Article Text
PDF
Abstract
Aims In cystic and solid pancreatic lesions, KRAS mutational status refines the diagnosis of uncertain endoscopic ultrasound (EUS) aspirates. This test should have a fast turnaround time and ideally be performed at the centre where the patient is diagnosed. The Idylla KRAS Mutation Test enables standardisation even in units without molecular expertise.
Methods The Idylla test was designed for use with formalin-fixed paraffin-embedded (FFPE) sections. However, we directly pipetted 3 µL (corresponding to 1/10th of a DNA preparation from the aspirate sample) in the cartridge, which was automatically run as if an FFPE sample had been inserted. The performance was compared with Sanger sequencing, Allele Specific Locked Nucleic Acid PCR (ASLNAqPCR), and 454 Next Generation Sequencing (454-NGS) in light of clinicopathological end points.
Results Idylla yielded valid results in 49/52 (94.2%) cases, in 2 h. A total of 18/49 cases showed mutation either in KRAS exon 2 (14/18) or in exon 3 (4/18). Idylla KRAS test had 100% specificity and a sensitivity (55.1%) higher than Sanger sequencing (41.3%) and identical to ASLNAqPCR (55.1%). When the low-abundant mutant allele (<5%) cases were excluded from the analysis, the Idylla KRAS Mutation Test clinical sensitivity increased to 61.9% approaching that of 454-NGS (66.6%).
Conclusions This is the first study that applied the novel Idylla KRAS test to the clinical setting of pancreatic cancer. In particular, this system can be easily implemented in the routine assessment of pancreatic EUS-fine-needle aspiration-derived DNA samples to quickly provide information on KRAS mutational status to supplement cytological evaluation.
- PANCREATIC CANCER
- MOLECULAR BIOLOGY
- CYTOLOGY
Statistics from Altmetric.com
Introduction
Endoscopic ultrasound-guided fine-needle aspiration (EUS-FNA) plays a crucial role in the management of pancreatic neoplastic lesions.1 More than 85% of pancreatic adenocarcinoma (PDAC) are diagnosed by EUS-FNA in an advanced stage when surgical pathological examination is precluded.2 As a general rule, EUS-FNA is a rapid, safe and cost-effective procedure.3 However, in up to 20% of cases, well-differentiated PDAC cannot reliably be distinguished from benign diseases such as chronic pancreatitis (CP) especially in its pseudo-tumoral form.3 In particular, EUS-FNA is less accurate, when PDAC is small, well vascularised or desmoplastic.4 Similarly, in patients with pancreatic cysts presenting for EUS evaluation, cyst fluid cytological examination has high specificity for malignancy but lacks sensitivity,5 not always being able to distinguish between benign cysts and premalignant or malignant mucinous cysts.6
KRAS mutations represent an early genetic event in PDAC pathogenesis.7 Even if molecular biology cannot replace cytology, the presence of a KRAS mutation in an inconclusive EUS-FNA specimen taken from a solid lesion suggests malignancy, reducing the false-negative (FN) rate by 55.6%, according to a recent meta-analysis;8 conversely, the presence of wild-type KRAS may be evocative of benignity.8–11 However, the presence of KRAS mutations is not entirely specific, occurring in a minority of patients with CP11 and in preinvasive dysplastic lesions.12
Since EUS-FNA smears often feature contaminating benign gastrointestinal cells, KRAS testing requires the use of a high analytically sensitive molecular technique. To date, real-time PCR (RT-PCR) assays have mostly been designed to target only exon 2 ‘hot-spot’ mutations.9 ,13 Conversely, next-generation sequencing (NGS) ensures analytical sensitivity similar to that of mutation-specific assays, allowing for the detection of common and uncommon mutations, including those of KRAS exons 3 and 4.13 However, the NGS procedure requires a complex validation procedure,14 being cost-effective only in large-volume centralised laboratories.15
The Idylla KRAS Mutation Test (Biocartis, Mechelen, Belgium) recently received CE-IVD certification for the detection of 21 mutations in codons 12 and 13 (exon 2), 59 and 61 (exon 3) and 117 and 146 (exon 4) of the KRAS gene, with a validated 5% limit of detection. This test, based on allele-specific RT-PCR and performed on the fully automated Idylla platform, provides sample to result functionality, enabling method standardisation even in those diagnostic units without molecular expertise and infrastructure.16 ,17 The aim of this study was to validate this novel technology on indeterminate pancreatic EUS-FNA.
Materials and methods
Selection of cases
In this retrospective study, the clinical performance of the Idylla KRAS test was assessed on archival DNA from a well-characterised series of EUS-FNAs that had already been tested for KRAS mutational status in a prior study by using three different techniques—Sanger sequencing, Allele Specific Locked Nucleic Acid PCR (ASLNAqPCR) and 454 Next Generation Sequencing (454-NGS)—as previously reported.13 Details regarding the modality of EUS-FNA sample collection and specimen handling and preparation for microscopic observation have been described.13 Briefly, in any single case an aliquot of the aspirated material, besides microscopic slide preparation, had directly been collected in a tube containing 100% ethanol for KRAS analysis (‘direct’ EUS-FNA material). Specimens were stored at room temperature for a period of up to two weeks and, then, DNA was extracted, as previously described.13 The criterion for including samples in this study, aiming to assess Idylla KRAS test performance, was the availability of at least 20 µL of archival DNA. Overall, a total of 52 cases were selected to undergo Idylla KRAS test. Results were compared with those previously obtained by Sanger sequencing, ASLNAqPCR and 454-NGS.
Idylla KRAS Mutation Test
Archived cases were processed between October and December 2015. The Biocartis' Idylla KRAS Mutation Test is a single-use cartridge-based test designed for the qualitative detection of 21 different mutations. Seven mutations are targeted in exon 2: p.G12C (c.34G>T), p.G12R (c.34G>C), p.G12S (c.34G>A), p.G12A (c.35G>C), p.G12D (c.35G>A), p.G12V (c.35G>T) and p.G13D (c.38G>A). Nine mutations can be detected in exon 3: p.A59E (c.176C>A), p.A59G (c.176C>G), p.A59T (c.175G>A), p.Q61K (c.181C>A; c.180_181TC>AA), p.Q61L (c.182A>T), p.Q61R (c.182A>G) and p.Q61H (c.183A>C; c.183A>T); whereas five mutations are targeted in exon 4: p.K117N (c.351A>C; c.351A>T), p.A146P (c.436G>C), p.A146T (c.436G>A) and p.A146V (c.437C>T).
The assay has been designed by the manufacturer to process 5–10 μm formalin-fixed paraffin-embedded (FFPE) tissue sections or FFPE slices. However, for other applications disposable cartridges have been applied to liquid sample types, including swab, blood, urine, stool, sputum and tissue.18 To perform the analysis on the archival EUS-FNA DNA, 3 µL (corresponding to 1/10th of a DNA preparation from the aspirate sample) of extracted genomic DNA was directly pipetted inside an Idylla KRAS Mutation Test cartridge. The cartridge was loaded onto the Idylla system for processing. Via microfluidic channels in the cartridge, nucleic acids are transported into five separate PCR chambers, which contain pre-deposited PCR reagents in dried form (ie, primers, probes, enzymes) for the analysis of the sample DNA that includes KRAS Total (wild-type gene acting as Sample Processing Control). Detection of these specific targets is performed using fluorescently labelled probes. The Idylla console auto-analyses the PCR curve to determine the presence or absence of a KRAS mutation and the results are presented on screen as either ‘no mutation detected’ or ‘KRAS mutation detected’.19 A quantification cycle value (Cq) value is calculated by Idylla software for every valid PCR curve. The presence of a mutant genotype is determined by calculating the ΔCq, that is, the difference between the KRAS wild-type Cq and the cut-offs for individual mutations. Mutant signal is considered valid if the ΔCq is within a validated range, and the sample will then be characterised as KRAS mutation positive, indicating the specific mutation. Noteworthy, the Idylla KRAS Mutation Test does not report double mutants, and in these cases, only the mutation with smallest ΔCq is called. All samples with a valid wild-type signal but a ΔCq value outside the validated range are characterised as ‘no mutation detected’.
Statistical measures of clinical performance
Idylla clinical performance was evaluated according to the final end point. This was represented by cytological, histological or, in cases of inoperable neoplasms, by clinical features. On this basis, we distinguished three different categories of lesions: (i) benign lesions, including non-neoplastic or benign cysts and a pancreatitis; (ii) adenocarcinomatous lesions or precursor lesions of adenocarcinoma, including PDAC, intraductal papillary mucinous neoplasms (IPMNs) and inoperable neoplasias with overt malignant clinical features; (iii) not adenocarcinomatous lesions, including pancreatic neuroendocrine tumour (pNET) and solid pseudopapillary tumours (SPPTs).
Idylla clinical performance was assessed by evaluating its clinical sensitivity, specificity, negative predictive value (NPV), positive predictive value and accuracy also in comparison to Sanger sequencing, ASLNAqPCR and 454-NGS performance. True positives were cases that showed a mutation in KRAS and that were PDAC, inoperable neoplasias or IPMNs according to final end point. False positives were cases in which a mutation was found but with a ‘benign’ end point or else diagnosed as SPPT or pNET. True negatives were cases with a wild-type KRAS result and a ‘benign’ end point or with an end point of neuroendocrine or pseudopapillary neoplasia. FNs were cases with a wild-type KRAS but were PDAC/inoperable neoplasias or IPMN at the final end point. Comparisons between clinical sensitivities were performed according to recommendations previously described.
Results
Results of Idylla KRAS Mutation Test
A total of 52 archival DNA from EUS-FNA pancreatic samples was tested at the University of Naples Federico II with the Idylla KRAS Mutation Test. Representative examples of PCR curves and the corresponding Cq values for different Idylla analysis results are shown in figure 1. The results were compared with the original assessments made by Sanger sequencing, ASLNAqPCR and 454-NGS at the University of Bologna (table 1).
Series of 49 endoscopic ultrasound-guided fine-needle aspiration tested for KRAS mutational status by using four different techniques: Allele Specific Locked Nucleic Acid PCR, 454 Next Generation Sequencing (454-NGS), Sanger sequencing and Idylla KRAS test
Example of concordant molecular results in a KRAS exon 2 mutated sample (case #12, table 1). (A) Electropherogram obtained using Sanger sequencing. The KRAS-G12V mutation (c.35G>T) is pointed by an arrow. (B) KRAS-G12V mutation was also detected using Allele Specific Locked Nucleic Acid PCR assay: two curves are detectable, one corresponding to wild-type allele and the other one to G12V-mutated allele. (C) Profile obtained using 454-Next Generation Sequencing. The KRAS-G12V mutation is identified by the vertical red bar. The percentage of mutated alleles is indicated on the left y-axis while the total number of reads on the right one. (D) Representative examples of Idylla analysis result showing a KRAS-G12V mutation detection call.
Idylla yielded valid results in 47/52 samples in a first run. The five invalid samples underwent a second run by increasing the DNA input up to 6 µL. Two cases gave a valid call, for a total of 49/52 (94.2%) adequate Idylla KRAS Mutation tests. KRAS mutation detection by four different techniques according to preoperative cytology evaluation and clinicopathological end points is reported in table 2. Briefly, KRAS mutations were found by Idylla in the 4/14 (28.6%) of inadequate samples (C1), in one of the two cases (50.0%) with atypical cells (C3), in 2/6 (33.3%) of the cases suspect for malignant neoplasia (C4) and in the 10/19 (52.6%) of samples diagnosed as malignant (C5). None of the benign (C2) cases showed KRAS gene mutations, and they were benign cysts (three cases) or pancreatitis (one case) on follow-up. One of four cases without available material for cytological evaluation was mutated for KRAS by Idylla KRAS Mutation Test.
KRAS mutation detection by four different techniques according to preoperative cytology evaluation and clinicopathological endpoints
Considering the final end point information available in 45 cases, we detected a KRAS mutation in 16/29 (55.2%) adenocarcinomatous and pre-neoplastic lesions (in 10/15 of PDAC, 3/8 of IPMNs and in the 3/6 of inoperable neoplasms), while no KRAS mutations were observed in not adenocarcinomatous or in benign lesions. In 15/49 (30.6%) cases, discordant results in KRAS mutational status were obtained using at least one of the four different techniques, as reported in table 1.
Discordant KRAS results in relation to mutant allele abundance
To evaluate the reason for discordant results among different techniques (15/49), the rate of mutant allele (MA) yielded by 454-NGS was taken into account when evaluating methods performance. In five discordant cases (#8, #9, #19, #42, #44 and #48) the MA abundance, as detected by 454-NGS was >10%. In cases #8 (MA=19%) and #48 (MA=12%), KRAS exon 2 gene mutations were missed by Sanger sequencing. Cases #9 (MA=15%) and #42 (MA=32%) harboured a mutation in KRAS exon 3, which was undetectable by ASLNAqPCR and missed in one instance (case #9) by Sanger sequencing. In case #44 (MA=34%), KRAS G12D mutation was missed by Idylla KRAS Mutation Test.
In 10 discordant cases (#10, #13, #19, #20, #22, #23, #27, #33, #36 and #45), <5% of MA was identified by 454-NGS. A representative example of discordant results (case #19) is shown in figure 2. Only case #33 was detected by Sanger sequencing, whereas ASLNAqPCR confirmed 454-NGS results, not considering cases #19, #22, #27 and #33 harbouring an exon 3 KRAS mutation that was undetectable by this technique. Also, 4 of 10 low-abundant discordant cases were detected by Idylla. In particular, in four cases that showed between 3% and 5% of MA (#13, #20, #27 and #45), two mutant cases (#13 and #45) were detected by Idylla KRAS Mutation Test. Out of two cases that showed between 2% and 3% of MA (#23 and #36), one mutant case (#23) was detected by Idylla KRAS Mutation Test. Among four cases that showed <2% of MA (#10, #19, #22 and #33), only one case (#33) was detected by Idylla KRAS Mutation Test.
Example of discordant molecular results in a KRAS exon 3 mutated sample (case #19, table 1). (A) Electropherogram obtained using Sanger sequencing, showing a native nucleotide sequence. Codon 61 is marked by two pink vertical bars. (B) No KRAS mutation (codons 12–13) was detected using Allele Specific Locked Nucleic Acid PCR assay: only the curve corresponding to the wild-type allele is visible. (C) Profile obtained using 454-Next Generation Sequencing. The KRAS-Q61R mutation is identified by the vertical black bar. The percentage of mutated alleles is indicated on the left y-axis while the total number of reads on the right one. (D) The KRAS-Q61R mutation was not detected by Idylla.
Statistical measures of performance
Idylla KRAS Mutation Test had 100% specificity, a clinical sensitivity (55.1%) higher than Sanger sequencing (41.3%), and identical to that of ASLNAqPCR (55.1%). Clinical sensitivity (71.1%), NPV (69.6% vs 55.1%) and accuracy (82.546% vs 71.1%) of 454-NGS were higher than Idylla KRAS Mutation Test. When the cases with <5% MA were excluded from the analysis, the Idylla KRAS Mutation Test clinical sensitivity increased up to 61.9% (with 100% specificity), higher than Sanger sequencing (52.3%) and ASLNAqPCR (57.1%). Clinical sensitivity (66.6% vs 61.9%), NPV (69.6% vs 66.6%) and accuracy (81.1% vs 78.3%) of 454-NGS were similar to Idylla KRAS Mutation Test.
Discussion
In both cystic and solid pancreatic lesions, the assessment of KRAS mutational status may be useful to refine the diagnosis of uncertain EUS-FNA samples;4 while in solid lesions, to state malignancy is crucial, in cystic lesions, the approach is different and is nowadays more to differentiating mucinous from non-mucinous lesions. In any case, KRAS testing should have a fast turnaround testing, in line with the need of urgent clinical actions. Thus, instead of outsourcing suspicious pancreatic EUS-FNA to a small number of referral molecular pathology laboratories, KRAS testing may be carried out in the same centre where the patient is being diagnosed, enabling the most cellular slide to be easily selected from in-house collected cytological material. However, KRAS testing is difficult to be implemented in cytopathology laboratories, with little expertise in molecular biology procedures and in-house validation of a laboratory-developed methods. The Idylla KRAS Mutation Test, a rapid and fully automated CE-IVD (European Conformity-In Vitro Diagnostic)-certified test, can represent a valid option for a wider number of pathological centres.16 ,17 ,20 The assay has been designed by the manufacturer to process FFPE tissue sections using high-frequency intensity-focused ultrasound technology to obtain amplifiable DNA.16 ,17 ,20 In this study, however, we have shown that the extracted DNA can directly be placed inside an Idylla KRAS Mutation Test cartridge.
Usually the EUS-FNA sampling of a pancreatic lesion yields abundant contaminating gastrointestinal cells; thus, a highly analytical sensitive technique is required to detect KRAS mutation in a minority of neoplastic cells.21 On the other side, a supersensitive test may detect early KRAS mutant clones even in small foci of pancreatic intraepithelial neoplasia, whose clinical significance is questionable.22 As stated by the manufacturer, the Idylla KRAS Mutation Test has a limit of detection of 5% (validated at the 95% CI) of MA, which seems to be well suited for clinical applications. Accordingly, in this study, the Idylla KRAS Mutation Test had a clinical sensitivity (55.1%) higher than Sanger sequencing (41.3%), but lower than 454-NGS (71.1%). In fact, 454-NGS had the capability to detect small mutant clones in 10 cases, which was achieved in only four cases by Idylla. Noteworthy, low abundance mutant cases usually were IPMN; only case #55, featuring as little as 4% of MA and detected only by 454 NGS was an overt malignant neoplasm. When the low-abundant mutant cases were excluded from the analysis, the Idylla KRAS Mutation Test clinical sensitivity increased from 55.1% up to 61.9% approaching that of 454 NGS (66.6%).
The allelic discrimination laboratory-developed technique can be applied to EUS-FNA with several advantages as it was very recently shown by Bournet et al,9 who developed TaqMan probe sets to identify the most prevalent KRAS codon 12 mutations (p.G12R—c.34G>C; p.G12D—c.35G>A; p.G12V—c.35G>T). In their experience, based on the analysis of solid suspect pancreatic lesions, the sensitivity of cytopathology alone increased from 73% to 88%.9 Compared with a similar qRT-PCR-based approach, ASLNAqPCR, Idylla KRAS Mutation Test was less sensitive in low-abundant MA cases, but detected exon 3 mutations in four instances. Only one of these cases harbouring a double G12V/Q61H mutation was detected by ASLNAqPCR. No mutation were detected in exon 4, confirming that these are uncommon in pancreatic cancer.13 Their inclusion in the reference range of the Idylla KRAS Mutation Test probably reflects its design for other common cancer types, such as colon cancer. Overall, six different mutation types (G12C, G12D, G12R, G12V, Q61H and Q61R/L) were detected by the Idylla KRAS Mutation Test, confirming that the whole spectrum of clinically relevant mutations for pancreatic cancer is covered.4 The clinical specificity of Idylla KRAS Mutation Test was 100% and no mutant cases were detected in benign lesions; conversely, Idylla KRAS Mutation Test missed only one G12D mutation detected by ASLNAqPCR and confirmed by Sanger sequencing. It may be important to note that the Idylla tests were performed on archived DNA that had been stored for several years and shipped to Naples before testing, while the other methods used freshly prepared DNA tested directly in Bologna. To assess whether a confirmation may be needed to avoid any possibility of FN results, a prospective study is required. Additionally, the performance of the Idylla KRAS Mutation test may be further improved by prior cytological observation of the harvested material and tumour cell enrichment.13
In conclusion, this is the first study that applied the novel Idylla KRAS test to the clinical setting. In particular, this system can be easily implemented in the routine assessment of pancreatic EUS-FNA samples to quickly provide information on KRAS mutational status that can supplement cytological evaluation.
Take home messages
This is the first study that applied the novel Idylla KRAS test to the clinical setting.
The Idylla test was designed for use with formalin-fixed paraffin-embedded sections. However, it can also be applied to pancreatic endoscopic ultrasound fine-needle aspiration by directly pipetting 3 µL of extracted DNA in the cartridge.
Idylla KRAS testing may be carried out in 2 h in the same centre where the patient is being diagnosed.
References
Supplementary materials
Abstract in Italian
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 Italian - Online abstract
Footnotes
Handling editor Runjan Chetty
Contributors DdB and CdL contributed equally to the article. DdB, CdL, UM, GTa and GT conceived the study and wrote the paper. GG and MV performed the experimental part. CB contributed as pathologists.
Funding The University of Napoli Federico II.
Competing interests None declared.
Patient consent Obtained.
Provenance and peer review Not commissioned; externally peer reviewed.