## Abstract

Objectives:

To compare 2 laboratory assays commonly used in the evaluation of epidermal growth factor receptor (EGFR) mutations in non–small cell lung cancer (NSCLC).

Methods:

Fifty-three formalin-fixed, paraffin-embedded NSCLC specimens were selected. Extracted DNA was analyzed using the EGFR RGQ Amplification Refractory Mutation System Scorpions probe-based real-time polymerase chain reaction (PCR) assay and the EGFR Pyro pyrosequencing assay.

Results:

Fourteen EGFR mutations were identified in 13 specimens using at least 1 of the assays, with a mutation concordance rate of 92.9%. Using dideoxy sequencing as the gold standard, clinical sensitivity was 73.7% and 68.4% by the RGQ and Pyro assays, respectively, but 100% by both for common drug sensitivity mutations. Performance observations included the following: the RGQ system requires higher DNA input, the RGQ system is a single-step procedure, the EGFR Pyro assay is a 2-step procedure, only the RGQ system can identify exon 20 insertions, the RGQ system is more sensitive, and the Pyro system can specify exact mutations for all interrogated sites.

Conclusions:

Both the RGQ real-time PCR and Pyro assays adequately detect common EGFR mutations; however, the RGQ system is more clinically and analytically sensitive. Performance characteristics should be considered when evaluating these EGFR mutation assays for clinical adoption.

The epidermal growth factor receptor (EGFR) is a type I receptor tyrosine kinase that, upon ligand binding, activates many important signaling pathways involved in cell proliferation, cell migration, and gene transcription, including the RAS-MAPK pathway and the phosphatidylinositol 3 kinase–Akt pathway. Activating mutations in the tyrosine kinase domain (exons 18–21) of the EGFR have been identified in 10% to 15% of pulmonary adenocarcinomas, especially in never smokers, females, and people of Asian descent.1–5 Approximately 80% of these mutations consist of either deletions in exon 19 (affecting the LREA motif) or a missense mutation in exon 21, resulting in an arginine substitution for a leucine at codon 858.1,2 Other less common mutations occur in exons 18 and 20. These mutations result in increased autophosphorylation of the receptor, unregulated tyrosine kinase activity, and stimulation of downstream proliferative pathways in an uncontrolled manner.6,7

Multiple studies have shown that the presence of certain EGFR mutations render adenocarcinomas sensitive to small molecular inhibitors of the receptor’s tyrosine kinase activity, such as erlotinib (Tarceva, Genentech, South San Francisco, CA) and gefitinib (Iressa, AstraZeneca, Wilmington, DE).1,2,5,8–10 Furthermore, tumors that harbor wild-type EGFR receptors often do not respond to EGFR inhibitor therapy and actually are more responsive to traditional chemotherapy.8–10 Thus, guidelines from the National Comprehensive Cancer Network now recommend testing for EGFR mutations prior to instituting anti-EGFR small molecular therapy in all adenocarcinomas and non–small cell lung carcinomas that cannot be further subtyped by histology.11 In addition, a joint venture between 3 societies (College of American Pathologists, Association for Molecular Pathology, and the International Association for the Study of Lung Cancer) has proposed guidelines to promote the most accurate and appropriate testing for molecular markers prior to initiation of targeted therapy in lung carcinoma.12 Therefore, as clinical molecular diagnostics laboratories develop EGFR mutation testing in their respective laboratories, it is highly important that they choose the most accurate, sensitive, and reliable test with the most appropriate mutation coverage. However, laboratory workflow, ease of analysis, and platform availability must also be taken into consideration.

Here, 2 platforms for EGFR mutation detection—the EGFR RGQ ARMS Scorpions probe-based real-time polymerase chain reaction (PCR) mutation assay (referred to as RGQ) and the pyrosequencing EGFR Pyro assay (referred to as Pyro)—were compared (QIAGEN, Valencia, CA). The 2 platforms were evaluated in our laboratory with regard to assay requirements, assay performance (accuracy, sensitivity, and precision), and laboratory workflow parameters.

## Materials and Methods

### Specimen Selection

Cases of lung adenocarcinoma or non–small cell lung cancer with greater than 20% tumor cells were identified for analysis between July 2007 and July 2011 in the archives at the University of Nebraska Medical Center, Department of Pathology and Microbiology. To enrich for EGFR mutations, we evaluated cases based on clinical history to identify never smokers or remote smokers. Fifty-three specimens, consisting of formalin-fixed, paraffin-embedded (FFPE) tissue (stored at room temperature) from 52 patients, were identified for the validation study. One patient had 2 separate specimens tested. Of the 53 cases, 22 had an EGFR mutation analysis study performed by dideoxy sequencing at a reference laboratory. Most cases identified were resection specimens (73.6%) as opposed to small biopsy specimens (26.4%). The characteristics of the specimens tested in the study are listed in Table 1. This study was performed with approval by the University of Nebraska Medical Center Institutional Review Board.

### Cell Lines and Dilutions

Four mutant cell lines were used, including H1975 (c.2369C>T, p.T790M; c.2573T>G, p.L858R dual mutation), H1650 (c.2235_2249del15, p.E746_A750del), HCC4006 (c.2239_2248TTAAGAGAAG>C, p.L747_A750>P), and HCC827 (c.2236_2250del15, p.E746_A750del) (kind gift of Hamid Band, MD, PhD, University of Nebraska Medical Center, Eppley Cancer Center).13 The cell lines were maintained in RPMI-1640 media with 5% or 10% fetal bovine serum. Dilution studies were performed using DNA extracted from fresh cell lines diluted at a designated percentage in the HT29 cell line wild-type DNA (American Type Culture Collection, Manassas, VA). For these studies, 5 ng/tube DNA was used in each initial amplification reaction prior to pyrosequencing, and 25 ng/tube DNA was used for each RGQ reaction. Since all amplification products are less than 200 nucleotides in the EGFR RGQ kit and less than 150 nucleotides in the Pyro kit (according to the manufacturer), fresh/frozen cell line DNA was used for the limit-of-detection (LOD) studies.

Table 1

Patient and Specimen Characteristicsa

### DNA Extraction

DNA was extracted from 1 cm2 of 10-mm-thick paraffin-embedded tissue containing at least 20% tumor tissue. In certain cases, less starting tissue was used due to limited specimen. When necessary, macrodissection was performed to obtain the threshold level of the tumor. Sections were first deparaffinized with CitraSolv (Fisher Scientific, Hanover Park, IL) and then digested in 180 mL QIAGEN ATL buffer and 20 mL QIAGEN proteinase K for a minimum of 4 hours to overnight. DNA was extracted with the QIAcube automated extractor (QIAGEN) using the QIAamp DNA FFPE tissue kit (QIAGEN) and eluted in ATE buffer (QIAGEN). DNA was extracted from the cell lines using organic phenol-chloroform extraction.

### Real-Time PCR

Simultaneous real-time PCR amplification and mutation detection was performed using Amplification Refractory Mutation System (ARMS) technology and Scorpions dual-primer probes available in the EGFR RGQ PCR kit (QIAGEN). Seven separate mutation amplification reactions plus 1 wild-type control reaction (8 total) were performed, which allowed for the detection of 29 somatic mutations using a FAM fluorochrome label. While the mutations targeted are located in exons 18 to 21, the normal template control (TC) reaction targets a region in exon 2 of the EGFR gene for calculation of a ΔCt. Overall, the DNA input ranged from 30 ng/tube (240 ng total) to 260 ng/tube (2.08 mg total) total DNA with a mean of 60 ng/tube (n = 53). We determined through a DNA concentration analysis of 3 specimens that 50 ng DNA per reaction tube (400 ng total for 8 tubes) would yield a TC Ct between 23.00 and 30.69, as suggested by the manufacturer protocol, and this amount of DNA was used for most reactions.

Real-time detection was performed on the Rotor-Gene Q 5plex HRM Instrument (QIAGEN). Cycling parameters were as follows: 95°C for 15 minutes for 1 cycle, 95°C for 30 seconds, and then 60°C for 60 seconds for 40 cycles (acquiring FAM-target [green] and HEX internal control [yellow]). Analysis was performed using the Rotor-Gene Q series built-in software, version 2.0.2 (QIAGEN). Real-time curves were generated using FAM-labeled primer probes for both the TC tube (exon 2) and each mutation in separate tubes. To calculate a ΔCt result for each mutation reaction, we used the following equation: [mutation Ct] – [TC Ct] = ΔCt. Manufacturer-supplied ΔCt thresholds were used to call a mutation for LOD studies (≤ ΔCt threshold is positive), with the exception of p.T790M (see Results and Discussion).

### Pyrosequencing

For pyrosequencing analysis, the protocol was followed according to the manufacturer instructions (EGFR Pyro Assay; QIAGEN). Briefly, genomic DNA was amplified in 4 separate PCR reactions with primers specific for each of 4 exon “hotspots” in exons 18 (codon 719), 19, 20 (codons 768 and 790), and 21 (codons 858–861). For the Pyro assay, 5 ng was used per reaction for the initial amplification prior to pyrosequencing (total 20 ng/specimen). The reverse primer was labeled with a biotin tag. The cycling parameters included an initial activation step for 15 minutes at 95°C, followed by 42 cycles of 20 seconds at 95°C, 30 seconds at 53°C, and 20 seconds at 72°C, with a final extension of 5 minutes at 72°C. The biotin-labeled amplicons were purified using Streptavidin Sepharose High Performance beads (GE Healthcare, Piscataway, NJ). The purified amplicons attached to the beads were washed and denatured to single strands per the manufacturer instructions with use of the PyroMark Q24 Vacuum Workstation (QIAGEN). The beads were added to the pyrosequencing primers in 5 separate reactions (codons 768 and 790 are sequenced in separate pyrosequencing reactions), and pyrosequencing was performed on the PyroMark Q24 Instrument (QIAGEN). The pyrosequencing results were analyzed using the PyroMark Q24 version 2.0.6 software (QIAGEN), which identifies the presence of a specific mutation and its percentage. Manufacturer-supplied LOD thresholds were used to call a mutation for LOD studies (≥ % LOD is positive).

### Dideoxy Sequencing

Amplification of exons 18 to 21 was performed using the following primers (sequences courtesy of Jason Foster, MD, University of Nebraska Medical Center): EGFRex18, 5′-ACTGCTTTCCAGCATGGTGAGG-3′ (forward) and 5′-CTTGCAAGGACTCTGGGCTCC-3′ (reverse); EGFRex19, 5′-GTGCATCGCTGGTAACATCCAC-3′ (forward) and 5′-GGGCCTGAGGTTCAGAGCCAT-3′ (reverse); EGFRex20, 5′-ATGCGTCTTCACCTGGAAGG-3′ (forward) and 5′-CGCAGACCGCATGTGAGGAT-3′ (reverse); and EGFRex21, 5′-CCTGAATTCGGATGCAGAGCTTC-3′ (forward) and 5′-GGAAGGCAGCCTGGTCCCTG-3′ (reverse). Optimal cycling parameters included an initial denaturation at 95°C for 4 minutes and 45 cycles at 95°C for 30 seconds, 58°C (exon 20) or 63°C (exons 18, 19, and 21) for 30 seconds, and 72°C for 30 seconds, followed by a final extension at 72°C for 4 minutes. The amplicons were subjected to a cleanup step with USB ExoSAP-IT (Affymetrix, Santa Clara, CA) and then sequenced bidirectionally with Big Dye Terminator version 3.1 (Applied Biosystems, Foster City, CA). After cleanup with Centri-Sep (Life Technologies, Grand Island, NY) or Performa DTR Gel Filtration (Edge Biosystems, Gaithersburg, MD) columns, sequencing products were analyzed on an ABI 3130xL or 3500 Genetic Analyzer (Applied Biosystems). Positive specimens without prior reference laboratory determination of the mutation status were confirmed by dideoxy sequencing. All specimens not previously submitted to a reference laboratory for dideoxy sequencing were subjected to exon 18 to 21 sequencing in our laboratory.

### Statistical Analysis

For sensitivity and specificity calculations, dideoxy sequencing was considered the gold standard. Averages and standard deviations were calculated using Microsoft Excel 2010 (Microsoft, Redmond, WA).

## Results

Fifty-three lung cancer specimens from 52 patients were tested for EGFR mutations using both the RGQ assay and the Pyro assay. Demographics showed that 39 (75.0%) of 52 patients were women, and 27 (51.9%) of 52 had a known smoking history (Table 1). Forty-three of the 53 tumors (81.1%) were from lung parenchyma, and 45 (85.0%) were diagnosed as primary pulmonary adenocarcinoma (Table 1).

In comparing the 2 assays (RGQ and Pyro), 14 EGFR mutations were identified in 13 specimens using at least 1 of the assays, with a mutation concordance rate of 92.9%. Using the RGQ assay, the Pyro assay, and/or dideoxy sequencing, 17 of the 53 specimens had at least 1 EGFR mutation. Because 2 specimens demonstrated 2 concomitant mutations, a total of 19 mutations were present in the specimens by any method Table 2. Representative images from the RGQ, Pyro, and dideoxy sequencing assays for individual specimens containing various mutations are presented in Figure 1.

Table 2

Mutation Detection by Platform

Figure 1

Examples of 5 different epidermal growth factor receptor mutations as detected by 3 methods: RGQ assay (upper left panel), dideoxy sequencing (lower left panel), and pyrosequencing (lower right panel). A wild-type pyrosequence is shown for comparison in the upper right panel. A, Exon 19 deletion (p.E746_A750; c.2235_2249del15 [estimated tumor percentage = 80%]; ΔCt = 1.28). B, Exon 18 mutation (p.G719A; c.2156G>C [estimated tumor percentage = 50%]; ΔCt = 5.23). C, Exon 20 mutation (p.S768I; c.2303G>T [estimated tumor percentage = 50%]; ΔCt = 4.21). D, Exon 20 mutation (p.T790M; c.2369C>T [estimated tumor percentage = 50%]; ΔCt = 3.61). E, Exon 21 mutation (p.L858R; c.2573T>G [estimated tumor percentage = 30%]; ΔCt = 5.25). The bent arrow marks the site of the deletion start site. Up arrows indicate nucleotide gain and down arrows indicate nucleotide loss for missense mutations. Control, template control curve; Δ, mutant curve; F, forward sequencing reaction; R, reverse sequencing reaction.

Figure 1

Examples of 5 different epidermal growth factor receptor mutations as detected by 3 methods: RGQ assay (upper left panel), dideoxy sequencing (lower left panel), and pyrosequencing (lower right panel). A wild-type pyrosequence is shown for comparison in the upper right panel. A, Exon 19 deletion (p.E746_A750; c.2235_2249del15 [estimated tumor percentage = 80%]; ΔCt = 1.28). B, Exon 18 mutation (p.G719A; c.2156G>C [estimated tumor percentage = 50%]; ΔCt = 5.23). C, Exon 20 mutation (p.S768I; c.2303G>T [estimated tumor percentage = 50%]; ΔCt = 4.21). D, Exon 20 mutation (p.T790M; c.2369C>T [estimated tumor percentage = 50%]; ΔCt = 3.61). E, Exon 21 mutation (p.L858R; c.2573T>G [estimated tumor percentage = 30%]; ΔCt = 5.25). The bent arrow marks the site of the deletion start site. Up arrows indicate nucleotide gain and down arrows indicate nucleotide loss for missense mutations. Control, template control curve; Δ, mutant curve; F, forward sequencing reaction; R, reverse sequencing reaction.

Using dideoxy sequencing as the gold standard, the clinical sensitivity was 73.7% for the RGQ system and 68.4% for the Pyro assay Table 3. Using the RGQ assay, 5 mutations were not detected, including a rare exon 18 deletion (c.2127_2129delAAC; p.E709_T710delinsD), 2 rare missense mutations in exon 20 (c.2379G>T; p.M793I and c.2320G>A; p.V774M), and 2 rare exon 20 insertions (c.2317_2319delCACins12; p.H773delHinsYNPY and c.2311_2313dupAAC; p.N771dup). In addition to these 5 mutations, an additional exon 20 insertion (c.2299_2307dup9; p.A767_V769dupASV) could not be identified by the Pyro assay because it does not detect exon 20 insertions. All undetected mutations were due to each assay’s specific design. However, for the common mutations (exon 19 deletions, p.G719 mutations, p.S768I, p.T790M, and p.L858R), the clinical sensitivity of both assays was 100%. In addition, 4 cell lines containing 5 different EGFR mutations Table 4 were also tested by both assays with concordant results.14 Clinical specificity was 100% for both the RGQ and Pyro assays (Table 3).

To determine the analytical sensitivity LOD of each assay, we analyzed 2 cell lines (H1975 and H1650) containing the most commonly identified EGFR mutations (p.E746_A750 and p.L858R) and the most common resistance mutation (p.T790M). Interestingly, neither cell line was purely homozygous (100%) or heterozygous (50%) by pyrosequencing analysis; instead, the respective mutation or mutations were present between 65% and 85%, depending on the cell line and mutation (Table 4). This is likely due to increased EGFR copy number and is in agreement with Gandhi et al,14 who identified that in these cell lines a majority of EGFR copies, but not all, were mutated, referred to as mutant allele–specific imbalance. Thus, LOD studies need to be evaluated in the context of these percentages.

In the RGQ assay, the p.E746_A750del and p.L858R mutations could be detected at 2% and 5% cell line DNA, respectively Table 5. Using the mutant allele percentages determined by pyrosequencing in Table 4, this would translate to mutant allele LODs of 1.3% (2% × 0.65) and 3.7% (5% × 0.73), respectively. The p.T790M resistance mutation was detected at 10% cell line DNA using the manufacturer-supplied ΔCt threshold of 6.38 (mutant allele LOD of 8.1% [10% × 0.81]). The low ΔCt threshold for this mutation is due to known mispriming that can occur with the p.T790M site (see Discussion and Figure 2). However, on the basis of our evaluation of late Ct and high ΔCt values for p.T790M in known wild-type specimens (Ct = 37.2 ± 1.5 and ΔCt = 13.16 ± 1.1), we changed the p.T790M threshold to 9.0, which yielded a mutant allele LOD of 1.6% (2% × 0.81).

The Pyro assay demonstrated that the p.E746_A750 mutation could be detected at 5% cell line DNA, resulting in a mutant allele LOD of 3.25% (5% × 0.65), and the p.L858R mutation could be detected at 10% cell line DNA, resulting in a mutant allele LOD of 7.3% (10% × 0.73) Table 6. The p.T790M mutation was detected at 15% cell line DNA (mutant allele LOD of 12.2% [15% × 0.81]). This is mainly due to the recurring artifact peak that occurs at the mutation position (see Discussion and Figure 2). Similar to the RGQ assay, we evaluated the raw data to see if an adjustment of the manufacturer LOD for the p.T790M mutation could be modified and determined that it could not be done. Overall, the RGQ real-time PCR assay demonstrated a higher analytical sensitivity than the Pyro assay (Tables 5 and 6) for 3 different mutations in 2 cell lines. Both systems showed the least sensitivity in their detection of the p.T790M mutation compared with the other mutations.

Table 3

Clinical Sensitivity and Specificity for Individual Mutation Detectiona

Precision studies were performed using 2 different patient specimens with the 2 most common mutations (p.E746_ A750del and p.L858R). Good reproducibility was identified in both interrun and intrarun analyses for both the RGQ ΔCt value (coefficient of variation, 0.59%–9.5%) and Pyro percent mutation value (coefficient of variation, 1.1%–15.8%) Table 7 and Table 8.

Table 4

Cell Lines in Analysis

Table 5

Comparison of Limit of Detection (LOD): RGQ Real-Time PCR

More FFPE-derived DNA input was required for the RGQ assay than for the Pyro assay (400 ng total vs 20 ng). The 400 ng (50 ng per tube) was optimized to give us adequate TC Ct values, averaging approximately 25 Cts. Three wild-type specimens were evaluated with TC Ct values later than 30.69 (starting total DNA input 240–280 ng or 30–35 ng/tube). All 3 specimens were wild type by the RGQ and Pyro assays and dideoxy sequencing. In addition, all 3 contained abundant tumor (60%–80%). A fourth specimen with a TC Ct value later than 30.69 was positive for a mutation (exon 19 deletion).

Figure 2

Examples of various artifacts identified in RGQ and Pyro assays. A, Artifacts associated with the p.T790M assays. Left panel, an RGQ curve with a late Ct value in a wild-type specimen for the p.T790M mutation. Upper right panel, the same specimen with an artifact peak at the site of the mutation (7% “mutant” peak, arrow). Lower right panel, the corresponding wild-type control with an artifact peak (4% “mutant” peak, arrow). The tumor percentage was 30%, and the specimen was wild type for all other mutations tested. B, An artifactual linear amplification curve that gave a positive Ct and ΔCt value for the p.L858R mutation (blue curve). Compare with the template control curve for that specimen (Control). Artifactual pyrosequencing peaks (arrows) for the p.G719C (C) and p.L861R (D) mutations. The tumor percentages were 60% and 80%, respectively. Both specimens were wild type for that specific mutation by either the RGQ system (p.G719C) or follow-up pyrosequencing (p.L861R).

Figure 2

Examples of various artifacts identified in RGQ and Pyro assays. A, Artifacts associated with the p.T790M assays. Left panel, an RGQ curve with a late Ct value in a wild-type specimen for the p.T790M mutation. Upper right panel, the same specimen with an artifact peak at the site of the mutation (7% “mutant” peak, arrow). Lower right panel, the corresponding wild-type control with an artifact peak (4% “mutant” peak, arrow). The tumor percentage was 30%, and the specimen was wild type for all other mutations tested. B, An artifactual linear amplification curve that gave a positive Ct and ΔCt value for the p.L858R mutation (blue curve). Compare with the template control curve for that specimen (Control). Artifactual pyrosequencing peaks (arrows) for the p.G719C (C) and p.L861R (D) mutations. The tumor percentages were 60% and 80%, respectively. Both specimens were wild type for that specific mutation by either the RGQ system (p.G719C) or follow-up pyrosequencing (p.L861R).

Table 6

Comparison of Limit of Detection (LOD): Pyrosequencing

Table 7

Precision Comparison (n = 3 for Each Mutation): RGQ Real-Time PCR Intrarun and Interrun Reproducibility

Table 8

Precision Comparison (n = 3 for Each Mutation): Pyrosequencing Intrarun and Interrun Reproducibility

## Discussion

Analysis of EGFR mutations is now recommended for all primary pulmonary adenocarcinomas and non–small cell lung cancers, and guidelines for testing have recently been published.12 The accurate detection of an EGFR mutation is important for directing targeted therapy with EGFR inhibitors to improve therapeutic efficacy and reduce adverse effects and cost. Since many lung carcinomas are detected at advanced stages, specimens for diagnosis and molecular analysis are often limited to small diagnostic biopsy specimens that have already been partially exhausted through morphologic and immunophenotypic analysis. Thus, a clinical assay to detect EGFR mutations should be highly sensitive and require as little tumor DNA as possible.

Although both assays could not detect rare mutations in exons 18 and 20 that were detected by dideoxy sequencing, due to the design of the platforms, the RGQ real-time PCR was more clinically and analytically sensitive than the Pyro platform. The RGQ system was designed to detect certain exon 20 insertions (approximately 60% of known exon 20 mutations4), whereas the Pyro assay does not interrogate exon 20 insertions. Exon 20 insertions are putatively thought to be resistance mutations and constitute approximately 5% to 10% of known EGFR mutations found in the lung.4,15,16 Since they are putatively resistant, a false-negative result would not necessarily affect clinical treatment, as EGFR inhibitors are usually given only when a sensitive mutation is identified. However, if there was a concomitant sensitive mutation with an exon 20 insertion resistance mutation, the pyrosequencing assay would not detect this resistance mutation, and the patient might be at risk for treatment failure.

Pyrosequencing does have the advantage over allele-specific–based real-time PCR assays (such as the RGQ system) in that other potential rare mutations present in the area of pyrosequencing interrogation may be detected or at least indicated, thus leading to potential dideoxy sequencing of the region to confirm the mutation. We have experienced this phenomenon with regard to our pyrosequencing-based KRAS and BRAF assays, in which rare mutations have been identified. As of yet, we have not been able to correlate clinician action with regard to targeted therapy with these rare variants, but our clinicians have expressed a desire to detect rare p.V600 mutations in BRAF. As next-generation sequencing becomes clinically available and technology improves to test rare variant biological function, it is likely that identifying rare variants will become important in the future.

The allele-specific design of the RGQ assay does allow for a higher analytical sensitivity than the Pyro assay (Tables 5 and 6). Allele-specific–based assays are generally thought to have sensitivities of less than 1% to 5%, whereas pyrosequencing assays are more on the order of 5% sensitivity. For the RGQ assay, the manufacturer has developed ΔCt cutoffs to determine a positive or negative result for each mutation. These correspond to sensitivities as low as less than 1.5% for some mutations (L. Williams, MT(ASCP)SH, QIAGEN, written communication, May 2012). The Pyro assay handbook, supplied by the manufacturer, lists the LOD for each mutation, which ranges from 0.6% (c.2238_2252>GCA; p.L747_T751>Q) to 10.7% (p.T790M mutation). Based on our dilutional studies of cell line DNA and use of pyrosequencing data to estimate the percent mutated allele, the sensitivity of the RGQ assay showed mutant allele LODs for p.L858R, p.E746_ A750del, and p.T790M of 3.7%, 1.3%, and 1.6%, respectively. The Pyro assay was less sensitive than the RGQ assay, with mutant allele LODs for p.L858R, p.E746_A750del, and p.T790M of 7.3%, 3.25%, and 12.2%, respectively. Thus, in specimens with minimal tumor, the RGQ system would be the preferred assay to prevent false-negative results.

Several artifacts were identified that required mandatory visual inspection of either pyrograms or amplification curves (Figure 2). The most problematic were those associated with the p.T790M mutation in both the RGQ assay and the Pyro assay (Figure 2A). This is likely due to the CG-rich nature of the exon 20 sequence region. In the RGQ assay, late curves with Ct values later than 35.0 (ΔCt > 11.0) were identified in more than half of the samples. The tendency for late curves and nonspecificity is likely the reason that the company’s recommended ΔCt threshold of 6.38 is much lower than the other mutations, requiring an earlier mutation Ct value for positivity. However, through our studies, we determined that this manufacturer-determined ΔCt value was too low and that a positive mutation may be missed with this threshold. Through our LOD analysis of p.T790M by the RGQ method and studies of wild-type specimens, we changed our analysis parameters so that any ΔCt between 6.38 and 9.0 should be considered positive. Tumor percentage of the specimen, the presence of other mutations, and the TC Ct values that give indication to the proper amount of DNA input should also be considered when evaluating a late Ct curve. The most problematic artifact in the Pyro assay reaction also occurs with the p.T790M mutation, at the dispensation specific for the mutation (Figure 2A). This explains the poor analytical sensitivity in the LOD studies and the reason that the manufacturer-recommended LOD is 10.7%. We were not able to adjust the Pyro LOD threshold of the p.T790M mutation supplied by the manufacturer as we did for the RGQ assay. Careful consideration of the presence of the p.T790M mutation is important, since this mutation is responsible for approximately 50% of acquired resistance to EGFR small molecular inhibitor therapy1719 but can also occur de novo.16 Identifying a true p.T790M resistance mutation is important to guiding optimally targeted therapy. If a p.T790M mutation is missed in the presence of a sensitivity mutation, an EGFR inhibitor would be given with likely therapeutic failure and unnecessary adverse effects. Alternatively, if the p.T790M mutation is identified as a false-positive result due to nonspecific artifacts, EGFR inhibitor therapy may be improperly withheld.

Other potential artifacts were also identified that could lead to a false-positive result if not properly considered. In the RGQ assay validation, a linear drift amplification curve occurred with the p.L858R mutation reaction mix in 1 specimen wild-type for that mutation (Figure 2B). This drift curve crossed the threshold, resulting in artifactually positive Ct and ΔCt values. If one does not follow standard protocol in real-time PCR by visualizing the actual curves, a false-positive result could occur. We have also subsequently identified linear drift with a no TC p.T790M reaction in a clinical specimen. In the Pyro assay, false-positive artifact peaks were also identified in pyrogram tracings for codon 719 (1.6%–1.9%) and codon 861 (4.4%–5.8%) (Figure 2C and Figure 2D). The percent mutation derived in these pyrosequencing assays for these artifact peaks did not mirror the estimated tumor percentage for the tumor-containing specimens (20%–60% tumor). Significant tumor heterogeneity would have to occur for such a discrepancy between percent tumor and percent mutation.

There are several other advantages and disadvantages of each system Table 9. We found that more DNA was optimal for the RGQ system (400 ng) vs the Pyro system. The Pyro assay requires only approximately 20 ng per specimen in our hands. Considering that many of the specimens received for EGFR testing are sparse needle core biopsy samples, the amount of DNA input required may be an important issue. However, clinically, we have found that the percentage of tumor cells within the biopsy specimen is a greater limiting factor than the amount of tissue. It should be stated that referral DNA specimens, not included in this study and with limited product, have been tested successfully by the RGQ assay with less DNA (data not shown). The amount chosen was 50 ng/tube (400 ng total), which would consistently give adequate control TC values, thereby streamlining workflow, whereas 30 to 35 ng/tube often gave Ct values later than the recommended 30.69 value.

The RGQ assay is a 1-step amplification and detection system as opposed to the 2-step amplification and pyrosequencing assay. However, the RGQ real-time setup can be tedious and complex with 8 reactions per specimen, allowing for the possibility of operator error. For laboratories with high-volume EGFR studies, a liquid handler may be beneficial in assay setup. The reagent cost for the RGQ assay is more expensive than the Pyro assay. Depending on the number of specimens evaluated per run with controls, the cost can range from approximately $170 to$250 for the Pyro assay as compared with $260 to$400 for the RGQ assay. Hands-on technician time is approximately 2 to 3 hours for each assay. Each system also requires a different piece of capital equipment (Rotor-Gene Q 5plex HRM or a PyroMark Q24 or Q96). Turnaround time is approximately 1 day for both assays, but the expense of the assays may dictate how often batch analysis is done to still reach the recommended 10 working days for both EGFR and EML4-ALK biomarkers.12 Both systems come with manufacturer-designed analysis software to aid in mutation analysis, which is especially helpful with the Pyro assay. Whereas the Pyro assay can identify specific mutations for exon 19 deletions and codon G719 base pair substitutions, the RGQ system cannot differentiate between these specific alterations. Although it is not currently a clinical necessity to determine the exact mutation, future studies or possible recommendations by the College of American Pathologists may dictate a need for proper mutation designation.

Table 9

Comparison of EGFR RGQ Real-Time PCR and EGFR Pyro Assay Systems

We used dideoxy (Sanger) sequencing as our gold standard in this study. Dideoxy sequencing is currently accepted as the gold standard when a laboratory is in the process of validating somatic mutations in cancer, and many laboratories use dideoxy sequencing as the method of primary detection of EGFR mutations in lung carcinoma. Dideoxy sequencing is useful in detecting EGFR mutations since it is a cost-effective way to evaluate all 4 EGFR kinase domain exons (18, 19, 20, and 21) in their entirety. However, it is commonly accepted that dideoxy sequencing has an analytical sensitivity of only approximately 15% to 20% (unless PCR enrichment strategies are used), requiring a tumor cell population of at least 30% to 40% (assuming a heterozygous mutation), so that mutant peaks can be identified on the electropherogram.20,21 Therefore, use of dideoxy sequencing as a gold standard must be viewed with caution because it is possible that low-level mutations would not be detected.

The current recommendation guidelines strongly encourage that a laboratory-employed method be able to detect a mutation in a 10% tumor cell population (5% mutant allele, assuming heterozygosity) and detect the secondary p.T790M mutation in as few as 5% of tumor cells.12 This is especially important in the setting of EGFR mutation analysis since many of the specimens submitted for EGFR testing are small needle core biopsy samples or pleural fluids with 20% or less tumor cells. When using the 10% tumor cell DNA percentage recommendation, both the RGQ and Pyro assays have adequate sensitivity for the exon 19 deletion mutations and the p.L858R mutation. However, the Pyro assay p.T790M LOD does not fulfill the 10% cell requirement. The RGQ assay p.T790M LOD does not fulfill the 5% requirement using the RGQ assay manufacturer-recommended cutoff of 6.38 (our cutoff adjustment to 9.0 would fulfill the 5% recommendation). Unlike dideoxy sequencing, neither the RGQ nor the Pyro assay evaluates the entire EGFR kinase domain, and both will miss rare mutations, as seen in this study. Some caution must be used when employing targeted mutation detection assays such as these since rare mutations will be missed. Both assays did identify all of the common EGFR mutations. Ultimately, laboratories must consider both analytical sensitivity and mutation coverage when choosing the most appropriate assay, taking into consideration current recommendations from professional societies.

In conclusion, both the EGFR RGQ ARMS Scorpions real-time PCR system and the EGFR Pyro pyrosequencing system are clinically sensitive for detecting common EGFR mutations, with the RGQ assay having a higher clinical and analytical sensitivity than the Pyro system with fewer artifacts. However, the pyrosequencing assay requires less DNA, is less complex in assay design and technician setup, and can determine specific exon 19 deletions and codon 719 substitutions. Individual laboratories should consider these performance characteristics and limitations when evaluating these EGFR mutation assays for adoption into the laboratory workflow.

This study was supported by a contract with QIAGEN, which supplied EGFR kit reagents and also fees for institutional review board, labor, and pathology services. All results and conclusions are solely those of the authors, without manufacturer editorial influence, as agreed upon contractually with QIAGEN. In addition, the personal communication with L. Williams (QIAGEN) has been approved by that individual. There are no other conflicts of interest for any of the authors.

## References

1.
Paez
JG
Janne
PA
Lee
JC
et al
EGFR mutations in lung cancer: correlation with clinical response to gefitinib therapy
.
Science
.
2004
;
304
:
1497
1500
.
2.
Lynch
TJ
Bell
DW
Sordella
R
et al
Activating mutations in the epidermal growth factor receptor underlying responsiveness of non–small-cell lung cancer to gefitinib
.
N Engl J Med
.
2004
;
350
:
2129
2139
.
3.
Kosaka
T
Yatabe
Y
Endoh
H
et al
Mutations of the epidermal growth factor receptor gene in lung cancer: biological and clinical implications
.
Cancer Res
.
2004
;
64
:
8919
8923
.
4.
Shigematsu
H
Lin
L
Takahashi
T
et al
Clinical and biological features associated with epidermal growth factor receptor gene mutations in lung cancers
.
J Natl Cancer Inst
.
2005
;
97
:
339
346
.
5.
Rosell
R
Moran
T
Queralt
C
et al
Screening for epidermal growth factor receptor mutations in lung cancer
.
N Engl J Med
.
2009
;
361
:
958
967
.
6.
Greulich
H
Chen
TH
Feng
W
et al
Oncogenic transformation by inhibitor-sensitive and -resistant EGFR mutants
.
PLoS Med
.
2005
;
2
:
e313
.
7.
Chen
YR
Fu
YN
Lin
CH
et al
Distinctive activation patterns in constitutively active and gefitinib-sensitive EGFR mutants
.
Oncogene
.
2006
;
25
:
1205
1215
.
8.
Pao
W
Miller
V
Zakowski
M
et al
EGF receptor gene mutations are common in lung cancers from “never smokers” and are associated with sensitivity of tumors to gefitinib and erlotinib
.
Proc Natl Acad Sci U S A
.
2004
;
101
:
13306
13311
.
9.
Mok
TS
Wu
YL
Thongprasert
S
et al
Gefitinib or carboplatin-paclitaxel in pulmonary adenocarcinoma
.
N Engl J Med
.
2009
;
361
:
947
957
.
10.
Hirsch
FR
Kabbinavar
F
Eisen
T
et al
A randomized, phase II, biomarker-selected study comparing erlotinib to erlotinib intercalated with chemotherapy in first-line therapy for advanced non–small-cell lung cancer
.
J Clin Oncol
.
2011
;
29
:
3567
3573
.
11.
National Comprehensive Cancer Network (NCCN)
.
NCCN Clinical Practice Guidelines in Oncology: Non–Small Cell Lung Cancer
.
Fort Washington, PA
:
NCCN
.
2013
; version 2.2013.
12.
Lindeman
NI
Cagle
PT
Beasley
MB
et al
Molecular testing guideline for selection of lung cancer patients for EGFR and ALK tyrosine kinase inhibitors: guideline from the College of American Pathologists, International Association for the Study of Lung Cancer, and Association for Molecular Pathology
.
J Mol Diagn
.
2013
;
15
:
1
39
.
13.
Chung
BM
Raja
SM
Clubb
RJ
et al
Aberrant trafficking of NSCLC-associated EGFR mutants through the endocytic recycling pathway promotes interaction with Src
.
BMC Cell Biol
.
2009
;
10
:
84
.
14.
Gandhi
J
Zhang
J
Xie
Y
et al
Alterations in genes of the EGFR signaling pathway and their relationship to EGFR tyrosine kinase inhibitor sensitivity in lung cancer cell lines
.
PLoS One
.
2009
;
4
:
e4576
.
15.
Wu
JY
Wu
SG
Yang
CH
et al
Lung cancer with epidermal growth factor receptor exon 20 mutations is associated with poor gefitinib treatment response
.
Clin Cancer Res
.
2008
;
14
:
4877
4882
.
16.
Wu
JY
Yu
CJ
Chang
YC
et al
Effectiveness of tyrosine kinase inhibitors on “uncommon” epidermal growth factor receptor mutations of unknown clinical significance in non-small cell lung cancer
.
Clin Cancer Res
.
2011
;
17
:
3812
3821
.
17.
Kobayashi
S
Boggon
TJ
Dayaram
T
et al
EGFR mutation and resistance of non–small-cell lung cancer to gefitinib
.
N Engl J Med
.
2005
;
352
:
786
792
.
18.
Pao
W
Miller
VA
Politi
KA
et al
Acquired resistance of lung adenocarcinomas to gefitinib or erlotinib is associated with a second mutation in the EGFR kinase domain
.
PLoS Med
.
2005
;
2
:
e73
.
19.
Kosaka
T
Yatabe
Y
Endoh
H
et al
Analysis of epidermal growth factor receptor gene mutation in patients with non–small cell lung cancer and acquired resistance to gefitinib
.
Clin Cancer Res
.
2006
;
12
:
5764
5769
.
20.
Tsiatis
AC
Norris-Kirby
A
Rich
RG
et al
Comparison of Sanger sequencing, pyrosequencing, and melting curve analysis for the detection of KRAS mutations: diagnostic and clinical implications
.
J Mol Diagn
.
2010
;
12
:
425
432
.
21.
Monzon
FA
Ogino
S
Hammond
ME
et al
The role of KRAS mutation testing in the management of patients with metastatic colorectal cancer
.
Arch Pathol Lab Med
.
2009
;
133
:
1600
1606
.