Mutations of EGFR or KRAS and expression of chemotherapy-related genes based on small biopsy samples in stage IIIB and IV inoperable non-small cell lung cancer

Abstract

Purposes

Epidermal growth factor receptor (EGFR) and KRAS mutations may predict the outcome of targeted drug therapy and also may be associated with the efficacy of chemotherapy in patients with non-small cell lung cancer (NSCLC). This report investigated the relation of EGFR or KRAS mutation and expression of chemotherapy-related genes, including excision repair cross-complementing 1 (ERCC1), thymidylate synthetase (TYMS), ribonucleotide reductase subunit M1 (RRM1) and class III β-tubulin (TUBB3), as a potential explanation for these observations.

Methods

A total of 143 patients with stage IIIB and IV NSCLC from bronchoscopy or percutaneous lung biopsy obtained tumor samples were analyzed concurrently for EGFR or KRAS mutations, and mRNA expression of ERCC1, TYMS, RRM1 and TUBB3. EGFR or KRAS mutations were detected with xTAG liquidchip technology (xTAG-LCT), and mRNA expression levels of four genes were detected by branched DNA-liquidchip technology (bDNA-LCT).

Results

Of 143 patients, 63 tumors were positive for EGFR-activating mutations, and 16 tumors were positive for KRAS mutations. EGFR-activating mutations are more frequent in females, adenocarcinoma and non-smokers patients, and KRAS mutations are more frequent in smoking patients. ERCC1 mRNA levels were significantly associated with histological type and tumor differentiation, whereas TYMS levels were significantly associated with age. NSCLC specimens that harboring EGFR-activating mutations are more likely to express low ERCC1 and high TUBB3 mRNA levels, whereas tumors from patients with NSCLC harboring KRAS mutation are more likely to express high ERCC1 mRNA levels.

Conclusions

Mutations and expression of chemotherapy-related genes may provide a basis for the selection of suitable molecular markers for individual treatment in a population with stage IIIB and IV NSCLC.

Electronic supplementary material

The online version of this article (doi:10.1007/s00432-014-1751-y) contains supplementary material, which is available to authorized users.

Introduction

At diagnosis, about 66 % of patients with NSCLC are at either locally advanced or advanced stage (phase IIIB or IV) (Bayman et al. 2014), and diagnosis is often based on a small biopsy (Ferretti et al. 2013). A satisfactory biopsy that allows for histological characterization and mutation or gene expression profiles analysis is becoming increasingly important for prediction of treatment response and individualized treatment in NSCLC (Chen et al. 2013a, b; Kerr 2012; Meert et al. 2004). So far, the outcome of stage IIIB and IV NSCLC remains very poor (median survival, 8–12 months) (Yamashita et al. 2013; Jemal et al. 2010; Stewart 2010). Of the therapeutic options primarily, radio-chemotherapy for this population has only slightly effective.

The application of selectively targeted therapy has altered the therapeutic model and shown promising clinical activity. EGFR tyrosine kinase inhibitors (EGFR–TKIs) were developed to block the EGFR pathway (Chen et al. 2013a, b; Kato et al. 2010) and have been used clinically for treatment of cancer. Studies using TKIs have revealed that greater benefit from treatment with those agents is seen in patients with NSCLC harboring EGFR-activating mutations than in patients with wild-type tumors (Verduyn et al. 2012a, b). However, not all patients benefit from EGFR–TKIs therapies. Some mutations on EGFR exon 18, 19 or 21 lead to alterations in downstream signaling pathways and make tumor cells more susceptible to TKI-mediated apoptosis, whereas other mutations on EGFR exon 20 contribute to resistance to therapy with TKI (Vazquez-Martin et al. 2013). Furthermore, KRAS mutations have been found in about 30 % of NSCLC cases and have been considered as predictive factors of poor prognostic biomarker (Lewandowska et al. 2013). Therefore, detecting somatic mutations in those biomarkers may provide direct and valuable information to guide treatment making. Gene expression has been successfully used for classification and prognosis of NSCLC, although recent review of the published data suggest that this test is not yet ready for clinical application (Subramanian and Simon 2010). Recent studies have firmly established that excision repair cross-complementing 1 (ERCC1), thymidylate synthase (TYMS), ribonucleotide reductase subunit M1 (RRM1) and class III β-tubulin (TUBB3) genes may be useful molecular markers to guide drug selection in patients with NSCLC (Zhang et al. 2012; Janku et al. 2011; Andrews et al. 2011a, b).

In this study, we applied liquidchip technology (LCT) to detect EGFR or KRAS mutation and mRNA expression of ERCC1, TYMS, RRM1 and TUBB3 in patients with stage IIIB and IV NSCLC from bronchoscopy and percutaneous lung biopsy obtained samples. In addition, the relationship between genetic mutation and expression above and clinicopathological characteristics are also investigated to guide the individualized treatment of NSCLC.

Materials and methods

Study population

A total of 143 patients with stage IIIB and IV NSCLC between March 2010 and December 2012 were recruited for this study. Samples were consecutively collected from the tissue specimen database of Second Affiliated Hospital of Harbin Medical University, and all participants had signed informed consent. Use of samples was approved by the Ethical Review Committee of Second Affiliated Hospital of Harbin Medical University. Only the patients with bronchoscopic or percutaneous needle lung biopsy accessible tumors, deemed inoperable at the initial diagnosis, were enrolled in this retrospective study. Preliminarily, archive slides of the collected formalin-fixed, paraffin-embedded (FFPE) tumor samples were reviewed by two pathologists for confirmation of tumor histology and tumor content. From each paraffin block of representative tumor areas, serial sections with a thickness of 10 μm were prepared and then stained with nuclear Fast Red (Sigma-Aldrich, St. Louis, MO). The specimen contained a minimum of 50 % tumor cells, and patients with insufficient or poor-quality tissue for molecular analyses were excluded from this study. The TNM classification was determined according to version 7 the International Association for the Study of Lung Cancer staging system (Mirsadraee et al. 2012). Patients were considered never-smokers in this study if they reported smoking <100 cigarettes in their lifetime. The patient had not received any anticancer drugs or radiotherapy. Detailed clinicopathological characteristics are listed in Table 1.

Table 1.

Clinicopathological characteristics of patients with NSCLC (N = 143)

Characteristics	No. of patients (%)
Specimen source
Bronchoscopic biopsy	62 (43.4)
CT-guided percutaneous needle lung biopsy	81 (56.6)
Age (years)
Median (range) (years)	56 (32–78)
≥60	61 (42.7)
<60	82 (57.3)
Gender
Male	65 (45.5)
Female	78 (54.5)
Stage
IIIB	80 (55.9)
IV	63 (44.1)
Histopathology
AD	77 (53.8)
SC	55 (38.5)
NSCLC-NOS	11 (7.7)
Tumor differentiation
Well-moderately differentiated	86 (60.1)
Poorly differentiated	57 (39.9)
Smoking status
Yes	82 (57.3)
No	61 (42.7)

AD adenocarcinoma, SC squamous cell carcinoma, NOS not otherwise specified

xTAG liquidchip technology (xTAG-LCT) for gene mutation

DNA from microdissected tumor samples was extracted with the Maxwell^® system (Promega, GA, USA). Two hundred nanogram of each DNA sample was used for multiplex PCR. The xTAG technology by Luminex (Luminex Corp., TX, USA) contain the following basic steps: (1) amplify the regions of target genes by multiplex PCR (H₂O 13.8 μl, 5× PCR buffer 10 μl, 25 mM MgCl₂ 4 μl, dNTPs 2 μl, primers 10 μl, DNA polymerase 0.2 μl, sample 10 μl, 94 °C for 30 s, 52 °C for 30 s and 72 °C for 30 s, 30 cycles); (2) digest by exonuclease and hydrolysis by alkaline phosphatase. The PCR mixture was treated using exonuclease I and shrimp alkaline phosphatase (EXO–SAP) to remove excess nucleotides and primers; (3) allele-specific primer extension (ASPE) (H2O 7.8 μl, 10 × PCR buffer 2 μl, 25 mM MgCl 0.5 μl, dNTPs (no dCTP) 1 μl, Biotin-dCTP 0.5 μl, primers 5 μl, DNA polymerase 0.2 μl, EXO–SAP-treated PCR product 3 μl, 94 °C for 30 s, 56 °C for 1 min and 74 °C for 2 min, 30 cycles). The EXO–SAP-cleaned PCR product was subjected to an ASPE step; (4) hybridization to beads (beads 12 μl, hybridization solution 33 μl, ASPE product 5 μl, 95 °C for 3 min, followed by 30 min incubation at 37 °C). The ASPE products were hybridized to specific anti-tag probes that were pre-coated on the polystyrene microspheres; and (5) the beads were then applied to the Luminex 200 and median fluorescence intensity (MFI) read.

Primers were designed against the sequences of the target in these genes (Supplementary Tables 1 and 2). The primers were synthesized by Invitrogen (Shanghai Invitrogen Biotechnology Co., Ltd. China).

Detection of mRNA expression levels by branched DNA-liquidchip technology (bDNA-LCT)

The FFPE sample was homogenized, and the homogenate was centrifuged. The supernatant was transferred to a fresh microcentrifuge tube. Forty microliter sample homogenate were added to each well of a 96-well plate that contains following reagents in each well: 18.5 μl of RNase-free water, 33.3 μl of lysis solutions, 2 μl of blocking reagent, 1 μl of capture beads and 5 μl of target gene-specific probe set. The plate was sealed and incubated for 18 h at 54 °C on a shaker with 750 rpm. Signals for the bound target mRNA were developed, and the fluorescence value of each sample was analyzed by the Luminex 200 system.

mRNA expression criteria: The mean is the distribution of the patient’s gene expression among the whole population, indicating the level of the mRNA expression in each patient. Using high, middle and low to describe the testing result, the expression greater than or equal to 75 % is high, greater than or equal to 25 % and less than 75 % is moderate, and expression greater than or equal to 0 % and less than 25 % is low, the data base comes from thousands of patients with lung cancer in China.

Statistical analysis

The association between mutation status and clinicopathological factors was evaluated using the χ ² test. To test significant associations between the continuous variable (i.e., gene expression) and dichotomous variables (i.e., patient’s age, sex, tumor stage, mutation status, etc.), the Mann–Whitney U test was used. Values were considered statistically significant at a level of P < 0.05.

Results

EGFR and KRAS mutation analysis

Sixty-three of 143 tumors (44.1 %) harbored EGFR kinase domain mutations. Among these, 23 (16.1 %) mutations were deletions in exon 19, 30 (21.0 %) were leucine to arginine codon 858 (L858R) missense changes, 4 (2.8 %) were threonine to methionine codon 790 (T790M) mutations, and 6 (4.2 %) were found with G719A mutations in exons 18. There are no tumors contained two different mutations in EGFR. Analysis of correlation between EGFR mutations and clinicopathological factors revealed that mutations in exons 19 and 21 of EGFR were correlated with gender, histopathology and smoking status of patients, 44 out of 78 female patients (56.4 %) were found with mutations in exons 19 or 21 of EGFR, which was much higher than that found in male patients, 13.8 % (9/65), P < 0.0001 (Table 2). The frequencies of EGFR mutation were higher in adenocarcinomas (40/77, 51.9 %), compared with those in NSCLC-not otherwise specified (NSCLC-NOS) (2/11, 18.2 %) and squamous cell carcinomas (11/55, 20.0 %). Considering the squamous cell carcinoma and NSCLC-NOS as one group, its mutation frequency was significant lower than that of the adenocarcinoma group (P < 0.0001) (Table 2). Moreover, a total of 29 out of 61 non-smoking patients (47.5 %) were found with EGFR active mutations, which was much higher than in smokers, 29.3 % (24/82), P = 0.025 (Table 2). EGFR mutation status had no significant association with other clinicopathological characteristics (see Table 2).

Table 2.

Correlation between EGFR active mutation (exons 19 and 21) or KRAS mutation and clinicopathological factors

Characteristics	N	EGFR		KRAS
		Mutation (%)	P value	Mutation (%)	P value
Age (years)
≥60	61	21 (34.4)	0.573	6 (9.8)	0.658
<60	82	32 (39.0)		10 (12.2)
Gender
Male	65	9 (13.8)	0.000	7 (10.8)	0.884
Female	78	44 (56.4)		9 (11.5)
Stage
IIIB	80	31 (38.8)	0.638	10 (12.5)	0.575
IV	63	22 (34.9)		6 (9.5)
Histopathology
AD	77	40 (51.9)	0.000	11 (14.3)	0.204
SC + NSCLC-NOS	66	13 (19.7)		5 (7.6)
Tumor differentiation
Well-moderately differentiated	86	34 (39.5)	0.452	10 (11.6)	0.838
Poorly differentiated	57	19 (33.3)		6 (10.5)
Smoking status
Yes	82	24 (29.3)	0.025	14 (17.1)	0.013
No	61	29 (47.5)		2 (3.3)

EGFR epidermal growth factor receptor, AD adenocarcinoma, SC squamous cell carcinoma, NOS not otherwise specified

KRAS mutations were observed in 12 out of 143 (8.4 %) and 4 out of 143 samples (2.8 %) in exons 2 and 3 of KRAS, respectively. Analysis of the correlation between KRAS mutations and clinicopathological factors revealed that KRAS mutations were correlated with smoking status of patients, a total of 14 out of 82 smoking patients (17.1 %) were found with KRAS mutations, which was much higher than in non-smokers, 3.3 % (2/61), P = 0.0126 (Table 2). KRAS mutations had no statistical correlation with other clinical characteristics (see Table 2). EGFR mutations and KRAS mutations were mutually exclusive among the 143 patients.

ERCC1, TYMS, RRM1 and TUBB3 mRNA expression levels

The mRNA expression levels were carried out in bDNA-LCT. ERCC1 expression levels ranged from 0.240 to 2.467 (median 0.781), TYMS from 0.011 to 0.763 [median 0.102], RRM1 from 0.059 to 0.812 (median 0.217) and TUBB3 from 0.026 to 2.471 (median 0.287). Analysis of the correlation between gene expression and clinicopathological features revealed that ERCC1 positive expression levels were significantly associated with histological type (P = 0.003) and tumor differentiation (P = 0.013). The expression of ERCC1 in poorly differentiated patients with adenocarcinoma was much higher than that found in well-moderately differentiated patients with non-adenocarcinoma (Table 3). In addition, the expression of TYMS in patients younger than 60-year old was significantly greater than that in patients at or older than 60-year old, P = 0.001 (Table 3). No association was observed between the expression of RRM1 and TUBB3 and clinicopathological features (Table 3).

Table 3.

Correlation between ERCC1, TYMS, RRM1 and TUBB3 mRNA expression and clinicopathological factors

Variable	ERCC1		TYMS		RRM1		TUBB3
	Median	Range	Median	Range	Median	Range	Median	Range
Age (years)
≥60	0.435	0.240–1.149	0.080	0.011–0.374	0.179	0.059–0.360	0.125	0.026–1.015
<60	0.835	0.335–2.467	0.115	0.032–0.763	0.217	0.111–0.812	0.348	0.029–2.471
P	0.145		0.001		0.068		0.082
Gender
Male	0.794	0.284–2.467	0.124	0.020–0.661	0.237	0.059–0.643	0.263	0.030–2.471
Female	0.781	0.240–2.465	0.094	0.011–0.763	0.194	0.080–0.812	0.342	0.026–1.236
P	0.607		0.375		0.212		0.089
Stage
IIIB	0.712	0.240–2.467	0.088	0.011–0.763	0.197	0.059–0.812	0.265	0.026–1.918
IV	0.740	0.335–2.434	0.082	0.033–0.654	0.211	0.065–0.789	0.301	0.123–2.471
P	0.445		0.678		0.098		0.178
Histopathology
AD	0.931	0.240–2.467	0.081	0.011–0.763	0.192	0.059–0.513	0.237	0.026–2.471
SC + NSCLC-NOS	0.360	0.284–0.959	0.096	0.071–0.374	0.217	0.191–0.812	0.151	0.110–0.454
P	0.003		0.116		0.101		0.447
Tumor differentiation
Well-moderately differentiated	0.536	0.240–0.863	0.106	0.011–0.376	0.277	0.194–0.725	0.172	0.121–1.457
Poorly differentiated	0.837	0.540–2.467	0.091	0.032–0.763	0.202	0.059–0.812	0.246	0.026–2.471
P	0.013		0.733		0.456		0.623
Smoking status
Yes	0.835	0.284–2.467	0.102	0.011–0.661	0.196	0.059–0.643	0.275	0.026–1.051
No	0.781	0.240–2.465	0.144	0.041–0.763	0.243	0.082–0.812	0.339	0.057–2.471
P	0.927		0.268		0.591		0.544

EGFR epidermal growth factor receptor, AD adenocarcinoma, SC squamous cell carcinoma, NOS not otherwise specified

The relationship of EGFR or KRAS mutation status and mRNA expression of ERCC1, TYMS, RRM1 and TUBB3

Using EGFR or KRAS mutation status to divide the patients into groups, we identified four groups: an EGFR active mutation group, an EGFR mutation-negative group, a KRAS mutation group and a group that was negative for both biomarkers. By adopting cut-off values according to median expression levels, the Mann–Whitney U test was used for correlation between gene expression and mutation status. The ERCC1 mRNA levels in patients with KRAS mutations were significantly higher than the levels in patients who EGFR mutation-positive (P = 0.001), EGFR mutation-negative (P = 0.031) and negative for both biomarkers (P = 0.037) (Table 4), whereas the ERCC1 mRNA levels in patients with EGFR mutation-positive were significantly lower than the levels in patients who EGFR mutation-negative (P = 0.012) (Table 4). Moreover, TUBB3 expression levels were significantly higher in patients who were positive for EGFR mutations than in the patients who were negative for both biomarkers (P = 0.042, see Table 4). However, by pairwise comparisons, the expression of TYMS and RRM1 was not correlated significantly with EGFR or KRAS mutation status (P > 0.05) (Table 4).

Table 4.

mRNA expression levels of the ERCC1, TYMS, RRM1 and TUBB3 genes according to different EGFR or KRAS mutation status

Genes	EGFR mutation-positive		EGFR mutation-negative		KRAS Mutation-positive		Both negative
	Median	Range	Median	Range	Median	Range	Median	Range
ERCC1	0.617**	0.240–1.278	0.849*,#	0.284–2.467	1.117	0.741–1.492	0.889*	0.284–2.263
TYMS	0.088	0.032–0.272	0.107	0.012–0.763	0.111	0.011–0.642	0.107	0.040–0.661
RRM1	0.198	0.082–0.360	0.221	0.059–0.812	0.222	0.169–0.275	0.197	0.102–0.643
TUBB3	0.433	0.029–2.471	0.330	0.026–2.471	0.404	0.227–0.581	0.231∆	0.044–1.051

Compared with KRAS mutation-positive group, ** P < 0.01; * P < 0.05; compared with EGFR mutation-positive group, ^# P < 0.05; compared with EGFR mutation-positive group, ^∆ P < 0.05

EGFR epidermal growth factor receptor, ERCC1 excision repair cross-complementing group 1 (ERCC1), TYMS thymidylate synthase, RRM1 ribonucleotide reductase subunit M1, TUBB3 class III β-tubulin

Discussion

In this study, we analyzed mutation status of EGFR and KRAS and mRNA expression of ERCC1, TYMS, RRM1 and TUBB3 in 143 patients with stage IIIB and IV NSCLC. Specimens were taken by bronchoscopic biopsy or percutaneous lung biopsy. Our objective was to determine whether mutations of EGFR or KRAS were closely related to the expression of genes above in small biopsy samples. We observed that NSCLC specimens with EGFR-activating mutations were more likely to express lower ERCC1 mRNA and higher TUBB3 mRNA, whereas specimens that were positive for KRAS mutations were more likely to express high levels of ERCC1, compared with other subgroups.

Nowadays, an activating EGFR mutation is one of the most important factors to be considered while making treatment programs. Randomized clinical trials have revealed that NSCLC patients harboring EGFR-activating mutations show a dramatic response rate to EGFR–TKIs and a longer PFS than those without EGFR mutations (Verduyn et al. 2012a, b; Rafael et al. 2012; Vadakara and Borghaei 2012). Therefore, the mutation status of EGFR should be detected. In our study, the biopsy samples of 23 patients (16.1 %) were found to have deletions in exon 19 and 30 patients (21.0 %) were found to have L858R point mutations in exon 21. The frequency of EGFR-activating mutations of our samples is comparable with those of numerous studies. In addition, previous studies have found that activating EGFR mutations are more frequent in females, non-smokers and adenocarcinoma patients (Chen et al. 2013a, b; Maemondo et al. 2010; Ren et al. 2012a, b). Our data in this study also support this observation. Thus, for NSCLC patients with activating EGFR mutations, the use of EGFR–TKI is crucial to success in treatment. In some previous clinical studies, however, EGFR–TKIs have no significant anti-tumor activity in NSCLC patients with KRAS mutations (Vadakara and Borghaei 2012; Zhou et al. 2013). Other studies showed the presence of KRAS mutation in NSCLC patients has indicated worse outcome in spite of the treatment they received (Mao et al. 2010; de Mello et al. 2013). However, till now, the role of KRAS mutation in predicting survival with EGFR–TKIs treatment remains unclear. Although most studies show that a KRAS mutation predicts a poor response to EGFR–TKIs, these data do not strongly support an association between KRAS mutation and survival outcome because the reported treatment outcomes are not survival outcome such as OS or PFS, but response rate. The main finding of our study was a lower incidence of the KRAS mutation in our Chinese study group compared to that reported for Western populations and no patient harbors a double mutation (KRAS and EGFR). Besides, regardless of histology, our study confirms the known fact that the rate of KRAS mutation was significantly more common in smokers than in non-smokers. Data from a recent meta-analysis show a significantly higher KRAS mutation frequency among current smokers or former smokers compared to that of never-smokers (26 vs. 6 %, P < 0.01) (Mao et al. 2010). However, in contrast, Kim et al. (2013) reported KRAS mutations were found in 5.8 % (7 of 120) of tumors from never-smokers, 15 % (6 of 40) from former smokers, and 7.5 % (3 of 40) from current smokers, and the frequency of KRAS mutations did not differ significantly according to smoking history (P = 0.435). So the association of KRAS mutation with smoking history awaits further study of large populations with detailed smoking histories.

Expression of chemotherapy-related genes has been successfully used for predicting therapeutic response and assessing risk of relapse of NSCLC, although recent study suggests that this approach is not yet ready for clinical application (Andrews et al. 2011a, b). Among these, ERCC1 plays an important role in the repair of lesions in DNA induced by platinum compounds. Patients with ERCC1-negative NSCLC appear to benefit from adjuvant cisplatin-based chemotherapy, unlike patients with ERCC1-positive tumors (Maus et al. 2013). As for others, TYMS is associated with the efficacy of pemetrexed, which is a multi-targeted antifolate drug that inhibits the enzyme thymidylate synthase (TYMS). Some studies indicated that high expression of TYMS is considered a resistance mechanism in NSCLC and may be a predictive biomarker of pemetrexed sensitivity (Kasai et al. 2013). RRM1 is a predictive marker of gemcitabine efficacy and also has role in DNA repair systems like ERCC1. Published data already suggested that patients with low RRM1 level benefited significantly from cisplatin/gemcitabine in NSCLC (Ren et al. 2012a, b). Expression of TUBB3 is associated with resistance of paclitaxel-based chemotherapy and poor prognosis (Leng et al. 2012). Levallet et al. (2012) also reported that TUBB3-negative patients derived more than 1 year of overall survival advantage compared with TUBB3-positive patients (84 vs. 71.7 months) in NSCLC patients who received paclitaxel-based chemotherapy. In this report, we have characterized the expression levels of ERCC1, TYMS, RRM1 and TUBB3 in tumor tissue of selected population. The results revealed that the ERCC1 expression levels were significantly correlated with histological type and tumor differentiation, whereas TYMS expression levels were significantly associated with age. Such estimate preliminarily indicates a relatively small but significant association between expression of chemotherapy-related genes and pathological factors of the tumor. Some technical and practical obstacles, including the absence of standardized interlaboratory quality control procedures and the use of snap frozen versus paraffin-embedded, etc., make the personalized medicine extremely difficult to move from research findings to clinical practice (Ceppi et al. 2006; Friboulet et al. 2013).

It is noteworthy that specific gene mutation not only was useful for predicting therapeutic response of the targeted drugs but also appeared to be associated with outcome of chemotherapy. An interesting aspect is that in the Iressa Pan-Asia Study (IPASS) (Mok et al. 2009), NSCLC patients with EGFR-activating mutations showed a higher overall response rate than patients with EGFR wild type (47.3 vs. 23.5 %) when they received carboplatin/paclitaxel chemotherapy. These results questioned that whether NSCLC patients with EGFR mutations are more responsive to cytotoxic agents like platinum-based compounds. An association between EGFR-activating mutations and ERCC1 expression levels would provide a partial explanation and a mechanism-based rationale for such observations. Lee et al. (2008) investigated that EGFR mutations were associated significantly with lower ERCC1 expression and EGFR mutations were more frequently found in ERCC1-negative tumors. Furthermore, Yamashita et al. (2013) reported that ERCC1-negative patients with exon 19 deletion had a longer PFS than ERCC1-positive patients without exon 19 deletion when they received platinum-based chemotherapy. In agreement with these results, we observed that ERCC1 levels were significantly lower in the EGFR-mutated group than that in the EGFR mutation-negative group, suggesting that NSCLC patients harboring EGFR-activating mutations may be more suitable to platinum-based chemotherapy. However, whether chemotherapy without platinum can be considered as a reasonable regimen for patients with positive ERCC1 remain to be further elucidated. Besides, in our study, we also found that the expression of ERCC1 mRNA was significantly higher in patients who were KRAS-positive, suggesting the interaction between ERCC1 and KRAS gene. However, it remains to be further studied how ERCC1 could affect mutations of the KRAS gene in lung cancer.

In addition to ERCC1, TYMS and RRM1 genes also are critical components in the DNA synthesis and DNA damage repair pathways. Regarding these genes, relatively little information exists on whether EGFR-mutant NSCLC cell lines are more sensitive to DNA-damaging chemotherapy or ionizing radiation in preclinical models (Das et al. 2006; Chen and Nirodi 2007). According to results reported by Ren et al. (2012a, b), NSCLC patients with EGFR mutations were more likely to express low TYMS mRNA levels, which, to some extent, may explain the finding that EGFR mutation-positive patients who received pemetrexed had a better response rate and longer PFS than the patients with wild-type EGFR (Wu et al. 2011). However, in our current study, no association was observed between EGFR or KRAS mutation status and the expression of TYMS and RRM1 genes. Taking into account the fact that how TYMS and RRM1 could interact with EGFR or KRAS genes in lung cancer, it is still a long way from understanding. Although the correlation between EGFR or KRAS mutations and DNA repair genes has been unclear, it can be postulated that an impaired capacity for DNA repair and synthesis may be correlated with increased genome instability and tumor mutations (Kosaka et al. 2004; Simon et al. 2005). Moreover, our results showed that TUBB3 was more frequently observed in patients with EGFR-activating mutations than in both EGFR and KRAS mutation-negative patients, indicating that NSCLC patients harboring EGFR-activating mutations may be more resistant to taxanes. However, these results are inconsistent with previous studies. Levallet et al. (2012) correlated TUBB3 expression with KRAS and EGFR mutations in 208 cryopreserved NSCLC specimens and indicated that EGFR mutations had no significant effect on TUBB3 expression and high TUBB3 expression was associated with KRAS mutation (P < 0.001). In addition, taking account of these clinical findings, researchers downregulated KRAS expression by short interfering RNA (siRNA) and discovered reducing TUBB3 expression in human bronchial cell lines. A limited number of samples in our study may partly explain these discordant results and a further study of large populations is needed.

In summary, we conclude that the analysis of gene mutation and expression profiles based on bronchoscopic biopsy or percutaneous lung biopsy obtained samples appeared feasible. NSCLC specimens that harboring EGFR-activating mutations are more likely to express low ERCC1 and high TUBB3 mRNA levels, whereas tumors from patients with NSCLC harboring KRAS mutation are more likely to express high ERCC1 mRNA levels. Our findings support the hypothesis that EGFR or KRAS mutation status affects the efficacy of chemotherapy through the pathways of chemotherapy-related genes. The combined detecting mutations of EGFR and KRAS and mRNA expression of chemotherapy-related genes might better individualize the efficacy of treatment in NSCLC.

Electronic supplementary material

Below is the link to the electronic supplementary material.