Lung Cancer 90 (2015) 509–515

Contents lists available at ScienceDirect

Lung Cancer
journal homepage: www.elsevier.com/locate/lungcan

EGFR mutation detection in ctDNA from NSCLC patient plasma: A

cross-platform comparison of leading technologies to support the
clinical development of AZD9291
Kenneth S. Thress a,∗ , Roz Brant b , T. Hedley Carr c , Simon Dearden d , Suzanne Jenkins e ,
Helen Brown d , Tracey Hammett f , Mireille Cantarini g , J. Carl Barrett a
a
Translational Science, Oncology iMED, AstraZeneca, 35 Gatehouse Drive, Waltham, MA 02451, USA
b
Translational Science, Oncology iMED, AstraZeneca, Alderley Park, Macclesfield SK10 4TF, UK
c
Translational Science, Oncology iMED, AstraZeneca, Milton Road, Cambridge CB4 0FZ, UK
d
Personalised Healthcare & Biomarkers, AstraZeneca, Milton Road, Cambridge CB4 0FZ, UK
e
Personalised Healthcare & Biomarkers, AstraZeneca, Alderley Park, Macclesfield SK10 4TF, UK
f
Early Clinical Development, Alderley Park, Macclesfield SK10 4TF, UK
g
Global Medicines Development, Alderley Park, Macclesfield SK10 4TF, UK

a r t i c l e i n f o a b s t r a c t

Article history: Objectives: To assess the ability of different technology platforms to detect epidermal growth factor recep-
Received 13 August 2015 tor (EGFR) mutations, including T790M, from circulating tumor DNA (ctDNA) in advanced non-small cell
Received in revised form lung cancer (NSCLC) patients.
28 September 2015
Materials and methods: A comparison of multiple platforms for detecting EGFR mutations in plasma
Accepted 4 October 2015
ctDNA was undertaken. Plasma samples were collected from patients entering the ongoing AURA trial
(NCT01802632), investigating the safety, tolerability, and efficacy of AZD9291 in patients with EGFR-
Keywords:
sensitizing mutation-positive NSCLC. Plasma was collected prior to AZD9291 dosing but following clinical
Non-small cell lung cancer
Circulating tumor DNA
progression on a previous EGFR-tyrosine kinase inhibitor (TKI). Extracted ctDNA was analyzed using
Epidermal growth factor receptor mutation two non-digital platforms (cobas® EGFR Mutation Test and therascreenTM EGFR amplification refrac-
T790M tory mutation system assay) and two digital platforms (Droplet DigitalTM PCR and BEAMing digital PCR
AZD9291 [dPCR]).
Results: Preliminary assessment (38 samples) was conducted using all four platforms. For EGFR-TKI-
sensitizing mutations, high sensitivity (78–100%) and specificity (93–100%) were observed using tissue
as a non-reference standard. For the T790M mutation, the digital platforms outperformed the non-digital
platforms. Subsequent assessment using 72 additional baseline plasma samples was conducted using
the cobas® EGFR Mutation Test and BEAMing dPCR. The two platforms demonstrated high sensitivity
(82–87%) and specificity (97%) for EGFR-sensitizing mutations. For the T790M mutation, the sensitivity
and specificity were 73% and 67%, respectively, with the cobas® EGFR Mutation Test, and 81% and 58%,
respectively, with BEAMing dPCR. Concordance between the platforms was >90%, showing that multiple
platforms are capable of sensitive and specific detection of EGFR-TKI-sensitizing mutations from NSCLC
patient plasma.
Conclusion: The cobas® EGFR Mutation Test and BEAMing dPCR demonstrate a high sensitivity for T790M
mutation detection. Genomic heterogeneity of T790M-mediated resistance may explain the reduced
specificity observed with plasma-based detection of T790M mutations versus tissue. These data support
the use of both platforms in the AZD9291 clinical development program.
© 2015 The Authors. Published by Elsevier Ireland Ltd. This is an open access article under the CC
BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

Abbreviations: ARMS, amplification refractory mutation system; BEAM, beads, emulsions, amplification, and magnetics; CR, complete response; ctDNA, circulating tumor
DNA; dPCR, digital polymerase chain reaction; ddPCR, Droplet Digital PCR; EGFR, epidermal growth factor receptor; M1a/M0, disease confined to the thoracic cavity; M1b,
extra-thoracic metastatic disease; NSCLC, non-small cell lung cancer; PCR, polymerase chain reaction; PR, partial response; SD, stable disease; TKI, tyrosine kinase inhibitor.
∗ Corresponding author.
E-mail address: kenneth.thress@astrazeneca.com (K.S. Thress).

http://dx.doi.org/10.1016/j.lungcan.2015.10.004
0169-5002/© 2015 The Authors. Published by Elsevier Ireland Ltd. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
510 K.S. Thress et al. / Lung Cancer 90 (2015) 509–515
1. Introduction
Nearly all patients with non-small cell lung cancer (NSCLC)
who initially respond to epidermal growth factor receptor (EGFR)-
tyrosine kinase inhibitor (TKI) therapy develop resistance. In such
patients, it is now recommended to obtain a biopsy in order to char-
acterize the mechanism of resistance. However, obtaining sufficient
tissue for mutation analysis in patients with advanced disease
is challenging, as invasive interventions may be ineffective and
unsafe. Moreover, detection of disease-relevant mutations from the
biopsy of a single tumor lesion may not be reflective of the patient’s
complete disease burden, especially in heterogeneous cancers [1,2].
In recent years, circulating tumor DNA (ctDNA) has emerged
as a specific and sensitive blood-based biomarker for detection of
EGFR mutations. T790M is the most common mechanism of resis-
tance to first-line EGFR-TKIs, detectable in nearly 60% of tissue
biopsies taken after resistance develops [3,4]. Although the mech-
anism by which ctDNA is released into the bloodstream is unclear
[5], it is thought to be related to physiological events related to
the tumor cells, including apoptosis and necrosis [6]. Several stud-
ies have demonstrated that mutations, including the EGFR T790M
mutation, detected in plasma ctDNA are highly concordant with
those detected in tumor tissue in patients [7–12], indicating that
ctDNA as a liquid biopsy is a feasible and minimally invasive alter-
native to tissue biopsy. Other clinical applications for ctDNA in
this setting include molecular assessment of patients at diagnosis
[11], serial (real-time) monitoring of patients for the develop-
ment of resistance mutations (which is not practical/possible
with repeat biopsies) [11,13], and for the clinical management
of patients [14]. Importantly, because ctDNA analysis does not
involve formaldehyde fixation, there is reduced frequency of a false
positive result due to deamination [15]. Although several method-
ologies are available for mutation analysis from ctDNA, including
digital polymerase chain reaction (dPCR), mutant-enriched PCR,
and peptide nucleic acid-locked nucleic acid PCR [16], there is
a paucity of widely accepted and approved methods for ctDNA
analysis.
AZD9291 is a highly selective, irreversible EGFR-TKI that tar-
gets both EGFR-TKI-sensitizing mutations (e.g., exon 19 deletion
and L858R) and the T790M resistance mutation [17]. AZD9291 has
shown promising clinical activity in the ongoing phase I AURA
trial (clinicaltrials. gov. identifier: NCT01802632) in patients with
EGFR-TKI-resistant advanced NSCLC [18]. To date, patients have
been selected for AURA trials using tissue testing for the T790M
resistance mutation with the cobas® EGFR Mutation Test (Roche
Molecular Systems, Inc., Pleasanton, CA, USA), the companion diag-
nostic being developed for AZD9291. However, as part of the clinical
development process for AZD9291, alternative selection strate-
gies are also being evaluated, namely EGFR mutation testing using
plasma ctDNA. To this end, a comparison was undertaken of mul-
tiple platforms for the detection of EGFR mutations in plasma
ctDNA to identify the most appropriate technology for use during
AZD9291 clinical development.
2. Methods
2.1. Patients
Plasma samples were collected from patients enrolled in the
ongoing phase I AURA study. In brief, AURA (NCT01802632) is
an open-label, multicenter study designed to assess the safety,
tolerability, and efficacy of AZD9291 in patients with EGFR
mutation-positive NSCLC who have progressed following prior
therapy with an EGFR-TKI agent [18]. All patients had mea-
surable disease at baseline, with radiological documentation of
disease progression while receiving continuous treatment with an
EGFR-TKI. Patients were required to have EGFRm-positive NSCLC,
confirmed by either the presence of a tumor harboring an EGFR
mutation known to be associated with EGFR-TKI sensitivity, or
evidence of a clinical benefit with an EGFR-TKI, followed by sys-
temic objective progression (according to Response Evaluation
Criteria In Solid Tumors or World Health Organization criteria)
while on continuous EGFR-TKI treatment as per Jackman criteria
[19].
2.2. Ethics statement
The study was conducted in accordance with the Declara-
tion of Helsinki, the International Conference on Harmonisa-
tion/Good Clinical Practice, applicable regulatory requirements,
and AstraZeneca’s policy on bioethics. All patients provide written
informed consent before any study-specific procedures, sampling,
and analyses.
The AURA protocol was reviewed and approved by Institutional
Review Boards at each trial site before patient enrollment.
2.3. Plasma sample collection and DNA extraction
Plasma samples were collected from patients enrolled into
AURA, following progression on a previous EGFR-TKI, but prior to
dosing with AZD9291; samples were taken from patients in both
dose escalation and dose expansion cohorts. Plasma samples from
identical timepoints were pooled per patient and 2 ml (where avail-
able) were assigned for each platform assessment; samples were
stored at −80 ◦ C and were tested in a blinded fashion.
Patient-matched tumor material (formalin-fixed and paraffin-
embedded) was available for 80% of plasma samples to enable
ctDNA–tumor concordance testing. Tumor tissue originated from
pre-study biopsies collected following progression on the last line
of therapy. Tumor tissue genotyping was conducted in one of three
central laboratories using a standardized cobas® EGFR Mutation
Test assay (Roche Molecular Systems, Inc.), but with a protocol
modification that allowed the testing laboratories to use more
slides than stated in the package insert if it was thought to be
deemed necessary. In the dose expansion cohorts, T790M muta-
tion status was identified either locally or centrally; all local results
were confirmed by a mandatory retrospective centralized tissue
result for each patient; in the dose escalation cohorts, a central tis-
sue test was not required but a local EGFRm tissue mutation result
was available for most patients.
The sensitivity of the plasma assays was calculated as follows:
100% × true positives/(true positives + false negatives), where true
positives and false negatives were defined according to the tissue-
based test. The specificity of the plasma assays was calculated
as follows: 100% × true negatives/(false positives + true negatives),
where true negatives and false positives were defined according
to the tissue-based test [20]. Overall concordance between each
plasma assay and the tissue-based test was also calculated.
2.4. Mutation detection
Extracted ctDNA was tested for three common EGFR mutations
(exon 19 deletion, L858R, and T790M) using multiple platforms.
This involved non-digital detection using the cobas® EGFR Muta-
tion Test and the therascreenTM EGFR amplification refractory
mutation system (ARMS) assay (QIAGEN, Venlo, The Netherlands)
and digital detection using the highly sensitive and quantitative
Droplet Digital PCR (ddPCRTM ; Bio-Rad/MolecularMD, Hercules,
CA, USA) and beads, emulsions, amplification, and magnetics
(BEAM)ing dPCR technique (Sysmex Inostics, Inc., Mundelein, IL,
USA).
 K.S. Thress et al. / Lung Cancer 90 (2015) 509–515 511

Fig. 1. Relationship of T790M plasma detectiona to NSCLC disease classification.

**P < 0.01; refers to the presence of T790M positive mutations in M1a/M0 versus M1b patients. a As determined by ddPCR assay (Bio-Rad/MolecularMD).
ctDNA, circulating tumor DNA; ddPCR, Droplet Digital PCR; M1a/M0, disease confined to the thoracic cavity; M1b, extra-thoracic metastatic disease; NSCLC, non-small cell
lung cancer.

Table 1 For the therascreenTM assay, DNA was extracted from plasma
Performance of four different plasma assays, using a tissuea test result as a non-
as described in the QIAamp Circulating Nucleic Acid Handbook,
reference standard, for detection of EGFR mutations from circulating tumor DNA in
a set of 38 plasma samples from the AURA clinical trial. second edition (QIAGEN) using the protocol, purification of circu-
lating nucleic acids from 1 ml, 2 ml, or 3 ml plasma. Modifications of
cobas® therascreenTM ddPCRTM BEAMingdPCR
the protocol and specific details are as follows: samples were cen-
EGFR EGFR
Mutation ARMS-PCR trifuged at 3000 rpm for 2 min and the supernatant was transferred
Test to a clean tube, prior to extraction (step 3); a brief centrifuge spin
Exon 19 deletion
was performed after the 60 ◦ C incubation (step 4); 55 ␮l buffer AVE
Sensitivity 86% 82% –b 93% was added at the elution step 15; DNA was stored at −80 ◦ C. Prior
(24/28) (23/28) (26/28) to mutation testing, all samples were assessed for total amplifiable
b
Specificity 100% 100% – 100% DNA and subsequently processed for the detection of EGFR muta-
(10/10) (10/10) (10/10)
tions and data analysis using the therascreenTM EGFR RGQ PCR Kit
Concordance 89% 87% –b 95%
Handbook version 1 (QIAGEN); a minor protocol modification was
L858R that 16 samples per run were tested.
Sensitivity 90% 78% 90% 100%
(9/10) (7/9) (9/10) (10/10)
Testing on the ddPCRTM and BEAMing dPCR platforms was per-
Specificity 100% 100% 100% 93% formed at MolecularMD (Cambridge, MA, USA) and Sysmex Inostics
(28/28) (28/28) (28/28) (26/28) (Hamburg, Germany), respectively.
Concordance 97% 95% 97% 95%

T790M
Sensitivity 41% 29% 71% 71%
3. Results
(7/17) (5/17) (12/17) (12/17)
Specificity 100% 100% 83% 67% 3.1. Initial platform comparison (n = 38)
(6/6) (6/6) (5/6) (4/6)
Concordance 57% 48% 74% 70%
A preliminary assessment using all four platforms was per-
ARMS, amplification refractory mutation system; dPCR, digital polymerase chain formed with 38 plasma samples. High sensitivity and specificity,
reaction; ddPCR, Droplet Digital polymerase chain reaction; EGFR, epidermal growth
using a tissue test result as a non-reference standard, was observed
factor receptor; PCR, polymerase chain reaction.
a with the cobas® EGFR Mutation Test, the therascreenTM EGFR
24/38 samples were tested using local tissue assays and 14/38 samples were
tested via central assays. ARMS assay, and the BEAMing dPCR for EGFR-sensitizing muta-
b
Comprehensive exon 19 deletion data not available. tions (Table 1). The same was true with the ddPCRTM for the L858R
EGFR mutation. No sensitivity or specificity data were available for
the exon 19 deletion mutation for the ddPCRTM (Table 1) as it was
Assays were performed according to each manufacturer’s pro- only designed to detect a minority of known exon 19 deletions.
tocols unless otherwise stated. Testing on the cobas® z 480 Digital platforms appeared to detect a higher percentage of
analyzer and therascreenTM platform was performed internally T790M mutations, compared with non-digital platforms (Table 1).
at AstraZeneca, with analysis of cobas® EGFR Mutation Test data In the single case where ddPCRTM detected a positive T790M muta-
performed by Roche using algorithms specific for ctDNA testing. tion that was not detected in the tissue sample, BEAMing dPCR also
Specifically, for the cobas® EGFR Mutation Test, DNA was extracted classified this sample as T790M positive.
from plasma using the cobas® cfDNA Sample Preparation Kit (Roche T790M mutations were more readily detected in the plasma of
Molecular Systems, Inc.). Extracted DNA was assayed for mutation patients with extra-thoracic metastatic disease (M1b) than patients
status using the protocol specific for cfDNA, with the cobas® EGFR with disease confined to the thoracic cavity (M1a/M0) (P < 0.01;
Mutation Test. Fig. 1).
512 K.S. Thress et al. / Lung Cancer 90 (2015) 509–515

Table 2 Table 4
Performance of two different plasma assays relative to using a tissue test result as a Discordant results with two different plasma assays for detection of the EGFR T790M
non-reference standard for detection of epidermal growth factor receptor mutations mutation from circulating tumor DNA.
from circulating tumor DNA in a set of 72 plasma samplesa from the AURA clinical
trial. Patient Tissuea Plasma

®
cobas EGFR Mutation Test BEAMing dPCR cobas® BEAMing cobas®
EGFR dPCR EGFR
Exon 19 deletion Mutation (% mutant) Mutation
Sensitivity 82% (23/28) 82% (23/28) Test Test
Specificity 97% (30/31) 97% (30/31)
L858R 1 Positive Positive Negative ‘False’
Sensitivity 87% (20/23) 87% (20/23) (0.021%) negatives by
Specificity 97% (35/36) 97% (35/36) 2 Positive Positive Negative cobas® EGFR
T790M (0.048%) Mutation Test
Sensitivity 73% (30/41) 81% (33/41) 3 Positive Positive Negative
Specificity 67% (16/24) 58% (14/24) (0.064%)
4 Positive Positive Negative
BEAM, beads, emulsions, amplification, and magnetics; dPCR, digital polymerase (0.202%)
chain reaction; EGFR, epidermal growth factor receptor. 5 Positive Negative Negative
a
13 with no exon 19 deletion/L858R tissue result and seven with no T790M tissue 6 Positive Negative Negative
result. 7 Positive Negative Negative
8 Positive Negative Negative
9 Positive Negative Negative
Table 3 10 Positive Negative Negative
Concordance between two different plasma assays for detection of EGFR mutations 11 Positive Negative Negative
from circulating tumor DNA. 12 Negative Positive Negative ‘False’
(0.026%) positives
EGFR mutation Concordance
13 Negative Positive Positive by
(cobas® EGFR
(0.027%) BEAMing
Mutation Test and
14 Negative Positive Positive dPCR
BEAMing dPCR)
(0.054%)
Exon 19 deletion 90% (65/72) 15 Negative Positive Negative
L858R 93% (67/72) (0.080%)
T790M 90% (65/72) 16 Negative Positive Positive
(0.283%)
dPCR, digital polymerase chain reaction; EGFR, epidermal growth factor receptor.
17 Negative Positive Positive
(0.340%)
18 Negative Positive Positive
3.2. Subsequent platform comparison (n = 72) (0.344%)
19 Negative Positive Positive
(0.491%)
A subsequent assessment using two of the original four plat- 20 Negative Positive Positive
forms was performed with a set of distinct baseline plasma samples (1.113%)
from 72 additional, non-overlapping patients on the AURA trial. Of BEAM, beads, emulsions, amplification, and magnetics; dPCR, digital polymerase
the digital platforms, BEAMing PCR was selected for further test- chain reaction; EGFR, epidermal growth factor receptor.
ing as ddPCRTM did not allow a full assessment of EGFR mutations. a
Central T790M tissue results.
Of the non-digital platforms, the cobas® EGFR Mutation Test was
selected for further testing as the therascreenTM EGFR ARMS-PCR
had a lower sensitivity observed for T790M. Subsequent testing A preliminary assessment of clinical response to AZD9291 as
using the cobas® EGFR Mutation Test was conducted at Roche a function of the T790M resistance mutation detected at baseline
Molecular Systems, Inc. while BEAMing PCR testing continued to (i.e., prior to initiation of AZD9291 but following the development
be done at Sysmex Inostics. of resistance to first-line EGFR-TKI) in plasma and tissue is shown in
As seen in the initial comparison, high sensitivity and specificity Fig. 2. The clinical response rate in patients positive for the T790M
were observed with the cobas® EGFR Mutation Test and BEAMing mutation in plasma was almost identical to that for patients posi-
dPCR for EGFR-TKI-sensitizing mutations (Table 2). Both the cobas® tive for the T790M mutation in tissue (59% vs. 61%).
EGFR Mutation Test and BEAMing dPCR demonstrated good sensi-
tivity for T790M mutation detection (Table 2). As observed in the 4. Discussion
initial comparison, BEAMing dPCR showed high sensitivity (81%)
but a lower specificity (58%). The sensitivity of the cobas® EGFR Several technologies have been described for the detection of
Mutation Test was markedly improved compared with the initial EGFR mutations using plasma ctDNA. In the current study, four such
comparison (Table 2). technologies were compared, with the aim of identifying robust
Strong overall concordance was demonstrated between the platform (s) to support the ongoing AZD9291 clinical development
cobas® EGFR Mutation Test and BEAMing dPCR for detection of program. The study was conducted in two parts: (i) initial com-
EGFR mutations from ctDNA (Table 3). parison of four platforms using a small set of plasma samples; (ii)
An analysis of the 20 discordant T790M results was undertaken. subsequent comparison of two platforms using a larger set of addi-
There were four ‘false’ plasma negative results detected by the tional plasma samples. All samples (plasma and patient-matched
cobas® EGFR Mutation Test; all four were called positive by BEAM- tumor tissue) originated from patients enrolled in the ongoing
ing, but three of these were very near the lower limit of detection phase I AURA trial. The cross-platform comparison showed that the
(0.020% mutant) (Table 4). There were seven cases where both cobas® EGFR Mutation Test and BEAMing dPCR had highly concor-
plasma assays did not detect any T790M mutation despite a posi- dant results, with high sensitivity (73–81%) for the detection of the
tive result in the patient matched tumor tissue. Finally, there were T790M mutation; good specificity was also achieved with these two
nine ‘false’ plasma positives detected by BEAMing dPCR; seven of platforms (58–67%). The objective response rate (ORR) to AZD9291
these were also scored positive by the cobas® EGFR Mutation Test was almost identical in patients positive for the T790M mutation in
(Table 4). plasma and in tissue (both assessed using the cobas® EGFR Muta-
 K.S. Thress et al. / Lung Cancer 90 (2015) 509–515
Fig. 2. Clinical response to AZD9291 according to EGFR T790M mutation at baseline (data cut-off date December 2014).
CR, complete response; EGFR, epidermal growth factor receptor; PR, partial response; SD, stable disease.
tion Test). Collectively, these results show that the cobas® EGFR
Mutation Test and BEAMing dPCR have the potential to identify
patients harboring T790M mutations from plasma ctDNA.
Four platforms were evaluated in the current study. The two
non-digital platforms comprised the cobas® EGFR Mutation Test
and the therascreenTM EGFR ARMS assay; both of these testing
methods arise from established tissue tests that have been adapted
for low-DNA-input plasma samples. The cobas® EGFR Mutation
Test and the therascreenTM EGFR RGQ PCR test detect 41 [21] and
21 [22] mutations, respectively, in exons 18–21 of the EGFR gene;
importantly, both tests detect the T790 M mutation. The two digi-
tal platforms were the BioRad ddPCRTM and BEAMing dPCR. Digital
PCR methods are purportedly more sensitive than other assays for
mutant sequence detection [23] and are fully quantitative, making
them suitable for quantification of longitudinal plasma samples,
which may allow for monitoring of disease/mutation evolution over
time. Both BEAMing and ddPCR were specifically included as they
have been used to profile plasma ctDNA for a variety of cancer-
related mutations [11,23–27]. Although the ddPCRTM generated
comparable T790M results to BEAMing in the initial comparison
of all four platforms, its ability to detect only a minority of known
exon 19 deletions led to its exclusion from the subsequent platform
comparison.
To the best of the authors’ knowledge, this is the first comparison
of the cobas® EGFR Mutation Test and BEAMing dPCR to be pub-
lished. The results support the potential use of both platforms for
the detection of EGFR mutations from patient plasma samples. The
cobas® EGFR Mutation Test showed high sensitivity for the T790M
mutation, and had comparable sensitivity to BEAMing dPCR. An
improvement in sensitivity was observed with the cobas® EGFR
Mutation Test between the initial and subsequent comparisons.
The difference in sensitivity between the two datasets may be a
result of the greater sample size in the subsequent comparison and
optimization of the assay by Roche between comparisons, resulting
in the identification of more T790M mutations. As a complemen-
tary approach, because of its quantitative nature, BEAMing dPCR
may allow for a more detailed assessment of T790M allelic levels
and potential correlation with clinical response. Such an analysis
513
with a large patient cohort from AURA is now ongoing and will be
reported elsewhere. BEAMing dPCR may also be useful for dynamic
monitoring of longitudinal samples where a binary outcome (i.e.,
simply the presence or absence of a signal) is not appropriate.
The T790M resistance mutation was more frequently detected
in the plasma of patients with metastatic versus locally advanced
disease. This has been observed in a similar study [11], and suggests
that tumor bulk and metastatic status may impact the presence of
plasma-mutant EGFR and should be further validated in a clinical
setting [11].
AZD9291 is an orally administered EGFR-TKI that is highly
potent against EGFR-TKI-sensitizing mutations and the T790M
resistance mutation, but with a margin of selectivity against wild-
type EGFR activity [17,28]. While currently approved EGFR-TKIs,
including gefitinib and erlotinib, are effective in the treatment of
NSCLC, nearly all patients who respond to these treatments will
eventually develop acquired resistance, and approximately 60% of
EGFR-TKI-resistant tumors will carry a T790M mutation in exon
20 [17,28,29]. Treatment strategies in patients with acquired resis-
tance are limited at present; although irreversible EGFR-TKIs such
as afatinib and dacomitinib have been developed, their clinical
utility is limited in the T790M-resistant setting, likely owing to
dose-limiting toxicities associated with non-selective inhibition
of wild-type EGFR [17,28,30,31]. The safety, tolerability, and effi-
cacy of AZD9291 is currently being investigated in the ongoing
AURA program [18]. Preliminary phase I data show that AZD9291
demonstrates promising clinical activity in patients with centrally
confirmed T790M mutation-positive disease (confirmed ORR of
61% across all doses) [18]. In the current study, the clinical ORR
(complete and partial) was almost identical between patients pos-
itive for the T790M mutation in plasma ctDNA and by tissue
genotyping, suggesting that plasma detection T790M mutation
may be a suitable alternative to tissue biopsies.
Overall, the concordance between the BEAMing dPCR and the
cobas® EGFR Mutation Test was 90% for the T790M mutation. This
high rate of agreement using two distinct assay platforms run in
a blinded fashion builds confidence in the overall plasma result.
Indeed, analysis of the 20 discordant results between plasma and
514 K.S. Thress et al. / Lung Cancer 90 (2015) 509–515
tissue from the 72-patient set reveals that the BEAMing dPCR and
cobas® EGFR Mutation Test outcomes were comparable in 14/20
cases. Furthermore, all six of those instances where the plasma
tests disagreed arose in cases where the T790M allelic fraction
was very low (≤0.2%) and perhaps below the detection limit of the
cobas® EGFR Mutation Test plasma assay. Taken together, these
data suggest that tumor heterogeneity, and not technological lim-
itations alone, may explain the tissue–plasma discordance and
the ‘reduced’ T790 M specificity observed with the plasma assays.
T790M tumor heterogeneity has been reported of late, with differ-
ent tumor biopsies within the same patient demonstrating differing
EGFR mutational profiles [1,2]. In all, these data suggest that plasma
ctDNA may be more precise and informative than tissue as the
blood mirrors the entire tumor burden [2]; additionally, a single
biopsy is often taken from a non-progressing lesion. However, het-
erogeneous tumors may harbor additional resistance mechanisms
and thus ultimately result in lower clinical response rates. Sup-
porting this concept, in patients with plasma positive but tumor
negative for T790M, the clinical ORR was 38% (three/eight patients)
and the disease control rate was 75% (six/eight patients). These data
appear more reflective of the clinical responses observed in patients
tumor negative for T790M and thus suggest the potential for tumor
heterogeneity, albeit with the caveat of involving a small number
of patient samples.
Plasma ctDNA has potential in the management of patients with
NSCLC in routine clinical practice, particularly considering that DNA
extraction and mutation detection using a commercially available
and approved test could be performed in as little as 1 day [14,32],
allowing for real-time disease testing. That said, robust validation
in prospective clinical trials [16,33], such as that planned in the
AURA program, is required.
In conclusion, multiple platforms are capable of sensitive and
specific detection of EGFR-TKI-sensitizing mutations from ctDNA
in patients with NSCLC. The cobas® EGFR Mutation Test assay and
BEAMing dPCR showed highly concordant T790M resistance muta-
tion detection, with 70–80% sensitivity, using a tissue test result
as a non-reference standard, for genotyping. Genomic heterogene-
ity of T790M-mediated resistance may account for the reduced
specificity observed with plasma- versus tissue-based detection.
The rate of clinical response to AZD9291 was almost identical in
patients positive for the T790M mutation in plasma and in tissue,
indicating that plasma detection may be a suitable alternative to
tissue genotyping in patients with NSCLC.
Collectively, the results support the potential use of both plat-
forms in the AZD9291 clinical development program.
Conflicts of interest
All authors are employees of, and hold shares in, AstraZeneca.
Role of the funding source
AstraZeneca funded this research, including the experimental
design, the collection, analysis and interpretation of data, and the
writing of the report. Authors (all employees of AstraZeneca) were
responsible for the decision to submit the article for publication.
Author contributions
KST: designed study, analyzed data, interpreted data; RB: per-
formed the experiments, read and approved final manuscript; THC:
primarily analysis of therascreenTM data; SD and SJ: study concep-
tion/design and data interpretation; HB: data acquisition (tumor)
and joint contributor to data interpretation of the 38 sample set;
TH: clinical study lead, clinical sample and data acquisition, read
and approved final manuscript; MC: designed and supervised clin-
ical study, interpreted data, read and approved manuscript; JCB:
contributed to design of the study, interpretation of data, and
manuscript preparation.
Acknowledgments
The study was funded by AstraZeneca. Medical writing services
were provided by Kerry Acheson, Ph.D., of iMed Comms, Maccles-
field, UK, and were funded by AstraZeneca. The authors thank Wei
Wen, Sean Chien, and Ha Bich Tran (Roche Molecular Systems, Inc.)
for generating the cobas® EGFR Mutation Test data as well as the
investigators and patients who participated in the AURA study.
References
[1] R. Fisher, L. Pusztai, C. Swanton, Cancer heterogeneity: implications for
targeted therapeutics, Br. J. Cancer 108 (2013) 479–485.
[2] B. Weber, P. Meldgaard, H. Hager, L. Wu, W. Wei, J. Tsai, et al., Detection of
EGFR mutations in plasma and biopsies from non-small cell lung cancer
patients by allele-specific PCR assays, BMC Cancer 14 (2014) 294.
[3] M.E. Arcila, G.R. Oxnard, K. Nafa, G.J. Riely, S.B. Solomon, M.F. Zakowski, et al.,
Rebiopsy of lung cancer patients with acquired resistance to EGFR inhibitors
and enhanced detection of the T790M mutation using a locked nucleic
acid-based assay, Clin. Cancer Res. 17 (2011) 1169–1180.
[4] L.V. Sequist, B.A. Waltman, D. Dias-Santagata, S. Digumarthy, A.B. Turke, P.
Fidias, et al., Genotypic and histological evolution of lung cancers acquiring
resistance to EGFR inhibitors, Sci. Transl. Med. 3 (2011) 75ra26.
[5] C. Bettegowda, M. Sausen, R.J. Leary, I. Kinde, Y. Wang, N. Agrawal, et al.,
Detection of circulating tumor DNA in early- and late-stage human
malignancies, Sci. Transl. Med. 6 (2014) 224ra24.
[6] H. Schwarzenbach, D.S. Hoon, K. Pantel, Cell-free nucleic acids as biomarkers
in cancer patients, Nat. Rev. Cancer 11 (2011) 426–437.
[7] J.Y. Douillard, G. Ostoros, M. Cobo, T. Ciuleanu, R. Cole, G. McWalter, et al.,
Gefitinib treatment in EGFR mutated caucasian NSCLC: circulating-free tumor
DNA as a surrogate for determination of EGFR status, J. Thorac. Oncol. 9
(2014) 1345–1353.
[8] J.Y. Douillard, G. Ostoros, M. Cobo, T. Ciuleanu, R. McCormack, A. Webster,
et al., First-line gefitinib in Caucasian EGFR mutation-positive NSCLC patients:
a phase-IV, open-label, single-arm study, Br. J. Cancer 110 (2014) 55–62.
[9] T. Mok, Y.L. Wu, J.S. Lee, C.-J. Yu, V. Sriuranpong, W. Wen, et al., Detection of
EGFR-activating mutations from plasma DNA as a potent predictor of survival
outcomes in FASTACT 2: a randomized phase III study on intercalated
combination of erlotinib (E) and chemotherapy (C), J. Thorac. Oncol. 31 (2013)
(Suppl:Abstract 8021).
[10] E.A. Punnoose, S. Atwal, W. Liu, R. Raja, B.M. Fine, B.G. Hughes, et al.,
Evaluation of circulating tumor cells and circulating tumor DNA in non-small
cell lung cancer: association with clinical endpoints in a phase II clinical trial
of pertuzumab and erlotinib, Clin. Cancer Res. 18 (2012) 2391–2401.
[11] J.M. Sun, C. Karlovich, W. Wen, H. Wakelee, S. Chien, E. Mann, et al., Serial
monitoring of EGFR mutations in plasma and evaluation of EGFR mutation
status in matched tissue and plasma from NSCLC patients treated with
CO-1686, Mol. Cancer Ther. 12 (2013) (Suppl Abstract B25).
[12] H. Gevensleben, I. Garcia-Murillas, M.K. Graeser, G. Schiavon, P. Osin, M.
Parton, et al., Noninvasive detection of HER2 amplification with plasma DNA
digital PCR, Clin. Cancer Res. 19 (2013) 3276–3284.
[13] J. Luo, L. Shen, D. Zheng, Diagnostic value of circulating free DNA for the
detection of EGFR mutation status in NSCLC: a systematic review and
meta-analysis, Sci. Rep. 4 (2014) 6269.
[14] A. Vallee, M. Marcq, A. Bizieux, C.E. Kouri, H. Lacroix, J. Bennouna, et al.,
Plasma is a better source of tumor-derived circulating cell-free DNA than
serum for the detection of EGFR alterations in lung tumor patients, Lung
Cancer 82 (2013) 373–374.
[15] M.G. Denis, A. Vallee, S. Theoleyre, EGFR T790M resistance mutation in non
small-cell lung carcinoma, Clin. Chim. Acta 444 (2015) 81–85.
[16] M. Ilie, V. Hofman, E. Long, O. Bordone, E. Selva, K. Washetine, et al., Current
challenges for detection of circulating tumor cells and cell-free circulating
nucleic acids, and their characterization in non-small cell lung carcinoma
patients. What is the best blood substrate for personalized medicine? Ann.
Transl. Med. 2 (2014) 107.
[17] D.A. Cross, S.E. Ashton, S. Ghiorghiu, C. Eberlein, C.A. Nebhan, P.J. Spitzler,
et al., AZD9291, an irreversible EGFR TKI, overcomes T790M-mediated
resistance to EGFR inhibitors in lung cancer, Cancer Discov. 4 (2014)
1046–1061.
[18] P.A. Jänne, J.C. Yang, D.W. Kim, D. Planchard, Y. Ohe, S.S. Ramalingam, et al.,
AZD9291 in EGFR inhibitor-resistant non-small-cell lung cancer, N. Engl. J.
Med. 372 (2015) 1689–1699.
[19] D. Jackman, W. Pao, G.J. Riely, J.A. Engelman, M.G. Kris, P.A. Janne, et al.,
Clinical definition of acquired resistance to epidermal growth factor receptor
tyrosine kinase inhibitors in non-small-cell lung cancer, J. Clin. Oncol. 28
(2010) 357–360.
 K.S. Thress et al. / Lung Cancer 90 (2015) 509–515
[20] US Department of Health and Human Services, Food and Drug Administration
(FDA), Center for Devices and Radiological Health. Guidance for industry and
FDA staff. Statistical guidance on reporting results from studies evaluating
diagnostic tests. Rockville, MD: FDA, (2007).
[21] Roche. cobas® EGFR Mutation Test, 2015. Available at http://molecular.roche.
com/assays/Pages/cobasEGFRMutationTest.aspx, (accessed on 13.03.15.).
[22] QIAGEN. therascreen EGFR RGQ PCR kit for the detection of mutations in the
EGFR oncogene, 2015. Available at www.qiagen.com/gb/products/catalog/
assay-technologies/complete-assay-kits/personalized-healthcare/
therascreen-egfr-rgq-pcr-kit/, (accessed on 13.03.15.).
[23] W.W. Chen, L. Balaj, L.M. Liau, M.L. Samuels, S.K. Kotsopoulos, C.A. Maguire,
et al., BEAMing and droplet digital PCR analysis of mutant IDH1 mRNA in
glioma patient serum and cerebrospinal fluid extracellular vesicles, Mol. Ther.
Nucleic. Acids 2 (2013) e109.
[24] M.J. Higgins, D. Jelovac, E. Barnathan, B. Blair, S. Slater, P. Powers, et al.,
Detection of tumor PIK3CA status in metastatic breast cancer using peripheral
blood, Clin. Cancer Res. 18 (2012) 3462–3469.
[25] M. Holdhoff, K. Schmidt, F. Diehl, N. Aggrawal, P. Angenendt, K. Romans, et al.,
Detection of tumor DNA at the margins of colorectal cancer liver metastasis,
Clin. Cancer Res. 17 (2011) 3551–3557.
[26] J.A. Beaver, D. Jelovac, S. Balukrishna, R.L. Cochran, S. Croessmann, D.J.
Zabransky, et al., Detection of cancer DNA in plasma of patients with
early-stage breast cancer, Clin. Cancer Res. 20 (2014) 2643–2650.
[27] G.R. Oxnard, C.P. Paweletz, Y. Kuang, S.L. Mach, A. O’Connell, M.M. Messineo,
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. 20 (2014) 1698–1705.
515
[28] M.R. Finlay, M. Anderton, S. Ashton, P. Ballard, P.A. Bethel, M.R. Box, et al.,
Discovery of a potent and selective EGFR inhibitor (AZD9291) of both
sensitizing and T790M resistance mutations that spares the wild type form of
the receptor, J. Med. Chem. 57 (2014) 8249–8267.
[29] H.A. Yu, M.E. Arcila, N. Rekhtman, C.S. Sima, M.F. Zakowski, W. Pao, 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. 19
(2013) 2240–2247.
[30] N. Katakami, S. Atagi, K. Goto, T. Hida, T. Horai, A. Inoue, et al., LUX-Lung 4: a
phase II trial of afatinib in patients with advanced non-small-cell lung cancer
who progressed during prior treatment with erlotinib, gefitinib, or both, J.
Clin. Oncol. 31 (2013) 3335–3341.
[31] S.S. Ramalingam, P.A. Janne, T. Mok, K. O’Byrne, M.J. Boyer, J. Von Pawel, et al.,
Dacomitinib versus erlotinib in patients with advanced-stage, previously
treated non-small-cell lung cancer (ARCHER 1009): a randomised,
double-blind, phase 3 trial, Lancet Oncol. 15 (2014) 1369–1378.
[32] M. Marcq, A. Vallee, A. Bizieux, M.G. Denis, Detection of EGFR mutations in the
plasma of patients with lung adenocarcinoma for real-time monitoring of
therapeutic response to tyrosine kinase inhibitors? J. Thorac. Oncol. 9 (2014)
e49–e50.
[33] J.A. Shaw, J. Stebbing, Circulating free DNA in the management of breast
cancer, Ann. Transl. Med. 2 (2014) 3.