Sage Journals: Discover world-class research

Abstract

Background:

Many studies have evaluated the accuracy of EGFR mutation status in blood against that in tumor tissues as the reference. We conducted this systematic review and meta-analysis to assess whether blood can be used as a substitute for tumor tissue in detecting EGFR mutations.

Methods:

Investigations that provided data on EGFR mutation status in blood were searched in the databases of Medline, Embase, Ovid Technologies and Web of Science. The detect efficiency of EGFR mutations in paired blood and tissues was compared using a random-effects model of meta-analysis. Pooled sensitivity and specificity and diagnostic accuracy were calculated by receiver operating characteristic curve.

Results:

A total of 19 studies with 2,922 individuals were involved in this meta-analysis. The pooled results showed the positive detection rate of EGFR mutations in lung cancer tissues was remarkably higher than that of paired blood samples (odds ratio [OR] = 1.47, p<0.001). The pooled sensitivity and specificity of blood were 0.65 and 0.91, respectively, and the area under the receiver operating characteristic curve was 0.89.

Conclusions:

Although blood had a better specificity for detecting EGFR mutations, the absence of blood positivity should not necessarily be construed as confirmed negativity. Patients with negative results for blood should decidedly undergo further biopsies to ascertain EGFR mutations.

Keywords

Blood EGFR mutations Meta-analysis NSCLC Non-small cell lung cancer Tissue

Introduction

The detection of epidermal growth factor receptor (EGFR) mutations is critical in the treatment of non-small cell lung cancer (NSCLC) patients receiving EGFR tyrosine kinase inhibitors (TKIs). For example, gefitinib and erlotinib are specially targeted at the binding site of EGFR, and have been widely recommended for treating advanced NSCLC, but research shows that only patients with positive results for EGFR mutations respond to this strategy of treatment (1). Up to now, EGFR mutations confirmed in lung cancer tissues have been considered as the gold standard in the prediction of TKI treatment response and prognosis (2). Unfortunately, lung tissue specimens are sometimes of limited availability for testing of EGFR mutations because of various subjective and objective factors (3, 4). Research shows that the detection of EGFR mutations in cell-free DNA is a feasible option in NSCLC patients (5) because the DNA of tumor cells from the death of those tumor cells and tissues can be released from primary tumors and/or local metastatic sites into the blood (6). Based on this, blood of patients may be an alternative for detecting EGFR mutations.

Although blood has been investigated as a feasible substitute for tumor tissues in testing for EGFR mutations, not all investigations have shown a stable consistency in their results for EGFR mutations between blood and cancer tissues. So far, there is still some controversy regarding whether blood can be an alternative for tumor tissues in detecting EGFR mutations to guide EGFR TKI treatment. We believe that the key to answering this question is that those samples of blood and tissues for detecting EGFR mutations should be of paired design (blood and tissue must be from the same patient). Only in this way, can we compare the advantages and disadvantages between blood and cancer tissues for detecting EGFR mutations, which is also a basic requirement of statistics for scientific research. However, so far the meta-analyses published regarding to this question have not done this very well. To clarify this problem, we carried out this systematic literature evaluation.

Materials and methods

Identification and eligibility of relevant studies

We searched the scientific literature published in the databases of Medline, Embase, Ovid Technologies and Web of Science, using free text and the Medical Subject Headings (MeSH) terms “non-small cell lung cancer,” “NSCLC,” “EGFR,” “epidermal growth factor receptor,” “mutation,” “plasma,” “serum,” “tissues,” “cell-free DNA,” “circulating tumour DNA,” “lung cancer” and “TKIs.” We also manually checked the references from the studies included. If necessary, we also contacted the authors using e-mail and telephone to verify critical data.

Criteria for inclusion and exclusion

Inclusion criteria: literature studies (i) must have compared the EGFR mutations in paired blood and tissue samples (blood and tissues from the same patients); (ii) must have reported the number of true positives (TP), true negatives (TN), false positives (FP) and false negatives (FN); and (iii) must have confirmed patients’ results histopathologically or cytologically. Exclusion criteria were (i) nonoriginal literature, including abstracts, letters, editorials and expert opinions and reviews were excluded; (ii) patients had received anticancer therapy before collecting the samples; (iii) did not contain a distinct group of patients with NSCLC; and (iv) did not pair the samples (the number of blood and tissue samples was not consistent).

Data extraction

The data we extracted were as follows: (i) title of study, author, publication date and country; (ii) study design, case number of paired blood and tissue samples and data for follow-up; and (iii) testing methods for EGFR mutations and predictive ability of EGFR mutations for prognosis in blood and tissues when patients received EGFR TKI treatment.

Methodological quality assessment

The quality of the studies was assessed by 2 evidence-based medicine experts (Gao Wenlong and Liu Xia) according to the guidelines from the Standards for Reporting of Diagnostic Accuracy (STARD) initiative, with a maximum score of 30 (7) and the Quality Assessment for Studies of Diagnostic Accuracy (QUADAS) tool, with a maximum score of 14 (8). The discrepancy of assessment was resolved by negotiating with one of the authors.

Statistical analysis

The chi-square and Fisher’s exact tests were used to detect statistically significant heterogeneity. Heterogeneity was also quantified with the I² statistic which describes the proportion of total variation in study estimates due to heterogeneity. A fixed-effect model was first performed, assuming the same homogeneity of the effect size across all of the studies, followed by a random-effect model under the assumption of heterogeneity between studies. The effect size was generated by the estimation of odds ratios (ORs) with a 95% confidence interval (95% CI). The overall effect was calculated by Z-scores, with significance being set at a p value of <0.05. In addition, sensitivity analysis, funnel plots, Begg’s and Egger’s tests were performed to evaluate the bias of publications. Moreover, the overall sensitivity, specificity and diagnostic odds ratio (DOR) were calculated by a summary receiver operating characteristic (SROC) curve analysis (9). Relevant analyses were performed using SPSS (version 15.0; SPSS Institute, Chicago, IL, USA), Meta DiSc statistical software (version 1.4; Madrid, Spain) and Stata version 12.0 (Stata Corp., College Station, TX, USA). All p values were defined as 2-sided, and a p value of <0.05 indicated a statistical significance.

Results

Process used to search the literature

Initially, a total of 2,568 studies regarding EGFR mutations and lung cancer were found. Of these, 2,200 references were identified to be unsuitable and were removed. Subsequently, 320 studies from the rest were excluded because they were not original literature such as review, abstract and meeting records. Of the 48 publications that remained, 29 studies had to be excluded for the following reasons: repeated data, lacking key data, too small a number of cases or poor quality of study design. Finally, a total of 19 publications with 2,922 individuals met the inclusion criteria and were accepted for further analysis (Fig. 1) (10 -28).

Fig. 1

Study selection algorithm.

Description of the studies

As shown in Table I, the 19 studies included a total of 2,922 individuals, most of the studies were performed in Asian populations, and partly in Whites (10, 11). The sample sizes of studies varied between 27 (12) and 822 patients (13). Eleven studies (12 -23) mainly included lung cancer patients with stages III–IV disease, and individuals in 6 studies (10, 11, 17, 24 -26) had disease ranging from stage I to IV. Of the studies, 7 studies were retrospective (10, 12, 14, 15, 19, 24, 25), 3 studies were prospective (13, 16, 18) and the other 9 did not describe whether they are retrospective or prospective (11, 17, 20 -23, 26 -28).

Table I

Description of the studies included

Authors (ref.)	Year	Age (range or median)	Research design	Asians/Whites	All (No.)	Sex	Histology	Smoking	Tumor stage
						Male/female	LAC/non-LAC	No /yes
Kimura et al (12)	2006	44-87	Retrospective	27/0	27	17/10	23/4	NA	IIIB—IV
Kimura et al (14)	2006	NA	Retrospective	30/0	30	18/12	25/5	NA	IIIB—IV
Kimura et al (15)	2007	40-81	Retrospective	42/0	42	28/14	31/1	14/28	IIIB—IV
He et al (24)	2009	NA	Retrospective	134/0	134	85/49	101/33	NA	I—IV
Bai et al (16)	2009	60.7	Prospective	230/0	230	123/107	171/59	127/103	IIIB—IV
Kuang et al (28)	2009	NA	NA	NA	54	10/44	NA	NA	NA
Sriram et al (10)	2011	33-85	Retrospective	6/498	522	345/177	294/228	35/489	I—IV
Jiang et al (17)	2011	43-80	NA	58/0	58	40/18	42/16	22/36	IIIB—IV
Xu et al (19)	2012	NA	Retrospective	51/0	51	31/20	43/8	32/19	IIIB—IV
Yam et al (23)	2012	NA	NA	51/0	51	27/24	47/4	NA	III—IV
Zhao et al (25)	2012	27-87	Retrospective	111/0	111	76/35	73/38	54/57	I—IV
Goto et al (18)	2012	NA	Prospective	86/0	86	NA	86/0	NA	IIIB—IV
Huang et al (13)	2012	58.4	Prospective	822/0	822	438/384	641/181	482/340	IIIB—IV
Zhang et al (20)	2013	21-80	NA	86/0	86	49/37	65/20	42/44	IIIB—IV
Jing et al (27)	2013	NA	NA	120/0	120	69/51	70/50	NA	I—IV
Akca et al (26)	2013	N/A	NA	52/0	52	39/13	36/16	15/37	I—IV
Liu et al (22)	2013	28-81	NA	86/0	86	56/30	85/1	38/47	IIIB—IV
Weber et al (11)	2014	33-87	NA	0/199	199	101/98	190/9	17/182	IIA—IV
Li et al (21)	2014	32-81	NA	161/0	161	96/68	128/33	84/80	IIIB—IV

LAC = lung adenocarcinoma; NA = unavailable; No. = number of cases; non-LAC = non-lung adenocarcinoma.

Methodology and quality assessment

As shown in Table II, most of the studies included adopted direct sequencing and the strategy of Scorpion Amplification Refractory Mutation System (SARMS) to detect the status of EGFR mutations. On the whole, the studies included had a moderate (11 studies) to high (5 studies) quality. As for patient recruitment, 9 studies (10, 11, 13, 15 -17, 19, 25, 28) used consecutive samples and applied the principle of randomization, and 3 studies prevented clinical design imbalances (23, 25, 28). According to the criteria of STARD and QUADAS (7, 8) the total methodological quality score ranged from 64% to 93% in the studies included, and 7 studies had QUADAS scores ≥10.

Table II

Methodology and quality of studies included

Authors (ref.)	Year	Mutations	Detection method			Study quality score
			Tumor tissues	Blood samples	Type	STARD	QUADAS
Kimura et al (12)	2006	Exons 19, 21, 18	Direct sequencing	Scorpion ARMS	Serum	13	9
Kimura et al (14)	2006	Exons 19, 21	Direct sequencing	Direct sequencing	Serum	15	8
Kimura et al (15)	2007	Exons 19, 21, 18	Direct sequencing	Direct sequencing	Serum	13	11
He et al (24)	2009	Exons 19, 21	ME-PCR	ME-PCR	Plasma	12	7
Bai et al (16)	2009	Exons 19, 21	DHPLC	DHPLC	Plasma	22	12
Kuang et al (28)	2009	Exons 19, 21	SARMS	SARMS	Plasma	19	8
Sriram et al (10)	2011	Exons 19, 21	ME-PCR	ME-PCR	Serum	13	8
Jiang et al (17)	2011	Exons 19, 21	ME sequencing	ME sequencing	Serum	18	8
Xu et al (19)	2012	Exon 19	Scorpion ARMS	Scorpion ARMS	Serum	21	11
Yam et al (23)	2012	Exons 19, 21, 18	PNA-LNA	PNA-LNA	Plasma	12	9
Zhao et al (25)	2012	Exons 19, 21	ME-PCR	ME-PCR	Plasma	20	8
Goto et al (18)	2012	Exons 19, 21, 20	SARMS	SARMS	Serum	19	9
Huang et al (13)	2012	NA	DHPLC	DHPLC	Plasma	18	8
Zhang et al (20)	2013	Exons 19, 21, 20	MEL	70plex liquid chip	Plasma	22	11
Jing et al (27)	2013	Exons 18-21	HRM	HRM	Plasma	15	15
Akca et al (26)	2013	Exons 19, 21	Pyrosequencing	Pyrosequencing	Plasma	16	8
Liu et al (22)	2013	29 mutations	ARMS	ARMS	Serum	20	9
Weber et al (11)	2014	Exons 19, 21, 20	Allele-specific PCR assay	Allele-specific PCR assay	Plasma	22	10
Li et al (21)	2014	Exons 19, 21, 20	ARMS	ARMS	Plasma	18	9

ARMS = amplification refractory mutation; DHPLC = denaturing high-performance liquid chromatography; HRM = high-resolution melt; ME = mutant-enriched; ME-PCR = mutant-enriched polymerase chain reaction; PCR = polymerase chain reaction; PNA-LNA = peptide nucleic acid-locked nucleic acid; QUADAS = Quality Assessment for Studies of Diagnostic Accuracy (maximum score 14); STARD = Standards for Reporting of Diagnostic Accuracy (maximum score 30).

Test of heterogeneity

At first, we used a fixed-effects model to perform the meta-analysis, and the chi-square value for heterogeneity was 32.6 with 18 degrees of freedom and a p value of 0.020, which indicated the possibility of heterogeneity. We inferred that the heterogeneity was mainly derived from the different test methods and sizes of samples. But we noticed that these studies had a quite good clinical homogeneity. In addition, the I² value for the heterogeneity test was 44.2%, which suggested that the heterogeneity was basically acceptable. However, we still used the random-effects model to do the following analysis to keep the conclusions reliable (29).

Comparison of detection rates of EGFR mutations in paired blood and lung cancer tissues

As shown in Table III and Figure 2, all studies provided data on the detection rate of EGFR mutations in paired blood and tumor tissues, which showed that the weight of the included studies in the meta-analysis ranged from 1.26% to 13.03%. The OR value of comparison on detection rate of EGFR mutations in paired blood and lung cancer tissues was 1.47 (95% CI, 1.20-1.80) and the test for Z value of overall effect was 5.20 (p<0.001) (Fig. 2), indicating that the detection rate of EGFR mutations in lung cancer tissue was remarkably higher than that of the paired blood samples.

Table III

Data for EGFR mutations in paired tumor tissue and blood of patients

Authors (ref.)	Year	All cases	Matched cases*	EGFR mutations (%)		Diagnostic test (2 × 2 table)
				Tissue	Blood	TP	FP	FN	TN
Kimura et al (12)	2006	27	11	4	3	2	1	2	6
Kimura et al (14)	2006	30	12	4	5	3	2	1	6
Kimura et al (15)	2007	42	42	8	7	6	1	2	33
He et al (24)	2009	134	18	8	9	8	1	0	9
Bai et al (16)	2009	230	230	77	79	63	16	14	137
Kuang et al (28)	2009	54	43	23	30	21	9	2	11
Sriram et al (10)	2011	522	64	6	3	3	0	3	58
Jiang et al (17)	2011	58	18	18	14	14	0	4	0
Xu et al (19)	2012	51	34	8	4	4	0	4	16
Yam et al (23)	2012	51	37	32	33	32	1	0	4
Zhao et al (25)	2012	111	111	45	19	16	3	29	63
Goto et al (18)	2012	86	86	51	22	22	0	29	35
Huang et al (13)	2012	822	822	296	269	188	81	108	445
Zhang et al (20)	2013	86	86	22	15	15	0	7	64
Jing et al (27)	2013	120	120	48	31	31	0	17	72
Akca et al (26)	2013	52	21	17	14	14	0	3	21
Liu et al (22)	2013	86	86	40	27	27	0	13	46
Weber et al (11)	2014	199	199	28	25	17	7	11	164
Li et al (21)	2014	161	121	56	30	27	3	29	62

FN = false negative; FP = false positive; TN = true negative; TP = true positive.

Matched cases: pairs of both tumor sample and serum sample were obtained from each patient.

Fig. 2

Meta-analysis using a random-effects model for EGFR mutations in blood and tissue samples. The detection rate of EGFR mutations in lung cancer tissue was significantly higher than in the paired blood samples. Test for overall effect: Z = 5.20, p<0.00001. CI = confidence interval; OR = odds ratio.

Predictive ability of EGFR mutations in different samples for treating NSCLC patients with EGFR TKIs

Five studies reported that EGFR mutations were associated with increased progression-free survival (PFS) (13, 15, 16, 19, 21), whether in paired blood or tumor tissues when the patients received TKI therapy. We found that the mean of the p value for prognosis in tumor tissues was 0.018, which was remarkably less than in blood samples (p = 0.254) (Tab. IV), suggesting that tissue EGFR mutations had a better predictive ability for PFS than blood EGFR mutations. Three studies concerned the overall survival (OS) (15, 16, 19) and showed that the patients with EGFR mutations had a longer OS than those without EGFR mutations, but the difference of predictive ability between detection in blood and tumor tissues did not show statistical significance (mean of p value in tissue: 0.163; in blood: 0.656). However, the mean of the p value in tumor tissues was still lower than that in blood, indicating that the predictive ability of EGFR mutations in tissues had a clear advantage (Tab. IV).

Table IV

Predictive ability of EGFR mutations in paired blood and tumor tissue treated with EGFR-TKI

Authors (ref.)	Year	Matched cases*	EGFR mutations (%)			Forecasting ability for PFS (p value)		Forecasting ability for OS (p value)
			Tissue	Blood		Tissue	Blood	Tissue	Blood
Kimura et al (15)	2007	42	8	7		0.016	0.044	0.07	0.489
Bai et al (16)	2009	230	77	79		0.023	0.044	0.405	0.910
Xu et al (19)	2012	34	8	4		0.013	0.092	0.015	0.571
Huang et al (13)	2012	822	296	269		0.038	0.138	NA	NA
Li et al (21)	2014	121	56	30		0.001	0.953	NA	NA
Overall no.		1,249	445	389	Median (p value)	0.018	0.254	0.163	0.656

EGFR-TKI = epidermal growth factor receptor tyrosine kinase inhibitor; NA = unavailable; OS = overall survival; PFS = progression-free survival.

Matched cases: pairs of both tumor sample and serum sample were obtained from each patient.

Analysis of sensitivity and publication bias

The sensitivity analysis showed that the deletion of any 1 study did not substantially modify the estimators and affect the statistical value, and the pooled OR varied between 0.52 and 45.85 (Fig. 3A). The shape of the funnel plot derived from the studies included appeared to be approximately symmetrical (Fig. 3B). Moreover, the Begg’s test (p = 0.463) and the Egger’s test (p = 0.67) both did not indicate that a publication bias existed (Fig. 3C).

Fig. 3

Analysis of sensitivity and publication bias. (A) Taking away any one study from the included studies did not change the overall effect. (B) The shape of the funnel plot from Begg’s test appeared to be approximately symmetrical. (C) Egger’s test showed publication biases did not have an influence on conclusion of meta-analysis.

Sensitivity and specificity of EGFR mutation detection in blood

Figure 4 shows the forest plot of the sensitivity and specificity of 19 studies for EGFR mutations in paired blood and tumor tissues. The pooled sensitivity of EGFR mutations in blood was 0.65 (95% CI, 0.65-0.68) (Fig. 4A) and specificity was 0.91 (95% CI, 0.89-0.92) (Fig. 4B), when the detection results in lung cancer tissues was served as a gold standard. The summary positive likelihood ratio (PLR) was 8.225 (95% CI, 5.127-13.196) and summary negative likelihood ratio (NLR) was 0.398 (95% CI, 0.325-0.487).

Fig. 4

Forest plots of sensitivity and specificity for EGFR mutation detection in blood samples. (A) The point estimate of sensitivity from each study is shown as solid circles. Error bars are 95% confidence intervals. (B) The point estimate of specificity from each study is shown as solid squares. Error bars are 95% confidence intervals.

Diagnostic accuracy for EGFR mutation detection in blood

The diagnostic odds ratio (DOR) of 19 studies on EGFR mutation detection in blood was 32.33 (95% CI, 17.6-59.4), and the chi-square value was 38.51, with all p values less than 0.05 (Fig. 5A). From the SROC curve analysis, we found most studies had a better specificity estimation close to 1.00 for blood EGFR mutations, only 1 study (28) showed a low specificity of EGFR mutations in peripheral blood (<0.60). In contrast, estimations of sensitivity were relatively lower in blood than in lung cancer tissues because the sensitivity of 6 studies being less than 0.60 (10, 12, 18, 19, 21, 25). Finally, the maximum join of sensitivity and specificity was 0.83 (standard error of the mean = 0.027) while the area under curve (AUC) was 0.894 (standard error of the mean = 0.026) (Fig. 5B).

Fig. 5

Diagnostic odds ratio (DOR) and summary receiver operating characteristic (SROC) curve for EGFR mutation detection in blood samples. (A) The mean DOR was a moderate 38.53, indicating that blood samples could be helpful in the identification of EGFR mutations. (B) Each solid circle represents a study included in the meta-analysis, and the size of each study is indicated by the size of the solid circle. The SROC curves summarize the overall diagnostic accuracy.

Discussion

The outcomes of TKI treatment for NSCLC are always dependent on the status of EGFR mutations (30) because studies has been showing that NSCLC patients with mutations of EGFR have a better response to EGFR TKIs than those without mutations (31, 32). In the clinic, EGFR mutations are always identified from operative resected tumor samples, as we know. However, it is sometimes difficult to obtain tumor samples from patients with inoperable NSCLC. Moreover, repeated biopsies for tissue detection are viewed to be impractical because of the poor condition of these patients’ bodies. Thus, it is necessary to develop other more readily accessible samples to detect EGFR mutations (33). Up to now, many relevant studies into detecting EGFR mutations have focused on the blood samples, but theirs conclusion are still not certain.

In the present study, we reviewed 19 studies which detected the EGFR mutations in paired samples of lung cancer tissues and blood. Through a strict quality assessment, we found that the quality of these studies is relatively good. (7, 8). The results showed that the OR for comparisons of EGFR mutations between lung cancer tissues and blood was 1.47, suggesting that the detection rate of EGFR mutations in lung cancer tissues was remarkably higher than that of blood. This result implied that the absence of blood positivity should not necessarily be construed as a confirmation of negativity. Even so, the results still showed that blood could be a valuable alternative when lung cancer tissue is difficult to obtain or is insufficient (34). This new strategy may help us select patients who are most likely to benefit from TKIs, although the different detection methods used in the studies had an obvious impact on the reliability of any conclusions, and these partly explained the heterogeneity in our study. However, we also found a good clinical homogeneity existed in those studies, and there was not a possibility of publication bias. Altogether, the conclusion of this present meta-analysis should be more certain.

Using the positive results of EGFR mutations determined from tissues with lung cancer as a gold standard, we found that the detection of EGFR mutations in blood samples had a relatively higher specificity of 0.91 (95% CI, 0.89-0.92) and a summary PLR of 8.225 (95% CI, 5.127-13.196), which suggested that choosing blood samples to determine EGFR mutations has a potential advantage of specificity, thus indicating that blood could be a potential alternative for lung cancer tissues. However, the summary estimate of sensitivity was only 0.65 (95% CI, 0.61-0.68), which implied that some real EGFR mutations can not be found by the detection of blood samples, But the detection of EGFR mutations in blood samples is still necessary because we can confirm the EGFR mutations as long as it is positive in the blood. In addition, we noticed that the blood detection of EGFR mutations had lower NLR values, which indicated that EGFR mutations could not be excluded even if a patient’s result for EGFR mutations was negative in blood. In other words, a negative test in blood does not mean the absence of EGFR mutations, and patients with negative results in blood still have a chance of having positive results in lung cancer tissues, which is an important issue that we still have to face.

In the present study, we found that tissue that was EGFR mutation positive had a higher predictive ability in terms of survival than the results of blood detection, which means that the advantages of EGFR mutations in tissues for prognosis of lung cancer treated with TKIs far outweigh those for blood samples. The mean DOR of 38.53 further indicated that blood samples had a better specificity for the identification of EGFR mutations in spite of its low detection rate. In clinical practice, the identification of EGFR mutations usually requires biopsies of adequate size, a 38.53 DOR for blood samples would help confirm the positive result. As soon as the EGFR mutations are confirmed to be positive in blood samples of patients, the conclusion is certain, which may avoid further invasive tests on patients, which has great significance. Nevertheless, although blood samples have the advantage of specificity, this choice still has certain restrictions due to its higher false-negative rate. The SROC curve represents the performance of a diagnostic test (35): an AUC of 0.93 to 0.96 is very good, and 0.75 to 0.92 is good (29, 35). In our analysis, the AUC of blood in detecting EGFR mutations was 0.89, which showed that the choice of blood samples has a relatively good accuracy.

Based on the facts mentioned above, our present study proposes that patients who intend to accept TKI therapy should firstly undergo evaluation of blood samples for EGFR mutations, which can provide a evidence that is certain and avoid further invasive biopsy. However, patients with negative results for EGFR mutations in blood should decidedly undergo further invasive tests to ascertain EGFR mutations from tumor tissues.

Finally, there were limitations to this systematic evaluation: first, some studies had small size; second, the statistical designs of some of the studies had flaws; and third, and most critically, the detection methods for EGFR mutations in the studies were not the same.

Therefore, in future we should further investigate the differences in EGFR mutations in paired blood and lung cancer tissues in a large population involving multiple clinical centers. We should also continue to focus on the optimization of the methodology for the detection of EGFR mutations.

Conclusion

The detection rate of EGFR mutations in tissues with lung cancer was remarkably higher than that in blood, which means blood detection may miss some positive results for EGFR mutations. However, the higher specificity of blood samples suggests that their investigation could still be a valuable alternative to the use of tumor samples when tumor tissue is absent or insufficient.

Footnotes

Acknowledgements

The authors wish to thank Drs. Wu Dianlei and Wang Xinhong for their helpful discussions and critical comments provided during the preparation of this manuscript. They also wish to thank Gao Wenlong and Liu Xia for the quality assessment of the included research.

Disclosures

Financial support: No grants or funding have been received for this study.

Conflict of interest: None.

Abbreviations

AUC Area under the curve

CI Confidence interval

DOR Diagnostic odds ratio

EGFR Epidermal growth factor receptor

FN False-negative

FP False-positive

NSCLC Non-small cell lung cancer

NLR Negative likelihood ratio

OR Odds ratios

OS Overall survival

PFS Progression-free survival

PLR Positive likelihood ratio

QUADAS Quality Assessment for Studies of Diagnostic Accuracy

SARMS Scorpion Amplification Refractory Mutation System

SROC Summary receiver operating characteristic curve

STARD Standards for Reporting Diagnostic Accuracy

TN True-negative

TP True-positive

TKI Tyrosine kinase inhibitor

References

Paez

Jänne

Lee

et al . EGFR mutations in lung cancer: correlation with clinical response to gefitinib therapy. Science. 2004;304(5676):1497-1500.

Zheng

Fang

. The presence of EGFR mutations predicts the response in Chinese non-small cell lung cancer patients treated with erlotinib. Int J Biol Markers. 2014;29(2):e112-e119.

Costa

Kobayashi

Tenen

Huberman

. Pooled analysis of the prospective trials of gefitinib monotherapy for EGFR-mutant non-small cell lung cancers. Lung Cancer. 2007;58(1):95-103.

Biaoxue

Shuanying

Wei

Zongjuan

. Maintenance therapy of gefitinib for non-small-cell lung cancer after first-line chemotherapy regardless of epidermal growth factor receptor mutation: a review in Chinese patients. Curr Med Res Opin. 2012;28(10):1699-1708.

Liu

Shi

Song

. Can EGFR mutations in plasma or serum be predictive markers of non-small-cell lung cancer? A meta-analysis. Lung Cancer. 2015;88(3):246-253.

Hong

. Detecting circulating tumor cells: current challenges and new trends. Theranostics. 2013;3(6):377-394.

Bossuyt

Reitsma

Bruns

et al . Standards for Reporting of Diagnostic Accuracy. Towards complete and accurate reporting of studies of diagnostic accuracy: the STARD initiative. Clin Radiol. 2003;58(8):575-580.

Whiting

Rutjes

Reitsma

Bossuyt

Kleijnen

. The development of QUADAS: a tool for the quality assessment of studies of diagnostic accuracy included in systematic reviews. BMC Med Res Methodol. 2003;3(1):25.

Walter

. Properties of the summary receiver operating characteristic (SROC) curve for diagnostic test data. Stat Med. 2002;21(9):1237-1256.

10.

Sriram

Tan

Savarimuthu

et al . Screening for activating EGFR mutations in surgically resected nonsmall cell lung cancer. Eur Respir J. 2011;38(4):903-910.

11.

Weber

Meldgaard

Hager

et al . Detection of EGFR mutations in plasma and biopsies from non-small cell lung cancer patients by allele-specific PCR assays. BMC Cancer. 2014; 14(1):294.

12.

Kimura

Kasahara

Kawaishi

et al . Detection of epidermal growth factor receptor mutations in serum as a predictor of the response to gefitinib in patients with non-small-cell lung cancer. Clin Cancer Res. 2006;12(13):3915-3921.

13.

Huang

Wang

Bai

et al . The detection of EGFR mutation status in plasma is reproducible and can dynamically predict the efficacy of EGFR-TKI. Thoracic Cancer. 2012;3(4):334-340.

14.

Kimura

Kasahara

Shibata

et al . EGFR mutation of tumor and serum in gefitinib-treated patients with chemotherapy-naive non-small cell lung cancer. J Thorac Oncol. 2006;1(3):260-267.

15.

Kimura

Suminoe

Kasahara

et al . Evaluation of epidermal growth factor receptor mutation status in serum DNA as a predictor of response to gefitinib (IRESSA). Br J Cancer. 2007;97(6):778-784.

16.

Bai

Mao

Wang

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(16):2653-2659.

17.

Jiang

Liu

Yang

et al . Serum detection of epidermal growth factor receptor gene mutations using mutant-enriched sequencing in Chinese patients with advanced non-small cell lung cancer. J Int Med Res. 2011;39(4):1392-1401.

18.

Goto

Ichinose

Ohe

et al . Epidermal growth factor receptor mutation status in circulating free DNA in serum: from IPASS, a phase III study of gefitinib or carboplatin/paclitaxel in non-small cell lung cancer. J Thorac Oncol. 2012;7(1): 115-121.

19.

Xue

et al . Comparison of different methods for detecting epidermal growth factor receptor mutations in peripheral blood and tumor tissue of non-small cell lung cancer as a predictor of response to gefitinib. Onco Targets Ther. 2012;5:439-447.

20.

Zhang

Liu

et al . Comparison of EGFR signaling pathway somatic DNA mutations derived from peripheral blood and corresponding tumor tissue of patients with advanced non-small-cell lung cancer using liquid chip technology. J Mol Diagn. 2013;15(6):819-826.

21.

Ren

et al . Peripheral blood for epidermal growth factor receptor mutation detection in non-small cell lung cancer patients. Transl Oncol. 2014;7(3):341-348.

22.

Liu

Zhu

et al . The diagnostic accuracy of pleural effusion and plasma samples versus tumour tissue for detection of EGFR mutation in patients with advanced non-small cell lung cancer: comparison of methodologies. J Clin Pathol. 2013;66(12):1065-1069.

23.

Yam

Lam

Chan

et al . EGFR array: uses in the detection of plasma EGFR mutations in non-small cell lung cancer patients. J Thorac Oncol. 2012;7(7):1131-1140.

24.

Liu

Zhou

et al . Detection of epidermal growth factor receptor mutations in plasma by mutant-enriched PCR assay for prediction of the response to gefitinib in patients with non-small-cell lung cancer. Int J Cancer. 2009;125(10): 2393-2399.

25.

Zhao

Han

Zhao

et al . Comparison of epidermal growth factor receptor mutation statuses in tissue and plasma in stage I-IV non-small cell lung cancer patients. Respiration. 2013;85(2): 119-125.

26.

Akca

Demiray

Yaren

et al . Utility of serum DNA and pyrosequencing for the detection of EGFR mutations in non-small cell lung cancer. Cancer Genet. 2013;206(3):73-80.

27.

Jing

Wang

Cao

. High resolution melting analysis for epidermal growth factor receptor mutations in formalin-fixed paraffin-embedded tissue and plasma free DNA from non-small cell lung cancer patients. Asian Pac J Cancer Prev. 2014;14(11):6619-6623.

28.

Kuang

Rogers

Yeap

et al . Noninvasive detection of EGFR T790M in gefitinib or erlotinib resistant non-small cell lung cancer. Clin Cancer Res. 2009;15(8):2630-2636.

29.

Biaoxue

Shuanying

Xiguang

Wei

. Differential diagnostic CYFRA 211 level for benign and malignant pleural effusions: a meta-analysis in the Chinese population. Arch Med Sci. 2012;8(5):756-766.

30.

Gazdar

. Activating and resistance mutations of EGFR in non-small-cell lung cancer: role in clinical response to EGFR tyrosine kinase inhibitors. Oncogene. 2009;28(Suppl 1): S24-S31.

31.

Lynch

Bell

Sordella

et al . Activating mutations in the epidermal growth factor receptor underlying responsiveness of non-small-cell lung cancer to gefitinib. N Engl J Med. 2004;350(21):2129-2139.

32.

Paez

Jänne

Lee

et al . EGFR mutations in lung cancer: correlation with clinical response to gefitinib therapy. Science. 2004;304(5676):1497-1500.

33.

Sozzi

Conte

Leon

et al . Quantification of free circulating DNA as a diagnostic marker in lung cancer. J Clin Oncol. 2003;21(21):3902-3908.

34.

Mao

Yuan

J-Q

Yang

Z-Y

Tang

. Blood as a substitute for tumor tissue in detecting EGFR mutations for guiding EGFR TKIs treatment of nonsmall cell lung cancer: a systematic review and meta-analysis. Medicine (Baltimore). 2015; 94(21):e775.

35.

Walter

. Properties of the summary receiver operating characteristic (SROC) curve for diagnostic test data. Stat Med. 2002;21(9):1237-1256.

Tissue or blood: which is more suitable for detection of EGFR mutations in non-small cell lung cancer?

Abstract

Background:

Methods:

Results:

Conclusions:

Keywords

Introduction

Materials and methods

Identification and eligibility of relevant studies

Criteria for inclusion and exclusion

Data extraction

Methodological quality assessment

Statistical analysis

Results

Process used to search the literature

Description of the studies

Methodology and quality assessment

Test of heterogeneity

Comparison of detection rates of EGFR mutations in paired blood and lung cancer tissues

Predictive ability of EGFR mutations in different samples for treating NSCLC patients with EGFR TKIs

Analysis of sensitivity and publication bias

Sensitivity and specificity of EGFR mutation detection in blood

Diagnostic accuracy for EGFR mutation detection in blood

Discussion

Conclusion

Footnotes

Acknowledgements

Disclosures

Abbreviations

References