- Email: [email protected]
EGFR mutations and clinical outcomes of chemotherapy for advanced non-small cell lung cancer: A meta-analysis
Lung Cancer 85 (2014) 339–345
Contents lists available at ScienceDirect
Lung Cancer journal homepage: www.elsevier.com/locate/lungcan
Review
EGFR mutations and clinical outcomes of chemotherapy for advanced non-small cell lung cancer: A meta-analysis Qiong Zhang a , Hong-Hai Dai a , Hong-Yun Dong b,c , Cheng-Tao Sun a , Zhe Yang a,∗,1 , Jun-Qing Han a,∗,1 a
Tumor Research and Therapy Center, Provincial Hospital Affiliated to Shandong University, 324 Jingwu Weiqi Road, Jinan, Shandong 250021, PR China School of Public Health, Shandong University, 44 Wenhuaxi Road, Jinan, Shandong 250012, PR China c Wanhua Chemical Group Co., Ltd., 7 South Xingfu Road, Yantai, Shandong 264002, PR China b
a r t i c l e
i n f o
Article history: Received 11 October 2013 Received in revised form 20 April 2014 Accepted 13 June 2014 Keywords: Non-small cell lung cancer EGFR Mutation status Chemotherapy Response rate Meta-analysis
a b s t r a c t Background: This meta-analysis was performed to assess whether epidermal growth factor receptor (EGFR) mutation status was associated with objective response rate (ORR), progression-free survival (PFS), and overall survival (OS) in patients with advanced non-small cell lung cancer (NSCLC) treated with chemotherapy. Method: We systematically identified eligible articles investigating the effects of chemotherapy in patients with NSCLC stratified by EGFR mutation status. The summary risk ratio (RR) for ORR and hazard ratios (HRs) for both PFS and OS were calculated using the inverse variance formula of meta-analysis. Results: Identification for the current meta-analysis: 5 prospective studies (n = 875) and 18 retrospective studies (n = 1934) for ORR; 2 prospective studies (n = 434) and 10 retrospective studies (n = 947) for PFS; 2 prospective studies (n = 438) and 7 retrospective studies (n = 711) for OS. The ORR was significantly higher in patients with EGFR mutations in prospective studies (RR = 1.42; 95% confidence interval [CI], 1.16–1.74; P = 0.001), but not in retrospective studies (RR = 1.12; 95% CI, 0.96–1.32; P = 0.146). There was no obvious association between EGFR mutations and PFS both in prospective (HR = 0.84; 95% CI: 0.65–1.09; P = 0.197) and retrospective (HR = 1.02; 95% CI: 0.87–1.18; P = 0.838) studies. Association between EGFR mutations and OS was also not seen in prospective studies (HR = 0.74; 95% CI: 0.27–2.05; P = 0.566), but was seen in retrospective studies (HR = 0.48; 95% CI: 0.33–0.72; P < 0.001; I2 = 75.9%; P < 0.001) with significant heterogeneity. Conclusion: EGFR mutations in advanced NSCLC may be associated with higher ORRs to chemotherapy, but may have nothing to do with PFS and OS. Further prospective studies are required to identify the influence of EGFR mutations on chemotherapy effects in advanced NSCLC. © 2014 Elsevier Ireland Ltd. All rights reserved.
1. Introduction The development and utilization of epidermal growth factor receptor (EGFR) tyrosine kinase inhibitors (TKIs) was a great advancement in the treatment of specific cases of advanced non-small cell lung cancer (NSCLC). EGFR mutations have been demonstrated to predict good efficacy of EGFR-TKIs in NSCLC [1–4]. Consequently, it is important to confirm the EGFR genotype of patients with NSCLC to determine whether EGFR-TKI therapy
∗ Corresponding authors at: Tumor Research and Therapy Center, Provincial Hospital Affiliated to Shandong University, 324 Jingwu Weiqi Road, Jinan, Shandong 250021, PR China. Tel.: +86 531 6877 6100; fax: +86 531 87068707. E-mail addresses: [email protected] (Z. Yang), [email protected] (J.-Q. Han). 1 These two corresponding authors contributed equally to this work. http://dx.doi.org/10.1016/j.lungcan.2014.06.011 0169-5002/© 2014 Elsevier Ireland Ltd. All rights reserved.
would be appropriate. Although EGFR-TKIs play a critical role in the treatment of advanced NSCLC, chemotherapy with conventional anticancer agents is still the gold-standard approach for patients with unresectable, locally advanced, or metastatic NSCLC. Nevertheless, the association between EGFR genotype and the effects of chemotherapy remains uncertain. With the emphasis on biomarkers in the treatment of NSCLC, many studies have been conducted to investigate the biomarkers predicting patient response to specific chemotherapy regimens [5–7]. In the Iressa Pan-Asia Survival Study (IPASS) [4], Asian patients with EGFR mutations had a significantly higher response rate (47.3% versus 23.5%) than patients without EGFR mutations when they received paclitaxel plus carboplatin as first-line therapy. Similar results were observed in 2 other studies [8,9]. However, 2 preclinical experiments demonstrated that lung cancer cells or tumors with EGFR mutations may tend to markedly
340
Q. Zhang et al. / Lung Cancer 85 (2014) 339–345
resist chemotherapeutic agents [10,11]. Consistent with this, several clinical studies have suggested patients with mutated EGFR have poorer responses to chemotherapy than patients with wildtype EGFR [12–14]. Furthermore, different studies using the same methods to detect EGFR mutation status and the same chemotherapeutic agents have yielded contradictory results [9,14]. Because of these inconsistent results, we conducted a meta-analysis to evaluate the role of EGFR mutation status in predicting the efficacy, progression-free survival (PFS), and overall survival (OS) of chemotherapy in patients with advanced NSCLC. In this article, the word “chemotherapy” represents chemotherapy with conventional anticancer agents.
2. Materials and methods 2.1. Literature search strategy We performed computerized searches of the PubMed database, the EMBASE database and the Cochrane library (Issue 6, 2013) to identify all published articles reporting on chemotherapy for advanced NSCLC with known EGFR mutation status using the following key words: non-small cell lung cancer, NSCLC, lung cancer, chemotherapy, and epidermal growth factor receptor. The search was performed on July16, 2013. The published languages were not limited. The references of all reviewed articles were also screened for additional studies.
2.2. Selection criteria Eligible studies had to meet the following criteria:(1) patients with advanced or recurrent NSCLC; (2) detected EGFR gene mutation status in primary lung cancer tissue, metastatic tissue or sera DNA; (3) received chemotherapy; (4) reported data of response rate, the hazard ratios (HRs) with the corresponding 95% confidence intervals (CIs) comparing OS, PFS or time-to-progression (TTP) stratified by EGFR mutation status; and (5) the number of patients in the EGFR mutation or wild-type group was not less than 5. Studies examining chemotherapy in combination with any other agents such as EGFR-TKIs, any investigational drug or radiotherapy were excluded.
2.3. Quality assessment Two investigators (QZ, HYD) independently assessed the quality of studies using the Newcastle Ottawa Quality Assessment Scale for cohort studies [15,16]. Discrepancies were resolved by consensus. This scale is composed of eight items which assess patient selection, study comparability and outcome. The scale was recommended by the Cochrane Non-Randomized Studies Methods Working Group [17].
2.4. Data extraction Data from each study were extracted independently by 2 investigators, using a standardized data extraction form. Disagreements between the 2 investigators were resolved by consensus. The following information was abstracted from each publication: authors, publication date, country, sample size, study type, characteristics of the patients, data about EGFR mutations, test methods, numbers of patients in the EGFR mutation and wild-type groups, chemotherapy information, objective response rates (ORRs), and HRs and 95%CIs for PFS (or TTP) and OS.
2.5. Statistical methods The risk ratio (RR) was calculated for ORRs, and the HR was calculated for PFS and OS. In studies without HRs, Kaplan–Meier plots were used to calculate the HRs based on the methods presented by Tierney [18]. STATA SE 12.0 package (StataCorp, College Station, TX, USA) was used for statistical analyses. The fixed-effects model (inverse variance formula) was initially used. If P (for I2 ) < 0.05, indicating significant heterogeneity among studies, then the random-effects model (I–V heterogeneity formula) was used. Two-tailed P values were used for all comparisons, and statistical significance was defined as P < 0.05. The possibility of publication bias was investigated by inspecting funnel plots and was statistically analyzed using Begg’s test. We planned additional sensitivity analyses to further detect and evaluate clinical heterogeneity. Most of the eligible studies used first-line chemotherapy with platinum-based double-agent regimens (17/23), with the second agent varying between studies (Supplementary Table 1). Consequently, it was difficult to perform subgroup analyses according to the therapeutic regimen. There also was not enough data about pathological types and sex to perform separate analyses of these subgroups. Finally, subgroup analyses were conducted to assess the effects of ethnicity (Asian, Caucasian, or mixed), therapy line (first-line, second-line, and mixed-line), and test methods of EGFR genotype (scorpions amplification refractory mutation system [Scorpions ARMS] or direct DNA sequencing or denaturing high performance liquid chromatography [DHPLC]). Therefore the categories used to make subgroup analysis were not pre-specified. We performed sensitivity analysis for studies providing HRs and for some categories without sufficient studies for subgroup analysis in the additional analysis of PFS and OS. Studies conducted in Korea, China, Japan and other Asian countries were classified as “Asian” and studies conducted in Spain, American, England, and other European countries were classified as “Caucasian.”
3. Results 3.1. Eligible studies The search yielded 1639 references. Ultimately, 23 studies with 2809 patients were used for the meta-analysis. Fig. 1 shows the reasons for excluding the other 1616 articles. A total of 1069 out of 2809 patients harbored EGFR mutations. Five prospective studies (n = 875) and 18 retrospective studies with 20 groups of data (n = 1934) were included in the pooled analysis of ORR, 2 prospective studies (n = 434) and 10 retrospective studies with 12 groups of data (n = 947) were included in the analysis of PFS, and 2 prospective studies (n = 438) and 7 retrospective studies (n = 711) were included in the analysis of OS. The characteristics of the eligible studies are summarized in Table 1. The Newcastle-Ottawa Scale was used to perform quality assessment on all 23 studies. This scale has been used in other non-randomized studies [19]. Studies that fulfill 5 or more of the 8 criteria (more than 5 stars) were defined as high-quality studies. All studies included in this meta-analysis scored highly (not shown). Two articles [20,21] both had first-line and second-line chemotherapy information; therefore, appropriate data were included separately in the analysis from these studies. Ultimately, 15 studies containing 17 groups of data were from Asians, 6 studies were from Caucasians, and 2 studies were considered to be “mixed”, containing both Asian and Caucasian patients. Seventeen
Q. Zhang et al. / Lung Cancer 85 (2014) 339–345
341
Fig. 1. Flow chart of studies included in this meta-analysis.
studies had information about first-line chemotherapy, 1 study had information about second-line chemotherapy, 4 studies had information about mixed-line chemotherapies (using more than one type of therapy line), 2 studies had information about both first-line and second-line chemotherapy, and 1 study did not have information on the therapy line [8]. EGFR genotypes were detected by direct DNA sequencing in 16 studies, by Scorpions ARMS in 2 studies, and by DHPLC in 3 studies. Two studies, did not report the method for detection of EGFR genotype [8,22]. Two studies divided EGFR mutations into classical mutations and non-classical mutations [9,23], in this instance, only data about overall EGFR mutations were included in this analysis. All the studies detected EGFR mutation status in exons 19 and 21 (at least), except 2 articles that did not provide information on the detected gene locus [8,22]. Two studies focused on adenocarcinomas only [4,9], 3 studies focused on non-squamous non-small cell lung cancer only [22,24,25], and the remaining 18 studies focused on non-small cell lung cancer.
3.2. Response rate Pooled analysis of 5 prospective studies showed the ORR of 42.22% for the EGFR mutation group (133/315) and 26.25% for the EGFR wild-type group (147/560), which demonstrated significantly improved ORR in the EGFR mutation group (RR = 1.42; 95% CI, 1.16–1.74; P = 0.001; I2 = 55.40%; P = 0.062). Similar result was not found in retrospective studies (RR = 1.12; 95% CI, 0.96–1.32; P = 0.146; I2 = 18.70%; P = 0.222; Fig. 2A) The subgroup analysis
results of ORRs (Table 2) indicated that EGFR mutations may predict a better ORR in patients treated with first-line chemotherapy, and there was no statistically significant difference in ORRs between EGFR mutation and wild-type groups in all the three subgroups categorized by EGFR test methods.
3.3. PFS Six studies reported HRs for PFS. One study used univariate analysis to analyze data from 2 groups, while 3 studies used multivariate analysis. 7 studies provided survival curves for PFS, which were used to calculate the HRs and their CIs. Moreover, 1 study provided an HR from multivariate analysis for first-line therapy and a survival curve for second-line therapy. Therefore, 2 prospective studies (n = 434) and 10 retrospective studies with 12 groups of data (n = 947) were included in the pooled analysis of PFS. Compared with the EGFR wild-type population, PFS was not significantly better in patients with mutated EGFR both in prospective (HR = 0.84; 95% CI: 0.65–1.09; P = 0.197; I2 = 0.00%, P = 0.72) and retrospective (HR = 1.02; 95% CI: 0.87–1.18; P = 0.838; I2 = 43.50%, P = 0.053) studies (Fig. 2B). Table 3 showed the sensitivity analysis results of PFS. The combined HR of the 6 studies providing HRs (HR = 0.80; 95% CI, 0.67–0.96; P = 0.018; I2 = 7.50%, P = 0.368) was quite different from that of prospective and retrospective studies, suggesting that EGFR mutations had a statistically significant impact on PFS. Significant association between EGFR mutation status and PFS were not
342
Q. Zhang et al. / Lung Cancer 85 (2014) 339–345
Table 1 Characteristics of studies included in the meta-analysis. Author
Campos-Parra et al. [32] Park et al. [30] Zhuo et al. [33] Lee et al. [24] Lin et al. [34]* Khambata-Ford et al. [12] Yoshida et al. [20]# Yoshida et al. [20]# Hotta et al. [21] Hotta et al. [21] Cappuzzo et al. [29] Lee et al. [35] Eberhard et al. [36] Mok et al. [4] Li et al. [25] Bell et al. [37]# Wu et al. [13]* Guan et al. [8] Camidge et al. [38] Douillard et al. [39] Wu et al. [9]* Lee et al. [14] Okamoto et al. [22]* Kalikaki et al. [23] Wu et al. [40]
Year
2013 2012 2011 2012 2010 2010 2010 2010 2007 2007 2007 2006 2005 2009 2011 2005 2010 2013 2011 2010 2011 2011 2013 2010 2010
Study type
Prospective Retrospective Retrospective Retrospective Retrospective Retrospective Retrospective Retrospective Retrospective Retrospective Retrospective Retrospective Retrospective Prospective Retrospective Retrospective Retrospective Prospective Retrospective Prospective Retrospective Retrospective Prospective Retrospective Retrospective
Ethnicity
Caucasian Asian Asian Asian Asian Caucasian Asian Asian Asian Asian Caucasian Asian Caucasian Asian Asian Mix Asian Asian Caucasian Mix Asian Asian Asian Caucasian Asian
Line
1 1 1 1 1 1 1 2 1 2 1 1 1 1 1 1 2 NR 3 3 3 3 1 1 1
Patients
353 217 145 71 105 87 75 49 54 28 65 75 113 214 128 71 65 81 46 142 156 80 85 159 145
ORR (CR + PR)
PFS
OS
Case
Control
HR (95%CI)
HR (95%CI)
48/111 46/137 20/54 11/34 25/56 1/9 8/25 4/20 3/14 5/11 7/24 6/14 3/14 61/129 19/55 2/5 6/41 11/32 3/10 4/19 12/93 2/43 9/24 12/40 19/55
88/242 28/80 29/91 13/37 15/49 17/78 14/50 2/29 6/40 2/17 15/41 21/61 27/99 20/85 16/73 26/66 9/24 5/49 5/36 12/123 1/63 6/37 22/61 26/119 30/90
0.82 (0.60–1.10) NR NR 0.89 (0.57–1.38) 1.11 (0.73–1.67) NR 1.095 (0.668–1.794) 0.954 (0.528–1.722) 0.422 (0.179–0.991) 1.42 (0.58–3.51) NR 1.8 (1.05–3.09) 1.34 (0.69–2.63) NR NR 0.85 (0.36–2.01) NR 0.91 (0.56–1.47) 1 (0.49–2.04) NR 0.677 (0.485–0.943) 1.52 (0.99–2.34) NR NR NR
0.5 (0.3–0.8) NR 0.187 (0.141–0.412) NR 0.73 (0.46–1.16) NR NR NR 0.263 (0.099–0.699) NR NR NR NR NR NR 0.48 (0.29–0.83) 0.51 (0.28–0.91) NR NR NR 0.81 (0.58–1.12) NR 1.46 (0.45–4.75) NR 0.598 (0.372–0.961)
Note: The Caucasian ethnicity means the majority of patients in this study are Caucasian. Asian is also the same. Abbreviations: NR, no report; 1, first-line chemotherapy; 2, second-line chemotherapy; 3, mixed-line chemotherapy. : HR for PFS was calculated according to survival curve. *: HR for OS was calculated according to survival curve. #: HRs for PFS and OS were from univariate analysis.
observed in any of the subgroups categorized by ethnicity, therapy line, or test method, and there was not enough data for analysis in the mixed, Scorpions ARMS, and DHPLC subgroups. 3.4. Overall survival The HRs for OS were available for 5 studies; 1study was from univariate analysis, and the others were from multivariate analysis. Four studies provided the survival curves. Therefore, 2 prospective studies (n = 438) and 7 retrospective studies (n = 711) were included in the pooled analysis of OS. There was no obvious association between EGFR mutations and OS in prospective studies (HR = 0.74; 95% CI: 0.27–2.05; P = 0.566; I2 = 63.10%, P = 0.1) (Fig. 2C). Although in retrospective studies (HR = 0.48; 95% CI: 0.33–0.72;
P < 0.001; I2 = 75.9%; P < 0.001) and studies providing HRs and CIs (HR = 0.39; 95% CI: 0.25–0.60; P < 0.001; I2 = 67.8%; P = 0.014), the HR for OS favored patients with EGFR mutations; the heterogeneities in these two groups were both significant (Table 3). In the subgroup analysis, similar to the results of retrospective studies were found in Asian, first-line, and direct DNA sequencing subgroups, but not in DHPLC subgroup.
3.5. Publication bias No publication bias for ORR, PFS and OS was found according to funnel plot and Begg’s test (P = 0.636, P = 0.438 and P = 0.455; Supplementary Figure 1A–C).
Table 2 Subgroup analysis results of ORRs. Number of data
RR (95%CI)
P
Heterogeneity test I2
P
Prospective study Retrospective study
5 20
1.42 (1.16–1.74) 1.12 (0.96–1.32)
0.001 0.146
55.40% 18.70%
0.062 0.222
Ethnicity Asian Caucasian Mixed
17 6 2
1.27 (1.00–1.60) 1.16 (0.93–1.45) 1.53 (0.72–3.25)
0.047 0.194 0.270
47.30% 0.00% 0.00%
0.016 0.618 0.329
Therapy line First-line Second-line Mix-line
17 3 4
1.2 (1.05–1.37) 1.49 (0.30–7.44) 1.70 (0.55–5.26)
0.006 0.630 0.354
0.00% 78.20% 61.60%
0.698 0.010 0.050
Test method Direct DNA sequencing Scorpions ARMS DHPLC Overall
18 2 3 25
1.13 (0.89–1.44) 1.51 (0.90–2.51) 1.2 (0.91–1.60) 1.23 (1.08–1.39)
0.305 0.117 0.202 0.001
27.40% 76.10% 0.00% 32.40%
0.137 0.041 0.525 0.061
Q. Zhang et al. / Lung Cancer 85 (2014) 339–345
343
Fig. 2. Forest plots of chemotherapy effects associated with EGFR mutation status in prospective and retrospective studies. (A) ORR: mutated vs. wild-type EGFR. (B) PFS: mutated vs. wild-type EGFR. (C) OS: mutated vs. wild-type EGFR.
Table 3 Sensitivity analysis results of PFS and OS. PFS
Heterogeneity
OS
Number of data
HR (95%CI)
P
I2
P
Number of data
HR (95%CI)
P
I2
Prospective study Retrospective study
2 12
0.84 (0.65–1.09) 1.02 (0.87–1.18)
0.197 0.838
0.00% 43.50%
0.72 0.053
2 7
0.74 (0.27–2.05) 0.48 (0.33–0.72)
0.566 <0.001
63.10% 75.90%
0.1 <0.001
Ethnicity Asian Caucasian Mixed
10 3 1
1.02 (0.81–1.28) 0.90 (0.70–1.17) –
0.889 0.447 –
52.10% 0.00% –
0.027 0.406 –
7 1 1
0.53 (0.34–0.83) – –
0.005 – –
77.90% – –
<0.001 – –
8 3 2 1
0.99 (0.83–1.17) 0.78 (0.59–1.03) 1.36 (0.94–1.96) –
0.867 0.081 0.102 –
39.60% 29.70% 0.00% –
0.115 0.241 0.324 –
7 1 1 –
0.48 (0.31–0.72) – – –
<0.001 – – –
71.10% – – –
0.002 – – –
12 1 0 1 6 14
1.02 (0.87–1.18) – – – 0.80 (0.67–0.96) 0.97 (0.85–1.10)
0.838 – – – 0.018 0.633
43.50% – – – 7.50% 38.30%
0.053 – – – 0.368 0.071
5 1 2 1 5 9
0.60 (0.44–0.82) – 0.34 (0.11–1.05) – 0.39 (0.25–0.60) 0.52 (0.37–0.73)
0.001 – 0.061 – <0.001 <0.001
44.50% – 90.10% – 67.80% 71.10%
0.125 – 0.001 – 0.014 0.001
Therapy line First-line Second-line Mix-line Not report Test method Direct DNA sequencing Scorpions ARMS DHPLC Not report HRs from studies Overall
Abbreviations: –, there were not enough studies to make sensitivity analysis in this group.
Heterogeneity P
344
Q. Zhang et al. / Lung Cancer 85 (2014) 339–345
4. Discussion In this meta-analysis, we evaluated the effects of EGFR mutation status on ORR, PFS (or TTP), and OS of patients with advanced NSCLC treated with chemotherapy. Considering that the evidence from prospective studies was more robust, we drew a conclusion of our analysis mainly according to the result of prospective studies. The pooled analysis of prospective studies indicated that the ORR to chemotherapy was significantly higher in patients with mutated EGFR than in those with wild-type EGFR. This result suggested that EGFR mutation may have some associations with responses to chemotherapy. Neither PFS nor OS following chemotherapy was significantly different between patients with EGFR mutations and with wild-type EGFR in prospective studies. In our analysis, the results of prospective and retrospective studies were not consistent in the analysis of ORR and OS. Generally, results from retrospective studies are more prone to bias. Therefore, results in prospective studies are always considered more robust. Moreover, the obvious heterogeneity in retrospective studies of OS may be another reason for the inconsistency. The subgroup analysis results of ORR showed that better ORR was observed in patients with EGFR mutations in first-line groups, indicating that the association of better ORR to chemotherapy and EGFR mutations may be just observed in patients treated with first-line chemotherapy. EGFR mutation status may change after first-line chemotherapy [26], which can lead to bias in our analysis, and this may be the reason that better response was not found in the second-line subgroup. The small number of studies in the second-line subgroup could also influence the results. With regard to PFS, neither the subgroup analysis nor the sensitivity analysis showed significant associations between EGFR and PFS, except for the 6 studies providing HRs and CIs. As for OS, though the sensitivity analysis favored patients with EGFR mutations in Asian, first-line, studies providing HRs and CIs and direct DNA sequencing subgroups, significant heterogeneities were also found in these subgroups except in direct DNA sequencing subgroup. There were 2 factors which could have an impact on these results. Firstly, estimating HRs and CIs based on survival curves could cause errors both in the analysis of PFS and OS. Secondly, many patients with advanced EGFR mutation-positive NSCLC took EGFR-TKIs before or after chemotherapeutics, which can exert an influence on OS. Further prospective clinical studies are needed to evaluate the influence of EGFR mutation on PFS and OS following chemotherapeutics. In the EGFR test method subgroup analysis, there were no significant differences comparing ORRs between patients with mutated and wild-type EGFR, a statistically significant difference was found in the direct DNA sequencing group for OS. As the gold standard of EGFR detection [27], direct DNA sequencing was the main method used in the included studies. However, it is not the most accurate method compared with Scorpions ARMS and DHPLC [28], and the differences in accuracies of the 3 methods may result in diverse influences on the relationships among ORR, PFS, OS, and EGFR mutation status. However, there were not enough studies to assess the associations of these 3 methods with PFS and OS. One study revealed that individuals with exon 19 deletions were the only responders to chemotherapy among patients with EGFR mutations [29]. The genetic locus of mutations used to differentiate mutated and wild-type EGFR was inconsistent among the included studies, and this can generate bias in the pooled results. In this meta-analysis, when HR and CI were not described in the studies, they were extracted from the survival curves based on the method of Tierney. We should keep in mind that estimated HR values and their CIs may produce bias to some extent. Some degree of heterogeneity across studies was identified mainly due to differences in study type, therapy lines, and populations. Moreover, in the
analysis of PFS and OS, HRs and CIs from univariate and multivariate analyses were pooled together to calculate the summary statistics, and this may be another cause of potential bias. Publication bias was not found in our analysis; therefore the summary statistics obtained may approximate the actual average. This study had a number of limitations that are worth noting. First, different chemotherapeutic regimens may generate diverse effects on NSCLC patients with the same EGFR mutations [30,31], and future studies are necessary to address this issue. Second, this analysis had limited ability to explore the potential factors that may contribute to the ORR, PFS and OS including sex, smoking status, histology, and aberrations in other genes, such as KRAS and anaplastic lymphoma kinase (ALK) translocation. Third, although 23 studies were included in this meta-analysis, only 10 focused on the association of EGFR mutation status and chemotherapy effects specifically, and most of them were retrospective observational studies REV (Review), which inherently contain greater potential for confounding than randomized controlled trials. Despite the limitations of this meta-analysis, our study indicated that EGFR mutations may be associated with better responses in patients with advanced NSCLC treated with chemotherapy. However, the association between EGFR mutation and PFS/OS after conventional chemotherapy should be investigated further. Additionally, considering the heterogeneity of the data, further analysis from prospective, large, randomized clinical studies are required. Conflict of interest statement None declared. Acknowledgements This study was not supported financially by any fund or institution. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.lungcan. 2014.06.011. References [1] Gao G, Ren S, Li A, Xu J, Xu Q, Su C, et al. Epidermal growth factor receptor-tyrosine kinase inhibitor therapy is effective as first-line treatment of advanced non-small-cell lung cancer with mutated EGFR: a meta-analysis from six phase III randomized controlled trials. Int J Cancer 2012;131:E822–9. [2] Dahabreh IJ, Linardou H, Siannis F, Kosmidis P, Bafaloukos D, Murray S. Somatic EGFR mutation and gene copy gain as predictive biomarkers for response to tyrosine kinase inhibitors in non-small cell lung cancer. Clin Cancer Res 2010;16:291–303. [3] Yang C-H, Yu C-J, Shih J-Y, Chang Y-C, Hu F-C, Tsai M-C, et al. Specific EGFR mutations predict treatment outcome of stage IIIB/IV patients with chemotherapy-naive non-small-cell lung cancer receiving first-line gefitinib monotherapy. J Clin Oncol 2008;26:2745–53. [4] Mok TS, Wu YL, Thongprasert S, Yang CH, Chu DT, Saijo N, et al. Gefitinib or carboplatin-paclitaxel in pulmonary adenocarcinoma. N Engl J Med 2009;361:947–57. [5] Gong W, Zhang X, Wu J, Chen L, Li L, Sun J, et al. RRM1 expression and clinical outcome of gemcitabine-containing chemotherapy for advanced non-small-cell lung cancer: a meta-analysis. Lung Cancer 2012;75:374–80. [6] Chen S, Zhang J, Wang R, Luo X, Chen H. The platinum-based treatments for advanced non-small cell lung cancer, is low/negative ERCC1 expression better than high/positive ERCC1 expression? A meta-analysis. Lung Cancer 2010;70:63–70. [7] Han Y, Wang X, Xiao N, Liu Z-D. mRNA expression and clinical significance of ERCC1, BRCA1, RRM1, TYMS and TUBB3 in postoperative patients with non-small cell lung cancer. Asian Pac J Cancer Prev 2013;14:2987–90. [8] Guan JL, Zhong WZ, An SJ, Yang JJ, Su J, Chen ZH, et al. KRAS mutation in patients with lung cancer: a predictor for poor prognosis but not for EGFR-TKIs or chemotherapy. Ann Surg Oncol 2013;20:1381–8.
Q. Zhang et al. / Lung Cancer 85 (2014) 339–345 [9] Wu SG, Yang CH, Yu CJ, Lee JH, Hsu YC, Chang YL, et al. Good response to pemetrexed in patients of lung adenocarcinoma with epidermal growth factor receptor (EGFR) mutations. Lung Cancer 2011;72:333–9. [10] Sordella R, Bell DW, Haber DA, Settleman J. Gefitinib-sensitizing EGFR mutations in lung cancer activate anti-apoptotic pathways. Science 2004;305:1163–7. [11] Yoshimasu T, Oura S, Ohta F, Hirai Y, Naito K, Nakamura R, et al. Epidermal growth factor receptor mutations are associated with docetaxel sensitivity in lung cancer. J Thorac Oncol 2011;6:1658–62. [12] Khambata-Ford S, Harbison CT, Hart LL, Awad M, Xu LA, Horak CE, et al. Analysis of potential predictive markers of cetuximab benefit in BMS099, a phase III study of cetuximab and first-line taxane/carboplatin in advanced non-small-cell lung cancer. J Clin Oncol 2010;28:918–27. [13] Wu JY, Shih JY, Yang CH, Chen KY, Ho CC, Yu CJ, et al. Second-line treatments after first-line gefitinib therapy in advanced nonsmall cell lung cancer. Int J Cancer 2010;126:247–55. [14] Lee JO, Kim TM, Lee SH, Kim DW, Kim S, Jeon YK, et al. Anaplastic lymphoma kinase translocation: a predictive biomarker of pemetrexed in patients with non-small cell lung cancer. J Thorac Oncol 2011;6:1474–80. [15] Wells G, Shea B, O’connell D, Peterson J, Welch V, Losos M, et al. The Newcastle-Ottawa Scale (NOS) for assessing the quality of nonrandomised studies in meta-analyses; 2003. [16] Schoenleber SJ, Kurtz DM, Talwalkar JA, Roberts LR, Gores GJ. Prognostic role of vascular endothelial growth factor in hepatocellular carcinoma: systematic review and meta-analysis. Br J Cancer 2009;100:1385–92. [17] The Cochrane Collaborative Review Group on HIV Infection and AIDS. Editorial Policy: Inclusion and Appraisal of Experimental and Non-experimental (Observational) Studies; 2009. [18] Tierney JF, Stewart LA, Ghersi D, Burdett S, Sydes MR. Practical methods for incorporating summary time-to-event data into meta-analysis. Trials 2007;8:16. [19] Millett GA, Flores SA, Marks G, Reed JB, Herbst JH. Circumcision status and risk of HIV and sexually transmitted infections among men who have sex with men. J Am Med Assoc 2008;300:1674–84. [20] Yoshida K, Yatabe Y, Park J, Ogawa S, Park JY, Shimizu J, et al. Clinical outcomes of advanced non-small cell lung cancer patients screened for epidermal growth factor receptor gene mutations. J Cancer Res Clin Oncol 2010;136:527–35. [21] Hotta K, Kiura K, Toyooka S, Takigawa N, Soh J, Fujiwara Y, et al. Clinical significance of epidermal growth factor receptor gene mutations on treatment outcome after first-line cytotoxic chemotherapy in Japanese patients with non-small cell lung cancer. J Thorac Oncol 2007;2:632–7. [22] Okamoto I, Aoe K, Kato T, Hosomi Y, Yokoyama A, Imamura F, et al. Pemetrexed and carboplatin followed by pemetrexed maintenance therapy in chemo-naive patients with advanced nonsquamous non-small-cell lung cancer. Invest New Drugs 2013;31:1275–82. [23] Kalikaki A, Koutsopoulos A, Hatzidaki D, Trypaki M, Kontopodis E, Stathopoulos E, et al. Clinical outcome of patients with non-small cell lung cancer receiving front-line chemotherapy according to EGFR and K-RAS mutation status. Lung Cancer 2010;69:110–5. [24] Lee JK, Park HS, Kim DW, Kulig K, Kim TM, Lee SH, et al. Comparative analyses of overall survival in patients with anaplastic lymphoma kinase-positive and matched wild-type advanced nonsmall cell lung cancer. Cancer 2012;118:3579–86. [25] Li F, Bai H, Li X, Wu M, Yu R, Shi A, et al. Role of EGFR mutation status in patients with stage III non-squamous non-small cell lung cancer treated with chemoradiotherapy. Zhongguo Fei Ai Za Zhi 2011;14:715–8.
345
[26] Bai H, Wang Z, Chen K, Zhao J, Lee JJ, Wang S, et al. Influence of chemotherapy on EGFR mutation status among patients with non-small-cell lung cancer. J Clin Oncol 2012;30:3077–83. [27] Bakker E. Is the DNA sequence the gold standard in genetic testing? Quality of molecular genetic tests assessed. Clin Chem 2006;52:557–8. [28] Qin L, Zhong W, Zhang L, Li LY, Wang MZ. Comparison of three methods for detecting epidermal growth factor receptor mutations in plasma DNA samples of Chinese patients with advanced non-small cell lung cancer. Chin Med J (Engl) 2011;124:887–91. [29] Cappuzzo F, Ligorio C, Toschi L, Rossi E, Trisolini R, Paioli D, et al. EGFR and HER2 gene copy number and response to first-line chemotherapy in patients with advanced non-small cell lung cancer (NSCLC). J Thorac Oncol 2007;2: 423–9. [30] Park JH, Lee SH, Keam B, Kim TM, Kim DW, Yang SC, et al. EGFR mutations as a predictive marker of cytotoxic chemotherapy. Lung Cancer 2012;77: 433–7. [31] Dong X, Zhao X, Hao Y, Wei Y, Yin Q, Du J. Response to first-line chemotherapy in patients with non-small-cell lung cancer according to epidermal growth factor receptor and K-RAS mutation status. Clin Lung Cancer 2013;14:680–7. [32] Campos-Parra AD, Zuloaga C, Manriquez ME, Aviles A, Borbolla-Escoboza J, Cardona A, et al. KRAS mutation as the biomarker of response to chemotherapy and EGFR-TKIs in patients with advanced non-small cell lung cancer: clues for its potential use in second-line therapy decision making. Am J Clin Oncol 2013 [Epub ahead of print]. [33] Zhuo ML, Wu MN, Zhao J, Song SW, Bai H, Wang SH, 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;124:3510–4. [34] Lin CC, Hsu HH, Sun CT, Shih JY, Lin ZZ, Yu CJ, et al. Chemotherapy response in East Asian non-small cell lung cancer patients harboring wild-type or activating mutation of epidermal growth factor receptors. J Thorac Oncol 2010;5:1424–9. [35] Lee KH, Han SW, Hwang PG, Oh DY, Kim DW, Chung DH, et al. Epidermal growth factor receptor mutations and response to chemotherapy in patients with non-small-cell lung cancer. Jpn J Clin Oncol 2006;36:344–50. [36] Eberhard DA, Johnson BE, Amler LC, Goddard AD, Heldens SL, Herbst RS, et al. Mutations in the epidermal growth factor receptor and in KRAS are predictive and prognostic indicators in patients with non-small-cell lung cancer treated with chemotherapy alone and in combination with erlotinib. J Clin Oncol 2005;23:5900–9. [37] Bell DW, Lynch TJ, Haserlat SM, Harris PL, Okimoto RA, Brannigan BW, et al. Epidermal growth factor receptor mutations and gene amplification in non-small-cell lung cancer: molecular analysis of the IDEAL/INTACT gefitinib trials. J Clin Oncol 2005;23:8081–92. [38] Camidge DR, Kono SA, Lu X, Okuyama S, Baron AE, Oton AB, et al. Anaplastic lymphoma kinase gene rearrangements in non-small cell lung cancer are associated with prolonged progression-free survival on pemetrexed. J Thorac Oncol 2011;6:774–80. [39] Douillard JY, Shepherd FA, Hirsh V, Mok T, Socinski MA, Gervais R, et al. Molecular predictors of outcome with gefitinib and docetaxel in previously treated non-small-cell lung cancer: data from the randomized phase III INTEREST trial. J Clin Oncol 2010;28:744–52. [40] Wu M, Zhao J, Song SW, Zhuo M, Wang X, Bai H, et al. EGFR mutations are associated with prognosis but not with the response to front-line chemotherapy in the Chinese patients with advanced non-small cell lung cancer. Lung Cancer 2010;67:343–7.
Contents lists available at ScienceDirect
Lung Cancer journal homepage: www.elsevier.com/locate/lungcan
Review
EGFR mutations and clinical outcomes of chemotherapy for advanced non-small cell lung cancer: A meta-analysis Qiong Zhang a , Hong-Hai Dai a , Hong-Yun Dong b,c , Cheng-Tao Sun a , Zhe Yang a,∗,1 , Jun-Qing Han a,∗,1 a
Tumor Research and Therapy Center, Provincial Hospital Affiliated to Shandong University, 324 Jingwu Weiqi Road, Jinan, Shandong 250021, PR China School of Public Health, Shandong University, 44 Wenhuaxi Road, Jinan, Shandong 250012, PR China c Wanhua Chemical Group Co., Ltd., 7 South Xingfu Road, Yantai, Shandong 264002, PR China b
a r t i c l e
i n f o
Article history: Received 11 October 2013 Received in revised form 20 April 2014 Accepted 13 June 2014 Keywords: Non-small cell lung cancer EGFR Mutation status Chemotherapy Response rate Meta-analysis
a b s t r a c t Background: This meta-analysis was performed to assess whether epidermal growth factor receptor (EGFR) mutation status was associated with objective response rate (ORR), progression-free survival (PFS), and overall survival (OS) in patients with advanced non-small cell lung cancer (NSCLC) treated with chemotherapy. Method: We systematically identified eligible articles investigating the effects of chemotherapy in patients with NSCLC stratified by EGFR mutation status. The summary risk ratio (RR) for ORR and hazard ratios (HRs) for both PFS and OS were calculated using the inverse variance formula of meta-analysis. Results: Identification for the current meta-analysis: 5 prospective studies (n = 875) and 18 retrospective studies (n = 1934) for ORR; 2 prospective studies (n = 434) and 10 retrospective studies (n = 947) for PFS; 2 prospective studies (n = 438) and 7 retrospective studies (n = 711) for OS. The ORR was significantly higher in patients with EGFR mutations in prospective studies (RR = 1.42; 95% confidence interval [CI], 1.16–1.74; P = 0.001), but not in retrospective studies (RR = 1.12; 95% CI, 0.96–1.32; P = 0.146). There was no obvious association between EGFR mutations and PFS both in prospective (HR = 0.84; 95% CI: 0.65–1.09; P = 0.197) and retrospective (HR = 1.02; 95% CI: 0.87–1.18; P = 0.838) studies. Association between EGFR mutations and OS was also not seen in prospective studies (HR = 0.74; 95% CI: 0.27–2.05; P = 0.566), but was seen in retrospective studies (HR = 0.48; 95% CI: 0.33–0.72; P < 0.001; I2 = 75.9%; P < 0.001) with significant heterogeneity. Conclusion: EGFR mutations in advanced NSCLC may be associated with higher ORRs to chemotherapy, but may have nothing to do with PFS and OS. Further prospective studies are required to identify the influence of EGFR mutations on chemotherapy effects in advanced NSCLC. © 2014 Elsevier Ireland Ltd. All rights reserved.
1. Introduction The development and utilization of epidermal growth factor receptor (EGFR) tyrosine kinase inhibitors (TKIs) was a great advancement in the treatment of specific cases of advanced non-small cell lung cancer (NSCLC). EGFR mutations have been demonstrated to predict good efficacy of EGFR-TKIs in NSCLC [1–4]. Consequently, it is important to confirm the EGFR genotype of patients with NSCLC to determine whether EGFR-TKI therapy
∗ Corresponding authors at: Tumor Research and Therapy Center, Provincial Hospital Affiliated to Shandong University, 324 Jingwu Weiqi Road, Jinan, Shandong 250021, PR China. Tel.: +86 531 6877 6100; fax: +86 531 87068707. E-mail addresses: [email protected] (Z. Yang), [email protected] (J.-Q. Han). 1 These two corresponding authors contributed equally to this work. http://dx.doi.org/10.1016/j.lungcan.2014.06.011 0169-5002/© 2014 Elsevier Ireland Ltd. All rights reserved.
would be appropriate. Although EGFR-TKIs play a critical role in the treatment of advanced NSCLC, chemotherapy with conventional anticancer agents is still the gold-standard approach for patients with unresectable, locally advanced, or metastatic NSCLC. Nevertheless, the association between EGFR genotype and the effects of chemotherapy remains uncertain. With the emphasis on biomarkers in the treatment of NSCLC, many studies have been conducted to investigate the biomarkers predicting patient response to specific chemotherapy regimens [5–7]. In the Iressa Pan-Asia Survival Study (IPASS) [4], Asian patients with EGFR mutations had a significantly higher response rate (47.3% versus 23.5%) than patients without EGFR mutations when they received paclitaxel plus carboplatin as first-line therapy. Similar results were observed in 2 other studies [8,9]. However, 2 preclinical experiments demonstrated that lung cancer cells or tumors with EGFR mutations may tend to markedly
340
Q. Zhang et al. / Lung Cancer 85 (2014) 339–345
resist chemotherapeutic agents [10,11]. Consistent with this, several clinical studies have suggested patients with mutated EGFR have poorer responses to chemotherapy than patients with wildtype EGFR [12–14]. Furthermore, different studies using the same methods to detect EGFR mutation status and the same chemotherapeutic agents have yielded contradictory results [9,14]. Because of these inconsistent results, we conducted a meta-analysis to evaluate the role of EGFR mutation status in predicting the efficacy, progression-free survival (PFS), and overall survival (OS) of chemotherapy in patients with advanced NSCLC. In this article, the word “chemotherapy” represents chemotherapy with conventional anticancer agents.
2. Materials and methods 2.1. Literature search strategy We performed computerized searches of the PubMed database, the EMBASE database and the Cochrane library (Issue 6, 2013) to identify all published articles reporting on chemotherapy for advanced NSCLC with known EGFR mutation status using the following key words: non-small cell lung cancer, NSCLC, lung cancer, chemotherapy, and epidermal growth factor receptor. The search was performed on July16, 2013. The published languages were not limited. The references of all reviewed articles were also screened for additional studies.
2.2. Selection criteria Eligible studies had to meet the following criteria:(1) patients with advanced or recurrent NSCLC; (2) detected EGFR gene mutation status in primary lung cancer tissue, metastatic tissue or sera DNA; (3) received chemotherapy; (4) reported data of response rate, the hazard ratios (HRs) with the corresponding 95% confidence intervals (CIs) comparing OS, PFS or time-to-progression (TTP) stratified by EGFR mutation status; and (5) the number of patients in the EGFR mutation or wild-type group was not less than 5. Studies examining chemotherapy in combination with any other agents such as EGFR-TKIs, any investigational drug or radiotherapy were excluded.
2.3. Quality assessment Two investigators (QZ, HYD) independently assessed the quality of studies using the Newcastle Ottawa Quality Assessment Scale for cohort studies [15,16]. Discrepancies were resolved by consensus. This scale is composed of eight items which assess patient selection, study comparability and outcome. The scale was recommended by the Cochrane Non-Randomized Studies Methods Working Group [17].
2.4. Data extraction Data from each study were extracted independently by 2 investigators, using a standardized data extraction form. Disagreements between the 2 investigators were resolved by consensus. The following information was abstracted from each publication: authors, publication date, country, sample size, study type, characteristics of the patients, data about EGFR mutations, test methods, numbers of patients in the EGFR mutation and wild-type groups, chemotherapy information, objective response rates (ORRs), and HRs and 95%CIs for PFS (or TTP) and OS.
2.5. Statistical methods The risk ratio (RR) was calculated for ORRs, and the HR was calculated for PFS and OS. In studies without HRs, Kaplan–Meier plots were used to calculate the HRs based on the methods presented by Tierney [18]. STATA SE 12.0 package (StataCorp, College Station, TX, USA) was used for statistical analyses. The fixed-effects model (inverse variance formula) was initially used. If P (for I2 ) < 0.05, indicating significant heterogeneity among studies, then the random-effects model (I–V heterogeneity formula) was used. Two-tailed P values were used for all comparisons, and statistical significance was defined as P < 0.05. The possibility of publication bias was investigated by inspecting funnel plots and was statistically analyzed using Begg’s test. We planned additional sensitivity analyses to further detect and evaluate clinical heterogeneity. Most of the eligible studies used first-line chemotherapy with platinum-based double-agent regimens (17/23), with the second agent varying between studies (Supplementary Table 1). Consequently, it was difficult to perform subgroup analyses according to the therapeutic regimen. There also was not enough data about pathological types and sex to perform separate analyses of these subgroups. Finally, subgroup analyses were conducted to assess the effects of ethnicity (Asian, Caucasian, or mixed), therapy line (first-line, second-line, and mixed-line), and test methods of EGFR genotype (scorpions amplification refractory mutation system [Scorpions ARMS] or direct DNA sequencing or denaturing high performance liquid chromatography [DHPLC]). Therefore the categories used to make subgroup analysis were not pre-specified. We performed sensitivity analysis for studies providing HRs and for some categories without sufficient studies for subgroup analysis in the additional analysis of PFS and OS. Studies conducted in Korea, China, Japan and other Asian countries were classified as “Asian” and studies conducted in Spain, American, England, and other European countries were classified as “Caucasian.”
3. Results 3.1. Eligible studies The search yielded 1639 references. Ultimately, 23 studies with 2809 patients were used for the meta-analysis. Fig. 1 shows the reasons for excluding the other 1616 articles. A total of 1069 out of 2809 patients harbored EGFR mutations. Five prospective studies (n = 875) and 18 retrospective studies with 20 groups of data (n = 1934) were included in the pooled analysis of ORR, 2 prospective studies (n = 434) and 10 retrospective studies with 12 groups of data (n = 947) were included in the analysis of PFS, and 2 prospective studies (n = 438) and 7 retrospective studies (n = 711) were included in the analysis of OS. The characteristics of the eligible studies are summarized in Table 1. The Newcastle-Ottawa Scale was used to perform quality assessment on all 23 studies. This scale has been used in other non-randomized studies [19]. Studies that fulfill 5 or more of the 8 criteria (more than 5 stars) were defined as high-quality studies. All studies included in this meta-analysis scored highly (not shown). Two articles [20,21] both had first-line and second-line chemotherapy information; therefore, appropriate data were included separately in the analysis from these studies. Ultimately, 15 studies containing 17 groups of data were from Asians, 6 studies were from Caucasians, and 2 studies were considered to be “mixed”, containing both Asian and Caucasian patients. Seventeen
Q. Zhang et al. / Lung Cancer 85 (2014) 339–345
341
Fig. 1. Flow chart of studies included in this meta-analysis.
studies had information about first-line chemotherapy, 1 study had information about second-line chemotherapy, 4 studies had information about mixed-line chemotherapies (using more than one type of therapy line), 2 studies had information about both first-line and second-line chemotherapy, and 1 study did not have information on the therapy line [8]. EGFR genotypes were detected by direct DNA sequencing in 16 studies, by Scorpions ARMS in 2 studies, and by DHPLC in 3 studies. Two studies, did not report the method for detection of EGFR genotype [8,22]. Two studies divided EGFR mutations into classical mutations and non-classical mutations [9,23], in this instance, only data about overall EGFR mutations were included in this analysis. All the studies detected EGFR mutation status in exons 19 and 21 (at least), except 2 articles that did not provide information on the detected gene locus [8,22]. Two studies focused on adenocarcinomas only [4,9], 3 studies focused on non-squamous non-small cell lung cancer only [22,24,25], and the remaining 18 studies focused on non-small cell lung cancer.
3.2. Response rate Pooled analysis of 5 prospective studies showed the ORR of 42.22% for the EGFR mutation group (133/315) and 26.25% for the EGFR wild-type group (147/560), which demonstrated significantly improved ORR in the EGFR mutation group (RR = 1.42; 95% CI, 1.16–1.74; P = 0.001; I2 = 55.40%; P = 0.062). Similar result was not found in retrospective studies (RR = 1.12; 95% CI, 0.96–1.32; P = 0.146; I2 = 18.70%; P = 0.222; Fig. 2A) The subgroup analysis
results of ORRs (Table 2) indicated that EGFR mutations may predict a better ORR in patients treated with first-line chemotherapy, and there was no statistically significant difference in ORRs between EGFR mutation and wild-type groups in all the three subgroups categorized by EGFR test methods.
3.3. PFS Six studies reported HRs for PFS. One study used univariate analysis to analyze data from 2 groups, while 3 studies used multivariate analysis. 7 studies provided survival curves for PFS, which were used to calculate the HRs and their CIs. Moreover, 1 study provided an HR from multivariate analysis for first-line therapy and a survival curve for second-line therapy. Therefore, 2 prospective studies (n = 434) and 10 retrospective studies with 12 groups of data (n = 947) were included in the pooled analysis of PFS. Compared with the EGFR wild-type population, PFS was not significantly better in patients with mutated EGFR both in prospective (HR = 0.84; 95% CI: 0.65–1.09; P = 0.197; I2 = 0.00%, P = 0.72) and retrospective (HR = 1.02; 95% CI: 0.87–1.18; P = 0.838; I2 = 43.50%, P = 0.053) studies (Fig. 2B). Table 3 showed the sensitivity analysis results of PFS. The combined HR of the 6 studies providing HRs (HR = 0.80; 95% CI, 0.67–0.96; P = 0.018; I2 = 7.50%, P = 0.368) was quite different from that of prospective and retrospective studies, suggesting that EGFR mutations had a statistically significant impact on PFS. Significant association between EGFR mutation status and PFS were not
342
Q. Zhang et al. / Lung Cancer 85 (2014) 339–345
Table 1 Characteristics of studies included in the meta-analysis. Author
Campos-Parra et al. [32] Park et al. [30] Zhuo et al. [33] Lee et al. [24] Lin et al. [34]* Khambata-Ford et al. [12] Yoshida et al. [20]# Yoshida et al. [20]# Hotta et al. [21] Hotta et al. [21] Cappuzzo et al. [29] Lee et al. [35] Eberhard et al. [36] Mok et al. [4] Li et al. [25] Bell et al. [37]# Wu et al. [13]* Guan et al. [8] Camidge et al. [38] Douillard et al. [39] Wu et al. [9]* Lee et al. [14] Okamoto et al. [22]* Kalikaki et al. [23] Wu et al. [40]
Year
2013 2012 2011 2012 2010 2010 2010 2010 2007 2007 2007 2006 2005 2009 2011 2005 2010 2013 2011 2010 2011 2011 2013 2010 2010
Study type
Prospective Retrospective Retrospective Retrospective Retrospective Retrospective Retrospective Retrospective Retrospective Retrospective Retrospective Retrospective Retrospective Prospective Retrospective Retrospective Retrospective Prospective Retrospective Prospective Retrospective Retrospective Prospective Retrospective Retrospective
Ethnicity
Caucasian Asian Asian Asian Asian Caucasian Asian Asian Asian Asian Caucasian Asian Caucasian Asian Asian Mix Asian Asian Caucasian Mix Asian Asian Asian Caucasian Asian
Line
1 1 1 1 1 1 1 2 1 2 1 1 1 1 1 1 2 NR 3 3 3 3 1 1 1
Patients
353 217 145 71 105 87 75 49 54 28 65 75 113 214 128 71 65 81 46 142 156 80 85 159 145
ORR (CR + PR)
PFS
OS
Case
Control
HR (95%CI)
HR (95%CI)
48/111 46/137 20/54 11/34 25/56 1/9 8/25 4/20 3/14 5/11 7/24 6/14 3/14 61/129 19/55 2/5 6/41 11/32 3/10 4/19 12/93 2/43 9/24 12/40 19/55
88/242 28/80 29/91 13/37 15/49 17/78 14/50 2/29 6/40 2/17 15/41 21/61 27/99 20/85 16/73 26/66 9/24 5/49 5/36 12/123 1/63 6/37 22/61 26/119 30/90
0.82 (0.60–1.10) NR NR 0.89 (0.57–1.38) 1.11 (0.73–1.67) NR 1.095 (0.668–1.794) 0.954 (0.528–1.722) 0.422 (0.179–0.991) 1.42 (0.58–3.51) NR 1.8 (1.05–3.09) 1.34 (0.69–2.63) NR NR 0.85 (0.36–2.01) NR 0.91 (0.56–1.47) 1 (0.49–2.04) NR 0.677 (0.485–0.943) 1.52 (0.99–2.34) NR NR NR
0.5 (0.3–0.8) NR 0.187 (0.141–0.412) NR 0.73 (0.46–1.16) NR NR NR 0.263 (0.099–0.699) NR NR NR NR NR NR 0.48 (0.29–0.83) 0.51 (0.28–0.91) NR NR NR 0.81 (0.58–1.12) NR 1.46 (0.45–4.75) NR 0.598 (0.372–0.961)
Note: The Caucasian ethnicity means the majority of patients in this study are Caucasian. Asian is also the same. Abbreviations: NR, no report; 1, first-line chemotherapy; 2, second-line chemotherapy; 3, mixed-line chemotherapy. : HR for PFS was calculated according to survival curve. *: HR for OS was calculated according to survival curve. #: HRs for PFS and OS were from univariate analysis.
observed in any of the subgroups categorized by ethnicity, therapy line, or test method, and there was not enough data for analysis in the mixed, Scorpions ARMS, and DHPLC subgroups. 3.4. Overall survival The HRs for OS were available for 5 studies; 1study was from univariate analysis, and the others were from multivariate analysis. Four studies provided the survival curves. Therefore, 2 prospective studies (n = 438) and 7 retrospective studies (n = 711) were included in the pooled analysis of OS. There was no obvious association between EGFR mutations and OS in prospective studies (HR = 0.74; 95% CI: 0.27–2.05; P = 0.566; I2 = 63.10%, P = 0.1) (Fig. 2C). Although in retrospective studies (HR = 0.48; 95% CI: 0.33–0.72;
P < 0.001; I2 = 75.9%; P < 0.001) and studies providing HRs and CIs (HR = 0.39; 95% CI: 0.25–0.60; P < 0.001; I2 = 67.8%; P = 0.014), the HR for OS favored patients with EGFR mutations; the heterogeneities in these two groups were both significant (Table 3). In the subgroup analysis, similar to the results of retrospective studies were found in Asian, first-line, and direct DNA sequencing subgroups, but not in DHPLC subgroup.
3.5. Publication bias No publication bias for ORR, PFS and OS was found according to funnel plot and Begg’s test (P = 0.636, P = 0.438 and P = 0.455; Supplementary Figure 1A–C).
Table 2 Subgroup analysis results of ORRs. Number of data
RR (95%CI)
P
Heterogeneity test I2
P
Prospective study Retrospective study
5 20
1.42 (1.16–1.74) 1.12 (0.96–1.32)
0.001 0.146
55.40% 18.70%
0.062 0.222
Ethnicity Asian Caucasian Mixed
17 6 2
1.27 (1.00–1.60) 1.16 (0.93–1.45) 1.53 (0.72–3.25)
0.047 0.194 0.270
47.30% 0.00% 0.00%
0.016 0.618 0.329
Therapy line First-line Second-line Mix-line
17 3 4
1.2 (1.05–1.37) 1.49 (0.30–7.44) 1.70 (0.55–5.26)
0.006 0.630 0.354
0.00% 78.20% 61.60%
0.698 0.010 0.050
Test method Direct DNA sequencing Scorpions ARMS DHPLC Overall
18 2 3 25
1.13 (0.89–1.44) 1.51 (0.90–2.51) 1.2 (0.91–1.60) 1.23 (1.08–1.39)
0.305 0.117 0.202 0.001
27.40% 76.10% 0.00% 32.40%
0.137 0.041 0.525 0.061
Q. Zhang et al. / Lung Cancer 85 (2014) 339–345
343
Fig. 2. Forest plots of chemotherapy effects associated with EGFR mutation status in prospective and retrospective studies. (A) ORR: mutated vs. wild-type EGFR. (B) PFS: mutated vs. wild-type EGFR. (C) OS: mutated vs. wild-type EGFR.
Table 3 Sensitivity analysis results of PFS and OS. PFS
Heterogeneity
OS
Number of data
HR (95%CI)
P
I2
P
Number of data
HR (95%CI)
P
I2
Prospective study Retrospective study
2 12
0.84 (0.65–1.09) 1.02 (0.87–1.18)
0.197 0.838
0.00% 43.50%
0.72 0.053
2 7
0.74 (0.27–2.05) 0.48 (0.33–0.72)
0.566 <0.001
63.10% 75.90%
0.1 <0.001
Ethnicity Asian Caucasian Mixed
10 3 1
1.02 (0.81–1.28) 0.90 (0.70–1.17) –
0.889 0.447 –
52.10% 0.00% –
0.027 0.406 –
7 1 1
0.53 (0.34–0.83) – –
0.005 – –
77.90% – –
<0.001 – –
8 3 2 1
0.99 (0.83–1.17) 0.78 (0.59–1.03) 1.36 (0.94–1.96) –
0.867 0.081 0.102 –
39.60% 29.70% 0.00% –
0.115 0.241 0.324 –
7 1 1 –
0.48 (0.31–0.72) – – –
<0.001 – – –
71.10% – – –
0.002 – – –
12 1 0 1 6 14
1.02 (0.87–1.18) – – – 0.80 (0.67–0.96) 0.97 (0.85–1.10)
0.838 – – – 0.018 0.633
43.50% – – – 7.50% 38.30%
0.053 – – – 0.368 0.071
5 1 2 1 5 9
0.60 (0.44–0.82) – 0.34 (0.11–1.05) – 0.39 (0.25–0.60) 0.52 (0.37–0.73)
0.001 – 0.061 – <0.001 <0.001
44.50% – 90.10% – 67.80% 71.10%
0.125 – 0.001 – 0.014 0.001
Therapy line First-line Second-line Mix-line Not report Test method Direct DNA sequencing Scorpions ARMS DHPLC Not report HRs from studies Overall
Abbreviations: –, there were not enough studies to make sensitivity analysis in this group.
Heterogeneity P
344
Q. Zhang et al. / Lung Cancer 85 (2014) 339–345
4. Discussion In this meta-analysis, we evaluated the effects of EGFR mutation status on ORR, PFS (or TTP), and OS of patients with advanced NSCLC treated with chemotherapy. Considering that the evidence from prospective studies was more robust, we drew a conclusion of our analysis mainly according to the result of prospective studies. The pooled analysis of prospective studies indicated that the ORR to chemotherapy was significantly higher in patients with mutated EGFR than in those with wild-type EGFR. This result suggested that EGFR mutation may have some associations with responses to chemotherapy. Neither PFS nor OS following chemotherapy was significantly different between patients with EGFR mutations and with wild-type EGFR in prospective studies. In our analysis, the results of prospective and retrospective studies were not consistent in the analysis of ORR and OS. Generally, results from retrospective studies are more prone to bias. Therefore, results in prospective studies are always considered more robust. Moreover, the obvious heterogeneity in retrospective studies of OS may be another reason for the inconsistency. The subgroup analysis results of ORR showed that better ORR was observed in patients with EGFR mutations in first-line groups, indicating that the association of better ORR to chemotherapy and EGFR mutations may be just observed in patients treated with first-line chemotherapy. EGFR mutation status may change after first-line chemotherapy [26], which can lead to bias in our analysis, and this may be the reason that better response was not found in the second-line subgroup. The small number of studies in the second-line subgroup could also influence the results. With regard to PFS, neither the subgroup analysis nor the sensitivity analysis showed significant associations between EGFR and PFS, except for the 6 studies providing HRs and CIs. As for OS, though the sensitivity analysis favored patients with EGFR mutations in Asian, first-line, studies providing HRs and CIs and direct DNA sequencing subgroups, significant heterogeneities were also found in these subgroups except in direct DNA sequencing subgroup. There were 2 factors which could have an impact on these results. Firstly, estimating HRs and CIs based on survival curves could cause errors both in the analysis of PFS and OS. Secondly, many patients with advanced EGFR mutation-positive NSCLC took EGFR-TKIs before or after chemotherapeutics, which can exert an influence on OS. Further prospective clinical studies are needed to evaluate the influence of EGFR mutation on PFS and OS following chemotherapeutics. In the EGFR test method subgroup analysis, there were no significant differences comparing ORRs between patients with mutated and wild-type EGFR, a statistically significant difference was found in the direct DNA sequencing group for OS. As the gold standard of EGFR detection [27], direct DNA sequencing was the main method used in the included studies. However, it is not the most accurate method compared with Scorpions ARMS and DHPLC [28], and the differences in accuracies of the 3 methods may result in diverse influences on the relationships among ORR, PFS, OS, and EGFR mutation status. However, there were not enough studies to assess the associations of these 3 methods with PFS and OS. One study revealed that individuals with exon 19 deletions were the only responders to chemotherapy among patients with EGFR mutations [29]. The genetic locus of mutations used to differentiate mutated and wild-type EGFR was inconsistent among the included studies, and this can generate bias in the pooled results. In this meta-analysis, when HR and CI were not described in the studies, they were extracted from the survival curves based on the method of Tierney. We should keep in mind that estimated HR values and their CIs may produce bias to some extent. Some degree of heterogeneity across studies was identified mainly due to differences in study type, therapy lines, and populations. Moreover, in the
analysis of PFS and OS, HRs and CIs from univariate and multivariate analyses were pooled together to calculate the summary statistics, and this may be another cause of potential bias. Publication bias was not found in our analysis; therefore the summary statistics obtained may approximate the actual average. This study had a number of limitations that are worth noting. First, different chemotherapeutic regimens may generate diverse effects on NSCLC patients with the same EGFR mutations [30,31], and future studies are necessary to address this issue. Second, this analysis had limited ability to explore the potential factors that may contribute to the ORR, PFS and OS including sex, smoking status, histology, and aberrations in other genes, such as KRAS and anaplastic lymphoma kinase (ALK) translocation. Third, although 23 studies were included in this meta-analysis, only 10 focused on the association of EGFR mutation status and chemotherapy effects specifically, and most of them were retrospective observational studies REV (Review), which inherently contain greater potential for confounding than randomized controlled trials. Despite the limitations of this meta-analysis, our study indicated that EGFR mutations may be associated with better responses in patients with advanced NSCLC treated with chemotherapy. However, the association between EGFR mutation and PFS/OS after conventional chemotherapy should be investigated further. Additionally, considering the heterogeneity of the data, further analysis from prospective, large, randomized clinical studies are required. Conflict of interest statement None declared. Acknowledgements This study was not supported financially by any fund or institution. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.lungcan. 2014.06.011. References [1] Gao G, Ren S, Li A, Xu J, Xu Q, Su C, et al. Epidermal growth factor receptor-tyrosine kinase inhibitor therapy is effective as first-line treatment of advanced non-small-cell lung cancer with mutated EGFR: a meta-analysis from six phase III randomized controlled trials. Int J Cancer 2012;131:E822–9. [2] Dahabreh IJ, Linardou H, Siannis F, Kosmidis P, Bafaloukos D, Murray S. Somatic EGFR mutation and gene copy gain as predictive biomarkers for response to tyrosine kinase inhibitors in non-small cell lung cancer. Clin Cancer Res 2010;16:291–303. [3] Yang C-H, Yu C-J, Shih J-Y, Chang Y-C, Hu F-C, Tsai M-C, et al. Specific EGFR mutations predict treatment outcome of stage IIIB/IV patients with chemotherapy-naive non-small-cell lung cancer receiving first-line gefitinib monotherapy. J Clin Oncol 2008;26:2745–53. [4] Mok TS, Wu YL, Thongprasert S, Yang CH, Chu DT, Saijo N, et al. Gefitinib or carboplatin-paclitaxel in pulmonary adenocarcinoma. N Engl J Med 2009;361:947–57. [5] Gong W, Zhang X, Wu J, Chen L, Li L, Sun J, et al. RRM1 expression and clinical outcome of gemcitabine-containing chemotherapy for advanced non-small-cell lung cancer: a meta-analysis. Lung Cancer 2012;75:374–80. [6] Chen S, Zhang J, Wang R, Luo X, Chen H. The platinum-based treatments for advanced non-small cell lung cancer, is low/negative ERCC1 expression better than high/positive ERCC1 expression? A meta-analysis. Lung Cancer 2010;70:63–70. [7] Han Y, Wang X, Xiao N, Liu Z-D. mRNA expression and clinical significance of ERCC1, BRCA1, RRM1, TYMS and TUBB3 in postoperative patients with non-small cell lung cancer. Asian Pac J Cancer Prev 2013;14:2987–90. [8] Guan JL, Zhong WZ, An SJ, Yang JJ, Su J, Chen ZH, et al. KRAS mutation in patients with lung cancer: a predictor for poor prognosis but not for EGFR-TKIs or chemotherapy. Ann Surg Oncol 2013;20:1381–8.
Q. Zhang et al. / Lung Cancer 85 (2014) 339–345 [9] Wu SG, Yang CH, Yu CJ, Lee JH, Hsu YC, Chang YL, et al. Good response to pemetrexed in patients of lung adenocarcinoma with epidermal growth factor receptor (EGFR) mutations. Lung Cancer 2011;72:333–9. [10] Sordella R, Bell DW, Haber DA, Settleman J. Gefitinib-sensitizing EGFR mutations in lung cancer activate anti-apoptotic pathways. Science 2004;305:1163–7. [11] Yoshimasu T, Oura S, Ohta F, Hirai Y, Naito K, Nakamura R, et al. Epidermal growth factor receptor mutations are associated with docetaxel sensitivity in lung cancer. J Thorac Oncol 2011;6:1658–62. [12] Khambata-Ford S, Harbison CT, Hart LL, Awad M, Xu LA, Horak CE, et al. Analysis of potential predictive markers of cetuximab benefit in BMS099, a phase III study of cetuximab and first-line taxane/carboplatin in advanced non-small-cell lung cancer. J Clin Oncol 2010;28:918–27. [13] Wu JY, Shih JY, Yang CH, Chen KY, Ho CC, Yu CJ, et al. Second-line treatments after first-line gefitinib therapy in advanced nonsmall cell lung cancer. Int J Cancer 2010;126:247–55. [14] Lee JO, Kim TM, Lee SH, Kim DW, Kim S, Jeon YK, et al. Anaplastic lymphoma kinase translocation: a predictive biomarker of pemetrexed in patients with non-small cell lung cancer. J Thorac Oncol 2011;6:1474–80. [15] Wells G, Shea B, O’connell D, Peterson J, Welch V, Losos M, et al. The Newcastle-Ottawa Scale (NOS) for assessing the quality of nonrandomised studies in meta-analyses; 2003. [16] Schoenleber SJ, Kurtz DM, Talwalkar JA, Roberts LR, Gores GJ. Prognostic role of vascular endothelial growth factor in hepatocellular carcinoma: systematic review and meta-analysis. Br J Cancer 2009;100:1385–92. [17] The Cochrane Collaborative Review Group on HIV Infection and AIDS. Editorial Policy: Inclusion and Appraisal of Experimental and Non-experimental (Observational) Studies; 2009. [18] Tierney JF, Stewart LA, Ghersi D, Burdett S, Sydes MR. Practical methods for incorporating summary time-to-event data into meta-analysis. Trials 2007;8:16. [19] Millett GA, Flores SA, Marks G, Reed JB, Herbst JH. Circumcision status and risk of HIV and sexually transmitted infections among men who have sex with men. J Am Med Assoc 2008;300:1674–84. [20] Yoshida K, Yatabe Y, Park J, Ogawa S, Park JY, Shimizu J, et al. Clinical outcomes of advanced non-small cell lung cancer patients screened for epidermal growth factor receptor gene mutations. J Cancer Res Clin Oncol 2010;136:527–35. [21] Hotta K, Kiura K, Toyooka S, Takigawa N, Soh J, Fujiwara Y, et al. Clinical significance of epidermal growth factor receptor gene mutations on treatment outcome after first-line cytotoxic chemotherapy in Japanese patients with non-small cell lung cancer. J Thorac Oncol 2007;2:632–7. [22] Okamoto I, Aoe K, Kato T, Hosomi Y, Yokoyama A, Imamura F, et al. Pemetrexed and carboplatin followed by pemetrexed maintenance therapy in chemo-naive patients with advanced nonsquamous non-small-cell lung cancer. Invest New Drugs 2013;31:1275–82. [23] Kalikaki A, Koutsopoulos A, Hatzidaki D, Trypaki M, Kontopodis E, Stathopoulos E, et al. Clinical outcome of patients with non-small cell lung cancer receiving front-line chemotherapy according to EGFR and K-RAS mutation status. Lung Cancer 2010;69:110–5. [24] Lee JK, Park HS, Kim DW, Kulig K, Kim TM, Lee SH, et al. Comparative analyses of overall survival in patients with anaplastic lymphoma kinase-positive and matched wild-type advanced nonsmall cell lung cancer. Cancer 2012;118:3579–86. [25] Li F, Bai H, Li X, Wu M, Yu R, Shi A, et al. Role of EGFR mutation status in patients with stage III non-squamous non-small cell lung cancer treated with chemoradiotherapy. Zhongguo Fei Ai Za Zhi 2011;14:715–8.
345
[26] Bai H, Wang Z, Chen K, Zhao J, Lee JJ, Wang S, et al. Influence of chemotherapy on EGFR mutation status among patients with non-small-cell lung cancer. J Clin Oncol 2012;30:3077–83. [27] Bakker E. Is the DNA sequence the gold standard in genetic testing? Quality of molecular genetic tests assessed. Clin Chem 2006;52:557–8. [28] Qin L, Zhong W, Zhang L, Li LY, Wang MZ. Comparison of three methods for detecting epidermal growth factor receptor mutations in plasma DNA samples of Chinese patients with advanced non-small cell lung cancer. Chin Med J (Engl) 2011;124:887–91. [29] Cappuzzo F, Ligorio C, Toschi L, Rossi E, Trisolini R, Paioli D, et al. EGFR and HER2 gene copy number and response to first-line chemotherapy in patients with advanced non-small cell lung cancer (NSCLC). J Thorac Oncol 2007;2: 423–9. [30] Park JH, Lee SH, Keam B, Kim TM, Kim DW, Yang SC, et al. EGFR mutations as a predictive marker of cytotoxic chemotherapy. Lung Cancer 2012;77: 433–7. [31] Dong X, Zhao X, Hao Y, Wei Y, Yin Q, Du J. Response to first-line chemotherapy in patients with non-small-cell lung cancer according to epidermal growth factor receptor and K-RAS mutation status. Clin Lung Cancer 2013;14:680–7. [32] Campos-Parra AD, Zuloaga C, Manriquez ME, Aviles A, Borbolla-Escoboza J, Cardona A, et al. KRAS mutation as the biomarker of response to chemotherapy and EGFR-TKIs in patients with advanced non-small cell lung cancer: clues for its potential use in second-line therapy decision making. Am J Clin Oncol 2013 [Epub ahead of print]. [33] Zhuo ML, Wu MN, Zhao J, Song SW, Bai H, Wang SH, 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;124:3510–4. [34] Lin CC, Hsu HH, Sun CT, Shih JY, Lin ZZ, Yu CJ, et al. Chemotherapy response in East Asian non-small cell lung cancer patients harboring wild-type or activating mutation of epidermal growth factor receptors. J Thorac Oncol 2010;5:1424–9. [35] Lee KH, Han SW, Hwang PG, Oh DY, Kim DW, Chung DH, et al. Epidermal growth factor receptor mutations and response to chemotherapy in patients with non-small-cell lung cancer. Jpn J Clin Oncol 2006;36:344–50. [36] Eberhard DA, Johnson BE, Amler LC, Goddard AD, Heldens SL, Herbst RS, et al. Mutations in the epidermal growth factor receptor and in KRAS are predictive and prognostic indicators in patients with non-small-cell lung cancer treated with chemotherapy alone and in combination with erlotinib. J Clin Oncol 2005;23:5900–9. [37] Bell DW, Lynch TJ, Haserlat SM, Harris PL, Okimoto RA, Brannigan BW, et al. Epidermal growth factor receptor mutations and gene amplification in non-small-cell lung cancer: molecular analysis of the IDEAL/INTACT gefitinib trials. J Clin Oncol 2005;23:8081–92. [38] Camidge DR, Kono SA, Lu X, Okuyama S, Baron AE, Oton AB, et al. Anaplastic lymphoma kinase gene rearrangements in non-small cell lung cancer are associated with prolonged progression-free survival on pemetrexed. J Thorac Oncol 2011;6:774–80. [39] Douillard JY, Shepherd FA, Hirsh V, Mok T, Socinski MA, Gervais R, et al. Molecular predictors of outcome with gefitinib and docetaxel in previously treated non-small-cell lung cancer: data from the randomized phase III INTEREST trial. J Clin Oncol 2010;28:744–52. [40] Wu M, Zhao J, Song SW, Zhuo M, Wang X, Bai H, et al. EGFR mutations are associated with prognosis but not with the response to front-line chemotherapy in the Chinese patients with advanced non-small cell lung cancer. Lung Cancer 2010;67:343–7.