The detection and significance of EGFR and BRAF in cell-free DNA of peripheral blood in NSCLC

Objective Although driver mutation status is crucial to targeted therapy decision-making in non-small cell lung cancer (NSCLC), due to unavailable or inadequate biopsies, there are still many patients with unknown mutation status. A promising way to solve this problem is liquid biopsy, such as cell-free DNA (cfDNA) in peripheral blood. Additionally, due to the little amount of cfDNA, detecting methods with high sensitivity, specificity and economy are required in clinical practice. Here, we explored the feasibility of Competitive Allele-Specific TaqMan® PCR (CastPCR) detecting driver mutations in cfDNA from plasma in lung adenocarcinoma patients. Results Sensitivity, specificity, concordance, PPV and NPV of CastPCR detecting EGFR mutations in cfDNA was 56.4% (31/55), 94.2% (49/52), 74.8% (80/107), 91.2% (31/34) and 67.1% (49/73), respectively. Notably, specificity and PPV for p.T790M both reached 100.0%. For BRAF detection, it was 28.6% (2/7), 93.0% (93/100), 88.8% (95/107), 22.2% (2/9) and 94.9% (93/98), respectively. Materials and Methods Plasma specimens of 107 lung adenocarcinoma patients and their matched tumor formalin fixed paraffin embedded (FFPE) samples were analyzed. CastPCR was used to detect EGFR (c.2235_2249del, c.2236_2250del, c.2369C>T p.T790M, c.2573T>G p.L858R) and BRAF (c.1406G>C p.G469A, c.1799T>A p.V600E, c.1781A>G p.D594G) mutations. Mutation results of tumor tissue was set as gold standard, and the sensitivity, specificity, concordance, positive predictive value (PPV) and negative predictive value (NPV) were calculated for each mutation. Conclusions For patients whose tumor tissue is unavailable or inadequate, EGFR mutation detection in cfDNA with CastPCR could be first choice. Mutation positive results may provide reference for further clinical medication. While negative results indicate that detection in tissue should be considered as the following step. In this way, tumor tissue could be economized to the maximum extent and the risk of repeated percutaneous transthoracic lung biopsy could also be lowered to the maximum extent. For BRAF detection in cfDNA, CastPCR is a specific method while the sensitivity needs further exploration.


INTRODUCTION
Thanks to EGFR tyrosine kinase inhibitors (EGFR-TKIs), the treatment pattern of metastatic NSCLC has been revolutionized. Patients' progress free survival, overall response rate and quality of life have been ameliorated significantly [1][2][3][4][5][6][7]. However, in clinical practice, detection of driver mutations in tumor tissue is often not enough. First, sufficient tumor tissue is not readily available. For instance, only 35.9% (437/1217) and 20.3% (297/1466) of the advanced NSCLC patients had biopsied tissue that was suitable for testing in IPASS study [1] and INTEREST study [8], respectively. Second, due to heterogeneity, landscape of molecular information can't be covered with just one or two spots of biopsy [9]. Third, since tumor is evolving under drug therapy, dynamic and real-time monitoring of driver mutations during treatment is required. However, repeated biopsy is not practically feasible in clinical work for its invasiveness and risk. Fortunately, liquid biopsy such as cfDNA from plasma [10][11][12] has come into the picture. cfDNA originates from apoptosis, necrosis, phagocytosis, oncosis and active secretion of cells [13]. Studies have demonstrated that in cancer patients, cfDNA contains representation of the entire tumor genome [14,15]. The mutation load was reported higher than 25.0% in one third of mutant plasma samples [13]. There have been multiple publications on cfDNA mutation detection in the last two decades [16][17][18][19][20][21][22][23][24][25]. According to Committee for Medicinal Products for Human Use (CHMP) of the European Medicines Agency (EMA) , to predict therapeutic response of Iressa® (gefitinib), the use of circulating tumor DNA obtained from a blood sample has been allowed to assess EGFR mutation status in patients where a tumor sample is not an option [13].Additionally, cfDNA was shown to be useful in dynamic monitoring of acquired resistance and prediction of relapse in various cancers, including NSCLC [26][27][28][29][30]. In clinical practice, methods with high sensitivity, specificity, convenience and economy are required. Development of molecular detection methods such as qPCR based methods, would allow for highly specific analysis of cfDNA [31]. Here, we evaluated the use of CastPCR detecting EGFR mutations (c.2235_2249del, c.2236_2250del, p.T790M, and p.L858R) in cfDNA form plasma in adenocarcinoma patients.
Reported incidence of BRAF mutations ranges from 0.5-9.0% in NSCLC [32]. Phase II clinical trial NCT01336634 has made its results public: BRAF targeted inhibitor dabrafenib has shown clinical activity in BRAF p.V600E-positive metastatic NSCLC. It suggested that for patients with limited therapeutic options, dabrafenib could represent a treatment option [33]. Detection of BRAF mutations in cfDNA hasn't been reported in NSCLC, thus we also evaluated CastPCR technology in detecting BRAF mutations (p.G469A, p.V600E and p.D594G).

EGFR mutations
Of the 107 plasma specimens, 31

DISCUSSION
As the flourishing development of precision medicine in lung cancer, there is a pressing need for realtime information of clinical specimens, as well as the establishment of a detection platform with convenience, efficiency and economy. With its relative noninvasiveness, high repeatability and real-time dynamic, liquid biopsy is breaking new ground. For the first time, we evaluated the feasibility of CastPCR detecting EGFR and BRAF mutations in plasma cfDNA of a cohort lung adenocarcinoma patients.
Studies employing different methods detecting EGFR sensitive mutations in paired NSCLC tumor tissue and plasma were listed in Table 6. Though Amplification refractory mutation system (ARMS) is extensively adopted to detect tissue EGFR mutations in clinical practice, its sensitivity in cfDNA is not stable [34][35][36]. For denaturing www.impactjournals.com/oncotarget high performance liquid chromatography (DHPLC), its complex detection procedures hinders its clinical promotion [10,37,38]. Several targeted or sequencing methods have shown good sensitivity (78.0-81.1%) and concordance (86.0-93.6% ) on EGFR mutation detection, such as droplet digital PCR (ddPCR) and next-generation sequencing (NGS) [21]. Compared with their prohibitive cost, complex laboratory manipulations and high demand for operational platform, CastPCR is more feasible and economical.
EGFR T790M plays a key role in selecting the right patient for third-generation EGFR-TKI [39]. Up to date, U.S. Food and Drug Administration approved Cobas ® as the only assay to detect EGFR specific mutations (i.e., exon 19 deletions and exon 21 L858R substitution mutations) in plasma specimens (www.fda.gov). In recent studies that adopted Cobas ® testing platform for p. T790M mutation detection, the sensitivity was (51.0%) [40]. In our study, sensitivity detecting p.T790M was 45.0% (9/20). Notably, the specificity and concordance     [40]. Since the PPV and NPV for p.T790M was 100.0% (9/9) and 88.8% (87/98) in our study, for patients who acquire drug resistance to first generation EGFR-TKI, plasma detection of EGFR p.T790M by CastPCR could be first consideration. For those with cfDNA tested positive, third generation EGFR-TKI could be a treatment choice. Similarly, considering the PPV (91.2%, 31/34) and NPV (67.1%, 49/73) of EGFR detection with CastPCR in our study, for patients whose tumor tissue is not enough or unavailable, EGFR mutation detection with CastPCR in cfDNA could be first choice. For those with EGFR mutations in cfDNA, the result may be taken as reference for further clinical medication. While for those detected without EGFR mutations in cfDNA, further detection in tumor tissue is recommended. In this way, tumor tissue could be economized to the maximum  Oncotarget 49777 www.impactjournals.com/oncotarget extent and the risk of repeated percutaneous transthoracic lung biopsy could also be lowered to the maximum extent.
BRAF is another important drugable driver mutation in NSCLC. Compared with EGFR, the detection of BRAF mutation both in tumor tissue and plasma still has a long way to go. To the best of our knowledge, BRAF detection of cfDNA in NSCLC hasn't been reported yet. We performed BRAF mutation detection in 107 pairs of lung adenocarcinoma plasma cfDNA and matched FFPE DNA samples. CastPCR was applied to detect BRAF mutations in tumor tissues and 6.5% (7/107) patients were identified as mutation positive, consistent to the result concluded by NGS [32,41]. Interestingly, p.V600E accounted for about half BRAF mutations in NSCLC in earlier studies [42][43][44][45][46], and recent studies employing NGS technology exhibited that non-p.V600E mutations is the majority (70.0-80.0%) of BRAF mutations in NSCLC [32,41,47]. In our study, the frequency of non-p.V600E was 85.7% (6/7), similar to that demonstrated by studies using NGS.
In BRAF mutation detection of cfDNA with CastPCR, the sensitivity, specificity and concordance was 28.6% (2/7), 93.0% (93/100) and 88.8% (95/107), respectively ( Table 3). The low sensitivity may mainly related to the following reasons: Firstly, tumor heterogeneity. In the 12 patients whose mutation conditions were inconsistent, 58.3% (7/12) had distant metastasis. Thus their cfDNA released in plasma could probably carry different gene information compared with primary tumors. Secondly, detection of BRAF non-p.V600E mutations in exon 15 of cfDNA needs more investigation. In our study, sensitivity for p.V600E (locates on exon15) and p.G469A (locates on exon11) both reached 100.0%, while none of the five p.D594Gs (locate on exon15) in FFPE samples were detected in their matched plasma. Similarly, Couraud et al. [48] applied NGS to detect mutations of exon 11 and 15 in about 60 tumors and corresponding plasma, and better result was produced for mutations of exon 11 mutations than that of exon 15.

Materials
All samples used in this study were obtained from Nanjing Drum-Tower Hospital form 2007 to 2016. 125 lung adenocarcinoma patients were enrolled, and data of 107 paired samples was analyzed in the end (Figure 1). Patients' clinicopathological features are provided in Table 7. 83.2% (89/107) blood samples was collected when patients were diagnosed, together with their paired FFPE samples were cut into slices and each slice was 4μm thick. For every FFPE sample, a random slice was performed with the HE staining, then two pathologists identified and marked the tumor tissue through the observation of microscope. 1-2 mL peripheral blood was collected into an EDTA-containing tube, centrifuged at 3000 rpm/min for 15 min at room temperature, then the plasma portion was pipetted carefully, aliquoted to 0.5 mL Eppendorf tubes and stored at −80°C. All patients enrolled in the study provided written informed consent for the use of their resected/biopsied tumor tissue and peripheral blood. The study was approved by the Ethics Committee of Nanjing Drum-Tower Hospital.  our preliminary experiment, we adopted the latter method (data not shown). NanoDrop 2000 spectrophotometer (Thermo Fisher Scientific, USA ) was used to quantify cfDNA.

CastPCR
CastPCR technology combines allele specific TaqMan ® qPCR with allele-specific blocker (ASB) oligonucleotides that effectively suppress nonspecific amplification from the off-target allele. In our preliminary experiment, we assessed the limit of detection (LOD, minimum percentage of mutant alleles in a wild type background required for reliable mutation detection) of some of the assays with corresponding mutated cell lines. It turned out to be that the sensitivity was consistent to that provided by the manufacturer:

Statistical analysis
Sensitivity, specificity, concordance, PPV and NPV were calculated as follows [52]: Sensitivity = number of true positives/(number of true positives + number of false negatives); Specificity = number of true negatives/ (number of true negatives + number of false positives); Concordance = (number of true positives + number of true negatives) / (number of total cases); PPV = number of true positives/(number of true positives + number of false positives); NPV = number of true negatives/(number of true negatives + number of false negatives).

CONCLUSIONS
For the first time, we evaluated the possibility of CastPCR detecting EGFR and BRAF mutations in cfDNA of plasma from 107 lung adenocarcinoma patients. For patients whose tumor tissue is not enough or unavailable, EGFR mutation detection with CastPCR in cfDNA could be first choice. For those with EGFR mutations in cfDNA, the result may be taken as reference for further clinical medication. While for those detected without EGFR mutations in cfDNA, further detection in tumor tissue is recommended. Additionally, CastPCR technology was firstly employed to detect BRAF mutations in plasma specimens of lung adenocarcinoma and their matched FFPE samples. We found that for BRAF mutation detection, CastPCR is a specific method while its sensitivity, especially for BRAF non-p. V600E mutations on exon 15, needs further exploration.

ACKNOWLEDGMENTS AND FUNDING
We would like to thank all the stuff of the comprehensive cancer center of Drum Tower Hospital.

CONFLICTS OF INTEREST
No potential conflicts of interest were disclosed.