Monitoring of EGFR mutations in circulating tumor DNA of non‑small
cell lung cancer patients treated with EGFR inhibitors
R. B. Verheijen1  · T. T. van Duijl1
 · M. M. van den Heuvel2,3 · D. Vessies4
 · M. Muller3
 · J. H. Beijnen1,5 · J. M. Janssen1
 ·
J. H. M. Schellens5,6 · N. Steeghs6
 · D. van den Broek4
 · A. D. R. Huitema1,7
Received: 28 May 2020 / Accepted: 5 January 2021 / Published online: 23 January 2021
© The Author(s), under exclusive licence to Springer-Verlag GmbH, DE part of Springer Nature 2021
Abstract
Purpose We studied EGFR mutations in circulating tumor DNA (ctDNA) and explored their role in predicting the progres￾sion-free survival (PFS) of non-small cell lung cancer (NSCLC) patients treated with erlotinib or geftinib.
Methods The L858R, T790M mutations and exon 19 deletions were quantifed in plasma using digital droplet polymerase
chain reaction (ddPCR). The dynamics of ctDNA mutations over time and relationships with PFS were explored.
Results In total, 249 plasma samples (1–13 per patient) were available from 68 NSCLC patients. The T790M and L858R or
exon 19 deletion were found in the ctDNA of 49 and 56% patients, respectively. The median (range) concentration in these
samples were 7.3 (5.1–3688.7), 11.7 (5.1–12,393.3) and 27.9 (5.9–2896.7) copies/mL, respectively. Using local polynomial
regression, the number of copies of EGFR mutations per mL increased several months prior to progression on standard
response evaluation.
Conclusion This change was more pronounced for the driver mutations than for the resistance mutations. In conclusion,
quantifcation of EGFR mutations in plasma ctDNA was predictive of treatment outcomes in NSCLC patients. In particular,
an increase in driver mutation copy number could predict disease progression.
Keywords ctDNA · EGFR · NSCLC · Liquid biopsy · Erlotinib · Geftinib
Introduction
Non-small cell lung cancer (NSCLC) is the single most com￾mon histological subtype of lung cancer. Approximately,
5–20% of patients present with an activating mutation in
the epidermal growth factor receptor (EGFR) gene [1, 2].
The most abundant activating (or driver) mutations are the
L858R point mutation on exon 21 and deletions on exon 19
of the EGFR gene [3–5]. Once such a mutation is found in
the tumor, frst-line therapy consists of a tyrosine kinase
inhibitor (TKI) specifcally targeting these EGFR mutations
[6]. Erlotinib and geftinib are TKIs that are commonly used
for this purpose in the treatment of NSCLC. Despite the
fact that patients often show impressive initial responses to
treatments with these TKIs, resulting in signifcant improve￾ments in progression-free and overall survival, the tumor
will inevitably develop resistance and relapse [7–9].
The occurrence of the T790M mutation has been identi￾fed as a crucial resistance mutation, which has been shown
to account for a large proportion of the acquired resistance
to frst- and second-generation EGFR inhibitors [10–12].
* R. B. Verheijen
[email protected]
1 Department of Pharmacy & Pharmacology, The Netherlands
Cancer Institute – Antoni Van Leeuwenhoek, Louwesweg 6,
1066 EC Amsterdam, The Netherlands
2 Department of Respiratory Disease, Radboud University
Medical Centre, Nijmegen, The Netherlands
3 Department of Thoracic Oncology, The Netherlands
Cancer Institute – Antoni Van Leeuwenhoek, Amsterdam,
The Netherlands
4 Department of Laboratory Medicine, The Netherlands
Cancer Institute – Antoni Van Leeuwenhoek, Amsterdam,
The Netherlands
5 Department of Pharmaceutical Sciences, Utrecht University,
Utrecht, The Netherlands
6 Department of Medical Oncology and Clinical
Pharmacology, The Netherlands Cancer Institute – Antoni
Van Leeuwenhoek, Amsterdam, The Netherlands
7 Department of Clinical Pharmacy, University Medical Center
Utrecht, Utrecht, The Netherlands
270 Cancer Chemotherapy and Pharmacology (2021) 87:269–276
1 3
Although standard tumor genotyping is still being per￾formed by taking a needle or surgical biopsy from a tumor
lesion, the possibilities for ‘liquid’ biopsies, i.e. tumor gen￾otyping through analyzing tumor cells or DNA fragments
in the systemic circulation, are expanding rapidly [13, 14].
These technologies could be used as a more patient friendly
alternative to determine the EGFR genotype of tumors, for￾going the need for traditional invasive biopsies. Furthermore,
the less invasive nature of these types of techniques allows
for repeated measurements over time to study quantitative
changes before, during and after treatment with anti-cancer
drugs. An example of this liquid biopsy technology is using
digital droplet polymerase chain reaction (ddPCR) to detect
mutations in circulating tumor DNA (ctDNA) in blood or
plasma of patients with cancer [15, 16]. Using ddPCR, quan￾titative monitoring of selected genes such as the EGFR gene
during treatment, could for instance predict which patients
are likely to respond to treatment with targeted EGFR inhibi￾tors. Because an increase in plasma ctDNA could be a proxy
for increased tumor growth before this would be visible on
classical imaging-based periodic tumor assessments. Addi￾tionally, patients that may beneft from switching to another
therapy can be selected, as specifc inhibitors of the most
common T790M EGFR resistance mutation have become
available [17]. Even though it is apparent that techniques like
these will impact and improve the manner in which patients
with EGFR inhibitors are treated, the precise ways in which
ctDNA monitoring could guide treatment remains unclear.
The aim of this study was to measure the most impor￾tant EGFR driver and resistance mutations in ctDNA in a
cohort of NSCLC patients treated with the EGFR inhibi￾tors erlotinib and geftinib. The ultimate goal was to analyze
the quantitative dynamics of these mutations over time and
explore the roles of EGFR driver and resistance mutations in
predicting disease progression at a population level.
Subjects and methods
Patient population
An observational study was performed in the outpatient
clinic of the Netherlands Cancer Institute, Amsterdam,
The Netherlands. All patients with NSCLC who received
erlotinib or geftinib as frst-line anti-EGFR therapy for
whom at least one plasma sample was available were
included. Clinical visits and response evaluations were
scheduled in accordance with standard treatment guide￾lines. Clinical characteristics including demographic
data, medical history, tumor characteristics (stage, EGFR
mutational status), erlotinib and/or gefitinib dose and
administration schedule, plasma sampling date, treatment
duration, reason for discontinuation and progression-free
survival (PFS) were collected retrospectively from medi￾cal records. For this retrospective observational study, no
informed consent was required in accordance with code of
conduct for responsible use of human tissue and medical
research [18].
Sample collection
Surplus plasma was collected from samples obtained during
treatment with an EGFR inhibitor as part of routine care.
These blood samples had been collected into K2EDTA tubes,
centrifuged to plasma using a single 5-min cycle at 1000 G
and stored at (at least) − 20 °C until DNA isolation.
Circulating DNA analysis
Before analysis, the plasma samples were thawed, briefy
vortexed and spun at room temperature at 380 G using the
Heraeus Multifuge 3SR. Then, for mutation analysis, cell￾free DNA was extracted from plasma samples (circa 1 mL)
in elution buffer using QIAsymphony circulating DNA
Nucleic Acid Kits (Qiagen). Quantifcation of the a priori
selected EGFR mutants (T790M, L858R and exon 19 dele￾tions) in purifed DNA was performed using ddPCR assays.
To detect EGFR target alleles, TaqMan hydrolysis mutant
(T790M or L858R) and wildtype probes labelled with FAM/
HEX were used (Bio-Rad). Variants of in-frame exon 19
deletions were detected using a FAM/HEX drop-of assay
(PrimeTime, Integrated DNA Technologies). This assay
enabled detection and quantifcation of the following frame
deletions and indels: c.2235_2249del15, c.2236_2250del15,
c.2239_2256del18, c.2239-2251del13 ins C, c.2240-
2254del15 and c.2240-2257del18 [19].
Samples were partitioned into circa 20.000 water in oil
droplets using a Droplet Generator (Bio-Rad). The target
gene was amplifed in each droplet with PCR thermocycling
using a C1000 Touch Thermal Cycler. The cycles tempera￾tures used were, 95 °C for polymerase activation and dena￾turation, 55 °C for annealing, 98 °C for incubation and 12 °C
for the fnal hold. FAM/HEX fuorescence intensity was
measured for every droplet with a QX100 Droplet Reader
(Bio-Rad). In a two-dimensional fuorescence intensity his￾togram, thresholds were set manually (lasso tool) accord￾ing to amplitude guidelines within our lab. ddPCR results
were evaluated based on the quantity of positive droplets for
mutation and wildtype, double positive droplets and total
accepted droplets in the assay. The read-out was converted to
allele concentration using the initial plasma sample volume
from which the ctDNA was extracted. Absolute quantifca￾tion of the mutation was presented as copies of mutant allele
per mL plasma.
Cancer Chemotherapy and Pharmacology (2021) 87:269–276 271
1 3
Data analysis and statistics
The distribution of the concentrations (mutant copies/mL
plasma) in ctDNA were studied for each of the selected
EGFR mutations. The dynamics of these ctDNA mutations
over time were examined by plotting the ctDNA concentra￾tions versus the normalized time to progression of disease
(as measured by routine CT scans). Trends over time for
non-zero values would be identifed using locally weighted
polynomial regression or LOESS regression fts in resulting
scatter plots.
Exploratory survival analyses were conducted to inves￾tigate the relationship between EGFR copy numbers in
ctDNA and PFS. Specifcally, the relationship between the
occurrence of the T790M mutation at any time during treat￾ment were studied. Also, comparisons were made between
patients whose EGFR driver mutations reached concentra￾tions below the limit of detection versus patients whose
mutations were above this limit in all available samples. All
statistical analyses were performed in R 3.2.2 [20].
Results
Patient population
From September 2012 to March 2016, 68 patients with
NSCLC were included. Descriptive characteristics of the
included patients are shown in Table 1. Based on solid biop￾sies at diagnosis, 37 patients (54%) carried a deletion in exon
19, 22 patients (32%) harbored an EGFR L858R mutation
and for 9 patients (13%) the original driver EGFR mutation
was uncommon or unknown. In our cohort, median PFS on
anti-EGFR therapy was 14.1 (range 1.2–93.7) months and
at time of analysis 10 patients had ongoing response on erlo￾tinib or geftinib.
EGFR mutations in ctDNA
Of the 68 evaluable patients, a total of 249 plasma samples
were collected during treatment and samples were distrib￾uted among the patients with a mean of 3.7 (range 1–13)
samples per patient.
The T790M and L858R or exon 19 deletion were found
in the ctDNA of 49 and 56% patients, respectively. The
median (range) concentrations were 7.3 (5.1–3688.7), 11.7
(5.1–12,393.3) and 27.9 (5.9–2896.7) for the T790M, exon
19 del and L858R mutation, respectively.
For 33 patients (49%) the EGFR T790M mutant was
detected in at least one plasma sample collected during frst
TKI treatment (. The median concentration of EGFR
T790M mutant positive samples was 7.3 (range 5.1–3688.7)
copies per mL plasma (
In total, 59 patients had the L858R or exon 19 del activat￾ing mutations detected in solid biopsies (Table 1). For 54
of these, additional plasma was available for quantitative
mutation analysis of the original EGFR driver mutation. In
30 patients of these 54 patients (56%), the L858R or exon
19 deletion mutant was detectable in at least one plasma
sample.
The median concentration of the mutant allele was 11.7
(range 5.1–12,393.3) and 27.9 (range 5.9–2896.7) copies
per mL plasma, for the exon 19 deletion and L858R mutant,
respectively . In 42 of the 54 patients (78%), the
concentration of the EGFR activating mutation was or even￾tually dropped under the limit of detection in ctDNA during
treatment.
Dynamics of EGFR mutations in ctDNA over time
To explore the association between occurrence of EGFR
mutations in plasma and treatment response, data of all
patients were combined and normalized at the date of pro￾gression on CT scan (t=0). Separate plots were made for the
EGFR resistance and activating mutations
Table 1 Characteristics of included patients, n=68
a
These were the EGFR activating mutations based on solid biopsies
taken as part of routine patient care
*G719A (c.2156G>C); L747P (2239_2240 TT>CC); A840T
(c.2518G>A); L861Q (c.2582 T>A)
b
These patients switched from erlotinib to geftinib or vice versa due
to toxicity
Patients, n (%), (range)
Age (median in years) 62 (37–83)
Gender
Female 50 (74%)
Male 18 (26%)
Stage at start treatment
IIIA 5 (7%)
IIIB 8 (12%)
IV 55 (81%)
EGFR driver mutationa
exon 19 del 37 (54%)
L858R 22 (32%)
Other/uncommon* 4 (6%)
Unknown 5 (7%)
Treatment
Erlotinib 38 (56%)
Geftinib 24 (35%)
Erlotinib & Geftinibb 6 (9%)
Previous lines of therapy
None 48 (71%)
Chemotherapy 15 (22%)
Other/unknown 5 (7%)
272 Cancer Chemotherapy and Pharmacology (2021) 87:269–276
1 3
After application of local regression to the data, a modest
yet distinct increase in T790M mutant concentrations was
noticed approximately 80–120 days before progression of
disease observed on CT scan . Remarkably, two
patients with very high T790M mutation concentrations
showed rapid relapse of the tumor as shown in
The same plot is shown for the L858R or exon19 deletion
in Here, a clear rise in EGFR copy numbers is seen
approximately 5 months before progression using standard
response evaluation.
Survival analyses
Survival analyses using the EGFR driver mutations did not
result in a signifcant relationship with PFS. In total, 42
patients reached concentrations below the limit of detec￾tion during treatment, whilst 12 did not. Median PFS was
14.0 months in the former versus 11.8 months in the latter
group (p=0.134). Detection of the T790M mutation at any
time during treatment was also unable to predict PFS in
Kaplan–Meier analysis (median 14.0 months, n=35 versus
14.2 months n=33, p=0.648).
Discussion
We show the feasibility of ddPCR quantifcation of EGFR
driver and resistance mutations in ctDNA of NSCLC
patients treated with erlotinib and/or geftinib. In contrast to
what may have been expected, the occurrence of detectable
(at any concentration above the limit of detection) T790M
levels in plasma did not seem to predict imminent treatment
failure (Fig. 2). In Kaplan–Meier analysis, measurement
of T790M (without specifying the sampling time point)
also did not predict shorter PFS. The presence of a T790M
resistant sub clone in the tumor apparently does not translate
Fig. 1 EGFR mutations in patients with a positive ddPRC tests (in
copies/mL plasma) plotted on a log scale for a cohort of non-small
cell lung cancer patients treated with EGFR inhibitors. The T790M
and L858R or exon 19 deletion were found in the ctDNA of 49 and
56% patients, respectively. The median (range) concentrations were
7.3 (5.1–3688.7), 11.7 (5.1–12,393.3) and 27.9 (5.9–2896.7) for the
T790M, exon 19 del and L858R mutation, respectively
Cancer Chemotherapy and Pharmacology (2021) 87:269–276 273
1 3
into immediate progression of tumor growth as measured
by standard imaging techniques. These observations are
supported by previous preclinical studies into the growth
dynamics of T790M mutated tumor cells which, though
resistant to therapy, exhibited remarkably slow growth rates
[21]. However, very high concentrations (>100 copies/mL)
of T790M in plasma did seem to result in rapid relapse of
the tumor, as illustrated by two cases in our cohort (Fig. 2).
In a separate analysis, changes over time in concentra￾tions of the EGFR mutations in ctDNA at a population level
seemed to be able to predict clinical progression several
months before progression of disease was determined using
standard CT scans (Figs. 2 and 3). This efect seemed par￾ticularly pronounced for the L858R mutation and exon 19
deletions (Fig. 3). This observation indicates that quantita￾tive monitoring of EGFR activating mutations could serve as
an early predictor of disease progression and could possibly
serve a role similar to that prostate specifc antigen in pros￾tate cancer. However, future prospective studies are needed
to confrm this hypothesis.
These early signals of progressive disease could enable
an early switch to second line treatment, in particular to an
EGFR inhibitor that also targets the T790M mutations such
as osimertinib [17]. If such an early change in treatment
translates into increased patient survival is still unknown.
This efect is studied in the currently ongoing APPLE trial.
In this trial, patients are randomized to start directly on
osimertinib, an EGFR inhibitor specifcally targeting the
T790M mutation (arm A), or to geftinib until the T790M
mutation is detected in ctDNA (arm B) or to geftinib until
progressive disease on CT is observed (arm C) [22]. Trials
like these will increasingly be needed to study the efects of
treatment changes based on ctDNA measurements on patient
outcomes.
Fig. 2 Semi-log scatter plot of the concentration (in copies/mL
plasma) of the EGFR T790M mutation, over time before disease
progression (in days). Each dot represents a single sample from a
patient. The line indicates the locally weighted polynomial regression
or LOESS regression plus 95% confdence interval (shaded area).
Approximately 125  days before clinical progression, an increase in
the mutant concentration is observable, suggesting the possibility of
early prediction of treatment failure using T790M concentrations
274 Cancer Chemotherapy and Pharmacology (2021) 87:269–276
1 3
Osimertinib has shown efficacy in untreated EGFR
mutant lung cancer [23]. Osimertinib showed similar PFS
for patients with both the L858R and exon 19 deletions, sup￾porting the combination of these two mutations as grouped
driver mutations in our analysis related to imaging-based
progression (although the efect of the EGFR driver muta￾tion on overall survival could be diferent between the two
driver mutations) [23].
Limitations of our study include the retrospective ‘real￾world’ nature of our analysis. This means ctDNA samples
and imaging-based tumor assessments were not taken/per￾formed at set prespecifed intervals. Therefore, although
our analysis clearly demonstrates the predictive potential of
ctDNA (Figs. 2 and 3), the exact timing should be inter￾preted with caution and would beneft from confrmation in
prospective cohort study. Other limitations of this real-world
analysis are the limited sample volume, as a higher sample
volume could have increased sensitivity and the fact that in
routine patient care clinical information is not captured as
comprehensively as it may have been in a prospective rand￾omized clinical trial.
Other studies support the value of monitoring ctDNA of
EGFR inhibitors. Many have focused on presenting indi￾vidual cases or case series, but some have also performed
population level analyses. For example, a cohort of NSCLC
patients treated with afatinib showed a rapid decline in
EGFR mutant alleles upon start of therapy and increasing
again upon disease progression [24]. A Korean cohort of
NSCLC patients treated with erlotinib and geftinib found
that patients that those patients who cleared ctDNA upon
initiation of treated had a signifcantly longer PFS and OS
[16]. These examples, together with our study support the
value of longitudinal EGFR monitoring in NSCLC patients
treated with EGFR inhibitors.
Fig. 3 Semi-log scatter plot of the concentration (in copies/mL
plasma) of the EGFR driver mutations, L858R and exon 19 del over
time before disease progression (in days). Each dot represents a single
sample from a patient. The line indicates the locally weighted poly￾nomial regression or LOESS regression plus 95% confdence interval
(shaded area). Approximately, 150  days before clinical progression,
an increase in the mutant concentration is observable, suggesting the
possibility of early prediction of treatment failure
Cancer Chemotherapy and Pharmacology (2021) 87:269–276 275
1 3
Although our study focused on quantifcation of the most
important EGFR activating and resistance mutations, future
studies could also take non-classical EGFR activating muta￾tions or mutations in non-EGFR genes such as KRAS into
account. As these mutations have previously been linked to
outcomes in NSCLC patients treated with EGFR inhibitors
[4, 25].
Future eforts could also focus on using more advanced
mathematical methods to explore the quantitative nature of
ctDNA dynamics and treatment outcomes. Here, the example
of population pharmacokinetic analysis could be followed. In
particular non-linear mixed efect modelling, which is at the
moment mainly employed for pharmacokinetic and pharmaco￾dynamic data, could be an attractive strategy to this end [26].
Conclusion
We show that ddPCR quantifcation at a population level of
EGFR activating and resistance mutations in plasma ctDNA
could be a relevant predictor of treatment outcomes in
NSCLC patients. In particular, an increase in the copies/mL
of the EGFR driver mutation over time may predict clinical
progression, suggesting ctDNA monitoring could be used as
an early read-out of treatment failure. ctDNA monitoring of
EGFR activating mutations could, therefore, possibly serve a
role similar to that prostate specifc antigen in prostate cancer.
However, future prospective studies are needed to confrm this
hypothesis.
Funding No funding was received for this research.
Data availability The data of this study are available from the corre￾sponding author upon reasonable request.
Compliance with ethical standards
Conflict of interest Remy Verheijen is an employee and shareholder of
AstraZeneca and Johnson & Johnson. All other authors declare they
have no conficts to disclose.
Informed consent to participate For this retrospective observational
study, no informed consent was required in accordance with code of
conduct for responsible use of human tissue and medical research [18].
Informed consent to publish For this retrospective observational study,
no informed consent was required in accordance with code of conduct
for responsible use of human tissue and medical research [18].
References
1. Dearden S, Stevens J, Wu YL, Blowers D (2013) Mutation
incidence and coincidence in non small-cell lung cancer: meta￾analyses by ethnicity and histology (mutMap). Ann Oncol
24:2371–2376
2. Boch C, Kollmeier J, Roth A, Stephan-Falkenau S, Misch D,
Gruning W et al (2013) The frequency of EGFR and KRAS
mutations in non-small cell lung cancer (NSCLC): routine
screening data for central Europe from a cohort study. BMJ
Open 3:e002560–e002560
3. Arcila ME, Nafa K, Chaft JE, Rekhtman N, Lau C, Reva BA
et al (2013) EGFR exon 20 insertion mutations in lung adeno￾carcinomas: prevalence, molecular heterogeneity, and clinico￾pathologic characteristics. Mol Cancer Ther 12:220–229
4. Kuiper JL, Hashemi SMS, Thunnissen E, Snijders PJF, Grünberg
K, Bloemena E et al (2016) Non-classic EGFR mutations in a
cohort of Dutch EGFR-mutated NSCLC patients and outcomes
following EGFR-TKI treatment. Br J Cancer 115:1504–1512
5. Murray S, Dahabreh IJ, Linardou H, Manoloukos M, Bafaloukos
D, Kosmidis P (2008) Somatic mutations of the tyrosine kinase
domain of epidermal growth factor receptor and tyrosine kinase
inhibitor response to TKIs in non-small cell lung cancer: an
analytical database. J Thorac Oncol 3:832–839
6. Novello S, Barlesi F, Califano R, Cufer T, Ekman S, Levra MG
et al (2016) Metastatic non-small-cell lung cancer: ESMO clini￾cal practice guidelines for diagnosis, treatment and follow-up.
Ann Oncol 27:V1–V27
7. Zhou C, Wu YL, Chen G, Feng J, Liu XQ, Wang C et al (2015)
Final overall survival results from a randomised, phase III study
of erlotinib versus chemotherapy as frst-line treatment of EGFR
mutation-positive advanced non-small-cell lung cancer (OPTI￾MAL, CTONG-0802). Ann Oncol 26:1877–1883
8. Zhao H, Fan Y, Ma S, Song X, Han B, Cheng Y et al (2015)
Final overall survival results from a phase III, randomized,
placebo-controlled, parallel-group study of geftinib versus pla￾cebo as maintenance therapy in patients with locally advanced
or metastatic non–small-cell lung cancer (INFORM; C-TONG
0804). J Thorac Oncol 10:655–664
9. Greenhalgh J, Dwan K, Boland A, Bates V, Vecchio F, Dun￾dar Y et al (2016) First-line treatment of advanced epidermal
growth factor receptor (EGFR) mutation positive non-squamous
non-small cell lung cancer (review). Cochrane Database Syst
Rev.
10. Sequist LV, Waltman BA, Dias-Santagata D, Digumarthy S,
Turke AB, Fidias P et al (2011) Genotypic and histological evo￾lution of lung cancers acquiring resistance to EGFR inhibitors.
Sci Transl Med 3:75ra26
11. Yu HA, Arcila ME, Hellmann MD, Kris MG, Ladanyi M, Riely
GJ (2014) Poor response to erlotinib in patients with tumors
containing baseline EGFR T790M mutations found by routine
clinical molecular testing. Ann Oncol 25:423–428
12. Wang Z, Chen R, Wang S, Zhong J, Wu M, Zhao J et al (2014)
Quantifcation and dynamic monitoring of EGFR T790M in
plasma cell-free DNA by digital PCR for prognosis of EGFR￾TKI treatment in advanced NSCLC. PLoS ONE 9:1–7
13. Crowley E, Di Nicolantonio F, Loupakis F, Bardelli A (2013)
Liquid biopsy: monitoring cancer-genetics in the blood. Nat Rev
Clin Oncol 10:472–484
14. Murtaza M, Dawson S-J, Tsui DWY, Gale D, Forshew T,
Piskorz AM et al (2013) Non-invasive analysis of acquired
resistance to cancer therapy by sequencing of plasma DNA.
Nature 497:108–112
15. Oxnard GR, Paweletz CP, Kuang Y, Mach SL, O’Connell A,
Messineo MM et al (2014) Noninvasive detection of response
and resistance in egfrmutant lung cancer using quantitative
next-generation genotyping of cell-free plasma DNA. Clin
Cancer Res 20:1698–1705
16. Lee JY, Qing X, Xiumin W, Yali B, Chi S, Bak SH et al (2016)
Longitudinal monitoring of EGFR mutations in plasma pre￾dicts outcomes of NSCLC patients treated with EGFR TKIs:
276 Cancer Chemotherapy and Pharmacology (2021) 87:269–276
1 3
Korean Lung Cancer Consortium (KLCC-12-02). Oncotarget
7:6984–6993
17. Mok TS, Wu Y-L, Ahn M-J, Garassino MC, Kim HR, Ramal￾ingam SS et al (2016) Osimertinib or platinum-pemetrexed in
EGFR T790M–positive lung cancer. N Engl J Med.
18. Federation of Dutch Medical Scientifc Societies (2011) Human
tissue and medical research: code of conduct for responsible use.

http://www.federa.org/sites/default/files/digital_version_first

_part_code_of_conduct_in_uk_2011_12092012.pdf
19. Yung TK, Chan KA, Mok TS, Tong J, To KF, Lo YD (2009)
Single-molecule detection of epidermal growth factor receptor
mutations in plasma by microfuidics digital PCR in non–small
cell lung cancer patients. Clin Cancer Res 15:2076–2084
20. R Core Development Team (2016) A Language and environment
for statistical computing. Vienna, Austria R Found. Stat. Comput.

https://www.r-project.org/

21. Chmielecki J, Foo J, Oxnard GR, Hutchinson K, Ohashi K, Som￾war R et al (2011) Optimization of dosing for EGFR-mutant non￾small cell lung cancer with evolutionary cancer modeling. Sci
Transl Med.
22. Remon J, Menis J, Hasan B, Peric A, De Maio E, Novello S et al
(2017) The APPLE trial: feasibility and activity of AZD9291 (osi￾mertinib) treatment on positive plasma T790M in EGFR -mutant
NSCLC patients. EORTC 1613. Clin Lung Cancer 9291:1–6
23. Soria JC, Ohe Y, Vansteenkiste J, Reungwetwattana T, Chewasku￾lyong B, Lee KH et al (2018) Osimertinib in untreated EGFR￾mutated advanced non–small-cell lung cancer. N Engl J Med
378:113–125
24. Iwama E, Sakai K, Harada T, Harada D, Nosaki K, Hotta K et al
(2017) Monitoring of somatic mutations in circulating cell-free
DNA by digital PCR and next-generation sequencing during
afatinib treatment in patients with lung adenocarcinoma positive
for EGFR activating mutations. Ann Oncol 28:136–141
25. Brugger W, Triller N, Blasinska-Morawiec M, Curescu S, Saka￾lauskas R, Manikhas GM et al (2011) Prospective molecular
marker analyses of EGFR and KRAS from a randomized, placebo￾controlled study of erlotinib maintenance therapy in advanced
non-small-cell lung cancer. J Clin Oncol 29:4113–4120
26. Barbolosi D, Ciccolini J, Lacarelle B, Barlési F, André N (2015)
Computational oncology—mathematical modelling of Erlotinib drug regi￾mens for precision medicine. Nat Rev Clin Oncol 13(4):242–254
Publisher’s Note Springer Nature remains neutral with regard to
jurisdictional claims in published maps and institutional afliations.