Aims Searching for EGFR and KRAS mutations within non-small cell lung carcinoma (NSCLC) samples remains time-consuming and can delay treatment choices in patients with acute deterioration. We evaluated the performances of the fully automated Idylla platform to quickly detect these mutations in NSCLC samples.
Methods We used the Idylla EGFR Mutation Assay and the Idylla KRAS Mutation Test to analyse 18 formalin-fixed paraffin-embedded NSCLC tumour samples with known EGFR and KRAS mutation status according to next-generation sequencing (NGS) and droplet digital PCR (ddPCR) for EGFRT790M mutations.
Results Idylla assays identified KRAS and EGFR activating mutations in 4 and 10 NSCLC samples, respectively. EGFRT790M resistance mutations were identified in only 1 sample using Idylla but in 4 and 14 samples using NGS and ddPCR, respectively. No false-positive result was noted with Idylla assays. Mutation written report was obtained after 130 min (KRAS assays) to 140 min (EGFR assays).
Conclusions The Idylla platform is an interesting ancillary first-line fast and fully automated tool to detect EGFR and KRAS mutations in NSCLC samples allowing rapid treatment choices in patients with acute deterioration.
- LUNG CANCER
- MOLECULAR PATHOLOGY
Statistics from Altmetric.com
Lung cancer remains the leading cause of cancer deaths.1 Treatment of non-small cell lung cancer (NSCLC) is being improved by a better understanding of the molecular mechanisms involved in tumour initiation and progression, mainly in adenocarcinomas. The discovery of EGFR activating mutations and ALK rearrangements in a subset of NSCLC led to major changes in the therapeutic strategy.2 ,3 Other genetic alterations are promising molecular targets as ROS1 or RET rearrangements or BRAF, PIK3CA, C-MET and HER2 mutations.4–8 Other biomarkers as KRAS mutations are associated with a poor prognosis but do not allow targeted therapies to date with nevertheless some recent encouraging results in this field.9 ,10 Driver genetics alterations are mostly mutually exclusive and, besides EGFR and ALK testing recommended in international guidelines, a rapid KRAS assay may be performed initially as a help to exclude KRAS-mutated tumours from EGFR mutation testing as a part of an algorithm designed to maximise testing efficiency.11 ,12
In a recent French nationwide study about 17 664 patients with NSCLC, KRAS and EGFR mutations were the most frequently diagnosed genetic alteration with EGFR, KRAS, BRAF, HER2, PIK3CA mutations and ALK rearrangements frequencies of 11%, 29%, 2%, 1%, 2% and 5% of NSCLC samples, respectively.13 In this study, the median interval between tissue specimen collection and the initiation of molecular analysis was 8 days and the median interval from the initiation of molecular analysis to the final written report of the analysis was 11 days.13 In some patients with acute deterioration, EGFR testing should be available as soon as possible to allow a first-line therapy with EGFR antagonists in patients with EGFR-mutated NSCLC. A long time between the initiation of the molecular analysis and provision of the written report could delay the treatment of patients. In this manner, fast methods of mutation diagnosis could be of interest helping to reduce the delay of treatment choices for the patients with advanced stages NSCLC.
In this study, we evaluated the performances of the fast and fully automated Idylla platform to diagnose EGFR and KRAS mutations in a subset of NSCLC samples with known genetic alterations.
We collected 18 formalin-fixed and paraffin-embedded (FFPE) lung adenocarcinoma samples from patients with known EGFR and KRAS mutation status selected from the cases analysed at the Brest University Hospital cancer molecular genetics platform in 2015 and 2016. EGFR and KRAS next-generation sequencing (NGS) analyses (Ion Torrent, Life Technologies) were conducted as part of the diagnostic workup for the therapeutic management of patients with advanced stages of NSCLC according the French National Cancer Institute recommendations as well as ALK, ROS1, RET and BRAF analyses. Our set of samples was enriched in EGFR mutated samples according to initial NGS results to evaluate the performances of the Idylla platform with different mutations. All samples were included in a registered tumour tissue collection and the present study was conducted after approval by our institutional review board (CHRU Brest, CPP no. DC-2008–214).
The proportion of tumour cells was established on each sample by a pathologist on a dedicated 3-µm-thick haematoxylin-eosin-saffron (HES) stained tissue slide. Serial 10 µm tissue sections were produced for molecular analyses. The tumour zones were macroscopically circled to allow macrodissection of tumour tissue for molecular analyses, that is, Idylla, NGS and droplet digital PCR (ddPCR) assays. The FFPE blocks used for molecular analyses were the same for NGS, Idylla and ddPCR assays and an additional HES stained slide was performed for each block to confirm that the tumour sample was not exhausted after these different analyses.
The Idylla platform (Biocartis, Mechelen, Belgium) is a fully cartridge-based automated quantitative PCR platform and uses microfluidics processing with all reagents on board. In our study, we studied the 18 lung adenocarcinoma samples with the Idylla EGFR Mutation Assay and the Idylla KRAS Mutation Test cartridges (Biocartis, see online supplementary data for the details of the detected mutations). For each analysis, one section (EGFR assay) or three sections (KRAS assay) were macrodissected and then transferred to a wetted (nuclease-free water) filter paper following the manufacturer's instructions. A second wetted filter paper was then added on top of the FFPE material and the sample with the two wetted filter papers was finally placed on the lysis pad in the Idylla cartridge and inserted in the instrument. Inside the cartridge, the sample was homogenised and cells lysed using a combination of high-intensity-focused ultrasound, enzymatic/chemical digestion and heat. The nucleic acids were liberated and ready for subsequent PCR amplification. The PCR was real time and used a fluorophore-based detection system. After a 130 min (KRAS assay) to 140 min (EGFR assay) run, all steps were automatically performed inside the cartridge and final reports were directly available on the system after an automatic on-board post-PCR curve analysis. Total EGFR and KRAS acted as sample processing controls (data not shown by the system).
Droplet digital PCR assays
ddPCR, a recently developed technique, involves emulsification and PCR amplification inside thousands of small droplets, each droplet containing one or no molecules of target DNA. Precise and absolute quantification of the number of target DNA molecules in reaction is simply achieved by counting the number of positive and negative droplets. The strategy reduces competitive amplification, allowing detection of 0.01% mutant fractions, which is 1000 times lower than real-time PCR. This technique appears as more sensitive than classical PCR and is useful for identification of predefined mutations expected to be present in a minor fraction of cell population.
Tumour DNA was extracted from FFPE slides by using Maxwell RSC DNA FFPE Kit (Promega, Courtaboeuf, France) according to the manufacturer's instructions. Bio-Rad QX200 ddPCR system (Biorad, Marne-la-Coquette, France) was used to analyse all samples. The ddPCR probe mastermix and primers targeting EGFRT790M mutation with EGFR wild type were all purchased from Bio-Rad. The primer sequences were proprietary to the company. We generated approximately 20 000 droplets from each sample. After PCR was performed on all droplets, a fluorescent droplet reader collected the FAM (mutant) and HEX (wild type) signals of the probes in each droplet. Each droplet is individually assigned a positive or negative value based on the fluorescent intensity. The mutant allele concentration (CMUT) and wild-type allele concentration (CWT) in the final PCR mix were calculated with QuantaSoft V.1.7 software (Bio-Rad) in copies per microliter. Each sample was analysed in duplicate with a quantity of input DNA. Results were reported as copies of mutant allele per microliter and in fractional abundance (%).
This assay was able to discriminate at least one copy of mutant allele for 10 000 wild-type copies (ie, an abundance ratio of 0.1%). The limit of blank (LOB) was the primary characteristic of assay that determined the lower limit of detection, and the LOB was defined by the frequency of positive droplets measured in wild-type samples as well as normal human genomic DNA. The number of false-positive droplet events using 400 ng of EGFR wild-type DNA (EGFR Wild-Type Reference Standard (HD709), Horizon Discovery, Cambridge, UK) were measured for 11 negative control experiments. The LOB was fixed at 0.047%.
To preserve us of the absence of contamination during ddPCR technique and DNA extraction, we tested in each assay of ddPCR one NTC and one blank of extraction. In each assay, we used EGFRT790M control (EGFR T790M Reference Standard, Horizon Discovery, Cambridge, UK) at 0.1% and 50% of sensibility.
The analysed set of 18 NSCLC samples consisted in 8 biopsies and 10 surgical samples (13 primary NSCLC tumour samples and 5 metastases). Mean tumour cells percentage was 55% (from 20% to 80%). Clinical, pathological and molecular results are summarised in table 1.
Idylla EGFR Mutation Assay detected every EGFR activating mutations in 10/10 EGFR mutated samples (four EGFRL858R, four EGFR deletions of exon 19, one sample with a double activating EGFRG719A and EGFRL861Q mutation and one sample with a double EGFRG719A and EGFRS768I mutation). In four samples with a known additional EGFRT790M mutation conferring resistance to the first-generation anti-EGFR tyrosine kinase inhibitor, Idylla EGFR Mutation Assay detected the EGFRT790M mutation in one surgical sample but failed to detect the resistance mutation in three EGFRT790M mutated biopsies according to NGS analyses (no data about the percentages of EGFRT790M mutations using NGS, see figure 1). According to ddPCR, 14 samples (78%) were positive for EGFRT790M mutation with mutant allelic frequencies ranging from 51.6% (in the Idylla and NGS EGFRT790M-positive case #10) to 0.13%. EGFRT790M mutant allele frequencies were 1.2%, 4.1% and 4.6% using ddPCR in the three EGFRT790M Idylla-negative but NGS-positive samples (cases #4, #7 and #8, see table 1 for details and figure 2 for examples of ddPCR results).
Idylla KRAS Mutation Test detected KRAS mutations in 4/4 samples with known KRAS mutations (one KRASG12A, one KRASG12C, one KRASG13D and one KRASQ61H mutations).
No false-positive result was noted with Idylla EGFR Mutation Assay and Idylla KRAS Mutation Test including the four EGFR- and KRAS- wild-type samples (one BRAFV600E mutated sample and one ALK, one ROS1 and one RET rearranged samples).
The fast and fully automated real-time PCR Idylla platform was developed to provide fast mutation diagnosis on the basis of FFPE or cytological tumour samples.14 ,15 In our small set of NSCLC FFPE samples, Idylla EGFR Mutation Assay and Idylla KRAS Mutation Test permitted to detect activating mutations in <1 day with high concordance in comparison with the NGS method. No false-positive result was obtained. In this manner, in patients with acute deterioration, it could permit a rapid treatment choice with a minimal delay between the initiation of molecular analysis and the final written report. It could also reduce the delay between tissue specimen collection and the initiation of molecular analysis because, as an easy-to-use platform, it does not require highly skilled staff and could be implemented even in inexperienced pathology units where patients are diagnosed avoiding external centralised testing. Nevertheless, additional ALK and ROS1 fluorescent in situ hybridisation (FISH) and immunohistochemistry analyses must also be performed on the same samples used for molecular assays and, in this manner, the best solution would be to perform the molecular (including Idylla), FISH and immunohistochemistry analyses in a same workflow in a single laboratory to prevent sample depletion, especially in small biopsies.
Idylla assays cover the vast majority of EGFR and KRAS mutations encountered in daily practice. Nevertheless, rare EGFR and KRAS variants and mutations in other molecular targets are not detected by this fast diagnosis platform to date. In previous studies, Idylla did not detect double mutants (eg, KRASG12V and KRASQ61H mutant) but only reported the mutation with the smallest ΔCq per assay/cartridge.16 In our study, Idylla identified every KRAS and EGFR activating mutations including two samples with a double mutation (EGFRG719X and EGFRS768I mutations in case #1; EGFRG719X and EGFRL861Q in case #9). None of our samples had a double KRAS mutation. Nevertheless, Idylla detected a single EGFRT790M mutation in a tyrosine kinase treatment naive patient in one EGFRL858R mutated pulmonary surgical sample containing about 80% of tumour cells with a EGFRT790M mutant fraction of 51.6% according to ddPCR. The EGFRT790M mutant fractions in three Idylla-EGFRT790M negative but NGS-EGFRT790M mutated samples were close to the 3.5% limit of detection recently reported about this EGFR resistance mutation that was higher than the limit of detection of other EGFR mutations screened by the Idylla EGFR Mutation Assay (about 1% of limit of detection for other mutations).14 ,17 Caution is in this manner required when analysing samples from patients treated with anti-EGFR tyrosine kinase inhibitors because, as encountered in our study in three small biopsies, Idylla may not detect some low amounts of resistance mutations that could require the switch from first-generation EGFR inhibitors to second-generation or third-generation ones in case of tumour progression. Caution is also required analysing low mutant fraction–liquid biopsy samples requiring highly sensitive methods, especially searching for the frequent EGFRT790M resistance mutation. Further studies are needed to determine the interest of the Idylla platform to detect EGFR mutations in liquid biopsies. Using NGS or ddPCR seems in this manner necessary after Idylla assays in case of Idylla wild-type result and/or in case of progression under targeted therapy. ddPCR is a particularly sensitive method to detect low mutant fractions and appeared in our study the most efficient method to detect EGFRT790M mutations. In a concordant manner with previous studies, it detected frequent low mutant fractions in patients with but also without associated EGFR activating mutations according to initial NGS diagnosis data.18 Nevertheless, the clinical significance of some very low mutant frequencies of EGFRT790M resistance mutations remains uncertain from a treatment choice point of view to date.
To conclude, the Idylla automate is a sensitive and rapid platform to detect EGFR and KRAS activating mutations in NSCLC tumour samples and it represents an interesting ancillary first-line tool for rapid treatment choices in patients ongoing acute deterioration prior to additional molecular analyses as NGS and/or ddPCR. Further studies are needed to determine the medico-economic limitations and advantages of integrating such a fast and fully automated platform in routine practice.
Take home messages
As it is now mandatory to determine EGFR and KRAS mutation status to treat patients with advanced non-small cell lung cancers (NSCLC), it is important to reduce the delay of molecular testing for rapid treatment choices in patients with acute deterioration.
This study shows that the Idylla™ EGFR Mutation Assay and the Idylla™ KRAS Mutation Test are easy to use and valuable ancillary diagnostic tools to detect EGFR and KRAS mutations in formalin-fixed paraffin-embedded NSCLC samples in 140 and 130 minutes respectively.
Caution is required searching for EGFRT790M mutations using the Idylla™ EGFR Mutation Assay because Idylla™ may not detect some low amounts of resistance mutations (1 EGFRT790M mutations detected using Idylla™ in our study versus 4 and 14 EGFRT790M mutations detected with next generation sequencing and droplet digital polymerase chain reaction respectively in this study).
The authors acknowledge Sandrine Duigou, Marina Pochic and Véronique Fainsin (CHRU Brest, Department of Pathology) and David Dejans (Biocartis) for their technical assistance and the local tumour tissue biobank BB-0033-00037 (‘CRB Santé/Tumorothèque de Brest’) for their collaboration in this work. The authors would also like to thank the molecular geneticists of the Brest University Hospital cancer molecular genetics platform for their daily collaboration in diagnosis practice.
Review history and Supplementary material
Abstract in French
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
- Abstract in French - Online abstract
Handling editor Runjan Chetty
Contributors Conceptualisation: LL and AU; sample preparation and Idylla assays: LL, FB, BG, IQ-R and AU; ddPCR assays: CC and J-PM; manuscript draft: LL, CC, PM, J-PM and AU.
Competing interests None declared.
Ethics approval CHRU Brest, CPP no. DC-2008-214.
Provenance and peer review Not commissioned; externally peer reviewed.
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.