected from a separate cohort of patients from those tested above. Eighteen of these cases were taken from the Idylla BRAF mutation study previously reported;15 the remaining two cases were not from this cohort. The slides were scored by two histopathologists, independently and blind to BRAF mutation status as assessed by molecular tests. A score was calculated using methods based on guidelines:27 a staining intensity (1–3, 0=no staining) and percentage of cells stained (0=<20%, 1=20–50%, 2=>50%) score were given by each pathologist for each case. These scores were multiplied together (intensity×%) and the mean of both pathologists' scores was calculated. A final score >1 was taken to be positive, based on how others have reported staining.13

A statistical analysis of concordance, sensitivity and specificity was carried out using conventional formulae in Microsoft Excel.

Results

Fifty-six Idylla test cartridges were run on 43 tumours, undertaking 74 test comparisons. Twenty-three cases were selected for comparison with Ion Torrent, 18 of these were tested with the NRAS-BRAF-EGFRS492R Mutation Assay and 18 were tested with the KRAS Mutation Test. The raw data are given in table 2. One cartridge failed (an NRAS-BRAF-EGFRS492R Mutation Assay) due to a software error; this brought the total number of comparisons down to 72. The comparison results and statistical analysis are summarised in table 3. Overall, there was concordance between Idylla and routine tests (NGS and IHC) for 69/72 comparison tests, making the concordance 96% (95% CI 88% to 99%), with only three discordant results.

Table 2

The raw data of the comparison between Idylla and Ion Torrent NGS for BRAF, KRAS and NRAS testing

Table 3

A summary of the results from the validation of Idylla against Ion Torrent (Ion) NGS and IHC for BRAF, KRAS and NRAS

There was 100% (95% CI 82% to 100%) concordance between Idylla and Ion Torrent for detecting BRAF and KRAS mutations. The concordance between Idylla and Ion Torrent for NRAS was lower at 94% (95% CI 73% to 99%) due to one discordant case which appeared mutated on the Idylla assay (Q61L), but wild type by Ion Torrent. This case was further tested by droplet digital polymerase chain reaction (ddPCR) and this confirmed a Q61L mutation in NRAS at 1.36–1.47%. Given this, the Idylla was 100% (95% CI 82% to 100%) concordant for NRAS detection when compared with established molecular techniques (Ion Torrent and ddPCR). The overall concordance between Idylla and established molecular tests (Ion Torrent/ddPCR) combined was 100% (93–100%) for BRAF, KRAS and NRAS.

The concordance between BRAF IHC and Idylla was lower at 90% (95% CI 70% to 97%) due to two discordant cases with apparent IHC-detected V600E mutations that appeared wild type on Idylla testing. We know from our previous study that the BRAF V600E Idylla results for these cases were concordant with Cobas PCR,15 indicating false-positive IHC staining. The scorers agreed in 17 of the 20 cases making the kappa value 0.7.

Discussion

Although we and others have previously published validation data for the Idylla system,15 ,21 ,27–29 this is the first publication to validate the Idylla NRAS-BRAF-EGFRS492R Mutation Assay for BRAF and NRAS targets. This is also the first publication to consider the full panel of colorectal molecular gene targets (BRAF, KRAS and NRAS) together. Two Idylla cartridges (KRAS and NRAS-BRAF-EGFRS492R) would cover all three genes in routine practice. We found that the concordance between Idylla and NGS was high at 100% for BRAF and KRAS mutations and 94% for NRAS. In addition to this, the Idylla identified one mutation that Ion Torrent did not, making it potentially more sensitive than Ion Torrent. This was not surprising, as, although the codon change is covered by the AmpliSeq NGS panel, the mutation frequency was below the 5% quoted sensitivity (table 1).24

Although this study did not follow the design of a diagnostic accuracy study, it gives a useful indication of the likely sensitivity and specificity of Idylla in this context. This is especially relevant for NRAS testing, for which a reliable commercial PCR-based test for use in CRC management is less established and for which NGS is being increasingly requested in clinical practice. A larger follow-up study for NRAS testing is needed to confirm this, however.

The Idylla KRAS Mutation Test covers all the common and some of the rarer CRC mutations in KRAS, however some rare mutations are not covered, such as the G13C codon change (full coverage details are given in table 1). The clinical implications of these rarer mutations for anti-EGFR mAb therapy are not clear at the moment; however, this is a potential limitation of the test.30

An important issue is the financial implications of using the Idylla system. The cost in our institution of Ion Torrent NGS is £300 per case, whereas the use of the KRAS and BRAF/NRAS Idylla cartridges in combination to cover all three actionable gene targets in CRC would cost around £400. Therefore, a move to using Idylla in an established molecular diagnostics laboratory may not be cost-effective. However, for institutions with no easy access to molecular testing, the fast turnaround times coupled with potential subscription models (no upfront cost) for Idylla use may be attractive.

The use of IHC to detect the BRAF V600E mutation is widespread. An example of the staining seen with VE1 is shown in figure 1. In our study, IHC showed poorer concordance (90%) with NGS than Idylla (100%), despite good agreement between pathologists. All cases used for IHC had previously shown 100% concordance for BRAF V600E mutation analysis between Cobas PCR and Idylla. Interestingly, the two discordant cases were scored slightly differently by the two pathologists. Thus, this likely reflects difficulties with VE1 IHC staining interpretation leading to poor interobserver agreement, commensurate with previous reports.13 ,31

Figure 1

Colorectal adenocarcinoma with BRAF V600E mutation, demonstrated by the AmpliSeq next-generation sequencing (NGS) panel and COBAS PCR, immunostained with mouse monoclonal antibody VE1 (anti-BRAF V600E mutation-specific antibody, Ventana Medical Systems). This demonstrates positive staining with an intensity score of 3 and a tumour percentage score of 2.

No EGFR mutations were detected by any of the testing modalities in this study. As there is currently no reliable comparison test for the S492R codon change commercially available we did not attempt to validate the Idylla for this target.

As discussed previously,15 the Idylla offers a compact alternative to conventional PCR-based and NGS-based assays and does not require specialist training to operate. The system also offers the advantages of being on-demand with minimal turnaround times and a reduced cost per test when compared with NGS or some conventional PCR assays.14 This system therefore may appeal to smaller, non-specialist centres and to larger regional molecular centres to complement available testing modalities and provide rapid turnaround times.15

Take away messages

  • Targeted mutation testing is becoming routine in colorectal carcinoma.

  • BRAF, KRAS and NRAS are the most clinically useful targets.

  • The need for specialist staff, techniques and equipment are barriers to universal NHS testing.

  • Automated PCR with the Idylla System offers simple, fast and cost-effective molecular testing which could be used on-site in any histopathology department in the UK.

  • The Idylla is accurate for detecting BRAF, KRAS and NRAS mutations in colorectal FFPE tissue.

Acknowledgments

The Idylla system was kindly loaned by Biocartis. Biocartis contributed some test cartridges. The authors thank the Oxford Molecular Diagnostics Centre for providing clinical results with conventional PCR and NGS.

References

View Abstract

Footnotes

  • Handling editor Runjan Chetty

  • Contributors RC devised the project and carried out the Idylla testing. All authors contributed to the final manuscript.

  • Funding Consumables, tissue handling costs and additional test cartridges were funded by the Oxfordshire Health Services Research Committee (Fund 8262 to RC/ES).

  • Competing interests None declared.

  • Ethics approval The National Research and Ethics Service (Oxfordshire Research and Ethics Committee A; reference 04/Q1604/21).

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.