Circulating cell-free DNA as a prognostic and predictive biomarker in non-small cell lung cancer

Abstract

Circulating cell-free DNA (cfDNA), which can be obtained from plasma or serum by non-invasive procedures, has showed great potential to predict treatment response and survival for cancer patients. Several studies have assessed the prognostic and predictive value of cfDNA in non-small cell lung cancer (NSCLC). However, these studies were often small and reported varying results. To address this issue, a meta-analysis was carried out. A total of 22 studies involving 2518 patients were subjected to the final analysis. Our results indicated that NSCLC patients with higher cfDNA concentration had shorter median progression-free survival (PFS) and overall survival (OS) time. In addition, high levels of cfDNA were significantly associated with poor PFS (hazard ratio or HR, 1.32; 95% CI, 1.02-1.71) and OS (HR, 1.64; 95% CI, 1.26-2.15). With respect to tumor specific mutations, we failed to reveal significant differences for PFS (HR, 1.30; 95% CI, 0.66-2.56) and OS (HR, 1.05; 95% CI, 0.49-2.25) when NSCLC patients were grouped according to KRAS genotype detected in cfDNA. However, NSCLC patients which harbored EGFR activating mutation in cfDNA had a greater chance of response to EGFR-TKIs (odds ratio or OR, 1.96; 95% CI, 1.59-2.42). No significant publication bias was detected in this study. In conclusion, cfDNA could act as a prognostic and predictive biomarker for patients with NSCLC.

Full text

Oncotarget44583
www.impactjournals.com/oncotarget
www.impactjournals.com/oncotarget/ Oncotarget, Vol. 7, No. 28
Circulating cell-free DNA as a prognostic and predictive
biomarker in non-small cell lung cancer
Bo Ai1, Huiquan Liu2, Yu Huang2, Ping Peng2
1Department of Thoracic Surgery, Tongji Hospital, Tongji Medical College, Huazhong University of Science and Technology,
Hubei, Wuhan 430030, People’s Republic of China
2Department of Oncology, Tongji Hospital, Tongji Medical College, Huazhong University of Science and Technology, Hubei,
Wuhan 430030, People’s Republic of China
Correspondence to: Ping Peng, email: pengpingtjh@163.com
Keywords: circulating cell-free DNA, non-small cell lung cancer, prognosis, biomarker, meta-analysis
Received: February 28, 2016 Accepted: May 29, 2016 Published: June 15, 2016
ABSTRACT
Circulating cell-free DNA (cfDNA), which can be obtained from plasma or serum
by non-invasive procedures, has showed great potential to predict treatment response
and survival for cancer patients. Several studies have assessed the prognostic and
predictive value of cfDNA in non-small cell lung cancer (NSCLC). However, these
studies were often small and reported varying results. To address this issue, a
meta-analysis was carried out. A total of 22 studies involving 2518 patients were
subjected to the nal analysis. Our results indicated that NSCLC patients with higher
cfDNA concentration had shorter median progression-free survival (PFS) and overall
survival (OS) time. In addition, high levels of cfDNA were signicantly associated with
poor PFS (hazard ratio or HR, 1.32; 95% CI, 1.02-1.71) and OS (HR, 1.64; 95% CI,
1.26-2.15). With respect to tumor specic mutations, we failed to reveal signicant
differences for PFS (HR, 1.30; 95% CI, 0.66-2.56) and OS (HR, 1.05; 95% CI, 0.49-
2.25) when NSCLC patients were grouped according to KRAS genotype detected in
cfDNA. However, NSCLC patients which harbored EGFR activating mutation in cfDNA
had a greater chance of response to EGFR-TKIs (odds ratio or OR, 1.96; 95% CI,
1.59-2.42). No signicant publication bias was detected in this study. In conclusion,
cfDNA could act as a prognostic and predictive biomarker for patients with NSCLC.
INTRODUCTION
Lung cancer is the most commonly diagnosed
cancer as well as the leading cause of cancer-related deaths
in the world [1]. Non-small cell lung cancer (NSCLC)
accounts for approximately 80% cases of lung cancer [2].
Most NSCLC patients are diagnosed with advanced or
distant stages and they are ineligible for curative surgery
and often suffer a poor survival. Identifying biomarkers
related to treatment response and prognosis may be helpful
to improve the clinical outcome of patients with NSCLC.
Circulating cell-free DNA (cfDNA), which can
be isolated from the plasma or serum by non-invasive
procedures, has been proposed as an attractive biomarker
to estimate treatment response, detect drug resistance
and predict clinical outcome for cancer patients [3–7].
It has been experimentally evidenced that tumor cells
can release genomic DNA into the blood and circulating
DNA can reect the tumor burden and tumor biologic
characteristics [6, 8]. A series of studies have shown that
NSCLC patients have higher levels of cfDNA in the blood
compared with healthy controls or patients with benign
diseases [9–11]. The quantitative assay of cfDNA may be
a screening tool for NSCLC. It has been shown that the
diagnostic accuracy of quantitative analysis of cfDNA is
not lower than conventional serum biomarkers for lung
cancer screening [12]. Furthermore, cancer-associated
genetic alterations, such as point mutations, deletions, and
copy number variations, can be detected in cfDNA [13].
In NSCLC, many studies have investigated the diagnostic
accuracy of cfDNA for detecting epithermal growth factor
receptor (EGFR) mutation [14–16]. Two recent meta-
analyses demonstrated that cfDNA was a highly specic
and effective biomarker to measure EGFR mutation
status in NSCLC [17, 18]. These evidences suggested that
genotype in cfDNA could be a promising tumor biomarker
for NSCLC.

Oncotarget44584
www.impactjournals.com/oncotarget
A large number of studies had investigated the
predictive or prognostic value of cfDNA concentration
in NSCLC patients in recent years [19–22] (see Table 1
for references). However, these studies were often small
and reported varying results. Some of them showed that a
higher cfDNA concentration was associated with poorer
survival in NSCLC patients [19, 20], whereas other studies
failed to demonstrate such correlation [21, 22]. On the
other hand, several studies had analyzed the association
between genotype detected in cfDNA with treatment
response or survival in NSCLC [23–26]. Some of them
suggested that tumor specic mutations such as KRAS or
EGFR presented in cfDNA might be useful prognostic
and predictive biomarkers for NSCLC [23, 24]. However,
some others indicated that such gene mutations in cfDNA
had no predictive or prognostic value [25, 26].
As the existing studies are conicting in their
results, it is still difcult to determine the predictive
and prognostic role of cfDNA in patients with NSCLC.
Therefore, a meta-analysis aimed to address this issue was
carried out.
RESULTS
Search results
Figure 1 illustrated the process of study selection.
298 studies were initially found by our search strategy.
30 articles were reviewed in detail after the article titles
and abstracts were checked [9, 16, 19–46]. Eight studies
were excluded from the meta-analysis [39–46], leaving
22 studies that fullled the eligibility criteria [9, 16, 19–
38] (Table 1). Among the 8 excluded studies, 7 did not
provide sufcient data for extracting odds ratio (OR) or
hazard ratio (HR) [39–45], and other 1 study was excluded
because the same cohort of patients was used in other
selected study [46]. The total number of patients included
in this study was 2518, ranging from 22 [21, 37] to 446
[29] cases per study. 12 studies evaluated the prognostic
role of cfDNA concentration in NSCLC [9, 19-22, 27-33].
4 studies reported the prognostic role of KRAS genotype
detected in cfDNA for NSCLC [20, 23, 25, 34]]. Another
7 studies dealt with the predictive role of EGFR genotype
presented in cfDNA for NSCLC patients who were treated
with tyrosine kinase inhibitors of EGFR (EGFR-TKIs)
[16, 24, 26, 35–38].
Impact of cfDNA concentration on the survival
of NSCLC
Six studies reported the median progression-free
survival (PFS) time in NSCLC patients according to
different cfDNA concentrations (high or low) [19, 20, 27,
29, 30, 32]. As showed in Figure 2A, patients with high
levels of cfDNA usually had shorter PFS time than those
with low cfDNA concentrations. In addition, the pooled
HR for PFS was 1.32 (95% CI, 1.02-1.71; P = 0.038),
suggesting that high cfDNA concentration was a good
predictor of poor PFS (Figure 2B). For overall survival
(OS), 6 of 7 studies reported shorter median OS times in
NSCLC patients with higher cfDNA concentration (Figure
3A). Similar to the results of PFS, higher levels of cfDNA
indicated lower overall survival rates with a pooled HR
of 1.64 (95% CI, 1.26-2.15; P = 0.000) (Figure 3B).
However, high heterogeneities were presented in these
analyses (I2 = 73.6%; P = 0.000 for PFS; I2 = 75.5%; P =
0.000 for OS).
As clinical stages and therapeutic regimens are
correlated with patient’s prognosis, they may bring
heterogeneity to the overall analysis. Consequently,
we focused on these two confounding variables in our
subgroup analysis. As showed in Table 1, the majority of
studies considered patients with advanced clinical stages
(stage III-IV). Thus, we combined studies that focused
on NSCLC patients with advanced stages to have a more
homogenic group. The pooled HRs for PFS and OS were
1.29 (95% CI, 1.02-1.65; P = 0.035; I2 = 71.7%; Figure
4A) and 1.64 (95% CI, 1.19-2.25; P = 0.002; I
2
= 81.1%;
Figure 4B), respectively. We further performed another
subgroup analysis according to the therapeutic regimens.
As chemotherapy was the most commonly used treatment
method in these studies, we then limited the analysis to
studies considering patients treated with chemotherapy.
The signicant association could also be observed for both
PFS (HR, 1.41; 95% CI, 1.06-1.89; P = 0.020; I
2
= 76.0%;
Figure 5A) and OS (HR, 1.83; 95% CI, 1.31-2.54; P =
0.000; I2 = 82.3%; Figure 5B).
Impact of KRAS genotype detected in cfDNA on
the survival of NSCLC
The correlation between KRAS genotype detected in
cfDNA with survival in NSCLC patients was evaluated in
four studies. The combined HR for PFS was 1.30 (95%
CI, 0.66-2.56; P = 0.450), suggesting that there were
no signicant differences between patients with KRAS
mutation and those with wild-type genotype with respect
to PFS (Figure 6A). Moreover, our study failed to reveal
signicant difference for OS when NSCLC patients were
grouped according to KRAS genotype detected in cfDNA
(HR, 1.05; 95% CI, 0.49-2.25; P = 0.892; Figure 6B).
Thus, KRAS genotype detected in cfDNA might not be a
prognostic factor for survival in NSCLC patients.
Impact of EGFR genotype detected in cfDNA on
response to EGFR-TKIs
Seven studies evaluated whether EGFR genotype
detected in cfDNA could act as a predictor of response to
EGFR-TKIs. As showed in Figure 7, the pooled OR for
objective response rates (ORR) was 1.96 (95% CI, 1.59-
2.42; P = 0.000; I2 = 71.4%). Our results suggested that

Oncotarget44585
www.impactjournals.com/oncotarget
Table 1: Characteristics of studies included in this meta-analysis
First author Country No. Clinical
stage
Therapeutic
regimen
cfDNA
assessments cfDNA analysis Clinical factors
Catarino(2012)9Portugal 104 I-IV chemotherapy qPCR(hTERT)
quantication(H/L)
OS
Tissot(2015)19 France 218 III-IV chemotherapy PicoGreen
dsDNA Kit
quantication(H/L)
PFS, OS
Nygaard(2014)20 Denmark 58 III-IV chemotherapy ARMS-qPCR
quantication(H/L)
PFS, OS
KRAS
mutation(+/-) PFS, OS
Bortolin(2015)21 Italy 22 I
stereotactic
body
radiotherapy
qPCR(hTERT)
quantication(H/L)
PFS, OS
Li(2016)22 America 101 III-IV chemotherapy qPCR(β-Actin)
quantication(H/L)
PFS, OS
Wang(2014)27 China 134 III-IV EGFR-TKI ARMS/
Scorpion assay
quantication(H/L)
PFS, OS
EGFR
mutation(+/-) PFS, OS
Vinayanuwattikun
(2013)28 Thailand 58 III-IV chemotherapy qPCR(GAPDH)
quantication(H/L)
OS
Sirera(2011)29 Spain 446 III-IV chemotherapy qPCR(hTERT)
quantication(H/L)
PFS, OS
Lee(2011)30 Korea 134 III-IV EGFR-TKI or
chemotherapy qPCR(β-Actin)
quantication(H/L)
PFS, OS
Ludovini(2008)31 Italy 76 I-III surgery+
chemotherapy qPCR(hTERT)
quantication(H/L)
PFS, OS
Camps(2006)32 Spain 99 III-IV chemotherapy qPCR(hTERT)
quantication(H/L)
PFS, OS
Gautschi(2004)33 Switzerland 185 I-IV chemotherapy Fluorogenic
qPCR
quantication(H/L)
OS
Nygaard(2013)23 Denmark 246 II-IV chemotherapy ARMS-qPCR KRAS
mutation(+/-) PFS, OS
Camps(2005)25 Spain 67 III-IV chemotherapy RFLP-PCR KRAS
mutation(+/-) PFS, OS
Gautschi(2007)34 Switzerland 175 I-IV surgery+
chemotherapy RFLP-PCR KRAS
mutation(+/-) OS
Bai(2009)16 China 102 III-IV EGFR-TKI DHPLC EGFR
mutation(+/-) ORR
Kimura(2007)24 Japan 42 III-IV EGFR-TKI DNA
sequencing
EGFR
mutation(+/-) ORR
Douillard(2014)26 Multicenter 102 III-IV EGFR-TKI EGFR RGQ
PCR kit
EGFR
mutation(+/-) ORR
He(2009)35 China 45 I-IV EGFR-TKI Mutant-
enriched PCR
EGFR
mutation(+/-) ORR
(Continued )

Oncotarget44586
www.impactjournals.com/oncotarget
Figure 1: Flow diagram of study selection.
First author Country No. Clinical
stage
Therapeutic
regimen
cfDNA
assessments cfDNA analysis Clinical factors
Kimura(2006)36 Japan 27 III-IV EGFR-TKI DNA
sequencing
EGFR
mutation(+/-) ORR
Kim(2013)37 Korea 22 III-IV EGFR-TKI PNA-LNA PCR EGFR
mutation(+/-) ORR
Li(2014)38 China 55 III-IV EGFR-TKI ARMS-qPCR EGFR
mutation(+/-) ORR
Abbreviation: No., number; cfDNA, circulating cell-free DNA; qPCR, quantitative polymerase chain reaction; hTERT,
human telomerase reverse transcriptase; ARMS, amplication refractory mutation system; GAPDH, glyceraldehyde-
phosphate dehydrogenase; RFLP, restricted fragment length polymorphisms; DHPLC, denaturing high-performance liquid
chromatography; PNA-LNA, peptide nucleic acid-locked nucleic acid; EGFR-TKI, epidermal growth factor receptor-
tyrosine kinase inhibitor; H/L, high/low; +/-, mutation/wide-type; OS, overall survival; PFS, progression-free survival;
ORR, objective response rate.

Oncotarget44587
www.impactjournals.com/oncotarget
patients with EGFR activating mutation in cfDNA had a
greater chance of response to EGFR-TKIs. Thus, EGFR
genotype detected in cfDNA may be a good predictor of
response to EGFR-TKIs for NSCLC patients.
Publication bias
We assessed the publication bias by visually
assessing a funnel plot for asymmetry and by
quantitatively performing Begg’s test and Egger’s test. As
shown in Figure 8, there was no clear evidence of funnel
plot asymmetry by visual assessment. Both Begg’s test
and Egger’s test revealed that no publication bias was
found when OS was analyzed (Begg’s test, p = 0.266,
Egger’s test, p = 0.286). The Egger’s test revealed a slight
publication bias when PFS was analyzed (Begg’s test,
p = 0.119, Egger’s test, p = 0.035). Thus, no signicant
publication bias existed in this study.
DISCUSSION
Non-invasive approaches, usually based on plasma
or serum samples, have showed great potential for
treatment monitoring in NSCLC patients [47]. cfDNA,
as an easily acquired liquid biomarker and a potential
surrogate for the entire tumor genome, may provide
Figure 2: Progression-free survival (PFS) according to cfDNA concentration in NSCLC patients. A. Median PFS time
according to cfDNA concentration. B. Forest plot of hazard ratio (HR) for the impact of cfDNA concentration on PFS.

Oncotarget44588
www.impactjournals.com/oncotarget
complementary roles in predicting treatment response
and survival of NSCLC patients. Many studies have
investigated the usefulness of cfDNA as a screening tool
for NSCLC. However, the predictive or prognostic role
of cfDNA remains to be conrmed. In this study, we
provided the evidence that high levels of cfDNA were
signicantly associated with poor survival in NSCLC. In
addition, our study indicated that cfDNA could act as a
promising predictive factor for response to EGFR-TKIs in
NSCLC patients.
To the best of our knowledge, this is the rst
comprehensive meta-analysis to conrm the prognostic
role of cfDNA concentration in NSCLC. Our study
suggested that NSCLC patients with higher levels of
cfDNA tend to have shorter PFS and OS time. One
explanation for our results might be that total cfDNA was
able to reect the underlying tumor burden. Many studies
had indicated that tumor cell lysate is the main source
of the DNA found in plasma or serum [12]. Besides, the
amount of cfDNA in the blood was signicantly higher
in NSCLC patients than that in healthy controls [9, 48].
What’s more, cfDNA levels were associated with tumor
volume, tumor stage, lymph node involvement and tumor
responses [13]. Newman et al. found that levels of cfDNA
Figure 3: Overall survival (OS) according to cfDNA concentration in NSCLC patients. A. Median OS time according to
cfDNA concentration. B. Forest plot of hazard ratio (HR) for the impact of cfDNA concentration on OS.

Oncotarget44589
www.impactjournals.com/oncotarget
signicantly correlated with tumor volume and provided
earlier response assessment than radiographic approaches
[49]. Thus, patients with higher tumor load might have
more intensive cfDNA released to the blood and cfDNA
levels could reect the tumor burden. On the other hand,
cfDNA levels can be regulated by treatment-caused cell
death. In NSCLC patients, an obvious transient rise
in cfDNA concentrations occurred immediately after
treatment and then it was followed by a rapid decrease
[50]. It suggested that cell death caused by treatment could
release cfDNA, which decreased as the tumor regressed.
These observations revealed that cfDNA levels in plasma
or serum were able to reect the tumor load. Thus, cfDNA
can be a surrogate for tumor burden, making it become a
valuable prognostic factor for patients with NSCLC.
Targeted therapy based on molecular
characterizations has greatly inuenced the treatment
strategies in NSCLC. Gene mutation analyses are the
commonly used predictive biomarkers for selecting
NSCLC patients to receive targeted agents. However,
the current mutation analyses are often based on tumor
tissues and have many limitations. First, the accessibility
of tumor tissues is not always satisfactory as most
NSCLC patients are diagnosed with advanced stages and
Figure 4: Forest plot of hazard ratio (HR) for the impact of cfDNA concentration on progression-free survival (PFS)
and overall survival (OS) in NSCLC patients with advanced stages. A. The impact of cfDNA concentration on PFS. B. The
impact of cfDNA concentration on OS.

Oncotarget44590
www.impactjournals.com/oncotarget
unsuitable to provide tissues through invasive surgery
or biopsy. Second, surgery and biopsy are not without
clinical complications. The adverse events rate for
thoracic biopsy was reported to be approximately 20%
[51]. Furthermore, some percentages of NSCLC patients
will develop resistance to molecular-targeted agents
[52, 53]. Assessing treatment resistance in real time by
repeated surgery or biopsy is not feasible. Considering
these limitations, exploring convenient and less invasive
techniques to monitor the therapeutic response and
effects in NSCLC is urgently needed. Due to its nature
of minimal invasiveness, cfDNA is a promising source
for gene mutation analyses. In this study, we analyzed the
impact of KRAS and EGFR genotype presented in cfDNA
on the survival and response to EGFR-TKIs in NSCLC
patients. Approximately 15-25% of patients with NSCLC
have KRAS mutations, resulting in constitutive activation
of KRAS signaling pathways. Many studies reported that
KRAS mutation could predict the poor outcomes of EGFR-
TKIs treatment and chemotherapy, but several studies
argued that KRAS mutation was not associated with the
outcome of NSCLC patients [54]. A meta-analysis aimed
to clarify the prognostic and predictive value of KRAS
mutation in NSCLC was carried out recently [54]. Its
Figure 5: Forest plot of hazard ratio (HR) for the impact of cfDNA concentration on progression-free survival (PFS)
and overall survival (OS) in NSCLC patients treated with chemotherapy. A. The impact of cfDNA concentration on PFS. B.
The impact of cfDNA concentration on OS.

Oncotarget44591
www.impactjournals.com/oncotarget
results showed that KRAS mutation was signicantly
associated with worse OS and disease-free survival
(DFS) in early stage NSCLC, and with inferior outcomes
of EGFR-TKIs treatment and chemotherapy. However,
statistical differences in DFS and PFS of chemotherapy
and response rates to EGFR-TKIs or chemotherapy were
not met when EGFR mutant patients were excluded.
Our results indicated that KRAS mutations detected in
cfDNA might not be a prognostic factor for the survival of
NSCLC patients. One explanation might be that mutations
of KRAS and EGFR were generally mutually exclusive in
NSCLC [55, 56]. Most EGFR mutations were existed in
KRAS wild-type patients, which might bias the results
toward an overestimation of the prognostic and predictive
value of KRAS mutation. Another reason might be that the
amount of studies which assessed the prognostic value of
KRAS mutation presented in cfDNA in NSCLC was small.
Thus, the clinical signicance of KRAS mutation detected
in cfDNA is yet under debate. Future large-scaled trails are
still needed to improve our results.
Nowadays, EGFR-TKIs are the most successful
example of targeted therapy in NSCLC. EGFR gene
mutations are the standard biomarkers for selecting
NSCLC patients to receive EGFR-TKIs treatment. As
a high degree of correlation between EGFR mutations
detected in tumors and those presented in cfDNA has
been conrmed by two recent meta-analyses [17, 18],
EGFR mutation presented in cfDNA may also be useful
predictive markers for guiding NSCLC patients to receive
EGFR-TKIs. Indeed, several studies have analyzed the
association between cfDNA EGFR mutation status and
clinical outcomes. Goto et al. [44] found a signicant
correlation between cfDNA EGFR mutation status and
PFS. In cfDNA EGFR activating mutation subgroup,
NSCLC patients had longer PFS when they were treated
with getinib. Another research demonstrated that EGFR
Figure 6: Forest plot of hazard ratio (HR) for the impact of KRAS genotype detected in cfDNA on progression-free
survival (PFS) and overall survival (OS). A. The impact of KRAS genotype detected in cfDNA on PFS. B. The impact of KRAS
genotype detected in cfDNA on OS.

Oncotarget44592
www.impactjournals.com/oncotarget
mutation status in cfDNA was a good predictor for
PFS after EGFR-TKIs treatment [57]. Consistently, our
results showed that EGFR activating mutation in cfDNA
indicated a greater chance of response to EGFR-TKIs
in NSCLC patients. Thus, cfDNA EGFR mutation test
had a good ability to predict the efcacy of EGFR-TKIs
treatment. cfDNA might be a reliable material to guide
EGFR-TKIs treatment for NSCLC patients.
However, there were some limitations in our present
meta-analysis. Firstly, our analyses were based on the
literature, making our results less reliable than individual
patient data-based analysis. Secondly, a signicant
heterogeneity was presented in this study. When subgroup
analyses were performed in terms of clinical stages and
therapeutic regimens, the heterogeneity between studies
did not change obviously. The heterogeneity might partly
Figure 7: Forest plot of odds ratio (OR) for the impact of EGFR genotype detected in cfDNA on response to EGFR-
TKIs.
Figure 8: Funnel plot for the assessment of publication bias in this study. A. Funnel plot for 8 studies reporting progression-free
survival (PFS). B. Funnel plot for 11 studies reporting overall survival (OS).

Oncotarget44593
www.impactjournals.com/oncotarget
come from other variations, such as techniques that were
adopted to detect cfDNA. Future standardization of
cfDNA assessment would hopefully solve this problem.
Thirdly, studies that could not provide sufcient data for
extracting OR or HR were excluded. The exclusion of
these studies might make the pooled estimates differ from
their true value on some level.
In view of this study, our ndings suggested that
cfDNA could act as a predictive and prognostic biomarker
for patients with NSCLC. High levels of cfDNA were
signicantly associated with poor PFS and OS in NSCLC.
In addition, EGFR activating mutation status in cfDNA
indicated a greater chance of response to EGFR-TKIs. In
conclusion, cfDNA had a prognostic and predictive value
for NSCLC patients, which might help to dene high risk
patients and guide clinical decision making. However,
considering the limitations of a literature-based meta-
analysis, these results need to be validated and updated by
future large-scaled researches.
MATERIALS AND METHODS
Literature searches
Electronic searches for relevant articles in PubMed,
Embase, and Web of Science databases were conducted
in January 2016. The search strategy was generated by
combining key words related to cfDNA (‘circulating
cell-free DNA’ or ‘plasma cell-free DNA’ or ‘serum cell-
free DNA’) and NSCLC (‘non small cell lung cancer’ or
‘NSCLC’). Moreover, we manually searched the reference
lists of relevant articles for additional publications.
Inclusion criteria
Studies were included in this meta-analysis if they
met the following criteria: (1) all patients recruited in the
study were diagnosed with NSCLC; (2) the predictive
or prognostic value of cfDNA was evaluated; (3) only
English-language studies were included; (4) the HR or
OR and their corresponding 95% CIs were described or
could be statistically extracted; (5) When several studies
reported the same patient population, the newest or most
informative study was included.
Data extraction
Data extraction was performed independently by 2
reviewers and disagreements among them were resolved
by consensus. The following information was extracted
from each study: rst author’s name, publication year,
country of origin, number of patients, therapeutic
regimen, cfDNA assessment (methods), cfDNA analysis
(quantication and molecular characterization) and
clinical factors (PFS, OS and ORR).
Statistical analysis
HR and its 95% CIs were used to estimate the
prognostic value of cfDNA. OR and its 95% CIs were
adopted to describe the correlation between cfDNA status
and objective response rates. The individual HR or OR
estimates were combined into an overall HR or OR, and
the results were presented in the form of a forest plot.
Pooled effect sizes were considered to be signicantly
different if their 95% CIs did not include 1 (p < 0.05).
HR > 1 implied a poor survival and OR > 1 indicated a
greater chance of objective response. Median pooled PFS
and OS were summarized using descriptive statistics.
The heterogeneity between studies was assessed by the
Cochran Q test and I2 test. When Cochran Q test P value
was ≤ 0.10 and I2 test I2 value was ≥ 50%, statistically
signicant heterogeneity was considered to be present.
Fixed effects models were employed when heterogeneity
was absent; otherwise, random effects models were
adopted. Funnel plots, Begg’s test, and Egger’s test were
performed to detect publication bias. All analyses were
carried out by using Stata Statistical Software, version
12.0 (Stata Corporation, College Station, TX, USA).
CONFLICTS OF INTEREST
The authors declared no conicts of interest.
REFERENCES
1.
Torre LA, Bray F, Siegel RL, et al. Global cancer statistics,
2012. CA Cancer J Clin. 2015; 65:87-108.
2.
Howlader N, Noone AM, Krapcho M, et al. SEER Cancer
Statistics Review, 1975-2012, National Cancer Institute.
Bethesda, MD, http://seer.cancer.gov/csr/1975_2012/,
based on November 2014 SEER data submission, posted to
the SEER web site, April 2015.
3.
De Mattos-Arruda L, Caldas C. Cell-free circulating tumour
DNA as a liquid biopsy in breast cancer. Mol Oncol. 2015.
doi: 10.1016/j.molonc.2015.12.001. [Epub ahead of print].
4.
Gray ES, Rizos H, Reid AL et al. Circulating tumor DNA to
monitor treatment response and detect acquired resistance
in patients with metastatic melanoma. Oncotarget. 2015;
6:42008-42018. doi: 10.18632/oncotarget.5788.
5.
Carpinetti P, Donnard E, Bettoni F et al. The use of
personalized biomarkers and liquid biopsies to monitor
treatment response and disease recurrence in locally advanced
rectal cancer after neoadjuvant chemoradiation. Oncotarget.
2015; 6:38360-38371. doi: 10.18632/oncotarget.5256.
6.
Jiang T, Ren S, Zhou C. Role of circulating-tumor DNA
analysis in non-small cell lung cancer. Lung Cancer. 2015;
90:128-134.
7.
Xia S, Kohli M, Du M et al. Plasma genetic and genomic
abnormalities predict treatment response and clinical

Oncotarget44594
www.impactjournals.com/oncotarget
outcome in advanced prostate cancer. Oncotarget. 2015;
6:16411-16421. doi: 10.18632/oncotarget.3845.
8.
Stroun M, Maurice P, Vasioukhin V et al. The origin and
mechanism of circulating DNA. Ann N Y Acad Sci. 2000;
906:161-168.
9.
Catarino R, Coelho A, Araujo A et al. Circulating DNA:
diagnostic tool and predictive marker for overall survival of
NSCLC patients. PLoS One. 2012; 7:e38559.
10. Szpechcinski A, Chorostowska-Wynimko J, Struniawski R
et al. Cell-free DNA levels in plasma of patients with non-
small-cell lung cancer and inammatory lung disease. Br J
Cancer. 2015; 113:476-483.
11.
Szpechcinski A, Rudzinski P, Kupis W et al. Plasma
cell-free DNA levels and integrity in patients with chest
radiological ndings: NSCLC versus benign lung nodules.
Cancer Lett. 2016. doi: 10.1016/j.canlet.2016.02.002. [Epub
ahead of print].
12.
Zhang R, Shao F, Wu X et al. Value of quantitative analysis
of circulating cell free DNA as a screening tool for lung
cancer: a meta-analysis. Lung Cancer. 2010; 69:225-31.
13. Nie K, Jia Y, Zhang X. Cell-free circulating tumor DNA in
plasma/serum of non-small cell lung cancer. Tumour Biol.
2015; 36:7-19.
14.
Zhao X, Han RB, Zhao J et al. Comparison of epidermal
growth factor receptor mutation statuses in tissue and
plasma in stage I-IV non-small cell lung cancer patients.
Respiration. 2013; 85:119-125.
15.
Ishii H, Azuma K, Sakai K et al. Digital PCR analysis
of plasma cell-free DNA for non-invasive detection of
drug resistance mechanisms in EGFR mutant NSCLC:
Correlation with paired tumor samples. Oncotarget. 2015;
6:30850-30858. doi: 10.18632/oncotarget.5068.
16.
Bai H, Mao L, Wang HS et al. Epidermal growth factor
receptor mutations in plasma DNA samples predict tumor
response in Chinese patients with stages IIIB to IV non-
small-cell lung cancer. J Clin Oncol. 2009; 27:2653-2659.
17.
Qiu M, Wang J, Xu Y et al. Circulating tumor DNA is
effective for the detection of EGFR mutation in non-
small cell lung cancer: a meta-analysis. Cancer Epidemiol
Biomarkers Prev. 2015; 24:206-212.
18.
Luo J, Shen L, Zheng D. Diagnostic value of circulating
free DNA for the detection of EGFR mutation status in
NSCLC: a systematic review and meta-analysis. Sci Rep.
2014; 4:6269.
19.
Tissot C, Toffart AC, Villar S et al. Circulating free DNA
concentration is an independent prognostic biomarker in
lung cancer. Eur Respir J. 2015; 46:1773-1780.
20.
Dowler Nygaard A, Spindler KL, Pallisgaard N et al. Levels
of cell-free DNA and plasma KRAS during treatment of
advanced NSCLC. Oncol Rep. 2014; 31:969-974.
21.
Bortolin MT, Tedeschi R, Bidoli E et al. Cell-free DNA
as a prognostic marker in stage I non-small-cell lung
cancer patients undergoing stereotactic body radiotherapy.
Biomarkers. 2015; 20:422-428.
22.
Li BT, Drilon A, Johnson ML et al. A prospective study
of total plasma cell-free DNA as a predictive biomarker
for response to systemic therapy in patients with advanced
non-small-cell lung cancersdagger. Ann Oncol. 2016;
27:154-159.
23.
Nygaard AD, Garm Spindler KL, Pallisgaard N et al.
The prognostic value of KRAS mutated plasma DNA in
advanced non-small cell lung cancer. Lung Cancer. 2013;
79:312-317.
24.
Kimura H, Suminoe M, Kasahara K et al. Evaluation of
epidermal growth factor receptor mutation status in serum
DNA as a predictor of response to getinib (IRESSA). Br J
Cancer. 2007; 97:778-784.
25.
Camps C, Sirera R, Bremnes R et al. Is there a prognostic
role of K-ras point mutations in the serum of patients with
advanced non-small cell lung cancer? Lung Cancer. 2005;
50:339-346.
26. Douillard JY, Ostoros G, Cobo M et al. Getinib treatment
in EGFR mutated caucasian NSCLC: circulating-free tumor
DNA as a surrogate for determination of EGFR status. J
Thorac Oncol. 2014; 9:1345-1353.
27.
Wang S, Han X, Hu X et al. Clinical signicance of
pretreatment plasma biomarkers in advanced non-small
cell lung cancer patients. Clin Chim Acta. 2014; 430:63-70.
28.
Vinayanuwattikun C, Winayanuwattikun P, Chantranuwat
P et al. The impact of non-tumor-derived circulating nucleic
acids implicates the prognosis of non-small cell lung cancer.
J Cancer Res Clin Oncol. 2013; 139:67-76.
29.
Sirera R, Bremnes RM, Cabrera A et al. Circulating DNA
is a useful prognostic factor in patients with advanced non-
small cell lung cancer. J Thorac Oncol. 2011; 6:286-290.
30.
Lee YJ, Yoon KA, Han JY et al. Circulating cell-
free DNA in plasma of never smokers with advanced
lung adenocarcinoma receiving getinib or standard
chemotherapy as rst-line therapy. Clin Cancer Res. 2011;
17:5179-5187.
31.
Ludovini V, Pistola L, Gregorc V et al. Plasma DNA,
microsatellite alterations, and p53 tumor mutations are
associated with disease-free survival in radically resected
non-small cell lung cancer patients: a study of the perugia
multidisciplinary team for thoracic oncology. J Thorac
Oncol. 2008; 3:365-373.
32.
Camps C, Sirera R, Bremnes RM et al. Quantication in
the serum of the catalytic fraction of reverse telomerase:
a useful prognostic factor in advanced non-small cell lung
cancer. Anticancer Res. 2006; 26:4905-4909.
33.
Gautschi O, Bigosch C, Huegli B et al. Circulating
deoxyribonucleic Acid as prognostic marker in non-small-
cell lung cancer patients undergoing chemotherapy. J Clin
Oncol. 2004; 22:4157-4164.
34.
Gautschi O, Huegli B, Ziegler A et al. Origin and prognostic
value of circulating KRAS mutations in lung cancer
patients. Cancer Lett. 2007; 254:265-273.

Oncotarget44595
www.impactjournals.com/oncotarget
35.
He C, Liu M, Zhou C et al. Detection of epidermal growth
factor receptor mutations in plasma by mutant-enriched
PCR assay for prediction of the response to getinib in
patients with non-small-cell lung cancer. Int J Cancer. 2009;
125:2393-2399.
36.
Kimura H, Kasahara K, Kawaishi M et al. Detection of
epidermal growth factor receptor mutations in serum as
a predictor of the response to getinib in patients with
non-small-cell lung cancer. Clin Cancer Res. 2006;
12:3915-3921.
37.
Kim ST, Sung JS, Jo UH et al. Can mutations of EGFR
and KRAS in serum be predictive and prognostic markers
in patients with advanced non-small cell lung cancer
(NSCLC)? Med Oncol. 2013; 30:328.
38.
Li X, Ren R, Ren S et al. Peripheral blood for epidermal
growth factor receptor mutation detection in non-small cell
lung cancer patients. Transl Oncol. 2014; 7:341-8.
39.
Camps C, Jantus-Lewintre E, Cabrera A et al. The
identication of KRAS mutations at codon 12 in plasma
DNA is not a prognostic factor in advanced non-small cell
lung cancer patients. Lung Cancer. 2011; 72:365-369.
40.
Karachaliou N, Mayo-de las Casas C, Queralt C et al.
Association of EGFR L858R Mutation in Circulating Free
DNA With Survival in the EURTAC Trial. JAMA Oncol.
2015; 1:149-157.
41.
Wang Z, Chen R, Wang S et al. Quantication 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. 2014; 9:e110780.
42.
Punnoose EA, Atwal S, Liu W 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. 2012; 18:2391-2401.
43.
Zhuo ML, Wu MN, Zhao J et al. Epidermal growth
factor receptor genotype in plasma DNA and outcome
of chemotherapy in the Chinese patients with advanced
non-small cell lung cancer. Chin Med J (Engl). 2011;
124:3510-3514.
44.
Goto K, Ichinose Y, Ohe Y et al. Epidermal growth factor
receptor mutation status in circulating free DNA in serum:
from IPASS, a phase III study of getinib or carboplatin/
paclitaxel in non-small cell lung cancer. J Thorac Oncol.
2012; 7:115-121.
45.
Thress KS, Brant R, Carr TH et al. EGFR mutation detection
in ctDNA from NSCLC patient plasma: A cross-platform
comparison of leading technologies to support the clinical
development of AZD9291. Lung Cancer. 2015; 90:509-515.
46.
Nygaard AD, Holdgaard PC, Spindler KL et al. The
correlation between cell-free DNA and tumour burden was
estimated by PET/CT in patients with advanced NSCLC. Br
J Cancer. 2014; 110:363-368.
47.
Sun W, Yuan X, Tian Y et al. Non-invasive approaches
to monitor EGFR-TKI treatment in non-small-cell lung
cancer. J Hematol Oncol. 2015; 8:95.
48.
Ulivi P, Mercatali L, Casoni GL et al. Multiple marker
detection in peripheral blood for NSCLC diagnosis. PLoS
One. 2013; 8:e57401.
49.
Newman AM, Bratman SV, To J et al. An ultrasensitive
method for quantitating circulating tumor DNA with broad
patient coverage. Nat Med. 2014; 20:548-554.
50.
Szpechcinski A, Chorostowska-Wynimko J, Kupis W et al.
Quantitative analysis of free-circulating DNA in plasma of
patients with resectable NSCLC. Expert Opin Biol Ther.
2012; 12:S3-9.
51.
Overman MJ, Modak J, Kopetz S et al. Use of research
biopsies in clinical trials: are risks and benets adequately
discussed? J Clin Oncol. 2013; 31:17-22.
52.
Janne PA. Challenges of detecting EGFR T790M in
getinib/erlotinib-resistant tumours. Lung Cancer. 2008;
60:S3-9.
53.
Russo A, Franchina T, Ricciardi GR et al. A decade of EGFR
inhibition in EGFR-mutated non small cell lung cancer
(NSCLC): Old successes and future perspectives. Oncotarget.
2015; 6:26814-26825. doi: 10.18632/oncotarget.4254.
54.
Pan W, Yang Y, Zhu H et al. KRAS mutation is a weak, but
valid predictor for poor prognosis and treatment outcomes
in NSCLC: A meta-analysis of 41 studies. Oncotarget.
2016;7:8373-88. doi: 10.18632/oncotarget.7080.
55.
Dogan S, Shen R, Ang DC et al. Molecular epidemiology
of EGFR and KRAS mutations in 3,026 lung
adenocarcinomas: higher susceptibility of women to
smoking-related KRAS-mutant cancers. Clin Cancer Res.
2012; 18:6169-6177.
56.
Boch C, Kollmeier J, Roth A et al. 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. 2013; 3:e002560.
57.
Weber B, Meldgaard P, Hager H et al. Detection of EGFR
mutations in plasma and biopsies from non-small cell lung
cancer patients by allele-specic PCR assays. BMC Cancer.
2014; 14:294.