Sung Lee, Jianli Cao, Theresa May, Jingchuan Li, Lilia Corona, Nitta Lee, Yiqiao Wu, Carrie Wong, Kelli DeMartin, Victoria H. Brophy, Stephen Soviero and John F. Palma*
Roche Molecular Systems, Inc. USA
Received Date: July 22, 2015 Accepted Date: July 24, 2015 Published Date: July 26, 2015
Citation: Lee S, Cao J, May T, Li J, Corona L, et al. (2015) Performance Characteristics of a PCR Assay for the Detection of KRAS Mutations in Formalin- Fixed Paraffin-Embedded Tissue Samples of Non-Small Cell Lung Cancer. J Mol Biomark Diagn 6:246. doi:10.4172/2155-9929.1000246
Copyright: © 2015 Lee S, et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Visit for more related articles at Journal of Molecular Biomarkers & Diagnosis
Introduction: Approximately 25% of non-small cell lung cancer (NSCLC) tumors contain mutations in KRAS. These tumors are insensitive to therapy directed against the epidermal growth factor receptor and appear to be resistant to adjuvant chemotherapy. The current study demonstrates the performance of the cobas® KRAS Mutation Test (cobas test), a TaqMelt polymerase chain reaction (PCR) assay designed to detect 19 mutations in codons 12, 13, and 61.
Methods: To reflect real-world testing conditions, the study used formalin-fixed, paraffin-embedded tissue (FFPET) NSCLC samples for the predominant mutations. DNA blends of cell lines and plasmids were used where FFPET samples were not available.
Results: In the limit of detection study, a correct mutation call rate of ≥95% was obtained with approximately 5% mutant sequences using 3.1-50.0 ng DNA per PCR reaction. Mutation levels as low as 2.4% consistently yielded correct mutation calls when 50.0 ng DNA was used for testing. The cobas test performance was compared to Sanger sequencing in a method correlation using two cobas reagent lots and 194 specimens. After resolution of discordant results using 454 sequencing, the positive, negative, and overall percent agreement for mutations in codons 12/13 and 61 between the cobas test and sequencing ranged from 97.0% to 100%. Mutation detection was 100% reproducible and showed greater specificity than Sanger sequencing. Test performance was not impaired by the presence of interfering substances or clinically relevant microbes, and inclusivity testing demonstrated the kit’s ability to detect rare mutations.
Conclusions: The cobas test is a robust, sensitive, and reproducible method for detecting KRAS mutations in FFPET tumor samples from NSCLC patients.
Non-small cell lung cancer; Polymerase chain reaction; KRAS; Formalin-fixed; Paraffin-embedded
KRAS is an intracellular signal transduction protein that has GTPase activity and functions downstream of the epidermal growth factor receptor (EGFR) . It normally functions as a switch to control intracellular signaling. In non-small cell lung cancer (NSCLC), approximately 90% of KRAS activating mutations occur in codon 12, with codons 13, and 61 accounting for the next most common mutation sites . Activating mutations in KRAS prevent GTP hydrolysis to GDP, thus inducing constitutive downstream signaling, and account for the gene’s function as an oncogenic driver [3,4]. In NSCLC, KRAS mutations are found in approximately 30% of adenocarcinomas and 5% of squamous cell tumors, and they occur more frequently in Caucasians and current or former smokers [5,6].
Effective therapies that directly target mutated KRAS have proven elusive. Nonetheless, assessing the mutational status of KRAS is important for several reasons. As KRAS acts downstream of EGFR, tumors with KRAS activating mutations are resistant to EGFR inhibition. Thus the presence of KRAS mutations in lung tumors contraindicates treatment with anti-EGFR tyrosine kinase inhibitors (TKIs) . Recent studies suggest that KRAS also confers resistance to adjuvant chemotherapy . Moreover, promising clinical results have recently emerged demonstrating the clinical utility of targeting MEK, a signaling molecule that acts downstream of KRAS. In a randomized phase 2 study, Jänne and colleagues demonstrated an extension of progression-free survival (PFS) when the MEK inhibitor, selumetinib,was combined with docetaxel for second-line treatment of NSCLC . The study evenly randomized 87 patients with mutated KRAS to receive docetaxel plus either selumetinib or placebo. Selumetinib added to docetaxel increased the overall response rate (ORR) to 37% versus 0% for placebo (P<0.001) and improved median PFS to 5.3 months versus 2.1 months for placebo (P=0`.014; HR, 0.58). Although not significant, median overall survival (OS) also improved in the selumetinib arm (9.4 months vs 5.2 months; HR, 0.80; P=0.21).With MEK established as an important target, other MEK inhibitors are being pursued for treating NSCLC. Trametinib is one such novel molecule that has shown promise in melanoma. It is currently approved for treating melanoma and has shown efficacy when treating NSCLC [9,10]. Cobimetinib (GDC-0973), which also inhibits MEK, has demonstrated efficacy and acceptable tolerability in a phase III trial in combination with vemurafenib in BRAF V600-mutated, treatment-naive, metastatic melanoma . Cobimetinib is also being investigated in early phase trials in combination with several investigational agents in NSCLC and other tumor types, as well as in cancer patients with a mutant KRAS gene .
Given the importance of KRAS mutations in the mitogen-activated protein kinase (MAPK) pathway, there is a need for reliable testing methods to determine KRAS mutation status that could eventually predict response to MEK inhibitors or determine eligibility for drugs that target other molecules in the MAPK pathway that are upstream or downstream of KRAS. The current study was undertaken to evaluate the cobas®KRAS Mutation Test (cobas test) for its ability to detect common mutations in KRAS codons 12, 13, and 61 in formalin-fixed, paraffin-embedded tissue (FFPET) from NSCLC tumors. The cobas test uses TaqMelt real-time polymerase chain reaction (PCR) technology, a patented method that uses a dual-labelled probe to differentiate between mutated and wild-type DNA sequences with high sensitivity and specificity. To assess clinical utility, the cobas test was evaluated for limit of detection (LOD), repeatability, correlation to Sanger sequencing, inclusivity, and the effect of substances in the sample that could potentially interfere with performance.
FFPET specimens from NSCLC tumors were purchased from commercial vendors (ABS Bio, Willington, DE, Asterand, Detroit, MI, Bioserve, Beltsville, MD, Conversant Bio, Huntsville, AL, Cureline Inc., Burlingame, CA, Cytomex, Lexington, MA, Discovery Life Sciences, Inc., Los Osos, CA, ILS Bio, Chestertown, MD, Indivumed Inc., Kensington, MD, and Protogenex, Culver City, CA) based on their origin of tissue and histological characteristics. Twelve adjacent 5-micron sections of each FFPET specimen were cut on a microtome. The first and the twelfth sections were stained with H&E and examined by a pathologist to confirm that they were NSCLC specimens and to assess the tumor content of each specimen. NSCLC cell lines with mutations in the KRAS gene (NCI-H23 with GGT>TGT in codon 12, NCI-H1734 with GGC>TGC in codon 13, and NCI-H460 with CAA>CAT in codon 61) were obtained from the American Type Culture Collection (www.atcc.org).
cobas® KRAS mutation test
The cobas test is a TaqMelt™-based PCR assay consisting of two major processes: (1) manual specimen preparation to obtain genomic DNA from a single 5-micron section from separate FFPET samples; and (2) PCR amplification and detection of target DNA using complementary primer pairs and two oligonucleotide probes labeled with a fluorescent dye. One probe is designed to detect the KRAS codon 12/13 sequence in exon 2, and the other probe is designed to detect the KRAS codon 61 sequence in exon 3 of the KRAS gene. Mutation detection is achieved by melting curve analysis using the cobas® z 480 analyzer. A mutant control, a negative control, and a calibrator are included in each run to confirm the validity of the run. The analysis of raw data and result reporting are fully automated. The DNA isolation, amplification/detection protocol, qPCR validation and result reporting can be performed in less than 8 hours and have been described previously .
Briefly, DNA is extracted from a single, deparaffinized 5 μm section of FFPET, quantified by Nanodrop (Thermo Scientific, Waltham, MA), and diluted with sample diluent to 2 ng/μL following the standard procedure described in the cobas test instructions for use . The target DNA is amplified and detected on the cobas®z 480 analyzer using the amplification and detection reagents provided in the cobas test. Testing for this study was conducted with the cobas® 4800 System Application Software 2.0 software. Analysis was performed with the cobas®4800 system KRAS Assay Specific Analysis Package v184.108.40.2063.
The 19 mutations detected by the kit are as follows, with the mutated nucleotide underlined:
G12A (GGT>GCT), G12C (GGT>TGT), G12D (GGT>GAT), G12R (GGT>CGT), G12S (GGT>AGT), G12V (GGT>GTT), G13A (GGC>GCC), G13C (GGC>TGC), G13D (GGC>GAC), G13R (GGC>CGC), G13S (GGC>AGC), G13V (GGC>GTC), Q61E (CAA>GAA), Q61H (CAA>CAT), Q61H (CAA>CAC), Q61K (CAA>AAA), Q61L (CAA>CTA), Q61P (CAA>CCA), and Q61R (CAA>CGA).
DNA extracts from FFPET specimens using the same extraction method as described for the cobas test were sequenced by 2X bidirectional Sanger sequencing by a Clinical Laboratory Improvement Amendments (CLIA)-certified laboratory (SeqWright, Houston, TX) using an internally validated protocol.
Quantitative massively parallel pyrosequencing
Specimens which gave discordant results with the cobas test versus Sanger sequencing were retested using a quantitative massively parallel pyrosequencing method (454 GS Titanium, 454 Life Sciences, Branford, CT) by a CLIA-certified laboratory (SeqWright, Houston, TX) using a validated protocol . Additionally, a randomly selected subset of specimens with concordant cobas test and Sanger sequencing results were tested by 454 sequencing to confirm the accuracy of mutation status. The validated LOD for the 454 KRAS protocol was 1%.
Limit of detection
Sixteen different point mutations in codons 12, 13 and 61 of the KRAS gene were examined for the LOD of the cobas test. Although the assay covers 19 mutations, three mutations with low frequency were not tested in our LOD study.
For the 5 predominant mutations in codons 12, 13 and 61 (one in codon 12, one in codon 13 and 3 in codon 61), FFPET specimens and cell lines were utilized. Testing was performed using three different lots of the cobas test and one lot of the cobas® DNA Sample Preparation Kit. DNA extracted from each mutant specimen was blended with genomic DNA extracted from a wild-type specimen to prepare various levels of mutant sequences (approximately 5% for FFPET specimens and 2.5%, 5%, and 10% for cell lines). 454 sequencing was used to assess and confirm the percentage of mutant sequences in DNA blends. The DNA blends were then serially diluted to input amounts ranging from 50.0 ng to 0.8 ng per PCR reaction. For FFPET samples, 5 different amounts of DNA were input into the reactions (50.0, 12.5, 3.1, 0.8 and 0.0 ng/PCR reaction) and were tested using 8 replicates and 3 reagent lots for each input level, yielding 24 results per DNA input level for each mutation tested. For cell line samples, 24 replicates for each of 5 DNA input levels (50.0, 12.5, 3.1, 0.8 and 0.0 ng per PCR reaction) for each of 3 reagent lots were tested, yielding a total of 72 results per DNA input level for each mutation and for each of 3 different percentages of mutated DNA sequences (2.5%, 5% and 10% mutant sequences). Sensitivity was determined as the lowest level of DNA input level that gave a correct mutation call rate with its one-sided upper limit of 95% confidence interval (CI) ≥ 95%.
In a separate analysis, the LOD was evaluated for 11 additional mutations in KRAS codons 12, 13, and 61. Due to the lack of naturally occurring specimens with 5% mutant sequences, FFPET DNA and plasmid blends at approximately 5% mutation (as assessed by 454 sequencing) were prepared including: seven FFPET specimens representing four KRAS codon 12 mutations, two FFPET specimens and three plasmids representing four KRAS codon 13 mutations, and one FFPET specimen and two plasmids representing three KRAS codon 61 mutations.
Before cobas testing, the mutation status and mutation percentage of the specimens were verified by 2X bi-directional Sanger sequencing and 454 sequencing, respectively.
Methods correlation with Sanger sequencing
Each of the two kit lots of the cobas test was used to test 194 NSCLC FFPET specimens for mutations in codons 12, 13, and 61. Positive, negative, and overall percent agreements (PPA, NPA and OPA) between the cobas test and 2X bi-directional Sanger test results were determined. For specimens that yielded discordant results for cobas testing and Sanger sequencing, discrepant analysis was performed by testing specimens with 454 sequencing for resolution. In addition, 54 randomly selected specimens with concordant Sanger and cobas test results were tested by 454 sequencing in order to confirm the accuracy of 454 sequencing as the resolution method.
The repeatability of the cobas test was evaluated across different reagent lots, days, operators and instruments with eight FFPET NSCLC specimens: two KRAS wild-type, two codon 12 mutants, two codon 13 mutants, and two codon 61 mutants. Three of the six KRAS mutants tested—one for each of codons 12, 13, and 61—had approximately 15% mutant sequences, which is 3 times the LOD of the cobas test, while the other three KRAS samples had greater than 15% mutant sequences. The materials used in the repeatability testing were: one cobas® DNA Sample Preparation Kit lot, two cobas test lots, two operators, and four cobas® 4800 System v.2.0.
Each operator performed two runs per day representing one run per reagent lot. The same test was conducted on each of 4 days using 4 instruments for a total of 8 runs with two replicates for each of the eight specimens. Therefore, the 16 results per run and 16 testing situations yielded a total of 256 analyses.
Interfering endogenous substances
The effects of endogenous substances, hemoglobin and triglycerides, on the performance of the cobas test were investigated in ten NSCLC FFPET specimens. Specimens were selected for mutation status based on Sanger and/or 454 sequencing. Three specimens were KRAS codon 12 mutation-positive, one was KRAS codon 13 mutation-positive, one was KRAS codon 61 mutation-positive, and five were wild-type. Four 5 μm sections were extracted for each specimen, one each in the following: Tissue Lysis Buffer + proteinase K (TLB/PK) (control), TLB/ PK + saline (saline control), TLB/PK + 2 g/L hemoglobin in saline (Clinical and Laboratory Standards Institute (CLSI)-recommended test concentrations ), and TLB/PK + 37 mM triglycerides in saline (CLSI-recommended test concentrations ).
Hemoglobin and triglycerides were added during the lysis step of DNA extraction and were not subjected to deparaffinization. All DNA extracts were tested for interference of mutation detection using the cobas test.
Interference by common lung-related microorganisms
The effects of lung-related microorganisms on the performance of the cobas test were investigated in 10 FFPET NSCLC specimens. Three specimens were KRAS codon 12 mutation-positive, one was KRAS codon 13 mutation-positive, one was KRAS codon 61 mutationpositive, and five were wild-type. Approximately 2.5 x 105 colony forming units of Streptococcus pneumoniae and Haemophilus influenzae were added during the lysis step of DNA extraction. Phosphate-buffered saline control was also tested. One 5 μm section of each specimen was extracted and all DNA extracts were tested for interference of mutation detection using the cobas test.
The intent of this study was to determine whether the configured cobas test could detect several additional rare mutations. In addition to the 16 single-base change mutations in codons 12, 13 and 61 tested for LOD studies, the inclusivity of the cobas test with a panel of rare KRAS mutations  in flanking sequences of codons 12, 13 and 61 (i.e., codons 11, 14, 15, 59, 60, 62, 63) and double or complex mutations in codons 12, 13 and 61 was also examined using plasmids and FFPET specimens that were available . Twenty-three plasmids, each containing one of 23 possible mutations, were blended separately with wild-type genomic DNA from K562 cells to obtain 5% mutant/95% wild type sequences. Six different NSCLC FFPET specimens were also tested, including five with the G12F mutation (mutant allele frequencies ranging from 24% to 60%) and one with the A59T mutation (with its mutant allele frequency at 22%). All 29 samples were tested in 4 replicates by the cobas test and mutation detectability was calculated as the percentage of correct results. For the samples that did not give 100% detectability at 5% mutant sequences, further testing was carried out using DNA blends containing 10% and 15% mutant sequences in replicates of 4.
Limit of detection
For the LOD analysis of the 5 predominant mutations in codons 12, 13, and 61, ≥ 95% correct mutation call rates were obtained across all specimen types with approximately ≥ 5% mutant sequences (as assessed by 454) at a DNA input of 0.8-12.5 ng per PCR reaction (Table 1). For all mutation levels examined that ranged from 2.4% to 12.1%, ≥ 95% correct mutation call rates were obtained at a DNA input of 50.0 ng per PCR reaction, which is the prescribed amount of DNA input for the cobas test, or less (Table 1). All of the 11 additional KRAS mutations in codons 12, 13, and 61 examined at approximately 5% mutant sequences were detected at rates whose one-sided upper limit of 95% CI were ≥ 95% correct mutation call rates at a DNA input of 50.0 ng per PCR reaction or less (Table 2).
|KRAS Mutation||Sample ID||Mutation
|Lowest DNA Input with Correct Call a (ng/PCR)|
|G12C, GGT>TGT||FFPET #1, DNA blend||6.2||12.5|
|FFPET #2, DNA blend||6.0||3.1|
|FFPET #3, DNA blend||5.6||3.1|
|Cell line (NCI-H23),
DNA blend #1
|Cell line (NCI-H23),
DNA blend #2
|Cell line (NCI-H23),
DNA blend #3
|G13C, GGC>TGC||FFPET #4, DNA blend||6.2||3.1|
|FFPET #5, DNA blend||4.4||3.1|
|FFPET #6, DNA blend||4.4||3.1|
|Cell line (NCI-H1734),
DNA blend #4
|Cell line (NCI-H1734),
DNA blend #5
|Cell line (NCI-H1734),
DNA blend #6
|Q61H, CAA>CAT||FFPET #7, DNA blend||4.5||3.1|
|Cell line (NCI-H460),
DNA blend #7
|Cell line (NCI-H460),
DNA blend #8
|Cell line (NCI-H460),
DNA blend #9
|Q61H, CAA>CAC||FFPET #8, DNA blend||6.4||3.1|
|Q61L, CAA>CTA||FFPET #9, DNA blend||6.5||3.1|
aLowest amount of input DNA (ng/PCR) tested that yielded a correct mutation call rate of ≥ 95% (upper one-sided 95% CI). Each test used 72 replicates for cell line DNA blends and 24 replicates for FFPET DNA blends.
Table 1: Summary of LOD results for the predominant KRAS mutations in codons 12, 13 and 61.
|KRAS Mutation||Sample ID||Mutation (%)||Lowest DNA Input with Correct Call
|G12A, GGT > GCT||FFPET #10, DNA blend||4.5||12.5|
|FFPET #11, DNA blend||5.1||12.5|
|G12D, GGT > GAT||FFPET #12, DNA blend||5.0||12.5|
|FFPET #13, DNA blend||5.2||12.5|
|G12S, GGT >AGT||FFPET #14, DNA blend||5.6||12.5|
|G12V, GGT > GTT||FFPET #15, DNA blend||2.9||12.5|
|FFPET #16, DNA blend||5.7||12.5|
|G13A, GGC > GCC||Plasmid, DNA blend||3.7||50.0b|
|G13D, GGC > GAC||FFPET #17, DNA blend||4.8||3.1|
|FFPET #18, DNA blend||6.8||3.1|
|G13S, GGC >AGC||Plasmid, DNA blend||1.6||50.0b|
|G13V, GGC > GTC||Plasmid, DNA blend||2.6||50.0b|
|Q61E, CAA >GAA||Plasmid, DNA blend||5.7||50.0b|
|Q61P, CAA > CCA||Plasmid, DNA blend||3.2||50.0b|
|Q61R, CAA > CGA||FFPET #19, DNA blend||4.8||50|
aLowest amount of input DNA (ng/PCR) tested with a correct mutation call rate of ≥95% (upper one-sided 95% CI) per 24 replicates. bPlasmid DNA blends were tested only at 50.0 ng per PCR reaction.
Table 2: Summary of LOD results for the additional 11 KRAS mutations in codons 12, 13 and 61.
One of the 194 FFPET NSCLC specimens tested gave an invalid result by the cobas test kit Lot 2 and another specimen was positive for mutations in both codon 12 and codon 61 by Sanger sequencing. As a consequence, there were 195 valid results for cobas test kit Lot 1 and 194 valid results for cobas test kit Lot 2 for the comparison between the cobas test and Sanger sequencing (Table 3). Concordance rates between the cobas test and Sanger sequencing for Lot 1 and Lot 2 were 93.8% PPA for both lots, 92.9% NPA for Lot 1 and 90.7% NPA for Lot 2, and 93.3% OPA for Lot 1 and 92.3% OPA for Lot 2 (Table 3A and 3B). Sixteen specimens with discordant results by one or both lots were subjected to discrepant analysis by 454 sequencing and their results are shown in Table 4. For discrepant analysis, each specimen that produced different results with cobas testing versus Sanger sequencing was subjected to 454 sequencing directed at the area of the DNA sequence in question.
Positive agreement, 100.0% (95% CI, 96.3%–100.0%); Negative agreement, 98.9% (95% CI, 94.3%–99.8%); Overall agreement, 99.5% (95% CI, 97.1%–99.9%)
|(A)||2X Bi-Directional Sanger Sequencing|
|cobas KRAS Kit Lot 1||MD Codon 12/13||90||0||6||96|
|MD Codon 61||0||1||1||2|
|Positive agreement, 93.8% (95% CI, 87.2%–97.1%); Negative agreement, 92.9% (95% CI, 86.0%–96.5%); Overall agreement, 93.3% (95% CI, 88.9%–96.1%)|
|2X Bi-Directional Sanger Sequencing|
|cobas KRAS Kit Lot 2||MD Codon 12/13||90||0||8||98|
|MD Codon 61||0||1||1||2|
|Positive agreement, 93.8% (95% CI, 87.2%–97.1%); Negative agreement, 90.7% (95% CI, 83.0%–95.0%); Overall agreement, 92.3% (95% CI, 87.6%–95.3%)|
|2X Bi-Directional Sanger Sequencing and
|cobasKRAS Kit Lot 1||MD Codon 12/13||96||0||0||96|
|MD Codon 61||0||1||1*||2|
|Positive agreement, 97.0% (95% CI, 91.5%–99.0%); Negative agreement, 98.9% (95% CI, 94.3%–99.8%); Overall agreement, 97.9% (95% CI, 94.8%–99.2%); *All four samples with discordant results were at <5% mutation frequency.|
|2X Bi-Directional Sanger Sequencing and
|cobas KRAS Kit Lot 2||MD Codon 12/13||98||0||0||98|
|MD Codon 61||0||1||1*||2|
*The sample with a discordant result was at <5% mutation frequency.MD mutation detected, NMD no mutation detected—wild-type for codons 12/13 and 61aN=194: One specimen is a double mutant (at codons 12 and 61) by Sanger sequencing. One of the 194 specimens produced an invalid result by cobas® Kit Lot 1.
Table 3: Method correlation between the cobas®KRAS Mutation Test and Sanger sequencing (Table 3A and 3B) and after discordant resolution with 454 sequencing (Tables 3C and 3D).
|Sample||Sanger Sequencing Results||cobasKRAS
|454 Sequencing Results (% Mutation)||Concordance After
|Lot 1||Lot 2||cobas
|1||codon 12||NMD||NMD||NMD (<1%)||Yes||Yes|
|2||codon 12||NMD||NMD||NMD (<1%)||Yes||Yes|
|3||codon 12||NMD||NMD||NMD (<1%)||Yes||Yes|
|4||codon 12||NMD||NMD||NMD (<1%)||Yes||Yes|
|5||codon 12||NMD||NMD||NMD (<1%)||Yes||Yes|
|6||codons 12 & 61||codon 12/13||codon 12/13||codon 12 (23.7%)||Yes||Yes|
|7||NMD||codon 12/13||codon 12/13||codon 12 (59.2%)||Yes||Yes|
|8||NMD||codon 12/13||codon 12/13||codon 12 (13.3%)||Yes||Yes|
|9||NMD||codon 12/13||codon 12/13||codon 12 (4.9%)||Yes||Yes|
|10||NMD||codon 12/13||codon 12/13||codon 12 (9.7%)||Yes||Yes|
|11||NMD||codon 12/13||codon 12/13||codon 12 (39.3%)||Yes||Yes|
|12||NMD||codon 12/13||codon 12/13||codon 12 (3.0%)||Yes||Yes|
|13||NMD||NMD||codon 12/13||codon 12 (1.2%)||No||Yes|
|14||NMD||NMD||codon 12/13||codon 12 (3.7%)||No||Yes|
|15||NMD||codon 61||codon 61||codon 59 (2.8%)||No||No|
|16||NMD||NMD||invalid||codon 13 (1.8%)||No||N/A|
NMD, no mutation detected
Table 4: Resolution of discordant results by 454 sequencing for method correlation
Based on the results from 454 sequencing, 12 (75%) of the 16 discordant specimens were correctly assessed by cobas test Lot 1, and 13 (87%) of the 15 discordant specimens were correctly assessed by cobas test Lot 2. The resulting PPA values for codon 12/13 were 97.0% for Lot 1 and 100.0% for Lot 2, the NPA values were 98.9% for both lots, and the OPA values were 97.9% for Lot 1 and 99.5% for Lot 2 (Tables 3C and 3D). It is noteworthy that each of the four specimens that showed discordant results after resolution by 454 sequencing had a mutation level below 5%, the LOD for the cobas test (Table 4). In addition to the 16 specimens with discordant results, 54 specimens with concordant cobas test and Sanger sequencing results were subjected to 454 sequencing. These included 19 specimens with mutations detected in codon 12/13 and 35 specimens with no mutation detected. The 454 sequencing results for all 54 specimens were concordant with the Sanger sequencing and cobas test results.
Repeatability was assessed by testing tissue sections from 8 different NSCLC FFPET samples, with each sample tested twice for a given run. All 16 repeatability runs were valid, yielding an overall accuracy of 100%, representing 256 valid and correct results from 256 cobas test analyses across all specimens, operators, reagent kit lots, and instruments (Table 5 and 6).
|Test Day||Operator||Run||Reagent Kit Lot||Instrument ID||Results||Accuracy
Table 5: Repeatability of the cobas®KRAS Mutation Test by day, operator, reagent kit lot and instrument.
|Amino Acid Substitution||KRAS DNA Mutation||Detectability
|5% Mutant KRAS DNA on Plasmid||p.A11P||GCT >CCT||100|
|p.A59E||GCA > GAA||100|
|p.A59G||GCA > GGA||100|
|p.G60A||GGT > GCT||100|
|p.E62D||GAG > GAT||100|
|p.G13D||GGC > GAC||100|
|DNA Insertions and Deletions||KRAS DNA Mutation||Detectability
|c.30_31insGGA||GGAGCT > GGAGGAGCT||0|
|c.36_37insGGT||GGTGGC > GGTGGTGGC||100|
|c.187_189delGAG||GGAGT > G^^^T||100|
|NSCLC FFPETd||Amino Acid Substitution||KRAS DNA Mutation||Detectability
aThese plasmids were tested in both the plasmid mixture and in NSCLC FFPET specimens.bThe p.G12F plasmid blended at 5% with 95% wild-type genomic DNA was not detectable. However, the same plasmid blended at 10% or 15% was detected at 100%.cThe p.G15S plasmid blended at 5% with 95% wild-type genomic DNA was not detectable. However, the same plasmid blended at 10% or 15% was detected at 100%.dFive different NSCLC FFPET specimens with the G12F mutation were tested and all 5 showed 100% detectability. One NSCLC FFPET specimen with the A59T mutation was tested and detected 100%.
Table 6: Inclusivity of the cobas®KRAS Mutation Test using 5% mutant KRAS DNA sequences blended with wild-type genomic DNA or FFPET NSCLC samples. The percent of KRAS mutant DNA in the FFPET samples is unknown.
In an effort to simulate analysis of real-world samples, substances that could be present in NSCLC FFPET specimens were added prior to testing. No interference was observed for any of the 10 FFPET specimens tested for potential endogenous interfering substances at CLSI-recommended test concentrations of 2 g/L hemoglobin and 37 mM triglycerides .
None of the lung-related micro-organisms tested interfered with the cobas test. All five of the KRAS wild-type specimens spiked with micro-organisms gave valid results of mutation not detected, while all five KRAS-mutant specimens spiked with micro-organisms gave valid results of mutation detected for the correct codon. For both endogenous interfering substances and micro-organisms, all study controls gave valid and accurate mutation results (data not shown).
Inclusivity testing was performed to determine the ability of the cobas test to detect rare mutations in the KRAS gene. KRAS mutations in codons 11, 14, 15, 59, 60, 62, and 63 were examined, as well as double or complex mutations in codons 12, 13 and 61. The cobas test detected 21 of the 23 mutations tested (Table 6). All plasmid samples were detected 100% of the time at the level of 5% mutant sequences, with the exceptions of p.G12F, p.G12L, p.G15S and c.30_31insGGA. The p.G12F and p.G15S were detected 100% of the time at the level of 10% and 15% mutant sequences. The remaining two mutations, p.G12L and c.30_31insGGA, were not detectable even at 15% mutant sequences. All of the six (6) FFPET specimens (five p.G12F and one p.A59T) were 100% detected.
In keeping with the rapid development of targeted therapies, there is a continuing need for standardized testing of involved genes to rapidly and consistently determine mutation status. Accurate assignment of therapy to oncology patients, including the withholding of treatment from those who are unlikely to benefit, is essential for optimizing treatment. In NSCLC, KRAS plays a pivotal role as a driver oncogene; moreover, KRAS mutations predict a lack of benefit in responding to EGFR inhibitors and may reflect eligibility for treatment with inhibitors of MEK or direct inhibitors of mutant KRAS as they become available [5,8]. Thus reliable KRAS testing is imperative to ensure accurate treatment assignment for NSCLC patients with these new agents.
The current study was undertaken to evaluate the capabilities of the cobas test in detecting mutations in NSCLC FFPET specimens. FFPET specimens were chosen to represent the intention to diagnose population. Sanger sequencing was used as the standard for assessing the cobas test performance. In addition, the cobas test was rigorously analyzed for mutation detection capabilities, reproducibility, and robustness during various steps of the testing workflow likely to be encountered in the clinic.
To evaluate the LOD, artificial blends of DNA extracts from cell lines and FFPET samples were examined. The cobas test correctly called at least 95% of mutations in clinical samples containing as little as 2.4% mutated KRAS and containing at least 50 ng DNA, the amount stipulated by the manufacturer for accurate test results. Furthermore, the vast majority of samples were correctly called with less than 5 ng input DNA, simulating the ability to accurately detect mutations in highly damaged specimens. Correct calls were also obtained for 11 additional KRAS mutations when samples were tested with 50 ng DNA and approximately 5% mutant sequences. These conditions established by the cobas test enable the laboratory to perform molecular testing while preserving precious clinical material for additional biomarker testing.
Testing two lots of the cobas test against Sanger sequencing for detecting mutations in KRAS codons 12 and 13 yielded OPA values of 97.9% for lot 1 and 99.5% for lot 2 after resolution of discordant specimens via 454 sequencing. Repeatability was assessed by comparing results from cobas test kits from different lots, 2 different operators, and 4 different cobas® 4800 System instruments. Each run analyzed 8 different FFPET NSCLC tumor specimens that included 2 KRAS wild-type, and 2 each with mutations in either codon 12, 13, or 61. Each run correctly identified either wild-type or mutated KRAS with 100% accuracy. The results indicate that the cobas test can produce consistent and accurate results reliably, which is essential for a clinical diagnostic assay. Similarly, the cobas test yielded accurate results when lung FFPET samples were deliberately contaminated with substances of clinical interest. In the presence of triglycerides or hemoglobin at the CLSI-recommended levels, as well as in the presence of Streptococcus pneumoniae or Haemophilus influenzae, the cobas test correctly identified KRAS mutations. An earlier study demonstrated the cobas test specificity by showing a complete lack of cross-reactivity against HRAS, NRAS, and the pseudogene KRASP1, three sequences that are highly homologous to KRAS .
Previously, the cobas test identified 19 single-base change mutations in KRAS codons 12, 13, and 61 in colorectal cancer . In this study of NSCLC, 16 of the 19 KRAS mutations in KRAS codons 12, 13, and 61 were detected at approximately 5% or below in mutant sequences. Furthermore, 23 additional mutations in codons 11, 14, 15, 60, 62, and 63 as well as double or complex mutations in codons 12, 13, and 61 were detected. In contrast, the therascreen KRAS test (Qiagen, Manchester, UK) identifies only 7 somatic mutations: 6 mutations in codon 12 and 1 mutation in codon 13 and none in codon 61. The additional 12 singlebase change mutations in codons 12, 13, and 61 that are detected by the cobas test and not by the therascreen KRAS test represent more than 3% of NSCLC patients with KRAS mutations . Furthermore, a recent report detailed the emergence and high frequency of codon 61 KRAS mutations, showing mutations in 16 (32%) of 50 plasma samples of colorectal cancer patients who became resistant to anti-EGFR antibodies; this finding underscores the importance of detecting codon 61 mutations in KRAS exon 3 .
In summary, the performance analysis presented in the current study demonstrates that the cobas test is a robust, sensitive, and reproducible method for detecting KRAS mutations in FFPET tumor samples from NSCLC patients.
The authors gratefully acknowledge the medical writing services of Maggie Merchant, PhD.
Make the best use of Scientific Research and information from our 700 + peer reviewed, Open Access Journals