Lung Cancer 96 (2016) 101–107

Contents lists available at ScienceDirect

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

EGFR mutation status on brain metastases from non-small cell lung

cancer
Fred Hsu a,∗ , Alex De Caluwe b , David Anderson a , Alan Nichol c , Ted Toriumi a , Cheryl Ho c
a
Abbotsford Centre, BC Cancer Agency, British Columbia, Canada
b
Department of Radiation Oncology, Jules Bordet Institute, Brussels, Belgium
c
Vancouver Centre, BC Cancer Agency, British Columbia, Canada

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

Article history: Objectives: The purpose of this study was to examine the impact of EGFR mutations on the incidence of
Received 16 December 2015 brain metastases in patients with advanced non-small cell lung cancer (NSCLC).
Received in revised form 3 March 2016 Materials and methods: A retrospective, population-based study was conducted using a provincial cancer
Accepted 5 April 2016
registry to identify patients with metastatic NSCLC. Patients with diagnostic EGFR mutation testing were
divided into EGFR mutation positive (EGFR+) and EGFR wild type (WT) cohorts. The primary endpoint
Keywords:
was the incidence of brain metastases. Cumulative incidence curves were estimated using the competing
EGFR
risk method. The secondary endpoint was overall survival.
Lung cancer
Brain metastases
Results: For 543 patients there were 121 EGFR+ and 422 EGFR WT. The cumulative incidence of brain
Chemotherapy metastases was 39.2% for EGFR+ patients compared to 28.2% for EGFR WT (p = 0.038; HR 1.4). In multi-
TKI variate analysis, younger age and EGFR+ status were significant factors for developing brain metastases.
The median survival for the EGFR+ and EGFR WT cohorts were 22.4 and 7.9 months (p < 0.001), respec-
tively. In multivariate analysis, poor performance status and brain metastases were factors significant
for worse survival.
Conclusions: There is a higher incidence of brain metastases for patients with EGFR+ metastatic NSCLC,
even when adjusted for differences in survival, compared to EGFR WT. For patients with and without
brain metastases, survival prognosis with stage IV NSCLC is much longer with EGFR+ disease.
© 2016 Elsevier Ireland Ltd. All rights reserved.

1. Introduction occur in approximately 15% of the advanced non-squamous NSCLC

Brain metastases are a common problem in patients with lung
cancer and are associated with a poor prognosis. For non-small
cell lung cancer (NSCLC), brain metastases have been reported to
affect about 20–40% of patients [1,2]. However, as improvements
are made in the treatment of NSCLC, patients are living longer with
a greater opportunity to develop spread to the brain. There is a lack
of contemporary data on the incidence of brain metastases, partic-
ularly in an era of molecular markers. Differences in tumor biology
may impact the pattern of metastases to the brain, putting some
patients at greater risk than others.
The epidermal growth factor receptor (EGFR) is a trans-
membrane receptor tyrosine kinase that is involved in various
cellular processes. Mutations in the EGFR tyrosine kinase domain
∗ Corresponding author at: 32900 Marshall Road, Abbotsford, British Columbia
V2S 0C2, Canada.
E-mail address: fhsu@bccancer.bc.ca (F. Hsu).
population. Exon 19 deletions and exon 21 mutations account
for the large majority of these mutations. Mutations in EGFR
are recognized as a positive prognostic factor and are associated
with improved treatment response and progression free survival
with EGFR tyrosine kinase inhibitors (TKI) compared to standard
chemotherapy in several randomized trials [3–6].
The impact of EGFR mutations on the incidence of brain metas-
tases could have implications for brain screening, surveillance,
and prophylactic treatment. Identification of high risk groups for
strategies of early brain metastases detection could have treatment
advantages. When brain metastases are detected while small in
size and few in number, stereotactic radiosurgery for individual
brain metastases can be used without whole brain radiotherapy
(WBRT) [7]. In this approach, patients could avoid the adverse
neuro-cognitive effects associated with WBRT. Prophylactic cranial
irradiation (PCI) for patients with locally advanced NSCLC decreases
the incidence of brain metastases [8], but is not routinely used
because it causes toxicity without improving survival. However,

http://dx.doi.org/10.1016/j.lungcan.2016.04.004
0169-5002/© 2016 Elsevier Ireland Ltd. All rights reserved.
102 F. Hsu et al. / Lung Cancer 96 (2016) 101–107
PCI might be advantageous for an identifiable subgroup of patients
with a high risk of brain metastases.
The purpose of this study was to examine the significance of
EGFR mutation status on the incidence of brain metastases in a pop-
ulation of patients with metastatic lung cancer. We also evaluate
the influence of systemic therapy (chemotherapy and EGFR-TKI)
on the development of brain metastases and the impact of differ-
ing survival with EGFR mutation status on the incidence of brain
metastases.
2. Methods
2.1. Patient selection
A retrospective, population-based study was conducted using
a Provincial cancer registry (British Columbia, Canada) to identify
patients with metastatic non-squamous NSCLC diagnosed between
March 2010 and March 2012, to give a minimum potential follow-
up of three years. The study inclusion criteria were stage IV disease
at initial diagnosis and conclusive EGFR mutation testing. Patients
with non-diagnostic mutation tests were excluded. There were
1373 patients diagnosed with stage IV NSCLC in British Columbia
for the period of study. After excluding 746 patients without EGFR
mutation testing and 84 patients with non-diagnostic EGFR test
results, there were 543 patients included for this study. Individual
patient medical records were reviewed for patient characteristics
(age, performance status, gender, Asian ethnicity, smoking history),
disease characteristics and imaging evidence of brain metastases
from the time of initial diagnosis to the time of death. At our
institution, the approved indication for first-line EGFR-TKI ther-
apy was in EGFR mutation positive (EGFR+) stage IV disease. Brain
imaging at initial presentation, using computed tomography (CT)
or magnetic resonance imaging (MRI), was done at the discretion
of the treating physician and as clinically indicated. This includes
brain screening for asymptomatic disease and diagnostic evalua-
tion of symptomatic disease. Brain imaging was not part of standard
follow-up unless patients were symptomatic.
2.2. Mutation analysis
Patients were divided into EGFR+ and EGFR wild type (WT)
cohorts for mutations on exon 19 and 21. Mutation analysis was
performed in a central laboratory using DNA extracted from tumor
tissue isolated from suitable blocks. DNA quantization was per-
formed by UV spectrophotometry (Nanodrop, USA). Detection of
EGFR exon 19 insertions/deletions and exon 21 L858R mutations
were performed with a minimum of 400 ng of DNA by polymerase
chain reaction (PCR) using gene specific PCR amplification primers,
and is previously described by Pan et al. [9].
2.3. Study endpoints
The primary endpoint was the cumulative incidence of brain
metastases. The time to brain metastases was measured from
the date of pathologic diagnosis to the date of brain imaging
with computed tomography or magnetic resonance imaging that
demonstrated metastases. The secondary endpoint was overall sur-
vival (OS), measured from the time of pathologic diagnosis to death.
2.4. Statistical analysis
Patient and disease characteristics were compared between
EGFR cohorts using the Fisher Exact test for categorical variables
and the Kruskal-Wallis rank-sum test for continuous variables. A
competing risk method was used to estimate the cumulative inci-
dence of brain metastases. In this analysis, patient death before
developing brain metastases was considered a competing risk
event. Patients who had not developed a brain metastasis and had
not died were censored at the time of last follow-up. Gray’s test
was used to test for differences in the cumulative incidence curves.
The association between brain metastases, EGFR mutation status,
and patient characteristics were assessed using a proportional haz-
ards model (Fine-Gray model). The OS was estimated using the
Kaplan-Meier method and survival was compared using the log-
rank test. The Cox proportional hazards model was used to assess
association between OS, EGFR mutation status, and patient charac-
teristics. All reported p-values are two-sided, with a p-value < 0.05
set at the level of significance. Confidence intervals (CI) are 95%.
Analyses were performed using the R Statistical Language (version
2.1.5; http://cran.r-project.org/). The study was approved by the
Research Ethics Board of the British Columbia Cancer Agency.
3. Results
3.1. Patient characteristics
For 543 patients there were 121 (22.3%) EGFR+ and 422 (77.7%)
EGFR WT. For the EGFR+ cohort, 73 (60%) patients had exon
19 deletion and 48 (40%) had exon 21 mutations. The median
follow-up time for living patients was 34.9 months. Patient age
(median 66 years, range 30–91) and gender were no different
between EGFR+ and EGFR WT cohorts. The proportion of patients
who received chemotherapy was no different between EGFR
cohorts. The most frequent first-line metastatic chemotherapy
regimens were: Cisplatinum-Gemcitabine (29.3%), Carboplatinum-
Gemcitabine (14.2%), and Carboplatinum-Pemetrexed (12.1%). The
proportion of patients who received an EGFR-TKI was different
between EGFR+ (87%) versus EGFR WT (28%), p < 0.001. First-line
EGFR-TKIs were Erlotinib in 56.8% and Gefitinib in 40.5%. The
patient characteristics are presented in Table 1.
3.2. Brain metastases
One hundred forty three (26.3%) patients had brain metas-
tases at initial presentation. This was not different between EGFR
cohorts: 37 (30.6%) for EGFR+ and 106 (25.1%) for EGFR WT, p = 0.36.
In 69 (12.7%) patients, the brain was the only site of metastases
at first presentation (13 (10.7%) EGFR+ vs. 56 (13.2%) EGFR WT,
p = 0.54). For the study population, the 1- and 3-year cumulative
incidence of brain metastases was significantly higher for the EGFR+
cohort compared to EGFR WT (39.0% vs. 28.1%, p = 0.041; and 39.2%
vs. 28.2%, p = 0.038), respectively. There was no difference in the
incidence of brain metastases between exon 19 and 21 subtypes.
In univariate and multivariate analysis, EGFR+ status (HR 1.41;
CI = 1.04–1.92; p = 0.03) and younger age (HR 2.08; CI = 1.54–2.83;
p < 0.001) were significant factors for brain metastases. Age < 66
years was chosen as the cut-off because it was the median age of
the study population. Performance status (p = 0.45), Asian ethnicity
(p = 0.54), and gender (p = 0.24) were not significant factors for brain
metastases in the univariate analysis. A Forest plot of the univariate
and multivariate analysis for brain metastases risk is presented in
Fig. 1. The cumulative incidence curves for brain metastases strat-
ified by EGFR mutation status and age are presented in Fig. 2. The
3-year cumulative incidence of brain metastases in EGFR+ patients
who were < 66 years was 59.4%.
In a subgroup analysis of patients without brain metas-
tases at initial diagnosis (n = 400), chemotherapy use (HR 1.05;
CI = 0.59–1.86; p = 0.88) and EGFR-TKI use (HR 1.48; CI = 0.84–2.63;
p = 0.18) were not significant factors in the subsequent devel-
opment of brain metastases (Fig. 1). This finding was the same
regardless of EGFR mutation status. For patients who developed
 F. Hsu et al. / Lung Cancer 96 (2016) 101–107 103

Table 1
Patient characteristics.

By EGFR mutation status

All (n = 543) EGFR WT (n = 422) EGFR+ (n = 121)

Characteristic No. % No. % No. % p-Value

Age at diagnosis (years)

Median 66 67 66 0.75a
Interquartile range 58–74 58–73 55–77

ECOG performance status

0 23 4 14 3 9 7 0.008b
1 229 42 166 39 63 52
2 160 30 129 31 31 26
3 116 21 100 24 16 13
4 15 3 13 3 2 2

Ethnicity
Asian 123 23 64 15 59 49 <0.001b
Non-Asian 420 77 358 85 62 51

Gender
Female 327 60 247 58 80 66 0.14b
Male 216 40 175 42 41 34

Smoking (pack years)

Median 21 30 0 <0.001a
Interquartile range 0–40 10–40 0–5

Chemotherapy
Yes 267 49 208 49 59 49 1b
No 276 51 214 51 62 51

TKI
Yes 222 41 117 28 105 87 <0.001b
No 321 59 305 72 16 13

Overall survival (months)

Median 10 7.9 22.4 <0.001c
95% CI 8.7–11.7 6.6–9.1 18.4–27.4

Abbreviations: EGFR, epidermal growth factor receptor; WT, wild type; +, mutation positive; n, number of patients; ECOG, Eastern Cooperative Oncology Group; TKI, tyrosine
kinase inhibitor; CI, confidence interval.
a
Using the Kruskal-Wallis rank-sum test.
b
Using the Fisher Exact test.
c
Using the log-rank test.

Fig. 1. Univariate and multivariate analysis for the risk of brain metastases. The Hazard Ratio of brain metastases was estimated in Fine-Gray’s Proportional Hazards
Competing Risks Regression Model. Inclusion for multivariate analysis was limited to variables significant in univariate analysis. *Subgroup analysis excluding patients with
brain metastases at initial presentation. Abbreviations: ECOG, Eastern Cooperative Oncology Group performance status; EGFR, epidermal growth factor receptor; +, mutation
positive; WT, wild type; TKI, tyrosine kinase inhibitor.
104 F. Hsu et al. / Lung Cancer 96 (2016) 101–107
Fig. 2. Cumulative incidence of brain metastases. The cumulative incidence was estimated using a competing risk model, with death as a competing risk event. Gray’s test
was used to compare cumulative incidence curves. Abbreviations: EGFR, epidermal growth factor receptor; +, mutation positive; WT, wild type.
brain metastases in this subgroup, the median time to brain
metastases for chemotherapy, EGFR-TKI, and no systemic therapy
patients were 14.7, 16.5, and 4.6 months, respectively.
3.3. Overall survival
The median OS was significantly longer for EGFR+ patients
compared to EGFR WT (22.4 vs. 7.9 months, p < 0.001). In uni-
variate and multivariate analysis (Fig. 3), EGFR+ status (HR of
death = 0.61; CI = 0.46–0.81; p < 0.001), chemotherapy use (HR 0.59,
CI = 0.48–0.73; p < 0.001), EGFR-TKI use (HR 0.70; CI = 0.54–0.89;
p = 0.004), and female gender (HR 0.80; CI = 0.66–0.96; p = 0.02)
were significant factors for longer survival. Poor performance
status (HR 1.53; CI = 1.27–1.85; p < 0.001) and brain metastases
(HR 1.45; CI = 1.19–1.77; p < 0.001) were significant for worse
survival. Younger age and Asian ethnicity was significant for
survival in univariate analysis but not in multivariate analysis
(p = 0.47 and p = 0.19, respectively). Kaplan-Meier curves showing
survival stratified by EGFR mutation status and systemic ther-
apy (chemotherapy + EGFR-TKI) use are presented in Fig. 4A. For
patients with brain metastases, the median survival from diagno-
sis was 12.4 months for EGFR+ and 5.8 months for EGFR WT cohorts,
p = 0.002 (Fig. 4B).
4. Discussion
In the modern era of chemotherapy and molecular-targeted
therapy, we report a cumulative incidence of brain metastases in
NSCLC of 28% for EGFR WT and 39% for EGFR mutation carriers
over a 3-year period. Because there is a significantly longer sur-
vival for EGFR+ patients, and therefore a longer period at risk for
the development of brain metastases, our analysis adjusted for the
differences in death rates between cohorts using a competing risk
method, with death as a competing risk event. With high mortal-
ity rates, the competing risk method has significant advantages in
providing a better estimation of incidence compared to the Kaplan-
Meier method or reporting in absolute numbers [10,11]. Our finding
with this methodology of analysis makes it likely that the underly-
ing biology of EGFR mutation status is responsible for the difference.
Further, differences in brain metastases rates were observed at both
1 and 3 years, with the indication that most brain metastases occur
within the first year of metastatic diagnosis regardless of EGFR
mutation status.
Literature about the relationship between EGFR mutation status
and brain metastases is limited. Doebele et al. [12] reported on a
cohort of 209 patients with metastatic NSCLC. For 39 EGFR+ patients
there were a higher number of liver metastases, but not pulmonary,
adrenal, bone or brain metastases. Stanic et al. [13] examined 629
Caucasian patients. Using the Kaplan-Meier method to estimate the
incidence of brain metastases, they found no difference between
EGFR+ and EGFR WT cohorts. In a case-control study by Hendriks
 F. Hsu et al. / Lung Cancer 96 (2016) 101–107 105

Fig. 3. Univariate and multivariate analysis for overall survival. Hazard Ratios were estimated in a Cox Proportional Hazards Regression Model. Inclusion for multivariate
analysis was limited to variables significant in univariate analysis. **Analyzed as a time-dependent variable. Abbreviations: ECOG, Eastern Cooperative Oncology Group
performance status; EGFR, epidermal growth factor receptor; +, mutation positive; WT, wild type; TKI, tyrosine kinase inhibitor; OS, overall survival.

A 1.0
ST, EGFR +
Median OS 95% CI
25.1 mo. 20.8−29.4 B 1.0
EGFR+
Median OS 95% CI
12.4 mo. 5.2−26.1
ST, EGFR WT 12.2 mo. 10.6−15.2 EGFR WT 5.8 mo. 4.4−7.5
No ST, EGFR + 4.7 mo. 1.1−13.5
No ST, EGFR WT 4.3 mo. 3.6−5.2
0.8 0.8
Proportion Surviving

Proportion Surviving

0.6 0.6

0.4 0.4

EGFR +
0.2 0.2

EGFR WT

0.0 p < 0.001 0.0 p = 0.002

0 1 2 3 4 0 1 2 3 4
Time from diagnosis (years) Time from diagnosis of brain metastasis (years)
Number at risk Number at risk
ST, EGFR + 109 88 55 24 4 EGFR + 47 24 16 8 3
ST, EGFR WT 211 109 58 20 3 EGFR WT 118 30 15 4 1
No ST, EGFR + 12 3 1 1 0
No ST, EGFR WT 211 42 14 7 2

Fig. 4. Kaplan-Meier curves showing overall survival. (A) Overall survival stratified by EGFR mutation status and systemic therapy use. (B) Overall survival after brain
metastases diagnosis by mutation status. Abbreviations: ST, systemic therapy; EGFR, epidermal growth factor receptor; +, mutation positive; WT, wild type; OS, overall
survival.

Table 2
Studies comparing brain metastases incidence and EGFR mutation status.

Author n (Total) n (EGFR+) Stage at initial Median Cumulative Statistical Association

diagnosis follow-up incidence of BM methodology between BM and
(months) (EGFR+) EGFR mutation

Doebele et al. [12] 209 39 72% IV n/a 23% (absolute number) ANOVA Not significant
Stanic et al. [13] 629 137 I–IV 53 39% (at 3 years from Fig. 1) Kaplan-Meier Not significant
Hendriks et al. [14] 189 62 87% IV n/a 32% (absolute number) ANOVA Not significant
Shin et al. [15] 314 138 I–IV 28.6 n/a Competing risk Significant
Current study 543 121 IV 34.9 39.2% (at 3 years) Competing risk Significant

Abbreviations: n, number of patients; EGFR, epidermal growth factor receptor; +, mutation positive; n/a, not available; BM, brain metastases; ANOVA, analysis of variance.
106 F. Hsu et al. / Lung Cancer 96 (2016) 101–107
et al. [14] the absolute number of brain and bone metastases was
not different between EGFR+ and EGFR WT cohorts. In contrast,
Shin et al. [15] reported a significant association between the fre-
quency of EGFR mutations and brain metastases, although they did
not estimate the overall incidence of brain metastases. In their sub-
group analysis, using a competing risk method, they also reported
an association between EGFR mutations and brain metastases risk
in patients who had curative resection. A summary of the studies
directly examining EGFR mutation status and brain metastases are
presented in Table 2.
Differences between studies could be due to differences in
patient selection or statistical methodology. We studied stage IV
patients at initial diagnosis because of the available EGFR mutation
data, which follows the approved indication for first-line EGFR-TKI
at our institution. Other studies were predominantly stage IV as
well, but did include a small number with earlier stages at pre-
sentation (Table 2). Consequently, we expect a higher incidence of
brain metastases at initial diagnosis compared to other series. In our
study, more than half of brain metastases diagnoses were at first
presentation. This is consistent with other reports that the brain
is one of the most frequent sites of initial disease failure [16,17].
However, this finding could also be explained by the detection
of asymptomatic brain metastases by brain screening during ini-
tial staging investigations. Despite differences in initial stage, our
cumulative incidence of brain metastases over 3 years was similar
to the comparably large study by Stanic et al. [13].
Supplemental evidence that EGFR mutation status affects the
biology of metastatic spread to the brain comes from studies report-
ing on brain imaging characteristics. A study by Eichler et al. [18]
reported that EGFR+ patients were more likely to have multiple
brain metastases compared to EGFR WT. Similarly, Sekine et al. [19]
reported that patients with exon 19 deletions were more likely to
have multiple and smaller brain metastases with less peritumoral
edema—a pattern resembling miliary brain metastases.
We also examined whether the higher rate of brain metastases
was the result of systemic therapy (chemotherapy and EGFR-
TKI). Patients who receive chemotherapy live longer, and effective
systemic control might drive failures intracranial if the brain is
viewed as a sanctuary site from drug therapy because of the blood-
brain-barrier. Conversely, there is evidence for the effectiveness of
EGFR-TKIs in the treatment of brain metastases, and EGFR-TKI use
may reduce the rates of symptomatic brain metastases [20]. Litera-
ture on the influence of EGFR-TKI use on brain metastases incidence
is conflicting. One study raised concern about the higher rate of
brain progression for stage III/IV patients who achieved a partial
response to EGFR-TKI treatment [21]. Another study concluded
a lower rate of brain progression, compared to historic reports,
for patients who were initially treated with an EGFR-TKI [22]. In
our subgroup analysis, which excluded patients with brain metas-
tases at initial diagnosis, chemotherapy and EGFR-TKI use did not
significantly impact the risk of developing brain metastases. How-
ever, the median time till brain metastases was much longer for
treated patients compared to untreated patients, suggesting that
chemotherapy and EGFR-TKIs delay the onset of symptomatic brain
metastases, although selection bias could partly explain this find-
ing.
Our study is population-based and included comparator EGFR+
and EGFR WT cohorts. It was different from other series that stud-
ied exclusively Asian or Caucasian cohorts because we examined
a population of mixed ethnicity. The other strength of our study
is the statistical methodology, which we feel is more robust than
reporting in absolute numbers. We also examined patient and dis-
ease subgroups to identify factors associated with a higher risk
of brain metastases. As with all retrospective analyses, interpre-
tation of these results is limited by bias. We only examined EGFR
mutations. The cofounding influence of other driver mutations on
brain metastasis risk is not clear. This potential bias is likely very
small because the frequencies of other driver mutations are com-
paratively low and more than one driver mutation is rarely found
concurrently in the same tumor. Another limitation is the small
size of some subgroups, which could have affected some analyses.
Because serial brain imaging was not part of standard follow-up,
our findings and estimation of incidence after initial presentation
are limited to symptomatic brain metastases. It is possible that
asymptomatic disease during the patient’s course was not detected.
As well, the use of CT imaging, instead of MRI, for some patients
may have left some small-volume disease undetected. However, in
the palliative setting, clinically symptomatic disease is most rele-
vant. Our study was limited to patients with initial stage IV disease
and most patients with brain metastases had them at initial pre-
sentation. This limited our ability to determine a time scale for
developing brain metastases.
With a better knowledge of brain metastases risk, this study
may help guide the use of brain screening, surveillance, and man-
agement. By far, most brain metastases were diagnosed at initial
metastatic presentation. Detection of brain metastases at an ear-
lier state with routine brain screening as part of initial diagnostic
investigations for locally advanced or metastatic NSCLC may allow
for more focused brain treatment. We identified the population at
greatest risk as younger patients with EGFR mutations, for which
the cumulative incidence approaches 60%. Future prospective stud-
ies will be needed to determine if surveillance or PCI in this group
of patients will offer a survival benefit.
5. Conclusions
There is a higher incidence of brain metastases for patients
with EGFR+ metastatic NSCLC, even when adjusted for better OS,
compared to patients with EGFR WT. Factors associated with the
development of brain metastases included younger age and the
presence of an EGFR mutation. For patients with and without brain
metastases, survival prognosis with stage IV NSCLC is much longer
with EGFR+ disease.
Conflicts of interest
None.
Disclosures
Fred Hsu—Research grant received from Varian Medical Systems
outside submitted work; Alan Nichol—Research grant and speaker
honorarium received from Varian Medical Systems outside sub-
mitted work; Cheryl Ho—Grants and honorarium from Boehringer
Ingelheim, Eli Lilly, Roche, Astra Zeneca, Bayer, and Pfizer outside
the submitted work; Alex De Caluwe, David Anderson, and Ted
Toriumi—no disclosures.
Acknowledgements
The first and second author contributed equally to the devel-
opment of this manuscript. Financial support received from the
Abbotsford Centre Radiation Therapy Academic Fund and Eleni
Skalbania Endowment for Lung Cancer Research, BC Cancer Foun-
dation.
References
[1] J.S. Barnholtz-Sloan, A.E. Sloan, F.G. Davis, F.D. Vigneau, P. Lai, R.E. Sawaya,
Incidence proportions of brain metastases in patients diagnosed (1973 to
2001) in the Metropolitan Detroit Cancer Surveillance System, J. Clin. Oncol.
22 (2004) 2865–2872.
 F. Hsu et al. / Lung Cancer 96 (2016) 101–107
[2] A. Mujoomdar, J.H. Austin, R. Malhotra, et al., Clinical predictors of metastatic
disease to the brain from non-small cell lung carcinoma: primary tumor size,
cell type, and lymph node metastases, Radiology 242 (2007) 882–888.
[3] T.S. Mok, Y.L. Wu, S. Thongprasert, et al., Gefitinib or carboplatin-paclitaxel in
pulmonary adenocarcinoma, N. Engl. J. Med. 361 (2009) 947–957.
[4] M. Fukuoka, Y.L. Wu, S. Thongprasert, et al., Biomarker analyses and final
overall survival results from a phase III, randomized, open-label, first-line
study of gefitinib versus carboplatin/paclitaxel in clinically selected patients
with advanced non-small-cell lung cancer in Asia (IPASS), J. Clin. Oncol. 29
(2011) 2866–2874.
[5] R. Rosell, E. Carcereny, R. Gervais, et al., Erlotinib versus standard
chemotherapy as first-line treatment for European patients with advanced
EGFR mutation-positive non-small-cell lung cancer (EURTAC): a multicentre,
open-label, randomised phase 3 trial, Lancet Oncol. 13 (2012) 239–246.
[6] J.C. Yang, Y.L. Wu, M. Schuler, et al., Afatinib versus cisplatin-based
chemotherapy for EGFR mutation-positive lung adenocarcinoma (LUX-Lung 3
and LUX-Lung 6): analysis of overall survival data from two randomised,
phase 3 trials, Lancet Oncol. 16 (2015) 141–151.
[7] American Society of Radiation Oncology, ASTRO releases second list of five
radiation oncology treatments to question, as part of national choosing
wisely® campaign. September 14, 2014; 2015.
[8] E.M. Gore, K. Bae, S.J. Wong, et al., Phase III comparison of prophylactic cranial
irradiation versus observation in patients with locally advanced
non-small-cell lung cancer: primary analysis of radiation therapy oncology
group study RTOG 0214, J. Clin. Oncol. 29 (2011) 272–278.
[9] Q. Pan, W. Pao, M. Ladanyi, Rapid polymerase chain reaction-based detection
of epidermal growth factor receptor gene mutations in lung
adenocarcinomas, J. Mol. Diagn. 7 (2005) 396–403.
[10] C. van Walraven, F.A. McAlister, Competing risk bias was common in
Kaplan-Meier risk estimates published in prominent medical journals, J. Clin.
Epidemiol. (2015), http://dx.doi.org/10.1016/j.jclinepi.2015.07.006 (Epub
ahead of print).
[11] H. Putter, M. Fiocco, R.B. Geskus, Tutorial in biostatistics: competing risks and
multi-state models, Stat. Med. 26 (2007) 2389–2430.
[12] R.C. Doebele, X. Lu, C. Sumey, et al., Oncogene status predicts patterns of
metastatic spread in treatment-naive nonsmall cell lung cancer, Cancer 118
(2012) 4502–4511.
107
[13] K. Stanic, M. Zwitter, N.T. Hitij, I. Kern, A. Sadikov, T. Cufer, Brain metastases in
lung adenocarcinoma: impact of EGFR mutation status on incidence and
survival, Radiol. Oncol. 48 (2014) 173–183.
[14] L.E. Hendriks, E.F. Smit, B.A. Vosse, et al., EGFR mutated non-small cell lung
cancer patients: more prone to development of bone and brain metastases?
Lung Cancer 84 (2014) 86–91.
[15] D.Y. Shin, I.I. Na, C.H. Kim, S. Park, H. Baek, S.H. Yang, EGFR mutation and brain
metastasis in pulmonary adenocarcinomas, J. Thorac. Oncol. 9 (2014)
195–199.
[16] A. Law, D.D. Karp, T. Dipetrillo, B.T. Daly, Emergence of increased cerebral
metastasis after high-dose preoperative radiotherapy with chemotherapy in
patients with locally advanced nonsmall cell lung carcinoma, Cancer 92
(2001) 160–164.
[17] A.M. Chen, T.M. Jahan, D.M. Jablons, J. Garcia, D.A. Larson, Risk of cerebral
metastases and neurological death after pathological complete response to
neoadjuvant therapy for locally advanced nonsmall-cell lung cancer: clinical
implications for the subsequent management of the brain, Cancer 109 (2007)
1668–1675.
[18] A.F. Eichler, K.T. Kahle, D.L. Wang, et al., EGFR mutation status and survival
after diagnosis of brain metastasis in nonsmall cell lung cancer,
Neuro-Oncology 12 (2010) 1193–1199.
[19] A. Sekine, T. Kato, E. Hagiwara, et al., Metastatic brain tumors from non-small
cell lung cancer with EGFR mutations: distinguishing influence of exon 19
deletion on radiographic features, Lung Cancer 77 (2012) 64–69.
[20] G.L. Ceresoli, F. Cappuzzo, V. Gregorc, S. Bartolini, L. Crinò, E. Villa, Gefitinib in
patients with brain metastases from non-small-cell lung cancer: a
prospective trial, Ann. Oncol. 15 (2004) 1042–1047.
[21] A.M. Omuro, M.G. Kris, V.A. Miller, et al., High incidence of disease recurrence
in the brain and leptomeninges in patients with nonsmall cell lung carcinoma
after response to gefitinib, Cancer 103 (2005) 2344–2348.
[22] S. Heon, B.Y. Yeap, G.J. Britt, et al., Development of central nervous system
metastases in patients with advanced non-small cell lung cancer and somatic
EGFR mutations treated with gefitinib or erlotinib, Clin. Cancer Res. 16 (2010)
5873–5882.