Are liquid biopsies a surrogate for tissue EGFR testing?

Are liquid biopsies a surrogate for tissue EGFR testing? Abstract Molecular profiling has changed the treatment landscape in advanced non-small-cell lung cancer. Accurately identifying the tumours that harbour sensitizing EGFR mutations, the most common targetable molecular alteration, as well as those with acquired resistance mutations (e.g. T790M) on treatment is a high clinical priority. The current clinical gold standard is genotyping of tumour specimens. However, the practical utility of this approach is limited by the lack of available tissue and the potential complications associated with biopsies. With the advent of newer sequencing assays, it has become feasible to assess tumour genomics via a blood sample, termed a ‘liquid biopsy’. In this review, we summarize the available techniques for liquid biopsies and their applicability for detecting sensitizing and resistance EGFR mutations and how these results may be used for making treatment decisions. cfDNA, ctDNA, lung cancer, EGFR, T790M, liquid biopsy Introduction Most advanced non-small-cell lung cancers (NSCLC) with mutations in the epidermal growth factor receptor (EGFR) are highly sensitive to EGFR tyrosine kinase inhibitors (TKIs). With targeted therapies, median overall survival (OS) has improved from 8–12 to 20–30 months for this patient subset [1]. Accurately identifying the tumours that harbour sensitizing EGFR mutations as well as those with acquired resistance mutations (T790M) that can be targeted by newer third-generation TKIs is a high clinical priority, with associated increases in OS of up to 46 months with personalized treatment at progression [2]. The current clinical gold-standard is genotyping of tumour biopsies. However, with the advent of newer sequencing assays, it has become feasible to assess tumour genomics via a blood sample, termed a ‘liquid biopsy’. Recent data suggest that circulating tumour DNA (ctDNA), that can be used for molecular profiles, derived from NSCLC can be detected with high sensitivity in urine as well as in plasma [3] and cerebral spinal fluid [4, 5], enabling complementary modes of tissue and liquid biopsies in this population. EGFR mutation profile by liquid biopsy is now accepted for making treatment decisions among NSCLC patients, and liquid biopsy tests have earned limited approvals in Europe, China, Japan, and the USA owing to several distinct advantages over tissue biopsy [6]. Tissue biopsy Lack of available tissue for performing molecular profiling, the location or size of the tumour, and the risk of complications (∼17% of adverse events with image guide-biopsies [7] and risk of pneumothorax requiring chest tube in up to 14% of patients [8]), are serious limitations for performing biopsies of NSCLC tumours. In EGFR-mutant NSCLC patients with acquired resistance to EGFR TKIs, re-biopsies are feasible in only half of the patients [9] as a consequence of inaccessible tumour sites, patient refusal, or physician discretion [10]. In prospective trials, the percentage of available tissue samples for EGFR genotyping can vary greatly; even when successfully obtained, ∼12%–30% of samples were not sufficient for genotyping [11–14]. In addition, after obtaining sufficient samples via invasive procedures, EGFR genotyping fails in ∼5%–10% of tests [15, 16]. Moreover, single site biopsies may not provide a representative profile of the overall predominant resistance mechanisms for a given patient [17]. Delays in tissue biopsy often occur as a result of scheduling, specimen delivery to molecular diagnostics, and laboratory processing time. In a prospective study of EGFR genotyping in advanced lung cancer patients, the median turnaround time for tissue biopsy was 12 days for new diagnosis of NSCLC patients and 27 days for those tested for acquired resistance. In contrast, the median time was 3 days for plasma based testing using an in-house assay for EGFR and KRAS hotspots [18]. There is also a significant difference in cost. An analysis of 8979 Medicare patients undergoing diagnostic assessment for lung cancer found that the median cost of a biopsy was $2235. However, this cost increased to a median of $14 824 if an adverse event occurred [8]. These are likely underestimates, as costs from Medicare are generally lower than from other payers. A study using simulation analytics found that a liquid-based alternative for EGFR genotyping could result in substantial financial savings [19]. There may also be advantages to a liquid biopsy that go beyond avoiding the risks and financial costs of a tumour biopsy (Table 1). NSCLCs are amongst the cancer types with the highest rates of point mutations, harbouring hundreds of mutations in their exome [20], and advanced cancers are highly heterogeneous [21]. Tumour heterogeneity results from numerous genomic alterations, catastrophic events, and clonal evolution during division. Dividing cells in the tumour acquire new mutations, creating sub-clones that differ from their founder cells so that geographically separated regions in the same tumour will generally also be more genomically distinct [22]. Seeding of varying sub-clones to other organs leads to further divergent evolution. In NSCLC, multi-region whole exome sequencing studies confirm that tumours show large spatial and temporal diversity in mutational profiles [23, 24]. Table 1. Concerns with tissue biopsy that are potentially addressed by liquid biopsy Concern  Advantage of liquid biopsy  Risk of complication  Least invasive or noninvasive.  High cost  Removes costs associated with procedure and complications.  Difficult tumor site  Blood, cerebrospinal fluid, or urine can be easily collected.  Specimens quality varies  More control over quantity and quality of specimen.  Lengthy turnaround time  Turnaround time is significantly less.  Tumor heterogeneity  Better captures the molecular state of tumours and metastases.  Impractical for disease monitoring  Repeat biopsies are easy and provide a ‘real-time snapshot’ to monitor response to therapy and development of resistance.  Concern  Advantage of liquid biopsy  Risk of complication  Least invasive or noninvasive.  High cost  Removes costs associated with procedure and complications.  Difficult tumor site  Blood, cerebrospinal fluid, or urine can be easily collected.  Specimens quality varies  More control over quantity and quality of specimen.  Lengthy turnaround time  Turnaround time is significantly less.  Tumor heterogeneity  Better captures the molecular state of tumours and metastases.  Impractical for disease monitoring  Repeat biopsies are easy and provide a ‘real-time snapshot’ to monitor response to therapy and development of resistance.  A tissue biopsy may therefore only demonstrate a minority of the genetic aberrations present in the tumour, as this single site may not be representative of intra-tumour heterogeneity nor interlesional heterogeneity (Figure 1). Prospective analysis of EGFR mutation status in advanced NSCLC patients who underwent multiple biopsies showed frequent changes in their mechanism of acquired resistance and T790M status (40% gained or lost the mutation) [17]. This complex and dynamic quality suggests that single site biopsies may not provide a representative profile of the overall predominant resistance mechanisms for a given patient. Figure 1. View largeDownload slide Liquid biopsy for discovery of actionable EGFR mutations, monitoring response to therapy, and detecting development of TKI resistance. Liquid biopsy carried out at diagnosis guides treatment with a TKI to target actionable mutations (A). ctDNA levels can correlate to response to TKIs (B). At progression, as ctDNA levels rise, one would ideally obtain a repeat tissue biopsy to guide next treatment decision (C). However, even when a tissue biopsy is carried out, due to tumour heterogeneity, it may miss mutations due to sampling bias (D). Figure 1. View largeDownload slide Liquid biopsy for discovery of actionable EGFR mutations, monitoring response to therapy, and detecting development of TKI resistance. Liquid biopsy carried out at diagnosis guides treatment with a TKI to target actionable mutations (A). ctDNA levels can correlate to response to TKIs (B). At progression, as ctDNA levels rise, one would ideally obtain a repeat tissue biopsy to guide next treatment decision (C). However, even when a tissue biopsy is carried out, due to tumour heterogeneity, it may miss mutations due to sampling bias (D). A study using direct DNA sequencing to detect EGFR mutations in 180 Asian patients with matched samples from primary tumour and distant sites found that nearly a fourth of those with multiple pulmonary nodules had discordant EGFR status [25]. These data suggest that re-biopsy or multiple biopsies would be advantageous, yet, as already described, this is impractical. Liquid biopsies based on ctDNA analysis are therefore effective surrogate samples for tumour molecular analysis, and also as potential dynamic markers for monitoring the efficacy of EGFR TKIs and early detection of resistance mutations [26]. ctDNA and techniques In 1977, the same year the eponymous Sanger sequencing method was described, Leon et al. reported elevated levels of circulating cell-free DNA (cfDNA) in the serum of cancer patients [27]. The fraction of cfDNA in serum or plasma derived from tumour cells, termed ctDNA, is hypothesized to originate from tumour secretion, necrosis, and apoptosis [28]. Circulating tumour cells (CTCs), exosomes, or platelets can also provide DNA or RNA for isolation. Plasma is preferred in the detection and characterization of ctDNA because of lower background levels of wild-type DNA [28]. The half-life of cfDNA in circulation has been estimated to be between 16 min and 2.5 h, which makes it akin to a ‘real-time snapshot’ of disease burden that can be monitored [28]. ctDNA can be specifically identified in cfDNA by the presence of somatic mutations. Methods for identifying these variations range from large-scale (up to whole genome) sequencing to analysis of individual mutations using polymerase chain reaction (PCR) with specific primers, often with complementary probes for known aberrations (Figure 2). The proportion of ctDNA at a specific locus is usually reported as mutant allele concentration (copies/ml) or mutant allele fraction (%). ctDNA often represents <1% of cfDNA in a sample and, consequently, reliable detection across cases requires sensitive analysis methods [29]. The sensitivity of sequencing is technically limited by the error rate of DNA polymerase, often considered to be in the range of 0.01% [30]. An additional limitation on detection of rare alleles results from Poisson sampling statistics, whereby an allele that is present at a very low fraction may not be present in a given blood sample. Therefore, it has been suggested that individual mutations present in ctDNA at levels <0.01% mutant allele fraction would be undetectable [31]. The sensitivity of classical Sanger (chain-termination) method to detect variant alleles is >10% [29], inadequate for wide clinical use where ctDNA levels can be far lower. In 1999, Kenneth Kinzler and Bert Vogelstein described digital PCR (dPCR) to identify rare cancer mutations [32] (Figure 2A1). The most common application of dPCR now relies on generating thousands to millions of water droplets in an oil emulsion rather than physically separated chambers, and is termed droplet dPCR (ddPCR), first exemplified by the same group in a technique called BEAMing (beads, emulsion, amplification, and magnetics) [33]. Both dPCR and ddPCR have a rapid turnaround time and are useful in detecting and quantifying recurrent hot-spot mutations or resistance mutations, and can, in principle, achieve a best-demonstrated detection below 0.001% [34]. In practice, the sensitivity will be limited by the amount of cfDNA analysed, which varies per patient and disease state from a few thousand to hundreds of thousands of copies, with a median around 15 000 amplifiable copies per 10 ml blood sample [35]. Figure 2. View largeDownload slide Common techniques used for ctDNA detection. Methods for studying ctDNA range from single locus to the much broader whole genome sequencing. (A1) Digital polymerase chain reaction (dPCR), and digital droplet PCR (ddPCR), are based on the concept that amplifying single molecules would be most sensitive if DNA templates could be diluted to the point where each reaction chamber contained either no amplified product or PCR product from a single amplification. This eliminates the noise of weak signals seen in analogue sequencing. (A2) The amplification refractory mutation system (ARMS) uses PCR amplification with complementary primers and fluorescent dye probes, relying on the terminal 3′ nucleotides of PCR primer being allele specific. The Scorpion ARMS (SARMS) method uses ‘scorpion’ probe sequences held in hairpin confirmation and then combined with ARMS along with fluorophores, and their incorporation is quantified by real-time PCR. Therefore, a fully matched primer can selectively amplify mutated sequences. (B) Targeted sequencing allows one to focus on specific regions of interest or alternatively, perform whole exome sequencing. Two main approaches to targeted sequencing include (B1) amplicon-based, using PCR amplification with primers for selected regions to generate targeted sequencing libraries; or (B2) ligation-based library preparation, followed by enrichment for targeted regions using hybridization to sequence baits (hybrid-capture). (C) Whole genome sequencing does not require a priori knowledge of genetic mutations and can be useful if ctDNA levels are high. Figure 2. View largeDownload slide Common techniques used for ctDNA detection. Methods for studying ctDNA range from single locus to the much broader whole genome sequencing. (A1) Digital polymerase chain reaction (dPCR), and digital droplet PCR (ddPCR), are based on the concept that amplifying single molecules would be most sensitive if DNA templates could be diluted to the point where each reaction chamber contained either no amplified product or PCR product from a single amplification. This eliminates the noise of weak signals seen in analogue sequencing. (A2) The amplification refractory mutation system (ARMS) uses PCR amplification with complementary primers and fluorescent dye probes, relying on the terminal 3′ nucleotides of PCR primer being allele specific. The Scorpion ARMS (SARMS) method uses ‘scorpion’ probe sequences held in hairpin confirmation and then combined with ARMS along with fluorophores, and their incorporation is quantified by real-time PCR. Therefore, a fully matched primer can selectively amplify mutated sequences. (B) Targeted sequencing allows one to focus on specific regions of interest or alternatively, perform whole exome sequencing. Two main approaches to targeted sequencing include (B1) amplicon-based, using PCR amplification with primers for selected regions to generate targeted sequencing libraries; or (B2) ligation-based library preparation, followed by enrichment for targeted regions using hybridization to sequence baits (hybrid-capture). (C) Whole genome sequencing does not require a priori knowledge of genetic mutations and can be useful if ctDNA levels are high. Methods that use next-generation sequencing (NGS), and generate sequencing libraries across selected regions of the genome, are commonly referred to as ‘targeted sequencing’ (Figure 2B). Forshew et al. described the first implementation of targeted sequencing to detect rare cancer mutations in ctDNA. Using a two-step amplification process to generate tagged (barcoded) amplicons, they were able to identify mutations across a gene panel at allele fractions as low as 2% in plasma, with sensitivity and specificity >97% [36]. This method, termed TAm-Seq (for Tagged-Amplicon Sequencing), could also identify predefined individual mutations (e.g. hot-spot mutations or previously characterized patient-specific alterations) at allele fractions as low as 0.14%. An enhanced version of TAm-Seq, eTAm-Seq, was developed to identify mutations in hot-spots (including EGFR) and entire exons across a gene panel, and an assay covering a panel of 35 genes was described in 2016, with detection of mutations at allele fractions of 0.06%–0.08% at nearly 40% sensitivity, and sensitivity of 90% for mutations present at allele fractions of 0.25% and greater [35]. Alternative methods use hybrid capture-based deep sequencing, such as the Guardant360 assay, which covers regions in 70 genes (including the actionable EGFR mutations), and uses a library of individually tagged cfDNA molecules to reduce false positives [37]. In a study utilizing the Guardant360 assay to profile ctDNA from patients with different cancer types, a subset with matched tissue tests were compared with results from the tissue sequencing project, TCGA, and high levels of accuracy were observed when blood and tumour were collected within 6 months of each other [38]. This demonstrates the potential of NGS technologies. Wider-scale analysis such as whole-exome sequencing (Figure 2C) and shallow whole-genome sequencing can be carried out to detect de novo mutations and chromosomal aberrations and study clonal evolution when the ratio of ctDNA to cfDNA is relatively high (∼5% or greater) [39, 40]. It is likely that future ctDNA studies will rely on both ultrasensitive targeted detection methods as well as wider-scale analysis at different stages of diagnosis, monitoring, and treatment. NGS-based methods allow for de novo sequencing of key regions in EGFR to identify mutations at low allele fractions. While these are now emerging as practical tools, the most widely used clinical tests have used PCR-based technologies to detect mutations in EGFR hot spots. Kits approved or endorsed by the US Food and Drug Administration (FDA) and the European Medicines Agency (EMA) have employed the Scorpion Amplification Refractory Mutation System (SARMS) (Figure 2A2). Clinical assays include the Cobas EGFR assay, which can detect mutations in exons 18, 19, 20, and 21 with at least 25–100 copies of the mutation per millilitres [41]. The therascreen SARMS kit has a median limit of detection of 1.42% [42]. The peptide nucleic acid-locked nucleic acid (PNA-LNA) clamp method exploits the tighter binding affinity of PNA and LNA probes to DNA sequences rather than to the DNA duplex, which prevents amplification of wild-type alleles, and has been employed in EGFR detection [43]. The commercially available PNAClamp technology has a stated limit of detection of <1% for EGFR mutations [44]. A range of other methods have been developed for detection of EGFR mutations at low allele fractions. A method that uses multiplex PCR amplification and detection based on mass spectroscopy can achieve detection of key hot-spot variants with as low as 0.1% minor allele frequency [45]. These alternative methods have been utilized in studies of EGFR mutations, but to the best of our knowledge have so far not been widely used. Clinical use Detection of activating EGFR mutations to guide initial treatment with TKIs The first trial to perform a pairwise comparison of tissue and liquid biopsy for EGFR mutations was conducted as part of a phase II study of 27 Japanese NSCLC patients in 2006 [46] (Table 2). When direct sequencing was used, serum EGFR mutational status was not correlated with response to gefitinib, but when SARMS was used to detect serum EGFR mutation there was a statistical correlation with treatment response. Direct sequencing, therefore, seemed unsatisfactory for detection of EGFR mutations in serum. In pairs of tumour and serum samples (n = 11), the EGFR mutation status in the tumours was consistent with those in the serum in 72.7% of the paired samples by SARMS. The following year, the same group tested 42 patients, and found an even higher concordance rate of 92.9%. They reported a strong correlation between either tissue or serum EGFR mutation status and response to gefitinib as well as progression-free survival (PFS) [47]. Table 2. Summary of selected clinical studies of liquid biopsy to detect EGFR mutations Clinical use  References  Summary  Initial detection of EGFR mutations to guide treatment  [46, 47, 61]  EGFR mutation status by liquid biopsy strongly correlates with tissue biopsy status.  [47, 62, 60]  EGFR mutation status correlates to TKI response, PFS, as well as OS.  [55, 56, 57]  Early studies showed mixed results in detecting EGFR ctDNA.  [58, 59, 61]  Improved methodology has greatly improved sensitivity and specificity of testing.  [48, 49, 50]  Results from two large multi-national trials suggest that optimization, validation, and standardization are needed to improve performance.  [51, 53, 54]  Meta-analyses show high diagnostic accuracy of ctDNA for EGFR mutation testing.  ctDNA as a marker of efficacy  [52, 71, 72]  ctDNA levels may predict progression before clinical progression.  [61, 68, 69]  Clearance of ctDNA may predict response and prognosis.  Identifying resistance mutations in patients progressing on TKI  [74, 75, 76]  Detection of the T790M mutation can be accurately carried out via liquid biopsy, and the presence of any detectable T790M ctDNA may be clinically relevant.  [24, 73, 74]  T790M status by liquid biopsy correlates well to response to third-generation TKIs.  Clinical use  References  Summary  Initial detection of EGFR mutations to guide treatment  [46, 47, 61]  EGFR mutation status by liquid biopsy strongly correlates with tissue biopsy status.  [47, 62, 60]  EGFR mutation status correlates to TKI response, PFS, as well as OS.  [55, 56, 57]  Early studies showed mixed results in detecting EGFR ctDNA.  [58, 59, 61]  Improved methodology has greatly improved sensitivity and specificity of testing.  [48, 49, 50]  Results from two large multi-national trials suggest that optimization, validation, and standardization are needed to improve performance.  [51, 53, 54]  Meta-analyses show high diagnostic accuracy of ctDNA for EGFR mutation testing.  ctDNA as a marker of efficacy  [52, 71, 72]  ctDNA levels may predict progression before clinical progression.  [61, 68, 69]  Clearance of ctDNA may predict response and prognosis.  Identifying resistance mutations in patients progressing on TKI  [74, 75, 76]  Detection of the T790M mutation can be accurately carried out via liquid biopsy, and the presence of any detectable T790M ctDNA may be clinically relevant.  [24, 73, 74]  T790M status by liquid biopsy correlates well to response to third-generation TKIs.  EGFR, epidermal growth factor receptor; TKI, tyrosine kinase inhibitor; PFS, progression-free survival; OS, overall survival; ctDNA, circulating tumor DNA. Two large multinational trials, IGNITE and ASSESS, have studied the efficacy of EGFR mutation testing in the real-world setting [48]. In the IGNITE trial, 3382 patients were enrolled and plasma and tissue data were available for 2581 patients. The techniques used for EGFR testing were not centralized and had limited sensitivity. The plasma sensitivity for detection of tissue-confirmed EGFR mutations was <50%. The ASSESS trial had data available from 1162 matched tissue and plasma samples and found a concordance rate of 89.1%, with a sensitivity for detecting mutations in plasma of 46% and specificity of 97.4% [49]. The low sensitivity suggests that wide implementation requires optimization, standardization, and validation. In fact, the study found large variation in the sensitivity of plasma testing amongst countries, from 36% to 100%. In subgroup analysis of only therascreen (an SARMS-based assay), the concordance rate was 95%, sensitivity increased to 73% and the specificity was 99% with a NPV of 95%. Similarly, subgroup analysis looking at Cobas EGFR testing alone found a 96% concordance rate, 75% sensitivity, 100% specificity, and NPV of 95% [50]. A meta-analysis examining 27 studies conducted between the years 2007 and 2015, including over 4000 patients, generated a receiver operating characteristic (ROC) curve, and found the area under the curve (AUC) to be 0.92, demonstrating high diagnostic accuracy (AUC > 0.9), with positive likelihood ratio >10. They found a pooled sensitivity of 64% for detection of the EGFR exon 19 deletions and 57% for L858R substitution mutations with 99% specificity for both [51]. As Oxnard et al. point out, in the context of a relatively uncommon mutation, specificity is more important than sensitivity. A sacrifice in test sensitivity with an accompanying small increase in specificity results in a dramatic increase in positive predictive value. They give the example that to detect an EGFR mutation present in 8.6% of patients, with a 10% decrease in sensitivity from 80% to 70%, if the specificity of 95% increased to 99% then the PPV would more than double, from 43% to 87% [52]. Two separate meta-analyses have found very similar results for pooled specificity (>93%), pooled sensitivity (>60%), and AUC (0.93 and 0.91) for EGFR analysis by ctDNA. These authors also concluded that the high diagnostic accuracy of ctDNA makes it a suitable screening test for EGFR mutations [53, 54]. The phase III IPASS trial, comparing gefitinib to carboplatin–paclitaxel amongst 233 Japanese patients with advanced NSCLC, retrospectively reported EGFR mutations in ctDNA detected by SARMS in 23.7% of patients versus 61% from tumour samples [55]. There was an unacceptably high false negative rate of 56.9%, which may be attributed to the use of serum rather than plasma, DNA extraction techniques, and a less refined SARMS kit. Another study of 54 patients with known clinical response to gefitinib or erlotinib found that only ∼47% of EGFR mutations were detected when an SARMS-based technique was used [56]. A similar study using SARMS detected only 39% of EGFR mutations in plasma ctDNA from 18 patients with known EGFR mutations [57]. These mixed results from earlier studies highlighted the challenges of ctDNA analysis, but also its great potential. With more data on pre-analytical variables and improvements in sample collection processes, alongside more robust, and sensitive methods for mutation detection, more recent studies have shown consistently greater sensitivity for ctDNA detection in a range of clinical applications [28]. In a first application of dPCR for analysis of EGFR in peripheral blood samples from NSCLC patients, Yung et al. developed probes to detect the exon Del19 and L858R at exon 21, which account for >85% of all clinically relevant EGFR mutations [58]. Their study of 35 patients reported a sensitivity of 92% and a perfect specificity [59]. More recently, Sacher et al. prospectively validated a ddPCR technique in a study of 180 patients to detect EGFR mutations at diagnosis or relapse. They reported that ddPCR for plasma genotyping of sensitizing EGFR mutations also had 100% specificity and positive predictive value given no false positives, and a sensitivity of 69%–80% [18]. The phase III EURTAC trial, comparing upfront erlotinib to platinum-based chemotherapy in 173 EGFR-mutant NSCLC patients, assessed the feasibility for testing EGFR mutations in a liquid biopsy by real-time PCR TaqMan assay as a surrogate of tissue testing, and compared outcomes according to detection of mutation in tissue versus liquid biopsy [60]. This study demonstrated that tumour and peripheral EGFR mutation status had similar ability to predict OS and PFS [60]. Similarly, to prospectively validate the use of blood-based EGFR testing, Mok et al. demonstrated high performance characteristics of the Cobas EGFR Blood test [61]. The phase III FASTACT-2 trial randomized advanced NSCLC patients to platinum-gemcitabine first-line chemotherapy intercalated with either erlotinib or placebo followed by maintenance treatment with erlotinib or placebo. Cobas EGFR testing had a high tissue-blood concordance of 88% with a sensitivity of 75%, specificity of 96%, and PPV of 94% [61]. The phase IV IFUM trial assessed the efficacy of gefitinib in Caucasian NSCLC patients and the clinical utility of the therascreen EGFR RGQ PCR kit for plasma ctDNA was tested. Concordance in mutation status amongst 652 matched samples was 94.3%, sensitivity was 65.7%, and specificity was 99.8%. The trial reported that patients’ outcome was similar independent of EGFR mutation being detected in plasma or in tissue [response rate (RR) 69.8% versus 76.9%, and PFS, 10.2 versus 9.7 months]. Of significance, 12 patients without available tissue samples for genotyping were discovered to have EGFR mutations via plasma testing [62]. In January 2015, the EMA granted marketing approval to the therascreen assay. Given the 34% false negative rate in the IFUM trial, they specified that the test should only be used for patients without a tumour sample. In June 2016, the Cobas EGFR mutation test v2 was approved by the FDA for the detection of exon 19 deletions and the L858 substitution mutation in plasma. In September 2016, this was extended to cover also the T790M resistance mutation. Given that a relatively low number, 76.7% of patients, who had a detectable EGFR mutation by tissue biopsy also had a detectable mutation in plasma, the FDA specified that those with negative plasma testing should be retested with tissue sampling. Recently, NGS-based ctDNA assays have been introduced as an important tool for molecular profiling in untreated NSCLC patients [63], detecting clinically relevant and actionable mutations in up to 23% of patients when tissue was unavailable, and also in previously treated NSCLC patients [64, 65]. The assay used, sensitivity of the test, and sample collection time point should be kept in mind when interpreting concordance between plasma and tumour analyses [65]. Indeed, liquid biopsies can be used longitudinally as a non-invasive approach to monitor patient treatment outcome and may provide a surrogate for response evaluation by radiographic RECIST assessment [63]. Also, recently ctDNA has been demonstrated as an important tool for detecting the emergence of resistance mutations on treatment [26, 66, 67]. Further analysis in well-defined patient populations is underway to provide supporting data for the clinical utility of such assays. ctDNA as a dynamic marker of efficacy and for the identification of resistance mutations in patients relapsing on targeted therapies ctDNA levels may predict progression before standard imaging progression by RECIST. The first study to demonstrate this in NSCLC found that serial plasma genotyping could detect resistance up to 16 weeks before radiographic progression [52]. Analysis of the FASTACT-2 data found that at cycle three, if a cfDNA sample was EGFR mutant positive, RR was only 33% compared with 66% for those who cleared their EGFR mutation positivity [61]. In a study of 81 patients with NSCLC treated with an EGFR-directed TKI therapy, PFS was 6.3 months for those with detectable levels after 2 months of treatment compared with 10.1 months for those without [68]. Similarly, in a study of 62 patients with EGFR mutated NSCLC, failure to clear plasma EGFR mutated ctDNA after 10 weeks of therapy was a predictor of lower PFS and OS [69]. A prospective cohort of 42 patients similarly demonstrated that ctDNA levels may be a predictive biomarker: EGFR mutant allele fractions provided early prediction of clinical response and significantly correlated with tumour shrinkage at 2 months [70]. The T790M mutation is the main mechanism of resistance in half of EGFR-mutant NSCLC patients treated with first- or second-generation EGFR TKIs. In a study of 117 patients who had acquired resistance to TKI therapy, nearly half of patients who were T790M positive by plasma ctDNA were identified at a median of 2.2 months before clinically progressive disease [71]. In an unselected cohort of 199 patients with NSCLC, monitoring for EGFR mutations allowed for the detection of the T790M acquired mutation as early as 344 days before clinical disease progression [72]. The hypothesis that switching treatment based on detection of resistance mutations in ctDNA instead of on RECIST radiographic criteria would have a positive impact on patient outcomes requires further evaluation. The ongoing randomized phase II APPLE trial aims to validate this hypothesis [73]. The identification of the acquired resistance T790M mutation triggered development of third generation EGFR TKIs that can overcome this mechanism of resistance. Indeed, this acquired EGFR mutation may also be detected in plasma [74, 75]. Jenkins et al. reported a 61% sensitivity for the Cobas plasma test in patients that were T790M-positive in a tissue sample; the authors recommended that in clinical practice, patients with a negative screening plasma test should undergo tissue biopsy testing if feasible [75]. The phase I/II TIGER-X study assessed the efficacy of the third-generation EGFR TKI, rociletinib, in previously treated EGFR-mutant advanced NSCLC patients [76]. Among 548 assessable patients, plasma testing was carried out by therascreen in tissue, BEAMing in plasma, and footprint mutation enrichment NGS in urine. In a retrospective analysis, they found that response rates were comparable regardless of sample type used (33.9% for tissue versus 32.1% for plasma versus 36.7% for urine). Duration of response and PFS were also comparable. When using tissue as a reference (N = 60), the sensitivity of urine or plasma testing for T790M, L858R, and Del19 mutations was: 72% for urine compared to tissue vs. 93% for plasma compared to tissue, 75% vs. 80%, and 67% vs. 80%, for the three EGFR mutations respectively; and the combination of urine and plasma testing improved sensitivity compared with either strategy alone [3]. Furthermore, the combination of urine and plasma testing detected more T790M mutations than tissue testing alone (92% versus 83%). Those with T790M detected by ctDNA (with or without tissue confirmation) had a demonstrable response, showing that the plasma or urine test was a true positive [3]. Efficacy of osimertinib according to T790M positivity in liquid biopsy by BEAMing was presented in a retrospective analysis, with a 30% false negative rate of plasma genotyping [74]. Prospective validation of osimertinib efficacy according to T790M positivity in a blood biopsy has been recently reported, and it confirms the retrospective data. Patient populations in whom T790M was detected in plasma had similar responses to osimertinib to patients when T790M was detected in tissue [26, 74]. Recent work suggests that at the time of clinically overt resistance, the simple presence of any detectable level of T790M in itself is clinically relevant, even with mutant allele fractions as low as 0.1% [26]. It is unknown whether some molecular alterations detected by ultrasensitive NGS methods represent clinically irrelevant sub-clones, especially those with an MAF of 0.01%–0.1% [77]. Additional studies are needed to confirm the minimum biological threshold with clinical relevance to guide treatment decisions. ctDNA testing has also allowed ongoing genomic analysis for patients on third generation EGFR inhibitors. With serial monitoring of patients on osimertinib, Thress et al. identified a novel tertiary EGFR resistance mutation, C797S, which prevents drug binding. In this manner, the resistance mechanism was recognized before the new drug was even approved [78]. While the T790M and C797S mutations in EGFR have been identified as key drivers of resistance, for the majority of patients the eventual progression will be driven by mechanisms outside of these hot-spot mutations in EGFR. The role of NGS-based liquid biopsy analysis is becoming increasingly important in discovering new mechanisms and in clinical assessment of patients to identify resistance to EGFR-targeted TKI. Some studies have reported that the ability to detect ctDNA varies by extent of disease [18, 79, 80]. These results suggest that one may be able to select a subset of patients where plasma ctDNA testing alone may obviate the need for tissue or other complementary testing methods. As in first-line treatment, clearance of plasma EGFR mutation after 6 weeks on osimertinib appears to be associated with improved outcome in T790M positive NSCLC patients, endorsing the dynamic predictive value of ctDNA on treatment [81]. The use of alternate methods, such as collected CTCs when combined with ctDNA, may also improve detection. In an exploratory analysis of 40 patients with EGFR-mutated tumours who progressed on TKI therapy, the use of ctDNA genotyping missed 30% of cases with tissue-identified T790M. However, when combined with genotyping from simultaneously collected CTCs, the two assays together identified all patients with this mutation. Additionally, the T790M mutation was identified in 35% of patients with a negative or indeterminate tissue biopsy when CTC and ctDNA analysis was combined [11]. Discussion Conclusion With the advent of newer ctDNA detection platforms, it is likely that sensitivity can be optimized with standardization and improved techniques. The relatively low sensitivity reported in some studies can likely be attributed to lack of consensus on choice of sample, storage, and technical approaches. Standardization should include the use of plasma over serum, avoidance of heparinized tubes, early centrifugation of tubes, and standardization of cell free extraction methods [28]. The comparative ease of ctDNA sample collection has accelerated clinical decision-making, as well as translational research, such as the identification of the novel acquired EGFR C797S mutation, exemplifying the power of ctDNA to track clonal evolution and tailor treatment. Identification of new mechanisms of acquired resistance in ctDNA in a dynamic manner may help to develop new personalized treatments among EGFR-mutant NSCLC patients. Liquid biopsies may be considered the new standard tool for detecting acquired resistance mutations, with tissue biopsies recommended only in cases without identifiable ctDNA. Following initial approvals by regulatory authorities, plasma-based testing for EGFR mutations is now entering clinical use. With improvements in methods and assays, it is likely that, in the near future, a subset of patients with advanced disease may only require plasma ctDNA testing, possibly with the addition of CTC or urine ctDNA testing, thus avoiding the need for percutaneous biopsy. Funding NR would like to acknowledge the support of The University of Cambridge, Cancer Research UK (grant numbers A11906, A20240), and the European Research Council under the European Union's Seventh Framework Programme (FP/2007-2013) / ERC Grant Agreement n. 337905. Disclosure JWG reports research funding from Clovis Oncology and AstraZeneca and honoraria from Vortex Biosciences. NR is co-founder, shareholder, and CSO of Inivata which commercializes ctDNA testing, an inventor of patent applications describing methods for analysis of cell-free DNA, and has received research funding from AstraZeneca. All remaining authors have declared no conflicts of interest. Key messages Liquid biopsies may be considered an alternative tool for detecting sensitising EGFR mutations and resistance mechanisms on treatment. Techniques and clinical applicability will be discussed in this review. References 1 Lin JJ, Cardarella S, Lydon CA et al.   Five-year survival in EGFR-mutant metastatic lung adenocarcinoma treated with EGFR-TKIs. J Thorac Oncol  2016; 11( 4): 556– 565. Google Scholar CrossRef Search ADS PubMed  2 Paz-Ares L, Tan E-H, O'Byrne K et al.   Afatinib versus gefitinib in patients with EGFR mutation-positive advanced non-small-cell lung cancer: overall survival data from the phase IIb LUX-Lung 7 trial. Ann Oncol  2017; 28( 2): 270– 277. Google Scholar CrossRef Search ADS PubMed  3 Reckamp KL, Melnikova VO, Karlovich C et al.   A highly sensitive and quantitative test platform for detection of NSCLC EGFR mutations in urine and plasma. J Thorac Oncol  2016; 11( 10): 1690– 1700. Google Scholar CrossRef Search ADS PubMed  4 Shingyoji M, Kageyama H, Sakaida T et al.   Detection of epithelial growth factor receptor mutations in cerebrospinal fluid from patients with lung adenocarcinoma suspected of neoplastic meningitis. J Thorac Oncol  2011; 6( 7): 1215– 1220. Google Scholar CrossRef Search ADS PubMed  5 Sasaki S, Yoshioka Y, Ko R et al.   Diagnostic significance of cerebrospinal fluid EGFR mutation analysis for leptomeningeal metastasis in non-small-cell lung cancer patients harboring an active EGFR mutation following gefitinib therapy failure. Respir Investig  2016; 54( 1): 14– 19. Google Scholar CrossRef Search ADS PubMed  6 Novello S, Barlesi F, Califano R et al.   Metastatic non-small-cell lung cancer: ESMO Clinical Practice Guidelines for diagnosis, treatment and follow-up. Ann Oncol  2016; 27(Suppl 5): v1– v27. Google Scholar CrossRef Search ADS   7 Overman MJ, Modak J, Kopetz S et al.   Use of research biopsies in clinical trials: are risks and benefits adequately discussed? J Clin Oncol  2013; 31( 1): 17– 22. Google Scholar CrossRef Search ADS PubMed  8 Lokhandwala T, Bittoni MA, Dann RA et al.   Costs of diagnostic assessment for lung cancer: a medicare claims analysis. Clin Lung Cancer  2017; 18( 1): e27– e34. Google Scholar CrossRef Search ADS PubMed  9 Hasegawa T, Sawa T, Futamura Y et al.   Feasibility of rebiopsy in non-small cell lung cancer treated with epidermal growth factor receptor-tyrosine kinase inhibitors. Intern Med  2015; 54( 16): 1977– 1980. Google Scholar CrossRef Search ADS PubMed  10 Kawamura T, Kenmotsu H, Taira T et al.   Rebiopsy for patients with non-small-cell lung cancer after epidermal growth factor receptor-tyrosine kinase inhibitor failure. Cancer Sci  2016; 107( 7): 1001– 1005. Google Scholar CrossRef Search ADS PubMed  11 Sundaresan TK, Sequist LV, Heymach JV et al.   Detection of T790M, the acquired resistance EGFR mutation, by tumor biopsy versus noninvasive blood-based analyses. Clin Cancer Res  2016; 22( 5): 1103– 1110. Google Scholar CrossRef Search ADS PubMed  12 Mok TS, Wu Y-L, Thongprasert S et al.   Gefitinib or carboplatin-paclitaxel in pulmonary adenocarcinoma. N Engl J Med  2009; 361( 10): 947– 957. Google Scholar CrossRef Search ADS PubMed  13 Kim ES, Hirsh V, Mok T et al.   Gefitinib versus docetaxel in previously treated non-small-cell lung cancer (INTEREST): a randomised phase III trial. Lancet  2008; 372( 9652): 1809– 1818. Google Scholar CrossRef Search ADS PubMed  14 Massard C, Michiels S, Ferté C et al.   High-throughput genomics and clinical outcome in hard-to-treat advanced cancers: results of the MOSCATO 01 Trial. Cancer Discov  2017; 7( 6): 586. Google Scholar CrossRef Search ADS PubMed  15 VanderLaan PA, Yamaguchi N, Folch E et al.   Success and failure rates of tumor genotyping techniques in routine pathological samples with non-small-cell lung cancer. Lung Cancer  2014; 84( 1): 39– 44. Google Scholar CrossRef Search ADS PubMed  16 Folch E, Yamaguchi N, VanderLaan PA et al.   Adequacy of lymph node transbronchial needle aspirates using convex probe endobronchial ultrasound for multiple tumor genotyping techniques in non-small-cell lung cancer. J Thorac Oncol  2013; 8( 11): 1438– 1444. Google Scholar CrossRef Search ADS PubMed  17 Piotrowska Z, Stirling K, Heist R et al.   Variation in mechanisms of acquired resistance among EGFR-mutant NSCLC patients with more than 1 postresistant biopsy. Int J Radiat Oncol Biol Phys  2017; 98( 1): 220. Google Scholar CrossRef Search ADS   18 Sacher AG, Paweletz C, Dahlberg SE et al.   Prospective validation of rapid plasma genotyping for the detection of EGFR and KRAS mutations in advanced lung cancer. JAMA Oncol  2016; 2( 8): 1014– 1022. Google Scholar CrossRef Search ADS PubMed  19 Sands J, Hornberger J. P3.02b-010 urine detection of EGFR T790M mutation in non-small-cell lung cancer: an outcomes and total cost of care analysis. J Thoracic Oncol  12( 1): S1191. CrossRef Search ADS   20 Lawrence MS, Stojanov P, Polak P et al.   Mutational heterogeneity in cancer and the search for new cancer-associated genes. Nature  2013; 499( 7457): 214– 218. Google Scholar CrossRef Search ADS PubMed  21 Jamal-Hanjani M, Wilson GA, McGranahan N et al.   Tracking the evolution of non-small-cell lung cancer. N Engl J Med  2017; 376( 22): 2109– 2121. Google Scholar CrossRef Search ADS PubMed  22 Yachida S, Jones S, Bozic I et al.   Distant metastasis occurs late during the genetic evolution of pancreatic cancer. Nature  2010; 467( 7319): 1114– 1117. Google Scholar CrossRef Search ADS PubMed  23 Zhang J, Fujimoto J, Zhang J et al.   Intratumor heterogeneity in localized lung adenocarcinomas delineated by multiregion sequencing. Science  2014; 346( 6206): 256– 259. Google Scholar CrossRef Search ADS PubMed  24 de Bruin EC, McGranahan N, Mitter R et al.   Spatial and temporal diversity in genomic instability processes defines lung cancer evolution. Science  2014; 346( 6206): 251– 256. Google Scholar CrossRef Search ADS PubMed  25 Chen Z-Y, Zhong W-Z, Zhang X-C et al.   EGFR mutation heterogeneity and the mixed response to EGFR tyrosine kinase inhibitors of lung adenocarcinomas. Oncologist  2012; 17( 7): 978– 985. Google Scholar CrossRef Search ADS PubMed  26 Remon J, Caramella C, Jovelet C et al.   Osimertinib benefit in EGFR-mutant NSCLC patients with T790M-mutation detected by circulating tumour DNA. Ann Oncol  2017; 28( 4): 784– 790. Google Scholar PubMed  27 Leon SA, Shapiro B, Sklaroff DM et al.   Free DNA in the serum of cancer patients and the effect of therapy. Cancer Res  1977; 37( 3): 646– 650. Google Scholar PubMed  28 Wan JCM, Massie C, Garcia-Corbacho J et al.   Liquid biopsies come of age: towards implementation of circulating tumour DNA. Nat Rev Cancer  2017; 17( 4): 223– 238. Google Scholar CrossRef Search ADS PubMed  29 Diaz LAJr, Bardelli A. Liquid biopsies: genotyping circulating tumor DNA. J Clin Oncol  2014; 32( 6): 579– 586. Google Scholar CrossRef Search ADS PubMed  30 Li M, Diehl F, Dressman D et al.   BEAMing up for detection and quantification of rare sequence variants. Nat Methods  2006; 3( 2): 95– 97. Google Scholar CrossRef Search ADS PubMed  31 Diehl F, Schmidt K, Choti MA et al.   Circulating mutant DNA to assess tumor dynamics. Nat Med  2008; 14( 9): 985– 990. Google Scholar CrossRef Search ADS PubMed  32 Vogelstein B, Kinzler KW. Digital PCR. Proc Natl Acad Sci USA  1999; 96( 16): 9236– 9241. Google Scholar CrossRef Search ADS PubMed  33 Dressman D, Yan H, Traverso G et al.   Transforming single DNA molecules into fluorescent magnetic particles for detection and enumeration of genetic variations. Proc Natl Acad Sci USA  2003; 100( 15): 8817– 8822. Google Scholar CrossRef Search ADS PubMed  34 Hindson BJ, Ness KD, Masquelier DA et al.   High-throughput droplet digital PCR system for absolute quantitation of DNA copy number. Anal Chem  2011; 83( 22): 8604– 8610. Google Scholar CrossRef Search ADS PubMed  35 Gale D, Plagnol V, Lawson A et al.   Analytical performance and validation of an enhanced TAm-Seq circulating tumor DNA sequencing assay. 2016; AACR 107th Annual Meeting 2016, Abstract 3639. 36 Forshew T, Murtaza M, Parkinson C et al.   Noninvasive identification and monitoring of cancer mutations by targeted deep sequencing of plasma DNA. Sci Transl Med  2012; 4( 136): 136ra68. Google Scholar CrossRef Search ADS PubMed  37 Lanman RB, Mortimer SA, Zill OA et al.   Analytical and clinical validation of a digital sequencing panel for quantitative, highly accurate evaluation of cell-free circulating tumor DNA. PLoS One  2015; 10( 10): e0140712. Google Scholar CrossRef Search ADS PubMed  38 Zill OA, Mortimer S, Banks KC et al.   Somatic genomic landscape of over 15, 000 patients with advanced-stage cancer from clinical next-generation sequencing analysis of circulating tumor DNA. J Clin Oncol  2016; 34(18 Suppl): LBA11501– LBA11501. Google Scholar CrossRef Search ADS   39 Murtaza M, Dawson S-J, Tsui DWY et al.   Non-invasive analysis of acquired resistance to cancer therapy by sequencing of plasma DNA. Nature  2013; 497( 7447): 108– 112. Google Scholar CrossRef Search ADS PubMed  40 Heitzer E, Ulz P, Belic J et al.   Tumor-associated copy number changes in the circulation of patients with prostate cancer identified through whole-genome sequencing. Genome Med  2013; 5( 4): 30. Google Scholar CrossRef Search ADS PubMed  41 US Food & Drug Administration—cobas EGFR Mutation Test v2. 2016; https://www.accessdata.fda.gov/cdrh_docs/pdf15/P150047c.pdf (4 November 2017, date last accessed). 42 QIAGEN. Therascreen EGFR RGQ PCR Kit Handbook Version 2. August 2016; https://www.qiagen.com/tn/resources/download.aspx? id=d217c9ca-b9f6-4891-8c46-32b71e83d88a&lang=en (4 November 2017, date last accessed). 43 Watanabe K, Fukuhara T, Tsukita Y et al.   EGFR mutation analysis of circulating tumor DNA using an improved PNA-LNA PCR clamp method. Can Respir J  2016; 2016: 5297329. Google Scholar CrossRef Search ADS PubMed  44 Panagene, PNAClamp EGFR Mutation Detection Kit (Ver.2), December 2013; https://secure.eurogentec.com/EGT/files/Panagene/TDS-PN-PNAC-3002.pdf (5 November 2017, date last accessed). 45 Mosko MJ, Nakorchevsky AA, Flores E et al.   Ultrasensitive detection of multiplexed somatic mutations using MALDI-TOF mass spectrometry. J Mol Diagn  2016; 18( 1): 23– 31. Google Scholar CrossRef Search ADS PubMed  46 Kimura H, Kasahara K, Kawaishi M et al.   Detection of epidermal growth factor receptor mutations in serum as a predictor of the response to gefitinib in patients with non-small-cell lung cancer. Clin Cancer Res  2006; 12( 13): 3915– 3921. Google Scholar CrossRef Search ADS PubMed  47 Kimura H, Suminoe M, Kasahara K et al.   Evaluation of epidermal growth factor receptor mutation status in serum DNA as a predictor of response to gefitinib (IRESSA). Br J Cancer  2007; 97( 6): 778– 784. Google Scholar CrossRef Search ADS PubMed  48 Bordi P, Del Re M, Danesi R et al.   Circulating DNA in diagnosis and monitoring EGFR gene mutations in advanced non-small cell lung cancer. Transl Lung Cancer Res  2015; 4( 5): 584– 597. Google Scholar PubMed  49 Han B, Tjulandin S, Hagiwara K et al.   96 odetermining the prevalence of EGFR mutations in Asian and Russian Patients (PTS) with advanced non-small-cell lung cancer (ANSCLC) of adenocarcinoma (ADC) and non-ADC histology: IGNITE study. Ann Oncol  2015; 26( Suppl 1): i29– i30. Google Scholar CrossRef Search ADS   50 Reck M, Hagiwara K, Han B et al.   ctDNA determination of EGFR mutation status in European and Japanese patients with advanced NSCLC: the ASSESS study. J Thorac Oncol  2016; 11( 10): 1682– 1689. Google Scholar CrossRef Search ADS PubMed  51 Qian X, Liu J, Sun Y et al.   Circulating cell-free DNA has a high degree of specificity to detect exon 19 deletions and the single-point substitution mutation L858R in non-small cell lung cancer. Oncotarget  2016; 7( 20): 29154– 29165. Google Scholar CrossRef Search ADS PubMed  52 Oxnard GR, Paweletz CP, Kuang Y et al.   Noninvasive detection of response and resistance in EGFR-mutant lung cancer using quantitative next-generation genotyping of cell-free plasma DNA. Clin Cancer Res  2014; 20( 6): 1698– 1705. Google Scholar CrossRef Search ADS PubMed  53 Luo J, Shen L, Zheng D. Diagnostic value of circulating free DNA for the detection of EGFR mutation status in NSCLC: a systematic review and meta-analysis. Sci Rep  2014; 4: 6269. Google Scholar CrossRef Search ADS PubMed  54 Qiu M, Wang J, Xu Y et al.   Circulating tumor DNA is effective for the detection of EGFR mutation in non-small cell lung cancer: a meta-analysis. Cancer Epidemiol Biomarkers Prev  2015; 24( 1): 206– 212. Google Scholar CrossRef Search ADS PubMed  55 Goto K, Ichinose Y, Ohe Y et al.   Epidermal growth factor receptor mutation status in circulating free DNA in serum: from IPASS, a phase III study of gefitinib or carboplatin/paclitaxel in non-small cell lung cancer. J Thorac Oncol  2012; 7( 1): 115– 121. Google Scholar CrossRef Search ADS PubMed  56 Kuang Y, Rogers A, Yeap BY et al.   Noninvasive detection of EGFR T790M in gefitinib or erlotinib resistant non-small cell lung cancer. Clin Cancer Res  2009; 15( 8): 2630– 2636. Google Scholar CrossRef Search ADS PubMed  57 Maheswaran S, Sequist LV, Nagrath S et al.   Detection of mutations in EGFR in circulating lung-cancer cells. N Engl J Med  2008; 359( 4): 366– 377. Google Scholar CrossRef Search ADS PubMed  58 Sharma SV, Bell DW, Settleman J et al.   Epidermal growth factor receptor mutations in lung cancer. Nat Rev Cancer  2007; 7( 3): 169– 181. Google Scholar CrossRef Search ADS PubMed  59 Yung TKF, Chan KCA, Mok TSK et al.   Single-molecule detection of epidermal growth factor receptor mutations in plasma by microfluidics digital PCR in non-small cell lung cancer patients. Clin Cancer Res  2009; 15( 6): 2076– 2084. Google Scholar CrossRef Search ADS PubMed  60 Karachaliou N, Mayo-de las Casas C, Queralt C et al.   Association of EGFR L858R mutation in circulating free DNA with survival in the EURTAC trial. JAMA Oncol  2015; 1( 2): 149– 157. Google Scholar CrossRef Search ADS PubMed  61 Mok T, Wu Y-L, Lee JS et al.   Detection and dynamic changes of EGFR mutations from circulating tumor DNA as a predictor of survival outcomes in NSCLC patients treated with first-line intercalated erlotinib and chemotherapy. Clin Cancer Res  2015; 21( 14): 3196– 3203. Google Scholar CrossRef Search ADS PubMed  62 Douillard J-Y, Ostoros G, Cobo M et al.   Gefitinib treatment in EGFR mutated Caucasian NSCLC: circulating-free tumor DNA as a surrogate for determination of EGFR status. J Thorac Oncol  2014; 9( 9): 1345– 1353. Google Scholar CrossRef Search ADS PubMed  63 Besse B, Remon J, Lacroix L et al.   Evaluation of liquid biopsies for molecular profiling in untreated patients with stage III/IV non-small cell lung cancer (NSCLC). J Clin Oncol  2017; 35(15 Suppl): 11540– 11540. 64 Remon J, Besse B, Lacroix L et al.   Evaluation of liquid biopsies for molecular profiling in patients with advanced non-small cell lung cancer (NSCLC) in the relapse treatment setting. J Clin Oncol  2017; 35(Suppl; abstr 11532). 65 Bradbury PA, Ding K, Ellis PM et al.   Evaluation of stored liquid biopsies for molecular profiling in patients with non-small cell lung cancer (NSCLC). J Clin Oncol  2017; 35(Suppl; abstr 9024). 66 Guibert NM, Paweletz C, Hu Y et al.   Early detection of competing resistance mutations using plasma next-generation sequencing (NGS) in patients (pts) with EGFR-mutant NSCLC treated with osimertinib. J Clin Oncol  2017; 35(Suppl; abstr 11529). 67 Jovelet C, Madic J, Remon J et al.   Crystal digital droplet PCR for detection and quantification of circulating EGFR sensitizing and resistance mutations in advanced non-small cell lung cancer. PLoS One  2017; 12( 8): e0183319. Google Scholar CrossRef Search ADS PubMed  68 Lee JY, Qing X, Xiumin W et al.   Longitudinal monitoring of EGFR mutations in plasma predicts outcomes of NSCLC patients treated with EGFR TKIs: Korean Lung Cancer Consortium (KLCC-12-02). Oncotarget  2016; 7( 6): 6984– 6993. Google Scholar PubMed  69 Tseng J-S, Yang T-Y, Tsai C-R et al.   Dynamic plasma EGFR mutation status as a predictor of EGFR-TKI efficacy in patients with EGFR-mutant lung adenocarcinoma. J Thorac Oncol  2015; 10( 4): 603– 610. Google Scholar CrossRef Search ADS PubMed  70 Marchetti A, Palma JF, Felicioni L et al.   Early prediction of response to tyrosine kinase inhibitors by quantification of EGFR mutations in plasma of NSCLC patients. J Thorac Oncol  2015; 10( 10): 1437– 1443. Google Scholar CrossRef Search ADS PubMed  71 Zheng D, Ye X, Zhang MZ et al.   Plasma EGFR T790M ctDNA status is associated with clinical outcome in advanced NSCLC patients with acquired EGFR-TKI resistance. Sci Rep  2016; 6: 20913. Google Scholar CrossRef Search ADS PubMed  72 Sorensen BS, Wu L, Wei W et al.   Monitoring of epidermal growth factor receptor tyrosine kinase inhibitor-sensitizing and resistance mutations in the plasma DNA of patients with advanced non-small cell lung cancer during treatment with erlotinib. Cancer  2014; 120( 24): 3896– 3901. Google Scholar CrossRef Search ADS PubMed  73 Remon J, Menis J, Hasan B et al.   The APPLE Trial: feasibility and activity of AZD9291 (osimertinib) treatment on positive PLasma T790M in EGFR-mutant NSCLC patients. EORTC 1613. Clin Lung Cancer  2017; 18( 5): 583– 588. Google Scholar CrossRef Search ADS PubMed  74 Oxnard GR, Thress KS, Alden RS et al.   Association between plasma genotyping and outcomes of treatment with osimertinib (AZD9291) in advanced non-small-cell lung cancer. J Clin Oncol  2016; 34( 28): 3375– 3382. Google Scholar CrossRef Search ADS PubMed  75 Jenkins S, Yang JC-H, Ramalingam SS et al.   Plasma ctDNA analysis for detection of the EGFR T790M mutation in patients with advanced non-small-cell lung cancer. J Thorac Oncol  2017; 12( 7): 1061– 1070. Google Scholar CrossRef Search ADS PubMed  76 Yu HA, Arcila ME, Rekhtman N et al.   Analysis of tumor specimens at the time of acquired resistance to EGFR-TKI therapy in 155 patients with EGFR-mutant lung cancers. Clin Cancer Res  2013; 19( 8): 2240– 2247. Google Scholar CrossRef Search ADS PubMed  77 Wang W, Karampini E, Correia LL et al.   Digital detection of T790M-yes or no to an ultrasensitive assay. Transl Lung Cancer Res  2016; 5( 3): 338– 342. Google Scholar CrossRef Search ADS PubMed  78 Thress KS, Paweletz CP, Felip E et al.   Acquired EGFR C797S mediates resistance to AZD9291 in advanced non-small cell lung cancer harboring EGFR T790M. Nat Med  2015; 21( 6): 560– 562. Google Scholar CrossRef Search ADS PubMed  79 Newman AM, Bratman SV, To J et al.   An ultrasensitive method for quantitating circulating tumor DNA with broad patient coverage. Nat Med  2014; 20( 5): 548– 554. Google Scholar CrossRef Search ADS PubMed  80 Zhao X, Han R-B, Zhao J et al.   Comparison of epidermal growth factor receptor mutation statuses in tissue and plasma in stage I-IV non-small cell lung cancer patients. Respiration  2013; 85( 2): 119– 125. Google Scholar CrossRef Search ADS PubMed  81 Thress KS, Markovets A, Barrett JC et al.   Complete clearance of plasma EGFR mutations as a predictor of outcome on osimertinib in the AURA trial. J Clin Oncol  2017; 35(Suppl abstract 9018). © The Author(s) 2018. Published by Oxford University Press on behalf of the European Society for Medical Oncology. All rights reserved. For Permissions, please email: journals.permissions@oup.com. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Annals of Oncology Oxford University Press

Are liquid biopsies a surrogate for tissue EGFR testing?

Loading next page...
 
/lp/ou_press/are-liquid-biopsies-a-surrogate-for-tissue-egfr-testing-qnSHTYZe0a
Publisher
Oxford University Press
Copyright
© The Author(s) 2018. Published by Oxford University Press on behalf of the European Society for Medical Oncology. All rights reserved. For Permissions, please email: journals.permissions@oup.com.
ISSN
0923-7534
eISSN
1569-8041
D.O.I.
10.1093/annonc/mdx706
Publisher site
See Article on Publisher Site

Abstract

Abstract Molecular profiling has changed the treatment landscape in advanced non-small-cell lung cancer. Accurately identifying the tumours that harbour sensitizing EGFR mutations, the most common targetable molecular alteration, as well as those with acquired resistance mutations (e.g. T790M) on treatment is a high clinical priority. The current clinical gold standard is genotyping of tumour specimens. However, the practical utility of this approach is limited by the lack of available tissue and the potential complications associated with biopsies. With the advent of newer sequencing assays, it has become feasible to assess tumour genomics via a blood sample, termed a ‘liquid biopsy’. In this review, we summarize the available techniques for liquid biopsies and their applicability for detecting sensitizing and resistance EGFR mutations and how these results may be used for making treatment decisions. cfDNA, ctDNA, lung cancer, EGFR, T790M, liquid biopsy Introduction Most advanced non-small-cell lung cancers (NSCLC) with mutations in the epidermal growth factor receptor (EGFR) are highly sensitive to EGFR tyrosine kinase inhibitors (TKIs). With targeted therapies, median overall survival (OS) has improved from 8–12 to 20–30 months for this patient subset [1]. Accurately identifying the tumours that harbour sensitizing EGFR mutations as well as those with acquired resistance mutations (T790M) that can be targeted by newer third-generation TKIs is a high clinical priority, with associated increases in OS of up to 46 months with personalized treatment at progression [2]. The current clinical gold-standard is genotyping of tumour biopsies. However, with the advent of newer sequencing assays, it has become feasible to assess tumour genomics via a blood sample, termed a ‘liquid biopsy’. Recent data suggest that circulating tumour DNA (ctDNA), that can be used for molecular profiles, derived from NSCLC can be detected with high sensitivity in urine as well as in plasma [3] and cerebral spinal fluid [4, 5], enabling complementary modes of tissue and liquid biopsies in this population. EGFR mutation profile by liquid biopsy is now accepted for making treatment decisions among NSCLC patients, and liquid biopsy tests have earned limited approvals in Europe, China, Japan, and the USA owing to several distinct advantages over tissue biopsy [6]. Tissue biopsy Lack of available tissue for performing molecular profiling, the location or size of the tumour, and the risk of complications (∼17% of adverse events with image guide-biopsies [7] and risk of pneumothorax requiring chest tube in up to 14% of patients [8]), are serious limitations for performing biopsies of NSCLC tumours. In EGFR-mutant NSCLC patients with acquired resistance to EGFR TKIs, re-biopsies are feasible in only half of the patients [9] as a consequence of inaccessible tumour sites, patient refusal, or physician discretion [10]. In prospective trials, the percentage of available tissue samples for EGFR genotyping can vary greatly; even when successfully obtained, ∼12%–30% of samples were not sufficient for genotyping [11–14]. In addition, after obtaining sufficient samples via invasive procedures, EGFR genotyping fails in ∼5%–10% of tests [15, 16]. Moreover, single site biopsies may not provide a representative profile of the overall predominant resistance mechanisms for a given patient [17]. Delays in tissue biopsy often occur as a result of scheduling, specimen delivery to molecular diagnostics, and laboratory processing time. In a prospective study of EGFR genotyping in advanced lung cancer patients, the median turnaround time for tissue biopsy was 12 days for new diagnosis of NSCLC patients and 27 days for those tested for acquired resistance. In contrast, the median time was 3 days for plasma based testing using an in-house assay for EGFR and KRAS hotspots [18]. There is also a significant difference in cost. An analysis of 8979 Medicare patients undergoing diagnostic assessment for lung cancer found that the median cost of a biopsy was $2235. However, this cost increased to a median of $14 824 if an adverse event occurred [8]. These are likely underestimates, as costs from Medicare are generally lower than from other payers. A study using simulation analytics found that a liquid-based alternative for EGFR genotyping could result in substantial financial savings [19]. There may also be advantages to a liquid biopsy that go beyond avoiding the risks and financial costs of a tumour biopsy (Table 1). NSCLCs are amongst the cancer types with the highest rates of point mutations, harbouring hundreds of mutations in their exome [20], and advanced cancers are highly heterogeneous [21]. Tumour heterogeneity results from numerous genomic alterations, catastrophic events, and clonal evolution during division. Dividing cells in the tumour acquire new mutations, creating sub-clones that differ from their founder cells so that geographically separated regions in the same tumour will generally also be more genomically distinct [22]. Seeding of varying sub-clones to other organs leads to further divergent evolution. In NSCLC, multi-region whole exome sequencing studies confirm that tumours show large spatial and temporal diversity in mutational profiles [23, 24]. Table 1. Concerns with tissue biopsy that are potentially addressed by liquid biopsy Concern  Advantage of liquid biopsy  Risk of complication  Least invasive or noninvasive.  High cost  Removes costs associated with procedure and complications.  Difficult tumor site  Blood, cerebrospinal fluid, or urine can be easily collected.  Specimens quality varies  More control over quantity and quality of specimen.  Lengthy turnaround time  Turnaround time is significantly less.  Tumor heterogeneity  Better captures the molecular state of tumours and metastases.  Impractical for disease monitoring  Repeat biopsies are easy and provide a ‘real-time snapshot’ to monitor response to therapy and development of resistance.  Concern  Advantage of liquid biopsy  Risk of complication  Least invasive or noninvasive.  High cost  Removes costs associated with procedure and complications.  Difficult tumor site  Blood, cerebrospinal fluid, or urine can be easily collected.  Specimens quality varies  More control over quantity and quality of specimen.  Lengthy turnaround time  Turnaround time is significantly less.  Tumor heterogeneity  Better captures the molecular state of tumours and metastases.  Impractical for disease monitoring  Repeat biopsies are easy and provide a ‘real-time snapshot’ to monitor response to therapy and development of resistance.  A tissue biopsy may therefore only demonstrate a minority of the genetic aberrations present in the tumour, as this single site may not be representative of intra-tumour heterogeneity nor interlesional heterogeneity (Figure 1). Prospective analysis of EGFR mutation status in advanced NSCLC patients who underwent multiple biopsies showed frequent changes in their mechanism of acquired resistance and T790M status (40% gained or lost the mutation) [17]. This complex and dynamic quality suggests that single site biopsies may not provide a representative profile of the overall predominant resistance mechanisms for a given patient. Figure 1. View largeDownload slide Liquid biopsy for discovery of actionable EGFR mutations, monitoring response to therapy, and detecting development of TKI resistance. Liquid biopsy carried out at diagnosis guides treatment with a TKI to target actionable mutations (A). ctDNA levels can correlate to response to TKIs (B). At progression, as ctDNA levels rise, one would ideally obtain a repeat tissue biopsy to guide next treatment decision (C). However, even when a tissue biopsy is carried out, due to tumour heterogeneity, it may miss mutations due to sampling bias (D). Figure 1. View largeDownload slide Liquid biopsy for discovery of actionable EGFR mutations, monitoring response to therapy, and detecting development of TKI resistance. Liquid biopsy carried out at diagnosis guides treatment with a TKI to target actionable mutations (A). ctDNA levels can correlate to response to TKIs (B). At progression, as ctDNA levels rise, one would ideally obtain a repeat tissue biopsy to guide next treatment decision (C). However, even when a tissue biopsy is carried out, due to tumour heterogeneity, it may miss mutations due to sampling bias (D). A study using direct DNA sequencing to detect EGFR mutations in 180 Asian patients with matched samples from primary tumour and distant sites found that nearly a fourth of those with multiple pulmonary nodules had discordant EGFR status [25]. These data suggest that re-biopsy or multiple biopsies would be advantageous, yet, as already described, this is impractical. Liquid biopsies based on ctDNA analysis are therefore effective surrogate samples for tumour molecular analysis, and also as potential dynamic markers for monitoring the efficacy of EGFR TKIs and early detection of resistance mutations [26]. ctDNA and techniques In 1977, the same year the eponymous Sanger sequencing method was described, Leon et al. reported elevated levels of circulating cell-free DNA (cfDNA) in the serum of cancer patients [27]. The fraction of cfDNA in serum or plasma derived from tumour cells, termed ctDNA, is hypothesized to originate from tumour secretion, necrosis, and apoptosis [28]. Circulating tumour cells (CTCs), exosomes, or platelets can also provide DNA or RNA for isolation. Plasma is preferred in the detection and characterization of ctDNA because of lower background levels of wild-type DNA [28]. The half-life of cfDNA in circulation has been estimated to be between 16 min and 2.5 h, which makes it akin to a ‘real-time snapshot’ of disease burden that can be monitored [28]. ctDNA can be specifically identified in cfDNA by the presence of somatic mutations. Methods for identifying these variations range from large-scale (up to whole genome) sequencing to analysis of individual mutations using polymerase chain reaction (PCR) with specific primers, often with complementary probes for known aberrations (Figure 2). The proportion of ctDNA at a specific locus is usually reported as mutant allele concentration (copies/ml) or mutant allele fraction (%). ctDNA often represents <1% of cfDNA in a sample and, consequently, reliable detection across cases requires sensitive analysis methods [29]. The sensitivity of sequencing is technically limited by the error rate of DNA polymerase, often considered to be in the range of 0.01% [30]. An additional limitation on detection of rare alleles results from Poisson sampling statistics, whereby an allele that is present at a very low fraction may not be present in a given blood sample. Therefore, it has been suggested that individual mutations present in ctDNA at levels <0.01% mutant allele fraction would be undetectable [31]. The sensitivity of classical Sanger (chain-termination) method to detect variant alleles is >10% [29], inadequate for wide clinical use where ctDNA levels can be far lower. In 1999, Kenneth Kinzler and Bert Vogelstein described digital PCR (dPCR) to identify rare cancer mutations [32] (Figure 2A1). The most common application of dPCR now relies on generating thousands to millions of water droplets in an oil emulsion rather than physically separated chambers, and is termed droplet dPCR (ddPCR), first exemplified by the same group in a technique called BEAMing (beads, emulsion, amplification, and magnetics) [33]. Both dPCR and ddPCR have a rapid turnaround time and are useful in detecting and quantifying recurrent hot-spot mutations or resistance mutations, and can, in principle, achieve a best-demonstrated detection below 0.001% [34]. In practice, the sensitivity will be limited by the amount of cfDNA analysed, which varies per patient and disease state from a few thousand to hundreds of thousands of copies, with a median around 15 000 amplifiable copies per 10 ml blood sample [35]. Figure 2. View largeDownload slide Common techniques used for ctDNA detection. Methods for studying ctDNA range from single locus to the much broader whole genome sequencing. (A1) Digital polymerase chain reaction (dPCR), and digital droplet PCR (ddPCR), are based on the concept that amplifying single molecules would be most sensitive if DNA templates could be diluted to the point where each reaction chamber contained either no amplified product or PCR product from a single amplification. This eliminates the noise of weak signals seen in analogue sequencing. (A2) The amplification refractory mutation system (ARMS) uses PCR amplification with complementary primers and fluorescent dye probes, relying on the terminal 3′ nucleotides of PCR primer being allele specific. The Scorpion ARMS (SARMS) method uses ‘scorpion’ probe sequences held in hairpin confirmation and then combined with ARMS along with fluorophores, and their incorporation is quantified by real-time PCR. Therefore, a fully matched primer can selectively amplify mutated sequences. (B) Targeted sequencing allows one to focus on specific regions of interest or alternatively, perform whole exome sequencing. Two main approaches to targeted sequencing include (B1) amplicon-based, using PCR amplification with primers for selected regions to generate targeted sequencing libraries; or (B2) ligation-based library preparation, followed by enrichment for targeted regions using hybridization to sequence baits (hybrid-capture). (C) Whole genome sequencing does not require a priori knowledge of genetic mutations and can be useful if ctDNA levels are high. Figure 2. View largeDownload slide Common techniques used for ctDNA detection. Methods for studying ctDNA range from single locus to the much broader whole genome sequencing. (A1) Digital polymerase chain reaction (dPCR), and digital droplet PCR (ddPCR), are based on the concept that amplifying single molecules would be most sensitive if DNA templates could be diluted to the point where each reaction chamber contained either no amplified product or PCR product from a single amplification. This eliminates the noise of weak signals seen in analogue sequencing. (A2) The amplification refractory mutation system (ARMS) uses PCR amplification with complementary primers and fluorescent dye probes, relying on the terminal 3′ nucleotides of PCR primer being allele specific. The Scorpion ARMS (SARMS) method uses ‘scorpion’ probe sequences held in hairpin confirmation and then combined with ARMS along with fluorophores, and their incorporation is quantified by real-time PCR. Therefore, a fully matched primer can selectively amplify mutated sequences. (B) Targeted sequencing allows one to focus on specific regions of interest or alternatively, perform whole exome sequencing. Two main approaches to targeted sequencing include (B1) amplicon-based, using PCR amplification with primers for selected regions to generate targeted sequencing libraries; or (B2) ligation-based library preparation, followed by enrichment for targeted regions using hybridization to sequence baits (hybrid-capture). (C) Whole genome sequencing does not require a priori knowledge of genetic mutations and can be useful if ctDNA levels are high. Methods that use next-generation sequencing (NGS), and generate sequencing libraries across selected regions of the genome, are commonly referred to as ‘targeted sequencing’ (Figure 2B). Forshew et al. described the first implementation of targeted sequencing to detect rare cancer mutations in ctDNA. Using a two-step amplification process to generate tagged (barcoded) amplicons, they were able to identify mutations across a gene panel at allele fractions as low as 2% in plasma, with sensitivity and specificity >97% [36]. This method, termed TAm-Seq (for Tagged-Amplicon Sequencing), could also identify predefined individual mutations (e.g. hot-spot mutations or previously characterized patient-specific alterations) at allele fractions as low as 0.14%. An enhanced version of TAm-Seq, eTAm-Seq, was developed to identify mutations in hot-spots (including EGFR) and entire exons across a gene panel, and an assay covering a panel of 35 genes was described in 2016, with detection of mutations at allele fractions of 0.06%–0.08% at nearly 40% sensitivity, and sensitivity of 90% for mutations present at allele fractions of 0.25% and greater [35]. Alternative methods use hybrid capture-based deep sequencing, such as the Guardant360 assay, which covers regions in 70 genes (including the actionable EGFR mutations), and uses a library of individually tagged cfDNA molecules to reduce false positives [37]. In a study utilizing the Guardant360 assay to profile ctDNA from patients with different cancer types, a subset with matched tissue tests were compared with results from the tissue sequencing project, TCGA, and high levels of accuracy were observed when blood and tumour were collected within 6 months of each other [38]. This demonstrates the potential of NGS technologies. Wider-scale analysis such as whole-exome sequencing (Figure 2C) and shallow whole-genome sequencing can be carried out to detect de novo mutations and chromosomal aberrations and study clonal evolution when the ratio of ctDNA to cfDNA is relatively high (∼5% or greater) [39, 40]. It is likely that future ctDNA studies will rely on both ultrasensitive targeted detection methods as well as wider-scale analysis at different stages of diagnosis, monitoring, and treatment. NGS-based methods allow for de novo sequencing of key regions in EGFR to identify mutations at low allele fractions. While these are now emerging as practical tools, the most widely used clinical tests have used PCR-based technologies to detect mutations in EGFR hot spots. Kits approved or endorsed by the US Food and Drug Administration (FDA) and the European Medicines Agency (EMA) have employed the Scorpion Amplification Refractory Mutation System (SARMS) (Figure 2A2). Clinical assays include the Cobas EGFR assay, which can detect mutations in exons 18, 19, 20, and 21 with at least 25–100 copies of the mutation per millilitres [41]. The therascreen SARMS kit has a median limit of detection of 1.42% [42]. The peptide nucleic acid-locked nucleic acid (PNA-LNA) clamp method exploits the tighter binding affinity of PNA and LNA probes to DNA sequences rather than to the DNA duplex, which prevents amplification of wild-type alleles, and has been employed in EGFR detection [43]. The commercially available PNAClamp technology has a stated limit of detection of <1% for EGFR mutations [44]. A range of other methods have been developed for detection of EGFR mutations at low allele fractions. A method that uses multiplex PCR amplification and detection based on mass spectroscopy can achieve detection of key hot-spot variants with as low as 0.1% minor allele frequency [45]. These alternative methods have been utilized in studies of EGFR mutations, but to the best of our knowledge have so far not been widely used. Clinical use Detection of activating EGFR mutations to guide initial treatment with TKIs The first trial to perform a pairwise comparison of tissue and liquid biopsy for EGFR mutations was conducted as part of a phase II study of 27 Japanese NSCLC patients in 2006 [46] (Table 2). When direct sequencing was used, serum EGFR mutational status was not correlated with response to gefitinib, but when SARMS was used to detect serum EGFR mutation there was a statistical correlation with treatment response. Direct sequencing, therefore, seemed unsatisfactory for detection of EGFR mutations in serum. In pairs of tumour and serum samples (n = 11), the EGFR mutation status in the tumours was consistent with those in the serum in 72.7% of the paired samples by SARMS. The following year, the same group tested 42 patients, and found an even higher concordance rate of 92.9%. They reported a strong correlation between either tissue or serum EGFR mutation status and response to gefitinib as well as progression-free survival (PFS) [47]. Table 2. Summary of selected clinical studies of liquid biopsy to detect EGFR mutations Clinical use  References  Summary  Initial detection of EGFR mutations to guide treatment  [46, 47, 61]  EGFR mutation status by liquid biopsy strongly correlates with tissue biopsy status.  [47, 62, 60]  EGFR mutation status correlates to TKI response, PFS, as well as OS.  [55, 56, 57]  Early studies showed mixed results in detecting EGFR ctDNA.  [58, 59, 61]  Improved methodology has greatly improved sensitivity and specificity of testing.  [48, 49, 50]  Results from two large multi-national trials suggest that optimization, validation, and standardization are needed to improve performance.  [51, 53, 54]  Meta-analyses show high diagnostic accuracy of ctDNA for EGFR mutation testing.  ctDNA as a marker of efficacy  [52, 71, 72]  ctDNA levels may predict progression before clinical progression.  [61, 68, 69]  Clearance of ctDNA may predict response and prognosis.  Identifying resistance mutations in patients progressing on TKI  [74, 75, 76]  Detection of the T790M mutation can be accurately carried out via liquid biopsy, and the presence of any detectable T790M ctDNA may be clinically relevant.  [24, 73, 74]  T790M status by liquid biopsy correlates well to response to third-generation TKIs.  Clinical use  References  Summary  Initial detection of EGFR mutations to guide treatment  [46, 47, 61]  EGFR mutation status by liquid biopsy strongly correlates with tissue biopsy status.  [47, 62, 60]  EGFR mutation status correlates to TKI response, PFS, as well as OS.  [55, 56, 57]  Early studies showed mixed results in detecting EGFR ctDNA.  [58, 59, 61]  Improved methodology has greatly improved sensitivity and specificity of testing.  [48, 49, 50]  Results from two large multi-national trials suggest that optimization, validation, and standardization are needed to improve performance.  [51, 53, 54]  Meta-analyses show high diagnostic accuracy of ctDNA for EGFR mutation testing.  ctDNA as a marker of efficacy  [52, 71, 72]  ctDNA levels may predict progression before clinical progression.  [61, 68, 69]  Clearance of ctDNA may predict response and prognosis.  Identifying resistance mutations in patients progressing on TKI  [74, 75, 76]  Detection of the T790M mutation can be accurately carried out via liquid biopsy, and the presence of any detectable T790M ctDNA may be clinically relevant.  [24, 73, 74]  T790M status by liquid biopsy correlates well to response to third-generation TKIs.  EGFR, epidermal growth factor receptor; TKI, tyrosine kinase inhibitor; PFS, progression-free survival; OS, overall survival; ctDNA, circulating tumor DNA. Two large multinational trials, IGNITE and ASSESS, have studied the efficacy of EGFR mutation testing in the real-world setting [48]. In the IGNITE trial, 3382 patients were enrolled and plasma and tissue data were available for 2581 patients. The techniques used for EGFR testing were not centralized and had limited sensitivity. The plasma sensitivity for detection of tissue-confirmed EGFR mutations was <50%. The ASSESS trial had data available from 1162 matched tissue and plasma samples and found a concordance rate of 89.1%, with a sensitivity for detecting mutations in plasma of 46% and specificity of 97.4% [49]. The low sensitivity suggests that wide implementation requires optimization, standardization, and validation. In fact, the study found large variation in the sensitivity of plasma testing amongst countries, from 36% to 100%. In subgroup analysis of only therascreen (an SARMS-based assay), the concordance rate was 95%, sensitivity increased to 73% and the specificity was 99% with a NPV of 95%. Similarly, subgroup analysis looking at Cobas EGFR testing alone found a 96% concordance rate, 75% sensitivity, 100% specificity, and NPV of 95% [50]. A meta-analysis examining 27 studies conducted between the years 2007 and 2015, including over 4000 patients, generated a receiver operating characteristic (ROC) curve, and found the area under the curve (AUC) to be 0.92, demonstrating high diagnostic accuracy (AUC > 0.9), with positive likelihood ratio >10. They found a pooled sensitivity of 64% for detection of the EGFR exon 19 deletions and 57% for L858R substitution mutations with 99% specificity for both [51]. As Oxnard et al. point out, in the context of a relatively uncommon mutation, specificity is more important than sensitivity. A sacrifice in test sensitivity with an accompanying small increase in specificity results in a dramatic increase in positive predictive value. They give the example that to detect an EGFR mutation present in 8.6% of patients, with a 10% decrease in sensitivity from 80% to 70%, if the specificity of 95% increased to 99% then the PPV would more than double, from 43% to 87% [52]. Two separate meta-analyses have found very similar results for pooled specificity (>93%), pooled sensitivity (>60%), and AUC (0.93 and 0.91) for EGFR analysis by ctDNA. These authors also concluded that the high diagnostic accuracy of ctDNA makes it a suitable screening test for EGFR mutations [53, 54]. The phase III IPASS trial, comparing gefitinib to carboplatin–paclitaxel amongst 233 Japanese patients with advanced NSCLC, retrospectively reported EGFR mutations in ctDNA detected by SARMS in 23.7% of patients versus 61% from tumour samples [55]. There was an unacceptably high false negative rate of 56.9%, which may be attributed to the use of serum rather than plasma, DNA extraction techniques, and a less refined SARMS kit. Another study of 54 patients with known clinical response to gefitinib or erlotinib found that only ∼47% of EGFR mutations were detected when an SARMS-based technique was used [56]. A similar study using SARMS detected only 39% of EGFR mutations in plasma ctDNA from 18 patients with known EGFR mutations [57]. These mixed results from earlier studies highlighted the challenges of ctDNA analysis, but also its great potential. With more data on pre-analytical variables and improvements in sample collection processes, alongside more robust, and sensitive methods for mutation detection, more recent studies have shown consistently greater sensitivity for ctDNA detection in a range of clinical applications [28]. In a first application of dPCR for analysis of EGFR in peripheral blood samples from NSCLC patients, Yung et al. developed probes to detect the exon Del19 and L858R at exon 21, which account for >85% of all clinically relevant EGFR mutations [58]. Their study of 35 patients reported a sensitivity of 92% and a perfect specificity [59]. More recently, Sacher et al. prospectively validated a ddPCR technique in a study of 180 patients to detect EGFR mutations at diagnosis or relapse. They reported that ddPCR for plasma genotyping of sensitizing EGFR mutations also had 100% specificity and positive predictive value given no false positives, and a sensitivity of 69%–80% [18]. The phase III EURTAC trial, comparing upfront erlotinib to platinum-based chemotherapy in 173 EGFR-mutant NSCLC patients, assessed the feasibility for testing EGFR mutations in a liquid biopsy by real-time PCR TaqMan assay as a surrogate of tissue testing, and compared outcomes according to detection of mutation in tissue versus liquid biopsy [60]. This study demonstrated that tumour and peripheral EGFR mutation status had similar ability to predict OS and PFS [60]. Similarly, to prospectively validate the use of blood-based EGFR testing, Mok et al. demonstrated high performance characteristics of the Cobas EGFR Blood test [61]. The phase III FASTACT-2 trial randomized advanced NSCLC patients to platinum-gemcitabine first-line chemotherapy intercalated with either erlotinib or placebo followed by maintenance treatment with erlotinib or placebo. Cobas EGFR testing had a high tissue-blood concordance of 88% with a sensitivity of 75%, specificity of 96%, and PPV of 94% [61]. The phase IV IFUM trial assessed the efficacy of gefitinib in Caucasian NSCLC patients and the clinical utility of the therascreen EGFR RGQ PCR kit for plasma ctDNA was tested. Concordance in mutation status amongst 652 matched samples was 94.3%, sensitivity was 65.7%, and specificity was 99.8%. The trial reported that patients’ outcome was similar independent of EGFR mutation being detected in plasma or in tissue [response rate (RR) 69.8% versus 76.9%, and PFS, 10.2 versus 9.7 months]. Of significance, 12 patients without available tissue samples for genotyping were discovered to have EGFR mutations via plasma testing [62]. In January 2015, the EMA granted marketing approval to the therascreen assay. Given the 34% false negative rate in the IFUM trial, they specified that the test should only be used for patients without a tumour sample. In June 2016, the Cobas EGFR mutation test v2 was approved by the FDA for the detection of exon 19 deletions and the L858 substitution mutation in plasma. In September 2016, this was extended to cover also the T790M resistance mutation. Given that a relatively low number, 76.7% of patients, who had a detectable EGFR mutation by tissue biopsy also had a detectable mutation in plasma, the FDA specified that those with negative plasma testing should be retested with tissue sampling. Recently, NGS-based ctDNA assays have been introduced as an important tool for molecular profiling in untreated NSCLC patients [63], detecting clinically relevant and actionable mutations in up to 23% of patients when tissue was unavailable, and also in previously treated NSCLC patients [64, 65]. The assay used, sensitivity of the test, and sample collection time point should be kept in mind when interpreting concordance between plasma and tumour analyses [65]. Indeed, liquid biopsies can be used longitudinally as a non-invasive approach to monitor patient treatment outcome and may provide a surrogate for response evaluation by radiographic RECIST assessment [63]. Also, recently ctDNA has been demonstrated as an important tool for detecting the emergence of resistance mutations on treatment [26, 66, 67]. Further analysis in well-defined patient populations is underway to provide supporting data for the clinical utility of such assays. ctDNA as a dynamic marker of efficacy and for the identification of resistance mutations in patients relapsing on targeted therapies ctDNA levels may predict progression before standard imaging progression by RECIST. The first study to demonstrate this in NSCLC found that serial plasma genotyping could detect resistance up to 16 weeks before radiographic progression [52]. Analysis of the FASTACT-2 data found that at cycle three, if a cfDNA sample was EGFR mutant positive, RR was only 33% compared with 66% for those who cleared their EGFR mutation positivity [61]. In a study of 81 patients with NSCLC treated with an EGFR-directed TKI therapy, PFS was 6.3 months for those with detectable levels after 2 months of treatment compared with 10.1 months for those without [68]. Similarly, in a study of 62 patients with EGFR mutated NSCLC, failure to clear plasma EGFR mutated ctDNA after 10 weeks of therapy was a predictor of lower PFS and OS [69]. A prospective cohort of 42 patients similarly demonstrated that ctDNA levels may be a predictive biomarker: EGFR mutant allele fractions provided early prediction of clinical response and significantly correlated with tumour shrinkage at 2 months [70]. The T790M mutation is the main mechanism of resistance in half of EGFR-mutant NSCLC patients treated with first- or second-generation EGFR TKIs. In a study of 117 patients who had acquired resistance to TKI therapy, nearly half of patients who were T790M positive by plasma ctDNA were identified at a median of 2.2 months before clinically progressive disease [71]. In an unselected cohort of 199 patients with NSCLC, monitoring for EGFR mutations allowed for the detection of the T790M acquired mutation as early as 344 days before clinical disease progression [72]. The hypothesis that switching treatment based on detection of resistance mutations in ctDNA instead of on RECIST radiographic criteria would have a positive impact on patient outcomes requires further evaluation. The ongoing randomized phase II APPLE trial aims to validate this hypothesis [73]. The identification of the acquired resistance T790M mutation triggered development of third generation EGFR TKIs that can overcome this mechanism of resistance. Indeed, this acquired EGFR mutation may also be detected in plasma [74, 75]. Jenkins et al. reported a 61% sensitivity for the Cobas plasma test in patients that were T790M-positive in a tissue sample; the authors recommended that in clinical practice, patients with a negative screening plasma test should undergo tissue biopsy testing if feasible [75]. The phase I/II TIGER-X study assessed the efficacy of the third-generation EGFR TKI, rociletinib, in previously treated EGFR-mutant advanced NSCLC patients [76]. Among 548 assessable patients, plasma testing was carried out by therascreen in tissue, BEAMing in plasma, and footprint mutation enrichment NGS in urine. In a retrospective analysis, they found that response rates were comparable regardless of sample type used (33.9% for tissue versus 32.1% for plasma versus 36.7% for urine). Duration of response and PFS were also comparable. When using tissue as a reference (N = 60), the sensitivity of urine or plasma testing for T790M, L858R, and Del19 mutations was: 72% for urine compared to tissue vs. 93% for plasma compared to tissue, 75% vs. 80%, and 67% vs. 80%, for the three EGFR mutations respectively; and the combination of urine and plasma testing improved sensitivity compared with either strategy alone [3]. Furthermore, the combination of urine and plasma testing detected more T790M mutations than tissue testing alone (92% versus 83%). Those with T790M detected by ctDNA (with or without tissue confirmation) had a demonstrable response, showing that the plasma or urine test was a true positive [3]. Efficacy of osimertinib according to T790M positivity in liquid biopsy by BEAMing was presented in a retrospective analysis, with a 30% false negative rate of plasma genotyping [74]. Prospective validation of osimertinib efficacy according to T790M positivity in a blood biopsy has been recently reported, and it confirms the retrospective data. Patient populations in whom T790M was detected in plasma had similar responses to osimertinib to patients when T790M was detected in tissue [26, 74]. Recent work suggests that at the time of clinically overt resistance, the simple presence of any detectable level of T790M in itself is clinically relevant, even with mutant allele fractions as low as 0.1% [26]. It is unknown whether some molecular alterations detected by ultrasensitive NGS methods represent clinically irrelevant sub-clones, especially those with an MAF of 0.01%–0.1% [77]. Additional studies are needed to confirm the minimum biological threshold with clinical relevance to guide treatment decisions. ctDNA testing has also allowed ongoing genomic analysis for patients on third generation EGFR inhibitors. With serial monitoring of patients on osimertinib, Thress et al. identified a novel tertiary EGFR resistance mutation, C797S, which prevents drug binding. In this manner, the resistance mechanism was recognized before the new drug was even approved [78]. While the T790M and C797S mutations in EGFR have been identified as key drivers of resistance, for the majority of patients the eventual progression will be driven by mechanisms outside of these hot-spot mutations in EGFR. The role of NGS-based liquid biopsy analysis is becoming increasingly important in discovering new mechanisms and in clinical assessment of patients to identify resistance to EGFR-targeted TKI. Some studies have reported that the ability to detect ctDNA varies by extent of disease [18, 79, 80]. These results suggest that one may be able to select a subset of patients where plasma ctDNA testing alone may obviate the need for tissue or other complementary testing methods. As in first-line treatment, clearance of plasma EGFR mutation after 6 weeks on osimertinib appears to be associated with improved outcome in T790M positive NSCLC patients, endorsing the dynamic predictive value of ctDNA on treatment [81]. The use of alternate methods, such as collected CTCs when combined with ctDNA, may also improve detection. In an exploratory analysis of 40 patients with EGFR-mutated tumours who progressed on TKI therapy, the use of ctDNA genotyping missed 30% of cases with tissue-identified T790M. However, when combined with genotyping from simultaneously collected CTCs, the two assays together identified all patients with this mutation. Additionally, the T790M mutation was identified in 35% of patients with a negative or indeterminate tissue biopsy when CTC and ctDNA analysis was combined [11]. Discussion Conclusion With the advent of newer ctDNA detection platforms, it is likely that sensitivity can be optimized with standardization and improved techniques. The relatively low sensitivity reported in some studies can likely be attributed to lack of consensus on choice of sample, storage, and technical approaches. Standardization should include the use of plasma over serum, avoidance of heparinized tubes, early centrifugation of tubes, and standardization of cell free extraction methods [28]. The comparative ease of ctDNA sample collection has accelerated clinical decision-making, as well as translational research, such as the identification of the novel acquired EGFR C797S mutation, exemplifying the power of ctDNA to track clonal evolution and tailor treatment. Identification of new mechanisms of acquired resistance in ctDNA in a dynamic manner may help to develop new personalized treatments among EGFR-mutant NSCLC patients. Liquid biopsies may be considered the new standard tool for detecting acquired resistance mutations, with tissue biopsies recommended only in cases without identifiable ctDNA. Following initial approvals by regulatory authorities, plasma-based testing for EGFR mutations is now entering clinical use. With improvements in methods and assays, it is likely that, in the near future, a subset of patients with advanced disease may only require plasma ctDNA testing, possibly with the addition of CTC or urine ctDNA testing, thus avoiding the need for percutaneous biopsy. Funding NR would like to acknowledge the support of The University of Cambridge, Cancer Research UK (grant numbers A11906, A20240), and the European Research Council under the European Union's Seventh Framework Programme (FP/2007-2013) / ERC Grant Agreement n. 337905. Disclosure JWG reports research funding from Clovis Oncology and AstraZeneca and honoraria from Vortex Biosciences. NR is co-founder, shareholder, and CSO of Inivata which commercializes ctDNA testing, an inventor of patent applications describing methods for analysis of cell-free DNA, and has received research funding from AstraZeneca. All remaining authors have declared no conflicts of interest. Key messages Liquid biopsies may be considered an alternative tool for detecting sensitising EGFR mutations and resistance mechanisms on treatment. Techniques and clinical applicability will be discussed in this review. References 1 Lin JJ, Cardarella S, Lydon CA et al.   Five-year survival in EGFR-mutant metastatic lung adenocarcinoma treated with EGFR-TKIs. J Thorac Oncol  2016; 11( 4): 556– 565. Google Scholar CrossRef Search ADS PubMed  2 Paz-Ares L, Tan E-H, O'Byrne K et al.   Afatinib versus gefitinib in patients with EGFR mutation-positive advanced non-small-cell lung cancer: overall survival data from the phase IIb LUX-Lung 7 trial. Ann Oncol  2017; 28( 2): 270– 277. Google Scholar CrossRef Search ADS PubMed  3 Reckamp KL, Melnikova VO, Karlovich C et al.   A highly sensitive and quantitative test platform for detection of NSCLC EGFR mutations in urine and plasma. J Thorac Oncol  2016; 11( 10): 1690– 1700. Google Scholar CrossRef Search ADS PubMed  4 Shingyoji M, Kageyama H, Sakaida T et al.   Detection of epithelial growth factor receptor mutations in cerebrospinal fluid from patients with lung adenocarcinoma suspected of neoplastic meningitis. J Thorac Oncol  2011; 6( 7): 1215– 1220. Google Scholar CrossRef Search ADS PubMed  5 Sasaki S, Yoshioka Y, Ko R et al.   Diagnostic significance of cerebrospinal fluid EGFR mutation analysis for leptomeningeal metastasis in non-small-cell lung cancer patients harboring an active EGFR mutation following gefitinib therapy failure. Respir Investig  2016; 54( 1): 14– 19. Google Scholar CrossRef Search ADS PubMed  6 Novello S, Barlesi F, Califano R et al.   Metastatic non-small-cell lung cancer: ESMO Clinical Practice Guidelines for diagnosis, treatment and follow-up. Ann Oncol  2016; 27(Suppl 5): v1– v27. Google Scholar CrossRef Search ADS   7 Overman MJ, Modak J, Kopetz S et al.   Use of research biopsies in clinical trials: are risks and benefits adequately discussed? J Clin Oncol  2013; 31( 1): 17– 22. Google Scholar CrossRef Search ADS PubMed  8 Lokhandwala T, Bittoni MA, Dann RA et al.   Costs of diagnostic assessment for lung cancer: a medicare claims analysis. Clin Lung Cancer  2017; 18( 1): e27– e34. Google Scholar CrossRef Search ADS PubMed  9 Hasegawa T, Sawa T, Futamura Y et al.   Feasibility of rebiopsy in non-small cell lung cancer treated with epidermal growth factor receptor-tyrosine kinase inhibitors. Intern Med  2015; 54( 16): 1977– 1980. Google Scholar CrossRef Search ADS PubMed  10 Kawamura T, Kenmotsu H, Taira T et al.   Rebiopsy for patients with non-small-cell lung cancer after epidermal growth factor receptor-tyrosine kinase inhibitor failure. Cancer Sci  2016; 107( 7): 1001– 1005. Google Scholar CrossRef Search ADS PubMed  11 Sundaresan TK, Sequist LV, Heymach JV et al.   Detection of T790M, the acquired resistance EGFR mutation, by tumor biopsy versus noninvasive blood-based analyses. Clin Cancer Res  2016; 22( 5): 1103– 1110. Google Scholar CrossRef Search ADS PubMed  12 Mok TS, Wu Y-L, Thongprasert S et al.   Gefitinib or carboplatin-paclitaxel in pulmonary adenocarcinoma. N Engl J Med  2009; 361( 10): 947– 957. Google Scholar CrossRef Search ADS PubMed  13 Kim ES, Hirsh V, Mok T et al.   Gefitinib versus docetaxel in previously treated non-small-cell lung cancer (INTEREST): a randomised phase III trial. Lancet  2008; 372( 9652): 1809– 1818. Google Scholar CrossRef Search ADS PubMed  14 Massard C, Michiels S, Ferté C et al.   High-throughput genomics and clinical outcome in hard-to-treat advanced cancers: results of the MOSCATO 01 Trial. Cancer Discov  2017; 7( 6): 586. Google Scholar CrossRef Search ADS PubMed  15 VanderLaan PA, Yamaguchi N, Folch E et al.   Success and failure rates of tumor genotyping techniques in routine pathological samples with non-small-cell lung cancer. Lung Cancer  2014; 84( 1): 39– 44. Google Scholar CrossRef Search ADS PubMed  16 Folch E, Yamaguchi N, VanderLaan PA et al.   Adequacy of lymph node transbronchial needle aspirates using convex probe endobronchial ultrasound for multiple tumor genotyping techniques in non-small-cell lung cancer. J Thorac Oncol  2013; 8( 11): 1438– 1444. Google Scholar CrossRef Search ADS PubMed  17 Piotrowska Z, Stirling K, Heist R et al.   Variation in mechanisms of acquired resistance among EGFR-mutant NSCLC patients with more than 1 postresistant biopsy. Int J Radiat Oncol Biol Phys  2017; 98( 1): 220. Google Scholar CrossRef Search ADS   18 Sacher AG, Paweletz C, Dahlberg SE et al.   Prospective validation of rapid plasma genotyping for the detection of EGFR and KRAS mutations in advanced lung cancer. JAMA Oncol  2016; 2( 8): 1014– 1022. Google Scholar CrossRef Search ADS PubMed  19 Sands J, Hornberger J. P3.02b-010 urine detection of EGFR T790M mutation in non-small-cell lung cancer: an outcomes and total cost of care analysis. J Thoracic Oncol  12( 1): S1191. CrossRef Search ADS   20 Lawrence MS, Stojanov P, Polak P et al.   Mutational heterogeneity in cancer and the search for new cancer-associated genes. Nature  2013; 499( 7457): 214– 218. Google Scholar CrossRef Search ADS PubMed  21 Jamal-Hanjani M, Wilson GA, McGranahan N et al.   Tracking the evolution of non-small-cell lung cancer. N Engl J Med  2017; 376( 22): 2109– 2121. Google Scholar CrossRef Search ADS PubMed  22 Yachida S, Jones S, Bozic I et al.   Distant metastasis occurs late during the genetic evolution of pancreatic cancer. Nature  2010; 467( 7319): 1114– 1117. Google Scholar CrossRef Search ADS PubMed  23 Zhang J, Fujimoto J, Zhang J et al.   Intratumor heterogeneity in localized lung adenocarcinomas delineated by multiregion sequencing. Science  2014; 346( 6206): 256– 259. Google Scholar CrossRef Search ADS PubMed  24 de Bruin EC, McGranahan N, Mitter R et al.   Spatial and temporal diversity in genomic instability processes defines lung cancer evolution. Science  2014; 346( 6206): 251– 256. Google Scholar CrossRef Search ADS PubMed  25 Chen Z-Y, Zhong W-Z, Zhang X-C et al.   EGFR mutation heterogeneity and the mixed response to EGFR tyrosine kinase inhibitors of lung adenocarcinomas. Oncologist  2012; 17( 7): 978– 985. Google Scholar CrossRef Search ADS PubMed  26 Remon J, Caramella C, Jovelet C et al.   Osimertinib benefit in EGFR-mutant NSCLC patients with T790M-mutation detected by circulating tumour DNA. Ann Oncol  2017; 28( 4): 784– 790. Google Scholar PubMed  27 Leon SA, Shapiro B, Sklaroff DM et al.   Free DNA in the serum of cancer patients and the effect of therapy. Cancer Res  1977; 37( 3): 646– 650. Google Scholar PubMed  28 Wan JCM, Massie C, Garcia-Corbacho J et al.   Liquid biopsies come of age: towards implementation of circulating tumour DNA. Nat Rev Cancer  2017; 17( 4): 223– 238. Google Scholar CrossRef Search ADS PubMed  29 Diaz LAJr, Bardelli A. Liquid biopsies: genotyping circulating tumor DNA. J Clin Oncol  2014; 32( 6): 579– 586. Google Scholar CrossRef Search ADS PubMed  30 Li M, Diehl F, Dressman D et al.   BEAMing up for detection and quantification of rare sequence variants. Nat Methods  2006; 3( 2): 95– 97. Google Scholar CrossRef Search ADS PubMed  31 Diehl F, Schmidt K, Choti MA et al.   Circulating mutant DNA to assess tumor dynamics. Nat Med  2008; 14( 9): 985– 990. Google Scholar CrossRef Search ADS PubMed  32 Vogelstein B, Kinzler KW. Digital PCR. Proc Natl Acad Sci USA  1999; 96( 16): 9236– 9241. Google Scholar CrossRef Search ADS PubMed  33 Dressman D, Yan H, Traverso G et al.   Transforming single DNA molecules into fluorescent magnetic particles for detection and enumeration of genetic variations. Proc Natl Acad Sci USA  2003; 100( 15): 8817– 8822. Google Scholar CrossRef Search ADS PubMed  34 Hindson BJ, Ness KD, Masquelier DA et al.   High-throughput droplet digital PCR system for absolute quantitation of DNA copy number. Anal Chem  2011; 83( 22): 8604– 8610. Google Scholar CrossRef Search ADS PubMed  35 Gale D, Plagnol V, Lawson A et al.   Analytical performance and validation of an enhanced TAm-Seq circulating tumor DNA sequencing assay. 2016; AACR 107th Annual Meeting 2016, Abstract 3639. 36 Forshew T, Murtaza M, Parkinson C et al.   Noninvasive identification and monitoring of cancer mutations by targeted deep sequencing of plasma DNA. Sci Transl Med  2012; 4( 136): 136ra68. Google Scholar CrossRef Search ADS PubMed  37 Lanman RB, Mortimer SA, Zill OA et al.   Analytical and clinical validation of a digital sequencing panel for quantitative, highly accurate evaluation of cell-free circulating tumor DNA. PLoS One  2015; 10( 10): e0140712. Google Scholar CrossRef Search ADS PubMed  38 Zill OA, Mortimer S, Banks KC et al.   Somatic genomic landscape of over 15, 000 patients with advanced-stage cancer from clinical next-generation sequencing analysis of circulating tumor DNA. J Clin Oncol  2016; 34(18 Suppl): LBA11501– LBA11501. Google Scholar CrossRef Search ADS   39 Murtaza M, Dawson S-J, Tsui DWY et al.   Non-invasive analysis of acquired resistance to cancer therapy by sequencing of plasma DNA. Nature  2013; 497( 7447): 108– 112. Google Scholar CrossRef Search ADS PubMed  40 Heitzer E, Ulz P, Belic J et al.   Tumor-associated copy number changes in the circulation of patients with prostate cancer identified through whole-genome sequencing. Genome Med  2013; 5( 4): 30. Google Scholar CrossRef Search ADS PubMed  41 US Food & Drug Administration—cobas EGFR Mutation Test v2. 2016; https://www.accessdata.fda.gov/cdrh_docs/pdf15/P150047c.pdf (4 November 2017, date last accessed). 42 QIAGEN. Therascreen EGFR RGQ PCR Kit Handbook Version 2. August 2016; https://www.qiagen.com/tn/resources/download.aspx? id=d217c9ca-b9f6-4891-8c46-32b71e83d88a&lang=en (4 November 2017, date last accessed). 43 Watanabe K, Fukuhara T, Tsukita Y et al.   EGFR mutation analysis of circulating tumor DNA using an improved PNA-LNA PCR clamp method. Can Respir J  2016; 2016: 5297329. Google Scholar CrossRef Search ADS PubMed  44 Panagene, PNAClamp EGFR Mutation Detection Kit (Ver.2), December 2013; https://secure.eurogentec.com/EGT/files/Panagene/TDS-PN-PNAC-3002.pdf (5 November 2017, date last accessed). 45 Mosko MJ, Nakorchevsky AA, Flores E et al.   Ultrasensitive detection of multiplexed somatic mutations using MALDI-TOF mass spectrometry. J Mol Diagn  2016; 18( 1): 23– 31. Google Scholar CrossRef Search ADS PubMed  46 Kimura H, Kasahara K, Kawaishi M et al.   Detection of epidermal growth factor receptor mutations in serum as a predictor of the response to gefitinib in patients with non-small-cell lung cancer. Clin Cancer Res  2006; 12( 13): 3915– 3921. Google Scholar CrossRef Search ADS PubMed  47 Kimura H, Suminoe M, Kasahara K et al.   Evaluation of epidermal growth factor receptor mutation status in serum DNA as a predictor of response to gefitinib (IRESSA). Br J Cancer  2007; 97( 6): 778– 784. Google Scholar CrossRef Search ADS PubMed  48 Bordi P, Del Re M, Danesi R et al.   Circulating DNA in diagnosis and monitoring EGFR gene mutations in advanced non-small cell lung cancer. Transl Lung Cancer Res  2015; 4( 5): 584– 597. Google Scholar PubMed  49 Han B, Tjulandin S, Hagiwara K et al.   96 odetermining the prevalence of EGFR mutations in Asian and Russian Patients (PTS) with advanced non-small-cell lung cancer (ANSCLC) of adenocarcinoma (ADC) and non-ADC histology: IGNITE study. Ann Oncol  2015; 26( Suppl 1): i29– i30. Google Scholar CrossRef Search ADS   50 Reck M, Hagiwara K, Han B et al.   ctDNA determination of EGFR mutation status in European and Japanese patients with advanced NSCLC: the ASSESS study. J Thorac Oncol  2016; 11( 10): 1682– 1689. Google Scholar CrossRef Search ADS PubMed  51 Qian X, Liu J, Sun Y et al.   Circulating cell-free DNA has a high degree of specificity to detect exon 19 deletions and the single-point substitution mutation L858R in non-small cell lung cancer. Oncotarget  2016; 7( 20): 29154– 29165. Google Scholar CrossRef Search ADS PubMed  52 Oxnard GR, Paweletz CP, Kuang Y et al.   Noninvasive detection of response and resistance in EGFR-mutant lung cancer using quantitative next-generation genotyping of cell-free plasma DNA. Clin Cancer Res  2014; 20( 6): 1698– 1705. Google Scholar CrossRef Search ADS PubMed  53 Luo J, Shen L, Zheng D. Diagnostic value of circulating free DNA for the detection of EGFR mutation status in NSCLC: a systematic review and meta-analysis. Sci Rep  2014; 4: 6269. Google Scholar CrossRef Search ADS PubMed  54 Qiu M, Wang J, Xu Y et al.   Circulating tumor DNA is effective for the detection of EGFR mutation in non-small cell lung cancer: a meta-analysis. Cancer Epidemiol Biomarkers Prev  2015; 24( 1): 206– 212. Google Scholar CrossRef Search ADS PubMed  55 Goto K, Ichinose Y, Ohe Y et al.   Epidermal growth factor receptor mutation status in circulating free DNA in serum: from IPASS, a phase III study of gefitinib or carboplatin/paclitaxel in non-small cell lung cancer. J Thorac Oncol  2012; 7( 1): 115– 121. Google Scholar CrossRef Search ADS PubMed  56 Kuang Y, Rogers A, Yeap BY et al.   Noninvasive detection of EGFR T790M in gefitinib or erlotinib resistant non-small cell lung cancer. Clin Cancer Res  2009; 15( 8): 2630– 2636. Google Scholar CrossRef Search ADS PubMed  57 Maheswaran S, Sequist LV, Nagrath S et al.   Detection of mutations in EGFR in circulating lung-cancer cells. N Engl J Med  2008; 359( 4): 366– 377. Google Scholar CrossRef Search ADS PubMed  58 Sharma SV, Bell DW, Settleman J et al.   Epidermal growth factor receptor mutations in lung cancer. Nat Rev Cancer  2007; 7( 3): 169– 181. Google Scholar CrossRef Search ADS PubMed  59 Yung TKF, Chan KCA, Mok TSK et al.   Single-molecule detection of epidermal growth factor receptor mutations in plasma by microfluidics digital PCR in non-small cell lung cancer patients. Clin Cancer Res  2009; 15( 6): 2076– 2084. Google Scholar CrossRef Search ADS PubMed  60 Karachaliou N, Mayo-de las Casas C, Queralt C et al.   Association of EGFR L858R mutation in circulating free DNA with survival in the EURTAC trial. JAMA Oncol  2015; 1( 2): 149– 157. Google Scholar CrossRef Search ADS PubMed  61 Mok T, Wu Y-L, Lee JS et al.   Detection and dynamic changes of EGFR mutations from circulating tumor DNA as a predictor of survival outcomes in NSCLC patients treated with first-line intercalated erlotinib and chemotherapy. Clin Cancer Res  2015; 21( 14): 3196– 3203. Google Scholar CrossRef Search ADS PubMed  62 Douillard J-Y, Ostoros G, Cobo M et al.   Gefitinib treatment in EGFR mutated Caucasian NSCLC: circulating-free tumor DNA as a surrogate for determination of EGFR status. J Thorac Oncol  2014; 9( 9): 1345– 1353. Google Scholar CrossRef Search ADS PubMed  63 Besse B, Remon J, Lacroix L et al.   Evaluation of liquid biopsies for molecular profiling in untreated patients with stage III/IV non-small cell lung cancer (NSCLC). J Clin Oncol  2017; 35(15 Suppl): 11540– 11540. 64 Remon J, Besse B, Lacroix L et al.   Evaluation of liquid biopsies for molecular profiling in patients with advanced non-small cell lung cancer (NSCLC) in the relapse treatment setting. J Clin Oncol  2017; 35(Suppl; abstr 11532). 65 Bradbury PA, Ding K, Ellis PM et al.   Evaluation of stored liquid biopsies for molecular profiling in patients with non-small cell lung cancer (NSCLC). J Clin Oncol  2017; 35(Suppl; abstr 9024). 66 Guibert NM, Paweletz C, Hu Y et al.   Early detection of competing resistance mutations using plasma next-generation sequencing (NGS) in patients (pts) with EGFR-mutant NSCLC treated with osimertinib. J Clin Oncol  2017; 35(Suppl; abstr 11529). 67 Jovelet C, Madic J, Remon J et al.   Crystal digital droplet PCR for detection and quantification of circulating EGFR sensitizing and resistance mutations in advanced non-small cell lung cancer. PLoS One  2017; 12( 8): e0183319. Google Scholar CrossRef Search ADS PubMed  68 Lee JY, Qing X, Xiumin W et al.   Longitudinal monitoring of EGFR mutations in plasma predicts outcomes of NSCLC patients treated with EGFR TKIs: Korean Lung Cancer Consortium (KLCC-12-02). Oncotarget  2016; 7( 6): 6984– 6993. Google Scholar PubMed  69 Tseng J-S, Yang T-Y, Tsai C-R et al.   Dynamic plasma EGFR mutation status as a predictor of EGFR-TKI efficacy in patients with EGFR-mutant lung adenocarcinoma. J Thorac Oncol  2015; 10( 4): 603– 610. Google Scholar CrossRef Search ADS PubMed  70 Marchetti A, Palma JF, Felicioni L et al.   Early prediction of response to tyrosine kinase inhibitors by quantification of EGFR mutations in plasma of NSCLC patients. J Thorac Oncol  2015; 10( 10): 1437– 1443. Google Scholar CrossRef Search ADS PubMed  71 Zheng D, Ye X, Zhang MZ et al.   Plasma EGFR T790M ctDNA status is associated with clinical outcome in advanced NSCLC patients with acquired EGFR-TKI resistance. Sci Rep  2016; 6: 20913. Google Scholar CrossRef Search ADS PubMed  72 Sorensen BS, Wu L, Wei W et al.   Monitoring of epidermal growth factor receptor tyrosine kinase inhibitor-sensitizing and resistance mutations in the plasma DNA of patients with advanced non-small cell lung cancer during treatment with erlotinib. Cancer  2014; 120( 24): 3896– 3901. Google Scholar CrossRef Search ADS PubMed  73 Remon J, Menis J, Hasan B et al.   The APPLE Trial: feasibility and activity of AZD9291 (osimertinib) treatment on positive PLasma T790M in EGFR-mutant NSCLC patients. EORTC 1613. Clin Lung Cancer  2017; 18( 5): 583– 588. Google Scholar CrossRef Search ADS PubMed  74 Oxnard GR, Thress KS, Alden RS et al.   Association between plasma genotyping and outcomes of treatment with osimertinib (AZD9291) in advanced non-small-cell lung cancer. J Clin Oncol  2016; 34( 28): 3375– 3382. Google Scholar CrossRef Search ADS PubMed  75 Jenkins S, Yang JC-H, Ramalingam SS et al.   Plasma ctDNA analysis for detection of the EGFR T790M mutation in patients with advanced non-small-cell lung cancer. J Thorac Oncol  2017; 12( 7): 1061– 1070. Google Scholar CrossRef Search ADS PubMed  76 Yu HA, Arcila ME, Rekhtman N et al.   Analysis of tumor specimens at the time of acquired resistance to EGFR-TKI therapy in 155 patients with EGFR-mutant lung cancers. Clin Cancer Res  2013; 19( 8): 2240– 2247. Google Scholar CrossRef Search ADS PubMed  77 Wang W, Karampini E, Correia LL et al.   Digital detection of T790M-yes or no to an ultrasensitive assay. Transl Lung Cancer Res  2016; 5( 3): 338– 342. Google Scholar CrossRef Search ADS PubMed  78 Thress KS, Paweletz CP, Felip E et al.   Acquired EGFR C797S mediates resistance to AZD9291 in advanced non-small cell lung cancer harboring EGFR T790M. Nat Med  2015; 21( 6): 560– 562. Google Scholar CrossRef Search ADS PubMed  79 Newman AM, Bratman SV, To J et al.   An ultrasensitive method for quantitating circulating tumor DNA with broad patient coverage. Nat Med  2014; 20( 5): 548– 554. Google Scholar CrossRef Search ADS PubMed  80 Zhao X, Han R-B, Zhao J et al.   Comparison of epidermal growth factor receptor mutation statuses in tissue and plasma in stage I-IV non-small cell lung cancer patients. Respiration  2013; 85( 2): 119– 125. Google Scholar CrossRef Search ADS PubMed  81 Thress KS, Markovets A, Barrett JC et al.   Complete clearance of plasma EGFR mutations as a predictor of outcome on osimertinib in the AURA trial. J Clin Oncol  2017; 35(Suppl abstract 9018). © The Author(s) 2018. Published by Oxford University Press on behalf of the European Society for Medical Oncology. All rights reserved. For Permissions, please email: journals.permissions@oup.com.

Journal

Annals of OncologyOxford University Press

Published: Jan 1, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 12 million articles from more than
10,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Unlimited reading

Read as many articles as you need. Full articles with original layout, charts and figures. Read online, from anywhere.

Stay up to date

Keep up with your field with Personalized Recommendations and Follow Journals to get automatic updates.

Organize your research

It’s easy to organize your research with our built-in tools.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

Monthly Plan

  • Read unlimited articles
  • Personalized recommendations
  • No expiration
  • Print 20 pages per month
  • 20% off on PDF purchases
  • Organize your research
  • Get updates on your journals and topic searches

$49/month

Start Free Trial

14-day Free Trial

Best Deal — 39% off

Annual Plan

  • All the features of the Professional Plan, but for 39% off!
  • Billed annually
  • No expiration
  • For the normal price of 10 articles elsewhere, you get one full year of unlimited access to articles.

$588

$360/year

billed annually
Start Free Trial

14-day Free Trial