Portal de Periódicos da CAPES

Analytical Validation of BRAF Mutation Testing from Circulating Free DNA Using the Amplification Refractory Mutation Testing System

2014; Elsevier BV; Volume: 16; Issue: 3 Linguagem: Inglês

10.1016/j.jmoldx.2013.12.004

1943-7811

Kyaw Lwin Aung, Emma Donald, Gillian Ellison, Sarah R. Bujac, Lynn M. Fletcher, Mireille Cantarini, Ged Brady, Maria Orr, Glen Clack, Malcolm Ranson, Caroline Dive, Andrew Hughes,

Renal cell carcinoma treatment

BRAF mutation testing from circulating free DNA (cfDNA) using the amplification refractory mutation testing system (ARMS) holds potential as a surrogate for tumor mutation testing. Robust assay validation is needed to establish the optimal clinical matrix for measurement and cfDNA-specific mutation calling criteria. Plasma- and serum-derived cfDNA samples from 221 advanced melanoma patients were analyzed for BRAF c.1799T>A (p.V600E) mutation using ARMS in two stages in a blinded fashion. cfDNA-specific mutation calling criteria were defined in stage 1 and validated in stage 2. cfDNA concentrations in serum and plasma, and the sensitivities and specificities of BRAF mutation detection in these two clinical matrices were compared. Sensitivity of BRAF c.1799T>A (p.V600E) mutation detection in cfDNA was increased by using mutation calling criteria optimized for cfDNA (these criteria were adjusted from those used for archival tumor biopsies) without compromising specificity. Sensitivity of BRAF mutation detection in serum was 44% (95% CI, 35% to 53%) and in plasma 52% (95% CI, 43% to 61%). Specificity was 96% (95% CI, 90% to 99%) in both matrices. Serum contains significantly higher total cfDNA than plasma, whereas the proportion of tumor-derived mutant DNA was significantly higher in plasma. Using mutation calling criteria optimized for cfDNA improves sensitivity of BRAF c.1799T>A (p.V600E) mutation detection. The proportion of tumor-derived cfDNA in plasma was significantly higher than in serum. BRAF mutation testing from circulating free DNA (cfDNA) using the amplification refractory mutation testing system (ARMS) holds potential as a surrogate for tumor mutation testing. Robust assay validation is needed to establish the optimal clinical matrix for measurement and cfDNA-specific mutation calling criteria. Plasma- and serum-derived cfDNA samples from 221 advanced melanoma patients were analyzed for BRAF c.1799T>A (p.V600E) mutation using ARMS in two stages in a blinded fashion. cfDNA-specific mutation calling criteria were defined in stage 1 and validated in stage 2. cfDNA concentrations in serum and plasma, and the sensitivities and specificities of BRAF mutation detection in these two clinical matrices were compared. Sensitivity of BRAF c.1799T>A (p.V600E) mutation detection in cfDNA was increased by using mutation calling criteria optimized for cfDNA (these criteria were adjusted from those used for archival tumor biopsies) without compromising specificity. Sensitivity of BRAF mutation detection in serum was 44% (95% CI, 35% to 53%) and in plasma 52% (95% CI, 43% to 61%). Specificity was 96% (95% CI, 90% to 99%) in both matrices. Serum contains significantly higher total cfDNA than plasma, whereas the proportion of tumor-derived mutant DNA was significantly higher in plasma. Using mutation calling criteria optimized for cfDNA improves sensitivity of BRAF c.1799T>A (p.V600E) mutation detection. The proportion of tumor-derived cfDNA in plasma was significantly higher than in serum. Circulating free DNA (cfDNA) levels are raised in the blood of patients with advanced cancers compared to that of healthy controls,1Fleischhacker M. Schmidt B. Circulating nucleic acids (CNAs) and cancer—a survey.Biochim Biophys Acta. 2007; 1775: 181-232Crossref PubMed Scopus (716) Google Scholar and tumor-specific somatic mutations can be examined in cfDNA.2Schwarzenbach H. Hoon D.S. Pantel K. Cell-free nucleic acids as biomarkers in cancer patients.Nat Rev Cancer. 2011; 11: 426-437Crossref PubMed Scopus (2020) Google Scholar BRAF is one of the most commonly mutated oncogenes in cancers, and its mutation is present in approximately 50% of cutaneous melanoma.3Davies H. Bignell G.R. Cox C. Stephens P. Edkins S. Clegg S. et al.Mutations of the BRAF gene in human cancer.Nature. 2002; 417: 949-954Crossref PubMed Scopus (8271) Google Scholar The c.1799T>A transversion resulting in p.V600E mutation, but also present in p.V600K and p.V600D changes in combination with additional flanking nucleotide changes, accounts for up to 90% of BRAF gene mutations.4Brose M.S. Volpe P. Feldman M. Kumar M. Rishi I. Gerrero R. Einhorn E. Herlyn M. Minna J. Nicholson A. Roth J.A. Albelda S.M. Davies H. Cox C. Brignell G. Stephens P. Futreal P.A. Wooster R. Stratton M.R. Weber B.L. BRAF and RAS mutations in human lung cancer and melanoma.Cancer Res. 2002; 62: 6997-7000PubMed Google Scholar Mutated BRAF protein has now been proven as a valid drug target because a BRAF inhibitor, vemurafenib, was shown to extend survival in patients with BRAF mutant advanced melanoma.5Chapman P.B. Hauschild A. Robert C. Haanen J.B. Ascierto P. Larkin J. Dummer R. Garbe C. Testori A. Maio M. Hogg D. Lorigan P. Lebbe C. Jouary T. Schadendorf D. Ribas A. O'Day S.J. Sosman J.A. Kirkwood J.M. Eggermont A.M. Dreno B. Nolop K. Li J. Nelson B. Hou J. Lee R.J. Flaherty K.T. McArthur G.A. BRIM-3 Study GroupImproved survival with vemurafenib in melanoma with BRAF V600E mutation.N Engl J Med. 2011; 364: 2507-2516Crossref PubMed Scopus (6115) Google Scholar On the other hand, in colorectal cancer, patients with BRAF mutant tumors do not derive clinical benefit from treatment with the epidermal growth factor receptor (EGFR) monoclonal antibodies cetuximab and panitumumab.6Bardelli A. Siena S. Molecular mechanisms of resistance to cetuximab and panitumumab in colorectal cancer.J Clin Oncol. 2010; 28: 1254-1261Crossref PubMed Scopus (561) Google Scholar Considering these findings, the importance of establishing BRAF mutation status of patients' tumors before they are treated with agents targeting components of the mitogen-activated protein kinase (MAPK) pathway cannot be overemphasized because this could have potential impact on their treatment-related clinical outcomes. Conventionally, somatic mutations are detected using archival formalin-fixed, paraffin-embedded (FFPE) tumor tissues obtained at diagnosis and/or from other biopsies or during surgery. However, using archival tumor material for mutation testing has inherent problems. The first challenge is the lack of tumor material in patients whose tumors are difficult to biopsy; eg, diagnosis in a significant proportion of patients with lung cancer is based purely on sputum cytology, and as a result, there is insufficient tumor material for comprehensive mutation testing.7Sequist L.V. Engelman J.A. Lynch T.J. Toward noninvasive genomic screening of lung cancer patients.J Clin Oncol. 2009; 27: 2589-2591Crossref PubMed Scopus (16) Google Scholar A second challenge is that archival tumor tissue from diagnostic biopsies often contains small amounts of tumor material mixed with normal stromal tissue, and DNA is usually degraded by formalin fixation.8Plesec T.P. Hunt J.L. KRAS mutation testing in colorectal cancer.Adv Anat Pathol. 2009; 16: 196-203Crossref PubMed Scopus (88) Google Scholar Furthermore, because a biopsy is usually taken from one small part of a tumor, potentially mutant clones could also be missed, and it is debatable whether tumor material from a core or needle biopsy adequately represents tumor heterogeneity.9Fleischhacker M. Schmidt B. Cell-free DNA resuscitated for tumor testing.Nat Med. 2008; 14: 914-915Crossref PubMed Scopus (27) Google Scholar, 10Gerlinger M. Rowan A.J. Horswell S. Larkin J. Endesfelder D. Gronroos E. Martinez P. Matthews N. Stewart A. Tarpey P. Varela I. Phillimore B. Begum S. McDonald N.Q. Butler A. Jones D. Raine K. Latimer C. Santos C.R. Nohadani M. Eklund A.C. Spencer-Dene B. Clark G. Pickering L. Stamp G. Gore M. Szallasi Z. Downward J. Futreal P.A. Swanton C. Intratumor heterogeneity and branched evolution revealed by multiregion sequencing.New Engl J Med. 2012; 366: 883-892Crossref PubMed Scopus (5730) Google Scholar By contrast, analysis of cfDNA might yield better coverage of subclones present in a tumor.9Fleischhacker M. Schmidt B. Cell-free DNA resuscitated for tumor testing.Nat Med. 2008; 14: 914-915Crossref PubMed Scopus (27) Google Scholar Lastly, in a tertiary cancer center, turnaround time for getting the tumor mutation result can take several weeks because tracing archival FFPE tumor material from referral hospitals can be logistically difficult. Testing the mutation from cfDNA, on the other hand, could significantly shorten the turnaround time. Studies have shown that BRAF c.1799T>A mutations can be assessed in cfDNA using real-time quantitative PCR (qPCR)-based assays,11Board R.E. Ellison G. Orr M.C. Kemsley K.R. McWalter G. Blockley L.Y. Dearden S.P. Morris C. Ranson M. Cantarini M.V. Dive C. Hughes A. Detection of BRAF mutations in the tumour and serum of patients enrolled in the AZD6244 (ARRY-142886) advanced melanoma phase II study.Br J Cancer. 2009; 101: 1724-1730Crossref PubMed Scopus (85) Google Scholar, 12Daniotti M. Vallacchi V. Rivoltini L. Patuzzo R. Santinami M. Arienti F. Cutolo G. Pierotti M.A. Parmiani G. Rodolfo M. Detection of mutated BRAFV600E variant in circulating DNA of stage III-IV melanoma patients.Int J Cancer. 2007; 120: 2439-2444Crossref PubMed Scopus (58) Google Scholar, 13Yancovitz M. Yoon J. Mikhail M. Gai W. Shapiro R.L. Berman R.S. Pavlick A.C. Chapman P.B. Osman I. Polsky D. Detection of mutant BRAF alleles in the plasma of patients with metastatic melanoma.J Mol Diagn. 2007; 9: 178-183Abstract Full Text Full Text PDF PubMed Scopus (30) Google Scholar highlighting the potential of using cfDNA mutation assays as a surrogate for tumor testing. However, before this approach could be adopted into oncology practice, a number of preanalytical and analytical questions need to be addressed. For instance, there are no robust data to support which clinical matrix, serum or plasma, is better suited for mutation testing from cfDNA. Although preliminary evidence suggests that plasma would be superior because serum contains higher background levels of wild-type DNA emanating from white blood cell lysis during the clotting process intrinsic to the preparation of serum,14Steinman C.R. Free DNA in serum and plasma from normal adults.J Clin Invest. 1975; 56: 512-515Crossref PubMed Google Scholar, 15Board R.E. Williams V.S. Knight L. Shaw J. Greystoke A. Ranson M. Dive C. Blackhall F.H. Hughes A. Isolation and extraction of circulating tumor DNA from patients with small cell lung cancer.Ann N Y Acad Sci. 2008; 1137: 98-107Crossref PubMed Scopus (89) Google Scholar both plasma and serum are being used widely for cfDNA-based mutation testing, and to our knowledge, there has been no systematic and definitive direct comparison of BRAF mutation pick-up rate between serum and plasma in an adequately powered study. Also pertinent to the overall goal of routine implementation of mutation testing from cfDNA, qPCR approaches validated for mutation testing from FFPE tumor tissues are being used for cfDNA mutation detection; they use identical mutation calling criteria even though these assays could have different performance characteristics in cfDNA. DNA cross-linking via formalin fixation of tumor is one of the culprits of mispriming in PCR reactions causing nonspecific amplification of wild-type DNA,16Srinivasan M. Sedmak D. Jewell S. Effect of fixatives and tissue processing on the content and integrity of nucleic acids.Am J Pathol. 2002; 161: 1961-1971Abstract Full Text Full Text PDF PubMed Scopus (943) Google Scholar and because such stringent mutation calling criteria are usually used for FFPE tumor DNA to prevent introduction of false-positive results. However, cfDNA is not formalin fixed, and as a result, less nonspecific amplification should be observed, potentially allowing more sensitive mutation calling criteria to be applied without affecting the assay specificity. In this study, two research hypotheses relating to the preanalytical and analytical phases of BRAF c.1799T>A mutation testing from cfDNA were examined: more BRAF c.1799T>A mutations will be detected in plasma compared to serum, and increased sensitivity of BRAF mutation detection from cfDNA using ARMS will be achieved by using mutation calling criteria specific to cfDNA, without compromising specificity. Two hundred and eight plasma samples and 208 serum samples were available for cfDNA quantification and BRAF c.1799T>A mutation analysis from 221 patients who were screened for participation in a randomized phase II study (Clinical Trials ID number NCT00936221). This study evaluated the efficacy of a specific MEK 1/2 inhibitor, selumetinib (AZD6244, ARRY-142886), in combination with dacarbazine versus dacarbazine alone in patients with BRAF mutant advanced melanoma in the first-line setting. The study was conducted according to Good Clinical Practice and the Declaration of Helsinki. Samples were analyzed in two stages in a blinded fashion, and analysts were also blinded to tumor mutation status. In the first stage, 50 serum and plasma samples from patients with BRAF mutant tumors and 50 serum and plasma samples from those with BRAF wild-type tumors were analyzed for BRAF c.1799T>A mutation using an ARMS allele-specific PCR. Of the 100 serum and plasma samples analyzed in the first stage, 90 of them were matched. The results from this first analysis stage were used as a training data set to derive a mutation calling criteria specific to cfDNA. The criteria derived were then validated using the remaining matched 108 plasma and serum samples in the second stage. For plasma, 4 mL of blood was drawn into a Becton-Dickinson Vacutainer Collection Tube (Becton, Dickinson and Company, Franklin Lakes, NJ) containing EDTA and centrifuged at 2000 × g for 10 minutes at 4°C within 30 minutes of blood collection. The supernatant was transferred to a 15-mL falcon tube and centrifuged at 2000 × g for 10 minutes at 4°C. The resultant plasma supernatant was separated and stored immediately at −80°C. For serum, 4 mL of blood was drawn into a Becton-Dickinson Vacutainer Serum Collection Tube. After allowing the blood to clot for 30 minutes at room temperature, it was centrifuged at 2000 × g for 10 minutes, and the resultant serum supernatant was subsequently separated and immediately stored at −80°C. Median sample storage time (from blood collection to cfDNA extraction) was 467 days (range, 107 to 690 days) for plasma and 478 days (range, 97 to 704 days) for serum. cfDNA was extracted from 2 mL of plasma and serum using QIAamp Circulating Nucleic Acids Kit (Qiagen, Hilden, Germany) according to the manufacturer's instructions, with one minor modification; after mixing plasma/serum samples with Proteinase K and lysis buffer (Buffer ACL; Qiagen), the resultant admixture was incubated at 60°C in a water bath for 1 hour instead of the 30 minutes instructed by the manufacturer. DNA was eluted in 100 μL of elution buffer (Buffer AVE; Qiagen) and subsequently quantified by using RNase P quantitative PCR (ABI; Life Technologies, Foster City, NJ) using 5 μL of DNA according to the manufacturer's instructions. PCR was performed using a Stratagene MX3000 Cycler (Agilent Technologies, Winnersh, Berkshire, UK). The R2 values of the standard curves should be >0.985 for results to be valid. BRAF c.1799T>A mutation in cfDNA was examined by an allele-specific PCR developed in AstraZeneca's Genetics Team (Alderley Park, Cheshire, UK) based on the ARMS technique. This is a single assay that identifies the BRAF c.1799T>A transversion in exon 15 of the BRAF proto-oncogene resulting in an amino acid change from valine to glutamic acid in codon 600 of BRAF protein (p.V600E). Less common additional flanking nucleotide changes also result in p.V600K and p.V600D amino acid changes, and although in theory this assay will still detect the c.1799T>A in the presence of the additional changes, it will not distinguish between them. As a control, the assay system detects a wild-type sequence in exon 17 of the BRAF gene. The PCR product size from mutant reaction is 91 bp, and that from the control reaction is 101 bp. This assay has been used and described previously by our group to detect BRAF c.1799T>A mutation in serum samples of melanoma patients collected from a separate clinical study (Clinical Trials ID number NCT00338130).11Board R.E. Ellison G. Orr M.C. Kemsley K.R. McWalter G. Blockley L.Y. Dearden S.P. Morris C. Ranson M. Cantarini M.V. Dive C. Hughes A. Detection of BRAF mutations in the tumour and serum of patients enrolled in the AZD6244 (ARRY-142886) advanced melanoma phase II study.Br J Cancer. 2009; 101: 1724-1730Crossref PubMed Scopus (85) Google Scholar In the current study, the same primers and probes [synthesized by Eurogentec (Seraing, Belgium)] used in the previous study were used except for the reverse control primer (Table 1). However, a different buffer and enzyme system, Platinum Quantitative PCR SuperMix-UDG (Invitrogen, Paisley, UK), that contained less magnesium than the previously used 1× Brilliant II PCR mix (Stratagene, Cedar Creek, TX), was used, and 40 cycles of PCR were performed instead of 50. These modifications to the previous protocol made the assay more specific and prevented false-positive results. Although they were relatively minor changes, the modified assay needed recharacterization and revalidation as described below.Table 1Primers and Probes Used for BRAF ARMS Allele-Specific Real-Time PCRDescription5′ ModificationSequence3′ ModificationBRAF ARMS primer5′-AAAAATAGGTGATTTTGGTCTAGCTACATA-3′Exon 15 common primer5′-CATCCACAAAATGGATCCAGACAA-3′Exon 15 probeYakima Yellow5′-GATGGAPTGGGTCLCATCEG-3′BHQ1BRAF control forward primer5′-CTCCAGATCTCAGTAAGGTACGG-3′BRAF control reverse primer5′-GGGAAAGAGTGGTCTCTCATCTC-3′BRAF control probeCy55′-CATGAEGEGATTAATGGCAGEGTGLC-3′DDQ2The probes contain locked nucleic acid (LNA)-modified bases. LNA base nomenclature: LNA-modified base A is indicated by E, C by L, G by P, and T by Z.BHQ1, Black Hole Quencher 1; Cy5, Indo-dicarbo Cyanine; DDQ2, Deep Dark Quencher 2. Open table in a new tab The probes contain locked nucleic acid (LNA)-modified bases. LNA base nomenclature: LNA-modified base A is indicated by E, C by L, G by P, and T by Z. BHQ1, Black Hole Quencher 1; Cy5, Indo-dicarbo Cyanine; DDQ2, Deep Dark Quencher 2. Each reaction contained 12 μL of Platinum Quantitative PCR SuperMix-UDG (Invitrogen), 2 μL of bovine serum albumin (10 mg/mL; New England Biolabs, Ipswich, MA), 0.5 μL of BRAF ARMS primer (2 μmol/L), 0.5 μL of BRAF common primer (2 μmol/L), 0.125 μL of BRAF probe (0.5 μmol/L), 0.125 μL of forward control primer (0.1 μmol/L), 0.125 μL of reverse control primer (0.1 μmol/L), 0.05 μL of control probe (0.2 μmol/L), and 4.075 μL of nuclease-free water. Five microliters of DNA was added to each reaction, and all clinical samples were tested in duplicate. PCR was performed in a 96-well optical PCR plate using Stratagene MX3000 Cycler (Agilent Technologies) using the following thermal cycling conditions: 95°C for 10 minutes, followed by 40 cycles of 94°C for 45 seconds, 60°C for 60 seconds, and 72°C for 45 seconds. Fluorescent data were captured at the 60°C annealing step. For each PCR run, three positive controls and one no template control were included in duplicate. Data were analyzed with MX Pro software version 4.1 (Agilent Technologies). No logarithmic increase in fluorescent signals should be seen in the no template control for results to be valid. Thresholds were then set manually above the no template control. If there was only control signal in the replicate tested, it was designated as mutation negative. If there were both control and diagnostic signals in the reaction, the delta Cq (ΔCq) value was calculated by subtraction of the control Cq value from the mutation Cq value. Presence or absence of the mutation in a sample was determined by the values of ΔCq in its replicates (described below). Limit of detection of the assay was 2% (20 mutant copies in 1000 wild-type copies) when it was assessed by using cell-line admixtures (made by HT 29 cell-line DNA and normal genomic DNA). Of the 2% admixtures tested in duplicate in seven different PCR runs, BRAF c.1799T>A mutation was detected in all, with a mean ΔCq value of 9.8 (median, 9.8; range, 7.4 to 12). Mean SD of the ΔCq between the replicates tested in the same PCR run was 0.9 (median, 0.8; range, 0.3 to 2.2). The performance of the assay on FFPE tumor DNA was assessed by using 72 archival melanoma tumor samples with known BRAF c.1799T>A mutation status established by Sanger sequencing. On the basis of the results from testing these samples with different lot batches of reagent and different operators and instruments, mutation calling criteria specific to FFPE tumor DNA were defined as positive if the ΔCq was ≤7.5. This enabled the detection of all known mutation-positive tumor samples and no known negatives, but minimized the risk of false positives that have been previously seen using other ARMS assays on FFPE extracted DNA (unpublished data). No background signal was detected in any of the known negative FFPE tumor or FFPE normal samples tested, including normal genomic DNA at 10-fold higher than the recommended DNA input for the assay. If there were both control and diagnostic signals in the replicate tested, and the ΔCq was ≤7.5, it was designated as mutation positive. Replicates with no diagnostic signal or those with a ΔCq value of >7.5 were designated as mutation negative. A sample was classified as BRAF mutant when both replicates of the sample tested were positive for mutation. The same samples were analyzed by a separate analyst, and all of the mutations were identified correctly using the criteria. The assay performance was also assessed on a separate cohort of 48 archival colorectal tumor samples with known BRAF c.1799T>A mutation status established by Sanger sequencing with visual inspection. Both the ARMS and sequencing analysis were performed in a blinded fashion with no knowledge of serum, plasma, or tumor result. The concordance between BRAF mutation results by ARMS and that by sequencing was 100% using the mutation calling criteria described above, achieving 100% sensitivity and 100% specificity. After confirmation of the percentage of tumor cells in an H&E-stained tumor section by a histopathologist, DNA from 8 × 5-μm unstained sections of FFPE tumor tissue was extracted using a QIAamp FFPE Tissue kit (Qiagen) according to the manufacturer's instructions. BRAF mutation status was evaluated by an AstraZeneca-appointed central laboratory (Cranford, NJ) by ARMS allele-specific PCR or Sanger sequencing or by AstraZeneca-appointed local laboratories using AstraZeneca-agreed methods that included Sanger sequencing, allele-specific PCR, pyrosequencing, and TaqMan PCR. BRAF mutation status was established by using the ARMS method [BRAF ARMS allele-specific PCR designed by AstraZeneca or Qiagen BRAF ARMS assay (Qiagen, Manchester, UK)] in 87% of the cases and by other methods in the remaining 13% of the cases. All statistical analyses, including calculation of 95% confidence intervals using a normal approximation, were performed by R software (version 2.14.1.; Vienna, Austria). Comparisons of cfDNA concentrations in paired plasma and serum samples were calculated using Wilcoxon signed rank test. Comparison of sensitivities and specificities of BRAF mutation detection from plasma and serum was analyzed by Fisher's exact test. A U-test was used to assess whether there was any difference in distribution of cfDNA concentration between patients with BRAF mutation in cfDNA and those without. All of the P values are two sided and considered significant if <0.05. One hundred plasma and serum samples were analyzed in this stage. To determine an optimal acceptance criterion to differentiate negative and positive samples, a range of ΔCq cutoffs were evaluated to determine the ΔCq cutoffs that maximized the number of positive serum or plasma samples that were also tumor positive (sensitivity), while trying to eliminate any plasma- or serum-positive, but tumor-negative samples (specificity) from being called positive (Table 2). In one patient with a BRAF wild-type tumor, a BRAF mutation was convincingly detected in both the plasma and serum, leading us to believe this was a tumor false negative, and is why we accepted a less stringent ΔCq cutoff than we would have if we had considered this a true negative. This sample accounts for the specificity of BRAF mutation detection in both matrices of 98% (95% CI, 89% to 99%) and not the desired 100%. A ΔCq cutoff of 9 or 10 was found to give the best sensitivity and specificity for BRAF mutation detection in cfDNA; 27 and 23 mutations were detected in plasma and serum, respectively, giving a sensitivity of BRAF mutation detection in plasma of 54% (95% CI, 39% to 68%) and in serum of 46% (95% CI, 32% to 61%). A ΔCq cutoff of lower than 9 resulted in fewer positives being detected in serum, and a ΔCq higher than 10 resulted in one more tumor-negative sample being detected, also in serum. However, other than this sample and the samples thought to be tumor false negatives, the remainder of cfDNA tumor-negative samples gave no background signal at all. A ΔCq cutoff of ≤10 was therefore adopted for validation in the second stage (Figure 1).Table 2Sensitivities and Specificities Obtained by Using Different ΔCq Cutoffs for Samples Analyzed in Stage 1ΔCq Cutoffs∗To classify a sample as a BRAF mutant, ΔCq values of both replicates of a sample tested should be ≤ the defined cutoff.SerumPlasmaSensitivity (%)Specificity (%)Sensitivity (%)Specificity (%)≤7.536984698≤840985498≤946985498≤1046985498≤1146965498∗ To classify a sample as a BRAF mutant, ΔCq values of both replicates of a sample tested should be ≤ the defined cutoff. Open table in a new tab Of the 108 matched samples analyzed, 74 were from patients with BRAF mutant tumors and 34 from patients with BRAF wild-type tumors. Using the mutation calling criteria specific to cfDNA (ΔCq cutoff of ≤10), 38 and 31 mutations were detected in plasma and serum, respectively, achieving a sensitivity of 51% (95% CI, 39% to 63%) for plasma and 42% (95% CI, 31% to 54%) for serum (Table 3). In two patients with BRAF wild-type tumors, BRAF mutation was detected in both serum and plasma, giving a specificity of 94% (95% CI, 80% to 99%) for both matrices.Table 3Sensitivities and Specificities Obtained by Using Different ΔCq Cutoffs for Samples Analyzed in Stage 2ΔCq Cut Offs∗To classify a sample as a BRAF mutant, ΔCq values of both replicates of a sample tested should be ≤ the defined cutoff.SerumPlasmaSensitivity (%)Specificity (%)Sensitivity (%)Specificity (%)≤7.528973194≤830974294≤934974594≤1042945194≤1150945394∗ To classify a sample as a BRAF mutant, ΔCq values of both replicates of a sample tested should be ≤ the defined cutoff. Open table in a new tab When results from both stages of this pilot study were combined (208 cases, of which 198 were matched), 54 mutations were detected in serum and 65 in plasma using the mutation calling criteria specific to cfDNA, giving a sensitivity of 44% for serum (95% CI, 35% to 53%) and 52% for plasma (95% CI, 43% to 61%) (Table 4). The difference in sensitivities of serum and plasma cfDNA assays was not statistically significant (P = 0.2). Specificity for both serum and plasma was 96% (95% CI, 90% to 99%). Concordance between tumor mutation status and cfDNA mutation status was 64% (95% CI, 58% to 71%) and 70% (95% CI, 63% to 76%) in serum and plasma, respectively. Further analysis of additional matched tumor serum and plasma will help provide additional confidence in this assay, as well as a suitable confirmatory method to validate the tumor-negative, cfDNA-positive samples.Table 4Sensitivities and Specificities Obtained by Using Different ΔCq Cutoffs for Samples Analyzed in Both StagesΔCq Cut Offs∗To classify a sample as a BRAF mutant, ΔCq values of both replicates of a sample tested should be ≤ the defined cutoff.SerumPlasmaSensitivity (%)Specificity (%)Sensitivity (%)Specificity (%)≤7.531984196≤834984796≤939984896≤1044965296≤1148965396∗ To classify a sample as a BRAF mutant, ΔCq values of both replicates of a sample tested should be ≤ the defined cutoff. Open table in a new tab Serum and plasma cfDNA concentrations were compared using the data from 198 paired samples. Mean cfDNA concentration in plasma was 23.1 ng/mL (median, 13.3 ng/mL; range, 0 to 247.2 ng/mL) and that for serum was 48.8 ng/mL (median, 29.7 ng/mL; range, 4.4 to 462.0 ng/mL). Median concentration of cfDNA was 2.2 times higher in serum compared to plasma (P value <10−6). However, for both plasma and serum, there was no significant difference in distribution of cfDNA concentration between patients with cfDNA mutation and those with no mutation (P = 0.2 for plasma, P = 0.5 for serum, U-test). Of the patient samples analyzed, 53 cases had BRAF mutation in both serum and plasma. In these cases, comparison of ΔCq values, which reflect the proportion of tumor-derived mutant DNA present in a sample (the higher the ΔCq, the lower the mutation fraction), between serum and plasma was performed. The ΔCq values were significantly higher in serum compared to plasma (P value A mutation detection in cfDNA by ARMS was achieved by using mutation calling criteria specific to cfDNA without compromising the specificity. More BRAF c.1799T>A mutations were detected in plasma than in serum (65 versus 54), although differences in assay sensitivity and specificity between them were not statistically significant. Nonetheless, these data are consistent with the contention that plasma contains less wild-type genomic DNA than serum and, as a result, contains a higher mutation fraction. The assay specificity is robust for both matrices at 96% (95% CI, 90% to 99%), which may even be higher considering that in three patients, BRAF c.1799T>A mutation was detected in cfDNA, but not in tumor, and mutation was found in both serum and plasma, suggesting these may be false-negative results in tumor rather than false-positive results in cfDNA. In two of these cases, tumor BRAF mutation was tested by ARMS using archival primary tumor excision material with a time gap between tumor sampling and plasma/serum sampling of 16 months in one patient and 33 months in the other. In the remaining case, the tumor mutation was tested by Sanger sequencing using archival tumor material from a metastatic lymph node biopsy, and the time gap between the biopsy and serum/plasma sampling was approximately 4 months. There are plausible biological and/or logistical explanations for mutation status discordance between tumor and cfDNA in these cases. The blood samples for cfDNA mutation testing were collected some time after the tumor biopsy, and the tumor may have evolved during that time such that mutant clones now predominate, as recently noted in colorectal cancer in which the emergence of KRAS mutations in cfDNA in patients with KRAS wild-type tumor was seen after treatment with EGFR targeted therapies.17Misale S. Yaeger R. Hobor S. Scala E. Janakiraman M. Liska D. Valtorta E. Schiavo R. Buscarino M. Siravegna G. Bencardino K. Cercek A. Chen C.T. Veronese S. Zanon C. Sartore-Bianchi A. Gambacorta M. Gallicchio M. Vakiani E. Boscaro V. Medico E. Weiser M. Siena S. Di Nicolantonio F. Solit D. Bardelli A. Emergence of KRAS mutations and acquired resistance to anti-EGFR therapy in colorectal cancer.Nature. 2012; 486: 532-536Crossref PubMed Scopus (1389) Google Scholar, 18Diaz Jr., L.A. Williams R.T. Wu J. Kinde I. Hecht J.R. Berlin J. Allen B. Bozic I. Reiter J.G. Nowak M.A. Kinzler K.W. Oliner K.S. Vogelstein B. The molecular evolution of acquired resistance to targeted EGFR blockade in colorectal cancers.Nature. 2012; 486: 537-540Crossref PubMed Scopus (1331) Google Scholar Another possibility is that the archival tumor biopsy did not fully represent the tumor heterogeneity, and mutant cfDNA may also have originated from a metastatic lesion rather than the site selected for the tumor biopsy. Taken together, it could be argued that specificity of BRAF mutation testing for serum or plasma is likely to be 100%. The sensitivity or pick-up of BRAF c.1799T>A in cfDNA remained relatively limited at 52% (95% CI, 43% to 61%) in plasma even by using the optimized mutation calling criteria we have developed for cfDNA. There are two main reasons for this limitation. The first one is that the limit of detection of the BRAF ARMS assay used in this study was 2% (20 mutant copies in 1000 wild-type copies), and the mutation fraction in a given cfDNA sample could significantly be <2%.19Li M. Diehl F. Dressman D. Vogelstein B. Kinzler K.W. BEAMing up for detection and quantification of rare sequence variants.Nat Methods. 2006; 3: 95-97Crossref PubMed Scopus (218) Google Scholar, 20Diehl F. Schmidt K. Choti M.A. Romans K. Goodman S. Li M. Thornton K. Agrawal N. Sokoll L. Szabo S.A. Kinzler K.W. Vogelstein B. Diaz Jr., L.A. Circulating mutant DNA to assess tumor dynamics.Nat Med. 2008; 14: 985-990Crossref PubMed Scopus (1866) Google Scholar The second reason is the limitation imposed by the amount of input cfDNA. Even if the mutation fraction is above 2% in a sample tested, the mutation will still not be detected if the absolute mutant DNA copy number is A mutation testing from cfDNA using ARMS should be considered as a first screening step in future clinical trials for positively selecting patients with BRAF mutant tumors in light of current trials of MAPK pathway–targeted drugs and as a surrogate for tumor testing in cases where tumor materials are not available.