Comparison of EGFR and K-RAS gene status between primary tumours and corresponding metastases in NSCLC

In non-small-cell lung cancer (NSCLC), epidermal growth factor receptor (EGFR) and K-RAS mutations of the primary tumour are associated with responsiveness and resistance to tyrosine kinase inhibitors (TKIs), respectively. However, the EGFR and K-RAS mutation status in metastases is not well studied. We compared the mutation status of these genes between the primary tumours and the corresponding metastases of 25 patients. Epidermal growth factor receptor and K-RAS mutation status was different between primary tumours and corresponding metastases in 7 (28%) and 6 (24%) of the 25 patients, respectively. Among the 25 primary tumours, three ‘hotspot’ and two non-classical EGFR mutations were found; none of the corresponding metastases had the same mutation pattern. Among the five (20%) K-RAS mutations detected in the primary tumours, two were maintained in the corresponding metastasis. Epidermal growth factor receptor and K-RAS mutations were detected in the metastatic tumours of three (12%) and five (20%) patients, respectively. The expressions of EGFR and phosphorylated EGFR showed 10 and 50% discordance, in that order. We conclude that there is substantial discordance in EGFR and K-RAS mutational status between the primary tumours and corresponding metastases in patients with NSCLC and this might have therapeutic implications when treatment with TKIs is considered.

Lung cancer is the most frequent solid tumour and represents the leading cause of cancer death throughout the developed world. Almost 70% of patients with non-small-cell lung carcinoma (NSCLC) present with locally advanced or metastatic disease at the time of diagnosis. Non-small-cell lung carcinoma is characterised by the accumulation of multiple genetic alterations (Marsit et al, 2004;Yokota and Kohno, 2004;Garnis et al, 2006). Mutations within the tyrosine kinase domain of epidermal growth factor receptor (EGFR) account for increased sensitivity to tyrosine kinase inhibitors (TKIs; gefitinib and erlotinib) and they are associated with prolonged overall survival (Lynch et al, 2004;Paez et al, 2004;Perez-Soler et al, 2004;Chou et al, 2005;Taron et al, 2005;Hirsch, 2006). However, the point mutation T790M and an insertion mutation in exon 20 were associated with resistance to TKIs (Gazdar and Minna, 2005). Furthermore, recent studies have shown that the expression of EGFR as assessed by gene copy number, mRNA and protein levels could be used to predict responsiveness to therapy with TKIs Taron et al, 2005;Dacic et al, 2006;Dziadziuszko et al, 2006;Endo et al, 2006). In addition, several clinicopathological characteristics, such as adenocarcinoma histology, non-smoking history, female gender and Asian origin, are also associated with a higher probability of response to TKIs, whereas the presence of K-RAS mutations seems to be correlated with primary resistance to these agents (Tokumo et al, 2005;Tsao et al, 2005;van Zandwijk et al, 2007). Thus, an emerging issue concerning EGFR-targeted therapy is to identify the best method for selecting patients who are more likely to benefit from EGFR inhibition.
Advanced NSCLC metastasises systemically to diverse sites, such as the brain, bone, adrenal glands and liver. The classical model for the metastatic process suggests that most cells of a given primary tumour have low metastatic potential and only a few cells acquire enough somatic mutations to become metastatic (Bernards and Weinberg, 2002). An alternative model proposes that the metastatic potential is encoded in the mass of a given primary tumour that has progressed to a pre-metastatic state, after which metastases may randomly occur without any further gene expression changes (van 't Veer et al, 2002;Hynes, 2003;Van't Veer and Weigelt, 2003). Taking into account these two models, a critical issue for the treatment of metastatic NSCLC is the question of genetic variability and differences between the primary tumour and the corresponding metastases. In the majority of studies, EGFR and K-RAS status was determined on the primary tumours and there are very few data concerning those of corresponding metastases (Italiano et al, 2006;Matsumoto et al, 2006). Therefore, it is unclear whether the same EGFR and K-RAS mutations are also present in the metastatic lesions or whether clones with different mutations are responsible for the generation of metastases.
In this study, the mutation status of EGFR and K-RAS as well as the EGFR and p-EGFR expressions on the primary tumours and the corresponding metastatic lesions were evaluated in 25 patients with advanced NSCLC. The objective of this study was to investigate the prevalence of EGFR and K-RAS mutations in metastases and to examine whether these mutations and the EGFR expression patterns are discordant between the primary tumours and the corresponding metastases. Secondary objectives were to explore whether the EGFR expression pattern correlated to EGFR and/or K-RAS mutations in both the primary tumours and corresponding metastases.

Patients
Patients, aged 418 years old, with histologically confirmed NSCLC who underwent biopsy or surgical excision of the primary tumour and the corresponding metastases were included in this retrospective analysis. Histological type was determined according to the World Health Organization criteria, and the stage of the disease corresponds to that of the time of primary diagnosis. Smoking history was obtained during the patient's first evaluation. All patients gave their informed consent for using their tumour sample for molecular and pathologic analysis. The study has been approved by the Ethics and Scientific Committees of our institution.
The sensitivity of our methodology was evaluated by determining the minimum frequency of EGFR and K-RAS mutations required for detection in our system. This was accomplished by performing mixing experiments using cell lines with and without EGFR (H2073-wt-EGFR and HCC827-Del19-EGFR) or K-RAS (H2073-wt-K-RAS and A549-G12D-K-RAS) mutations. These experiments demonstrated that the Del19 and K-RAS mutations could be detected when present in 10 and 20% of the cells in the sample, respectively (data not shown).

Immunohistochemistry for EGFR
The paraffin-embedded tissues were cut at 4 mm thickness and were deposited on SuperFrost/Plus Slides (O.Kindler GmbH, Freiburg, Germany). After deparaffinisation, the slides for EGFR were treated with proteinase K (Code S3020, DakoCytomation, Glostrup, Denmark) for 5 min at room temperature, and those for p-EGFR were treated with EDTA at pH 8 in a microwave oven three times for 5 min at 500 W for antigen retrieval. The primary anti-EGFR antibody (mouse monoclonal, clone H11, code M3563, Dako-Cytomation) was used at a dilution of 1 : 50 (v/v) and incubated for 1 h at room temperature. The primary anti-phospho-EGFR antibody (rabbit monoclonal, pY1173-EGFR, code no. 4407, Cell Signalling, Danvers, MA, USA) was used at a dilution of 1 : 200 (v/ v) and incubated overnight at 41C. For the detection of antigenantibody reaction, the UltraVision detection system AP Polymer kit (Cat no. TL-125-AL, Lab Vision, Fremont, CA, USA) was used according to the manufacturer's instructions. Fast red was used as chromogen for 20 min; the sections were counterstained with Mayer's haematoxylin for 3 min, subsequently rinsed in ammonium and finally mounted with glycergel. A positive (an NSCLC tumour specimen with known positivity) and a negative (omission of primary antibody) control were used. The stained sections were independently evaluated by two pathologists (AK and ES). Immunoreactions of EGFR were graded as 3 þ when strong complete membrane staining was observed in more than 10% of the tumour cells, as 2 þ when more than 10% of the tumour cells showed weak-to-moderate complete membrane staining, as 1 þ when partially, faint membrane stain was detected in more than 10% of the tumour cells and as 0 when no staining at all or membrane staining in less than 10% of the tumour cells was observed. In this study, we arbitrarily classified EGFR expression status in two subsets; the 2 þ or 3 þ signals were considered as EGFR overexpression and the 0 or 1 þ signals as non-expression .

Statistical analysis
McNemar test was used to compare the EGFR and K-RAS status between primary tumours and related metastatic sites. Differences were considered statistically significant when the P-value was o0.05. All statistical tests were two-sided.

EGFR mutation status of the primary tumours and the corresponding metastasis
The EGFR mutation status of the primary tumours and the corresponding metastases is presented in Table 2. Epidermal growth factor receptor mutations were detected in the primary tumours of five (20%) patients; of these, three of them were the well-characterised 'hotspot' mutations in exon 19 (Del746-750 and E746V; case nos. 20, 23 and 18) and the remaining two were novel point mutations in exons 18 and 21 (L692P and G857E; case nos. 17 and 19), respectively. The corresponding metastases of the Del746-750 (case no. 23), E746V (no. 18), L692P (no. 17) and G857E (no. 19) mutant primary tumours were wild type with respect to EGFR mutation status. Epidermal growth factor receptor mutations were detected in the metastatic tumours of three (12%) patients ( Table 2). The metastasis of one of these patients showed the same EGFR mutation (Del19) as the primary tumour as well as an additional one, the T790M in exon 20 (no. 20); conversely, the other two patients carried two novel mutations in exon 18 (L692P and V717A; no. 12) and the T847A (no. 13) in exon 21, which could not be detected in the patients' primary tumour samples. We have confirmed that the non-classical mutations detected in our series are not single nucleotide polymorphisms by mutation analysis of matched normal tissue or blood (data not shown).

EGFR and p-EGFR expression by IHC in the primary tumour and the corresponding metastases
Én 19 patients sufficient tumour tissue was available for immunohistochemical analysis of EGFR (Tables 2 and 4). The incidence of EGFR overexpression (grade 2 þ , 3 þ ) was 26.5% in both primary and metastatic tumours. Concordance between the primary tumour and the corresponding metastases was observed in 17 (89.5) patients (Cohen's Kappa ¼ 0.729, P ¼ 0.001) and among them EGFR was overexpressed in four (21.1%). Discordance was observed in two (10.5%) patients (nos. 14 and 15) (McNemar test, P ¼ 1). Evaluable paired tissue specimens for p-EGFR analysis by IHC were available in 16 patients. Three out of 16 (18.8%) primary  (Tables 2 and 4). There was no correlation between the expression of EGFR and pY1173-EGFR. In order to confirm our findings, we repeated the immunohistochemistry from serial sections in all tumour specimens that exhibited EGFR-negative and p-EGFR-positive staining and the obtained results were identical. Epidermal growth factor receptor gene copy number was also investigated by fluorescence in situ hybridisation in eight patients (nos. 2, 5, 7, 9, 10, 13, 19 and 24) for whom paired tissue specimens were available; amplification of EGFR gene was not detected in any of these tumour specimens.

K-RAS mutation status of the primary tumours and the corresponding metastases
Primary and metastatic tumours were also assessed for K-RAS mutations (Table 2). K-RAS mutations were detected in the primary tumours of five (20%) patients (nos. 8, 10, 22, 23 and 25) and in the metastatic tumours of five (20%) patients (nos. 9, 10, 11, 16 and 23), respectively. Two patients (nos. 10 and 23) carried the same K-RAS mutations in both primary and metastatic tumours ( Table 2). One of them (case no. 23) carried the Del746-750 EGFR mutation in the primary tumour but not in metastasis. This was confirmed by three independent PCRs from three genomic DNAs extracted from serial sections of the paraffin blocks. Discordance in K-RAS mutation status between the primary tumours and the corresponding metastases was observed in six (24%) patients (McNemar test, P ¼ 1) (Table 3).

Response to gefitinib according to EGFR and K-RAS mutation status
Nine patients received gefitinib as first (nos. 23 and 18), third (nos. 15, 16, 19 20 and 22) or fourth (nos. 14 and 17) line treatment. Three patients received gefitinib before the biopsy of metachronous metastases (nos. 17, 18 and 20) and six received gefitinib after the biopsy of metastases (nos. 14, 15, 16, 19, 22 and 23). Five patients (56%) achieved stable disease and four (44%) progressive disease ( Table 2). All patients who experienced progressive disease on gefitinib were wild type regarding the EGFR mutation status in both primary tumours and metastases (nos. 14, 15, 16, and 22), of these, two patients carried K-RAS mutations in the primary

DISCUSSION
Several studies have shown that activating EGFR mutations in exons 18, 19 and 21 have been associated with a 75 -95% objective response rate with EGFR TKIs, whereas the K-RAS mutations were associated with a lack of sensitivity to these agents (Pao et al, 2005;Sharma et al, 2007). Epidermal growth factor receptor gene amplification and protein expression have also been considered as predictors of clinical benefit with gefitinib . However, in clinical studies, a significant percentage of patients presented clinical benefit when treated with TKIs irrespective of the expression and mutational status of EGFR (Sharma et al, 2007). As, in most studies, EGFR expression and mutations were determined on the primary tumour, the observed clinical benefit of patients with wild-type EGFR or the absence of response to TKIs of patients with EGFR mutations could be due to discordance in the EGFR mutation status or expression between the primary tumour and the corresponding metastasis. This study demonstrated the existence of a significant discordance of EGFR and K-RAS mutations occurring in primary tumours and their corresponding metastases in patients with NSCLC. The discordance in EGFR mutation status was 28% and that in K-RAS was 24%. Similarly, two other studies in paired NSCLC tumours showed a discordance of 32 and 27% regarding the EGFR gene copy number (Italiano et al, 2006;Bozzetti et al, 2008), whereas another study including six NSCLC patients of Asian ethnicity reported a 100% concordance in regard to EGFR mutation status (Matsumoto et al, 2006). This discrepancy could be related to the different sites of the metastatic tumours analysed in this study (five different distant metastases), whereas in the Matsumoto's et al study, only tumour samples from brain metastases were included. Diverse sites of metastases in NSCLC probably represent different clonal outgrowths. Alternatively, we cannot exclude that the different ethnicity of patients and/or the different types of EGFR mutations could be the reason for this discrepancy (Tsao et al, 2006;Pallis et al, 2007). In this study including only Caucasian patients, three classical activating (Del19 and E746V) and four non-classical mutations (T847A, L692P-V717A, L692P and G857E) were detected, whereas in Matsumoto's et al study, only classical activating Del19 and L858R mutations were reported. Concerning the non-classical mutations, it is unlikely to represent PCR artifacts, as L692F, T847I and G857E EGFR mutations have been previously reported (Fujimoto et al, 2005;Tsao et al, 2005;Hsieh et al, 2006). Furthermore, the expression of phosphorylated EGFR (pY1173-EGFR-positive) on tumour cells from metastatic lesions carrying the T847A and L692P-V717A mutations strongly suggests that the former might be activating EGFR mutation. The time elapsed between diagnoses of the primary tumour and corresponding metastasis in patients (nos. 13 and 12) carrying the abovementioned mutations was 74.5 and 10 months, respectively. Therefore, acquisition of new mutations may be developed during the evolution of the metastatic process.
The administration of TKIs could be an additional explanation of the observed discordance of EGFR mutations. Three out of 25 patients received gefitinib before the development of a metachronous metastasis, whereas none of the patients reported in the Matsumoto's et al study had exposed to TKIs. It is known that NSCLC patients with EGFR-dependent primary tumours when treated with TKIs can develop metastases, in which either the EGFR signalling is negated or resistance is acquired due to secondary EGFR mutations like T790M or MET amplification (Daneshmand et al, 2003;Gazdar and Minna, 2005;Engelman et al, 2007;Lutterbach et al, 2007). It is interesting to note that in patients who had not been exposed to TKIs before biopsy of metastatic lesions, a 18% (4 out of 22 patients) discordance was observed between primary tumours and related metastases. However, a final explanation of the observed discordance, which cannot be excluded, concerns the low-frequency intratumoral heterogeneity for the occurrence of EGFR mutations (Sakurada et al, 2008).
Our findings concerning the K-RAS mutation status are in agreement with a previous study demonstrating that the K-RAS mutational status of the primary tumour does not always predict the status of bone metastasis in NSCLC (Badalian et al, 2007). A similar phenomenon was also reported in patients with colorectal cancer (Tortola et al, 2001). Although K-RAS mutations seem to be associated with the early development of NSCLC, it cannot be excluded that K-RAS mutations are lost later during tumour progression (Burmer and Loeb, 1989;Li et al, 1994). This may, in part, explain the discordance in the K-RAS mutation status between primary tumours and metachronous metastases.
Another possibility for the observed discordance in the EGFR and K-RAS mutation status could be related to the administered chemotherapy. However, as shown in Table 1, among the 10 patients who had not received any treatment before the mutation analysis of the metastatic lesions, 5 developed metastases with different mutation status from that of the corresponding primary tumours. Thus, although tumour clone selection through the various treatments could be an explanation for the different molecular pattern in the primary tumour and metastatic site, our findings suggest that the metastasis genotype could be different from that of the corresponding primary tumour irrespectively of administered chemotherapy.
Previous studies have shown that EGFR and K-RAS mutations are mutually exclusive, suggesting the presence of different pathways of lung carcinogenesis. However, as previously reported, our data show that K-RAS mutation may coexist with EGFR mutation (Fujimoto et al, 2005;Han et al, 2006). Among the five patients with EGFR mutations in primary tumours, one patient concomitantly had K-RAS mutation (G12C with deletion in exon 19). In the metachronous metastasis, only the K-RAS mutation was retained. This patient received gefitinib after the biopsy of metastatic lesion and had stable disease lasting for 3.5 months. The limited duration of response is compatible with the knowledge that the presence of K-RAS mutations is associated with resistance to TKIs.
In this study, the expression of pY1173-EGFR was different between primary tumours and corresponding metastases in eight (50%) patients, whereas EGFR expression was discordant in two (10%) patients. The apparent lack of correlation between EGFR and p-EGFR expression has been previously reported and could be due to the different scoring systems and different sensitivities of antibodies used to evaluate the EGFR and p-EGFR expressions (Han et al, 2005). It is well known that overexpression of EGFR is associated with Tyr phosphorylation of the receptor proteins and that mutations in the kinase domain may cause constitutive phosphorylation of EGFR . However, biochemical studies have shown that variable phosphorylation rates were associated with different tyrosine phosphorylation sites in the receptor; G719S mutant receptor had less EGF-induced phosphorylation at Y845, Y992, Y1068 and Y1173 than did wildtype EGFR, whereas L858R mutant receptor preferentially phosphorylated at the Y1068 but not at Y1173 Chen et al, 2006;Liu et al, 2007).
Clinical investigations suggested a correlation between a high EGFR gene copy number and EGFR mutations (Cappuzzo et al, 2005;Takano et al, 2005). Our study failed to demonstrate such a correlation probably due to limited number of cases analysed. Evaluable paired tissue specimens for fluorescence in situ hybridisation analysis were available for eight patients. Among them, two (nos 13 and 19) patients presented different EGFR mutation status in the primary tumours and corresponding metastases; however, these two patients harbour non-classical EGFR mutations (T847A and G857E) and did not show different EGFR amplification patterns between paired samples. On the basis of these findings, at the present time, we consider that only the genetic differences unequivocally distinguish EGFR-dependent tumours, which are likely to be sensitive to TKIs from the tumours that could be resistant to these agents (Papadopoulos et al, 2006).
In conclusion, our findings indicate a substantial discordance of EGFR and K-RAS mutations between the primary tumours and the corresponding metastases in NSCLC and underline the need to consider the genotype of both primary and metastatic tumours for selecting patients who will respond to therapy with TKIs.