SEARCH

SEARCH BY CITATION

Keywords:

  • lung cancer;
  • growth factor receptors;
  • gene amplification;
  • therapeutic marker

Abstract

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. References
  8. Supporting Information

Growth factor receptors (GFRs) are amenable to therapeutic intervention in cancer and it is important to select patients appropriately. One of the mechanisms for activation of GFRs is gene amplification (GA) but discrepancies arising from the difficulties associated with data interpretation and the lack of agreed parameters confound the comparison of results from different laboratories. Here, we attempt to establish appropriate conditions for standardization of the determination of GA in a panel of GFRs. A NSCLC tissue microarray panel containing 302 samples was screened for alterations at ALK, FGFR1, FGFR2, FGFR3, ERBB2, IGF1R, KIT, MET and PDGFRA by FISH, immunostaining and/or real-time quantitative RT-PCR. Strong amplification was found for FGFR1, ERBB2, KIT/PDFGRA and MET, with frequencies ranging from 1 to 6%. Thresholds for overexpression and GA were established. Strong immunostaining was found in most tumors with ERBB2, MET and KIT amplification, although some tumors underwent strong immunostaining in the absence of GA. KIT and PDFGRA were always coamplified, but only one tumor showed PDGFRA overexpression, indicating that KIT is the main target. Amplification of FGFR1 predominated in squamous cell carcinomas, although the association with overexpression was inconclusive. Interestingly, alterations at ALK, MET, EGFR, ERBB2 and KRAS correlated with augmented levels of phospho-S6 protein, suggesting activation of the mTOR pathway, which may prove useful to pre-select tumors for testing. Overall, here, we provide with parameters for the determination of GA at ERBB2, MET, KIT and PDGFRA which could be implemented in the clinic to stratify lung cancer patients for specific treatments.

The continuing endeavors to decipher cancer genomes guide the design of novel anticancer therapies. In the particular case of lung cancer, the most common cause of cancer deaths throughout the world, our better understanding of its genetic architecture is starting to pay off in this respect. To date, most of the novel drugs have as a target tyrosine kinase growth factor receptors (GFRs) that are genetically altered in tumors. Such is the case of patients with lung adenocarcinomas carrying EGFR-mutated proteins or translocations of ALK, RET and ROS proteins, which are candidates for specific treatments.[1-7] Furthermore, studies in cancer cell lines serve as preclinical tools for evaluating how the genotype of cancer cells determines the response to specific drugs. In this regard, in most cases, the growth of lung cancer cells with translocations at ALK and ROS or with strong amplification at MET, EGFR, ERBB2, FGFR1 or PDGFRA genes is significantly decreased when the cells are treated with their respective specific inhibitors.[8-12] Some of these observations have already been confirmed in lung cancer patients. This gives rise to a new scenario in which the analysis of the gene alteration patterns of tumors becomes essential to predict a patient's primary response to a given specific therapy.

In addition, increasing our knowledge of how the genetics of tumors is associated with different clinical, etiological and histopathological characteristics, and the understanding of how the biological pathways are affected by a given genetic pattern is essential for improving patient selection and cancer treatment. Alterations of some of the genes encoding GFRs or their downstream effectors cluster in specific histopathological types, as is the case of ALK, BRAF, KRAS, NRAS and EGFR, found mainly in lung adenocarcinomas (ACs)[9, 13] or amplification at FGFR1, found in lung squamous cell carcinomas (SCCs).[11] Moreover, some alterations like KRAS, NRAS and EGFR, are never simultaneously altered in the same tumor, which suggests a functional relationship. Functionally, the genetic alterations at these GFRs or signal transduction molecules trigger a constitutive activation of the associated molecular pathway with dire consequences for the cell. For example, the mutated EGFR triggers constant downstream signaling of PI3K/AKT/mTOR and RAS/ERK/MAPK, which are involved in cell survival and cell proliferation.[13, 14]

Oncogenic activation in cancer occurs by different mechanisms, for example, gene mutations, reciprocal translocations and gene amplification (GA). While the first two can be easily standardized for routine screening in diagnostic laboratories, determination of GA is a source of controversy and subjectivity because the distinction of true GA from increases in gene copy number is often unclear. A clear distinction between these two entities is essential, since the former is a bona fide indicator of a genetic activation event in an oncogene, while the latter may merely reflect the occurrence of wide and nonspecific chromosome aberrations, which are intrinsic to cancer development.

In response to the increasing clinical relevance of GFRs as therapeutic targets, we attempted an in-depth characterization of alterations at various GFRs in lung cancer. By combining the analysis of gene alterations and protein or mRNA levels, here, we provide with parameters to enable lung tumors with activation though GA at KIT, PDGFRA, MET and ERBB2 to be distinguished with the aim of facilitating appropriate treatment with specific targeted therapies in nonsmall cell lung cancer (NSCLC) patients.

Methods

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. References
  8. Supporting Information

Patients and lung tumors

A total of 302 tumors from NSCLC patients were analyzed. Paraffin-embedded tumors, from all these NSCLC patients, and additional paired fresh frozen from 144 of them, were retrospectively provided by the Hospital 12 de Octubre and the Hospital Virgen de la Arrixaca, Spain, through the CNIO Tumour Bank. Informed consent was obtained from patients and the work was reviewed by the relevant institutional review boards. Information about tumors and patient's is provided in Supporting Information Table S1.

DNA and RNA extraction and tissue microarray (TMA) construction

For DNA and RNA extraction, tumors were meticulously macrodissected to ensure at least 75% of tumor cells. Sections of 10–20 μm were collected from the samples and processed as previously described.[13, 15] For immunostaining and FISH analysis, we used 10 TMAs containing 302 unselected lung tumors, mostly comprising NSCLCs, several normal lungs and other tissue controls.[13, 15] Of the 302 tumors, some had to be discarded because it was not possible to conduct most of the analyses. Each sample was present in duplicate or triplicate cores.

Fluorescence in situ hybridization analysis

To determine gene copy number, we performed FISH analysis for the FGFR1, FGFR2, FGFR3, IGF1R, KIT, MET and PDGFRA genes on the TMAs. Dual-color FISH was performed in 4-µm sections of the TMAs following previously described protocols.[13, 15] Briefly, we used BAC clones, labeled in Spectrum Red, covering the genes of interest and control BACs, labeled in Spectrum Green, located on the same arm of the gene of interest or pericentromerically (see Supporting Information Table S2). BAC clones were obtained from the BACPAC Resource Centre at the Children's Hospital Oakland Research Institute (Oakland, CA). FISH was also carried out in the tumors with HER2 overexpression, as previously described.[13] The ALK translocation was determined in the tumors with ALK overexpression, using the LSI ALK Dual Color, Break Apart Rearrangement Probe (Abbott-Vysis) following the manufacturer's protocol. Fluorescence signals were scored in each sample by counting the number of single-copy genes and control probe signals in an average of 100 (60–150) well-defined nuclei.

Immunohistochemistry and real-time quantitative PCR

Immunohistochemistry was performed on TMAs sections following previously described protocols to determine protein expression for ALK, FGFR1, FGFR3, HER2, IGF1R, KIT, MET, phospho-AKT, phospho-S6 and phospho-mTOR.[13] Sections were counterstained with hematoxylin and eosin. The antibodies used, their source and the scoring criteria employed are summarized in the Supporting Information Table S3.

Due to the lack of reliable antibodies for immunostaining, we measured the mRNA levels of FGFR1, FGFR2, PDGFRA and KDR by real-time quantitative PCR using SYBR green and an ABI Prism 7900 Sequence detector (Applied Biosystems). Two internal controls (human IPO8 and GAPDH transcripts) were used in the analysis.[16] Primer sequences are available are summarized in the Supporting Information Table S4.

Mutation analysis

For the mutational screening, tumor DNA was amplified by PCR and sequenced, following previously described protocols.[15] Those exons containing known mutational hotspots or encoding the tyrosine kinase domain of the protein were analyzed: MET (exons 2 and 15–19), KIT (exons 9, 11–13 and 17), PDGFRA (exons 12, 14 and 18), FGFR1 (exons 9–18), FGFR2 (exons 6–8, 10, 11 and 14), FGFR3 (exons 6, 7, 10, 15 and 16) and IGF1R (exons 16–21). Primer sequences are available upon request.

Statistical analysis

Categorical variables were described using frequency distributions, and continuous variables were summarized as medians and ranges. Associations between categorical variables were assessed with the χ2 test (or Fisher's exact test for pairs of dichotomous variables). For the comparison of the relative levels of gene expression of FGFR1 and PDGFRA with different gene copy number groups we used the Mann-Whitney U test. Both test were performed with the SPSS statistics software. A nonparametric statistical hypothesis test was used because expression levels do not follow a Gaussian distribution. The median level of gene expression for each group was calculated.

Results

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. References
  8. Supporting Information

Screening and characterization of amplification at multiple GFR genes in lung primary tumors

First, we determined the copy number of the FGFR1, FGFR2, FGFR3, IGF1R, KIT, MET and PDGFRA genes using FISH analysis. Tumors were assigned to one of four categories on the basis of their gene copy number: (i) normal copy (NC), for <4 copies; (ii) minimally increased (MI), 4–6 copies; (iii) highly increased (HI), for 7–12 copies; and (iv) GA, >12 copies of the gene per cell. It is important to mention that while MI and HI always depicted individual and countable spots (<12) scattered throughout the genome, GA commonly appeared as one or several gene copy clusters (Fig. 1a and Supporting Information Fig. S1). This is compatible with an intrachromosomal GA mechanism, as observed in other known oncogenes (e.g., NMYC and ERBB2). Since cut-off values could not be applied in these cases, the presence of gene clusters was taken to indicate the occurrence of GA. The results are summarized in Table 1. GA was observed only for FGFR1, KIT/PDGFRA and MET, with total frequencies ranging from 1 to 6%. Examples are depicted in Figure 1a. Amplification at FGFR1 was significantly more frequent in SCCs while MET amplification was found in one AC from a smoker and in one large cell carcinoma. PDGFRA and KIT are located only about 0.35-kb apart on chromosome 4q and, in our tumor set, a GA event always affected both genes and occurred in SCCs and ACs. MI was the only copy number alteration observed for IGF1R, FGFR2 and FGFR3. In the case of IGF1R, MI significantly predominated in SCC.

Table 1. Distribution of gene copy number alterations and the indicated tumor and patients characteristics
 FGFR1FGFR2FGFR3IGF1RKIT/PDGFRAMET
(n = 265) (n = 266) (n = 263) (n = 253) (n = 258) (n = 249)
NCMIHIGANCMINCMINCMINCMIHIGANCMIGA
  1. AC, adenocarcinoma; GA, gene amplification (>12); HI, highly increased (7–12); LCC, large cell carcinoma; MI, minimally increased (4–6); NC, normal copies (<4); SCC, squamous cell carcinoma.

  2. a

    p < 0.01.

  3. b

    p < 0.05.

  4. In bold are indicated the groups that show statistically significant differences when compared with the NC.

Total219245172541225211206472351274229182
%839269559648119915329271
AC76811903881729823328151
SCC11813415a1398138711035b12972212680
LCC23201221233213212201951
Other21003030303000300
Male8018210109410628024967549863
Female17100190170182181001810
Ever35511451420338401324111
Never9100110100120101001010
image

Figure 1.  Detection of chromosomal alterations for the indicated genes and gene or protein expression levels. (a) Examples of interphase nuclei from lung tumors with NC number or GA at the indicated genes (probes in red). Control probe in green. (b) Representative examples of negative (upper panel) and positive (lower panel) immunostaining for the indicated proteins (original magnification, 200×). (c) mRNA levels of the indicated genes, within the different histopathologies, AC, adenocarcinoma; LCC, large cell carcinoma; SCC, squamous cell carcinoma, assessed by real-time quantitative RT-PCR. Five normal lung (NL) and positive control cell lines (H520 and DMS-114 for FGFR1 and H1703 for PDGFRA) and negative control from lung cancer cell lines (white bars) are also shown. The red and blue dashed lines indicate the mean level of gene expression in the normal lung and the threshold for gene overexpression, respectively.

Download figure to PowerPoint

We also determined the ALK translocations and ERBB2 amplification on the basis of protein immunostaining and verification by FISH, because these are known to produce an increase in their respective protein levels.[3, 13]

Determination of GFR protein and mRNA levels

Immunostaining was performed to determine the levels of ALK, FGFR1, FGFR3, HER2, IGF1R, KIT and MET proteins. Except for FGFR1, that was tested at the protein and mRNA level, real-time quantitative RT-PCR was performed only in those GFRs without reliable antibodies. Examples are included in Figures 1b and 1c and the overall results are provided in Table 2. Respectively, four and one of the analyzable lung tumors exhibited increased HER2 and ALK protein levels, caused by GA and translocation [ALK-NMP; (2;5)(p23;q25)], respectively (Supporting Information Fig. S2). All of them were adenocarcinomas and, in the case of the ALK-positive tumor, the specific subtype was a signet ring cell adenocarcinoma, from a woman of unknown smoking habit. Moderate or strong immunostaining of IGF1R was significantly higher in the SCC subtype, whereas FGFR1 (negative vs. moderate or strong), KIT (negative vs. weak) and MET (negative vs. moderate or strong) immunostaining was significantly more frequent in ACs (p < 0.0001; Chi-square test). Only five tumors, all of them SCCs, featured increased levels of FGFR3 protein.

Table 2. Distribution of the immunostaining of GFRs among lung tumors, by histopathological characteristics
  FGFR1 FGFR3 HER2 IGF1R KIT MET
(N = 266) (N = 275) (N = 282) (N = 277) (N = 289) (N = 284)
NWSNPNMSNMSNWSNMS
  1. M, moderate; N, negative; P, positive; S, strong; W, weak.

  2. a

    p < 0.001.

  3. b

    P < 0.0001.

  4. c

    p < 0.05.

  5. In bold are indicated the groups that show statistically significant differences when compared with the N group.

No. Tumors13693372705266124121102542265672143931
Frequency (%)513514982944144371978192751411
Histopathological subtypes
AC2739a24b9509754c6224b11b6435b35424b20b
SCC954211145514150397439138152136144
LCC13102270242017442522117
Other12030200300212300
Gender
Female3710a1133103213441c37c91236831516
Male5339171901812125315601352
Smoking status
Ever13171445045222415230172221311
Never1561201001821840822

We used real-time quantitative RT-PCR to determine the mRNA levels of FGFR1, FGFR2 and PDGFRA in about 60 lung tumors with available mRNA. In addition to normal lungs, we used lung cancer cell lines carrying known GA and concomitant overexpression at FGFR1 (NCI-H520 and DMS114) and PDGFRA (NCI-H1703) as positive controls, as well as negative controls. Taking these into consideration, we defined the threshold for gene overexpression as those relative values of gene expression more than ten times the median level for normal lung (Fig. 1c). According to this criterion, two and one of the lung tumors tested exhibited FGFR1 and PDGFRA gene overexpression, respectively. None of the tumors overexpressed FGFR2. The mean gene expression levels for FGFR1, FGFR2 and PDGFRA were similar in the various tumor histopathologies.

Correlation between gene copy number and expression levels

A functionally relevant GA event should trigger strong overexpression of the target oncogene. To determine whether GA is associated with strong expression, we compared copy number and mRNA or immunostaining data. We found no association between strong FGFR1 immunostaining and GA at the FGFR1 gene (Fig. 2a). Since the percentage of lung tumors, specially lung AC, exhibited very high levels of FGFR1 protein, we tested the FGFR1 expression at the mRNA level. Tumors with GA had significantly higher levels of mRNA than tumors with NC (Fig. 2b). However, of the nine available lung tumors with GA for FGFR1 and mRNA, only one featured overexpression, suggesting that FGFR1 is not always the target oncogene. Conversely, one AC with NC overexpressed FGFR1 (Fig. 2b). We discarded the presence of a FGFR1 translocation in this single tumor, using the FISH breakapart strategy (data not shown). MI at FGFR2 was observed in a few cases but the correlation with increased gene expression levels could not be determined because RNA was not available for most of these tumors. MI at FGFR3 was not correlated with positive immunostaining of FGFR3, while MI at IGF1R was associated with moderate and with strong immunostaining of IGF1R (Figs. 2a2c).

image

Figure 2.  Correlation between GA and gene or protein expression. (a) Protein immunostaining correlations, for the indicated proteins in the different copy number group of tumors. p-Values from the two-sided Fisher's exact test. (b) RNA levels comparison for the indicated genes and groups of tumors. The medians of mRNA levels for each group are indicated (horizontal line within the boxes). (c) Comparison of the gene copy number (purple-colored gradient) and protein levels (blue-colored gradient) of each individual tumor. *p < 0.05; **p < 0.01; ***p<0.001. NC, normal copy; MI, minimal increase; HI, high increase, GA, gene amplification; AC, adenocarcinomas; LCC, large cell carcinomas; SCC, squamous cell carcinomas; N, negative staining; M, moderate staining; S, strong staining.

Download figure to PowerPoint

PDGFRA and KIT are well-documented oncogenes and are often coamplified in lung cancer, making it difficult to determine which is the target. Four tumors depicted amplification at PDGFRA/KIT loci, and three of them (75%) exhibited strong KIT immunostaining (Figs. 2a2c), although other four tumors also did so in the absence of GA. Of the 63 tumors that could be analyzed, only one of the four that were PDGFRA-amplified showed gene overexpression. Intriguingly, the tumor with the strongest GA at the PDGFRA/KIT loci was negative for both KIT immunostaining and PDGFRA overexpression. It is unlikely that this GA is a random event, and probably indicates the presence of another oncogene target. Since KDR, which encodes one of the two VEGF receptors, is located very close to the KIT locus, we tested for overexpression of its transcript in our panel of tumors. We detected no KDR overexpression in any of the lung primary tumors tested (Supporting Information Fig. S3). As a whole, our data indicate that KIT is the main oncogene target of the amplification at the KIT/PDGFRA loci. Strong KIT immunostaining is highly reproducible and easily distinguishable from the negative and weak categories, which leads us to propose it as a surrogate marker for tumor preselection in KIT amplification analysis.

With respect to MET, the two tumors with GA featured concomitant strong MET immunostaining (100%) (Fig. 2a), although, as with KIT, strong immunostaining was also evident in tumors without GA or MI. This implies that while MET is the target of the GA event, MET overexpression can also arise from other mechanisms. In spite that strong MET immunostaining was observed in many tumors without GA, mostly adenocarcinomas, we believe that determination of MET protein levels by immunostaining could be used for tumor preselection in MET amplification analysis.

Overall, of all the GFRs analyzed here, MET, HER2 and KIT were the most closely correlated with the presence of GA and strong protein immunostaining. Figure 2c depicts a summary of the correlation gene copy number and protein immunostaining and Table 3 shows the specific histopathology of these alterations.

Table 3. Clinical, histopathological and molecular characteristics of patients whose tumors exhibited activation at the indicated GFRs
Patient NoGFR-alteredAge (yrs)SexSmokingStageMetastasisHistologyTypeOther mutations1
  1. NI, no information; NOS, not otherwise specified.

  2. Other genes tested for mutations (KRAS, EGFR and LKB1) were negative.

1ALK54FemaleNIT2N2 Invasive adenocarcinomaPapillar, signet ring cell featuresTP53
2ERBB268MaleNOT1N0NOInvasive adenocarcinomaSolidTP53
3ERBB272FemaleYEST3N0YESInvasive adenocarcinomaColloidTP53
4ERBB272FemaleNOT2N0NIInvasive adenocarcinomaPredominantly lepidic 
5ERBB2NININININIInvasive adenocarcinomaPredominantly lepidicNI
6KIT69MaleNIT2N0NOSquamous cell carcinoma  
7KIT82MaleYEST2N0NOInvasive adenocarcinomaPredominantly acinarTP53
8KIT/PDGFRA75MaleYEST3N0NOAdenocarcinomaSolid 
9MET65MaleYEST2N0NOInvasive adenocarcinomaPredominantly acinarTP53
10MET52MaleNIT1N0NOLarge cell carcinoma TP53

Because some tumors featured strong protein expression in the absence of GA, we aimed to rule out the possibility that point mutations cause the gene overexpression. We screened for FGFR1, FGFR2, FGFR3, IGF1R, KIT, MET and PDGFRA mutations at the exons coding for the respective tyrosine kinase domains and at known mutational hotspots. We selected about 50 lung tumors, including those that had strong immunostaining but were negative for GA. No mutations were found at any of the genes examined.

Activation of GFRs leads to constitutive downstream signaling that promotes cell growth. It is widely accepted that genes acting in the same pathway are not concurrently altered. We combined the data on GA with those on mutations at KRAS and EGFR, which were available for about 80 lung ACs. We did not detect simultaneous alterations at ALK, EGFR, ERBB2, KRAS, MET or KIT/PDGFRA (Table 3). Although this suggests a mutually exclusive activation of these oncogenes in cancer, the observation needs to be confirmed definitively in a larger set of tumors. This is especially important because concomitant activation of MET, ALK and BRAF with either EGFR or KRAS has been reported in lung cancer cell lines[9] (Sanger cell line project: www.sanger.ac.uk/genetics/CGP/CellLines/). Finally, and following the line of our previous observations,[13] there was a positive association between alterations in either of the ALK, EGFR, ERBB2, KRAS or MET genes and augmented levels of phospho-S6 protein (p = 0.0005; Fisher's Exact Test). This supports that alterations of these oncogenes in cancer trigger a constitutive activation of the mTOR pathway (Fig. 3a). We also tested for the correlation of activation at other proteins involved in the GFRs pathways, such as phospho-mTOR or phospho-AKT (Supporting Information Fig. S4a) and alterations at these GFR genes but not statistically significant associations were found (Supporting Information Fig. S4b). Furthermore, no correlations were observed among the immunostaining of phospho-S6, phospho-mTOR or phospho-AKT (Supporting Information Fig. S4c).

image

Figure 3.  (a) Distribution of the immunostaining of phospho-S6 protein among the lung tumors carrying alterations at the indicated oncogenes. High levels of phospho-S6 immunostaining (dark blue) were more common in tumors carrying activation at any of the oncogenes (p = 0.0005; Fisher's Exact Test). (b) Proposed diagram for testing the indicated GFRs in adenocarcinomas and squamous cell carcinomas of the lung. The initial determination of strong protein levels (or mRNA overexpression) will preselect tumors for gene copy number analysis by FISH. A positive association between two parameters will determine the oncogenic activation of the GFR. In the case of KIT/PDGFRA the situation is more complex in those tumors carrying simultaneous increases in the KIT and PDGFRA protein and mRNA levels, respectively. NC, normal copy; MI, minimal increase; HI, high increase; N, negative staining; W, weak staining; M, moderate staining.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. References
  8. Supporting Information

Our comprehensive analysis of gene copy number and gene or protein expression of multiple GFRs in a large panel of tumors is the first of that kind in NSCLC. Increased gene copy number but not GA was found at FGFR2, FGFR3 or IGF1R, indicating that GA is an uncommon mechanism for the oncogenic activation of these GFRs, at least in lung cancer. Similarly to previous reports, extra copies of IGF1R were frequently observed, especially in the histopathology of SCC, in association with positive protein immunostaining.[17, 18] Gains of copies of some chromosomal regions are common in cancer and often trigger increased gene expression within the region, although their contribution to carcinogenesis is unclear.[19] In contrast, GA and concomitant gene overexpression was found for FGFR1, KIT/PDGFRA, MET and ERBB2, implying the involvement of these receptors in the carcinogenesis of a subset of lung tumors. As other studies have shown,[11, 17] we found FGFR1 overexpression only in a few lung SCCs and ACs, whereas amplification of FGFR1 was more common and mainly associated with SCCs. The imperfect association of FGFR1 amplification with overexpression suggests that another oncogene within the region is also a target of the amplification. In contrast to previous reports, FGFR1 protein expression was significantly stronger in lung AC which is a novel observation.[18, 20] This is potentially interesting and would need further analysis. Mutations or chromosomal translocations involving KIT and PDGFRA are common in gastrointestinal stromal tumors and hematological neoplasias[21-23] and the inhibitors, sunitinib and imatinib, which target KIT and PDGFRA, can yield clinical benefit in both settings. Although these alterations have rarely been found in most solid tumors,[24] KIT and PDGFRA are strongly expressed in lung tumors of neuroendocrine origin.[22-25] We found KIT and PDGFRA amplification in fewer than 2% of the lung tumors, but this was positively associated with strong KIT immunostaining. Although both genes were coamplified, PDGFRA overexpression was negative in all cases but one, which implies that KIT is the main target oncogene of this amplicon in lung cancer. In contrast to our observations, others have reported a high frequency of amplification at the KIT/PDGFRA loci in lung tumors, but failed to detect a positive correlation with strong KIT immunostaining.[26] These discrepancies can probably be attributed to the lower restrictive threshold used by these researchers to define GA, compared with our stricter criteria. Intriguingly, we detected one lung tumor that was affected by strong GA at KIT/PDGFRA loci but was negative for overexpression at the two genes, suggesting the presence of another oncogene in this region. We were able to discount the possibility that this oncogene was KDR, encoding one VEGF receptor, because it did not overexpress KDR. However, and similarly to lung cancer cell lines,[9] amplification and overexpression at MET and ERBB2 affected a low percentage of lung adenocarcinomas.

Some lung tumors had FGFR1 overexpression and FGFR1, FGFR3, KIT or MET strong immunostaining in the absence of GA. We ruled out somatic mutations as the underlying cause, but other genetic alterations such as reciprocal translocations, which are known to activate membrane receptors such as ALK, RET and ROS in lung cancer,[3-7] can also trigger overexpression. Alternatively, feedback mechanisms arising from gene alterations at related biological pathways may also cause overexpression. In the latter scenario, therapies against these GFR are not expected to be effective.

Finally, activation of the ALK, EGFR, ERBB2, KRAS, MET and PDGFRA oncogenes did not co-occur in lung adenocarcinomas, consistent with most current observations.[3, 8, 9] However, this should be interpreted with caution because simultaneous alterations of ALK and KRAS/EGFR and of MET and KRAS occur in lung cancer cell lines and in a few lung primary tumors,[27, 28] refuting the generally held belief that alterations between these GFRs and the components of the EGFR/KRAS pathway are mutually exclusive. These findings have important clinical implications, since combinatorial treatments can be administrated to patients carrying tumors with concurrent EGFR and ALK or MET alterations. However, supporting our previous observations for EGFR and KRAS mutant tumors,[13] we did observe that lung adenocarcinomas carrying activating gene alterations in ALK, EGFR, ERBB2, KRAS, MET and PDGFRA have higher levels of phospho-S6, a ribosomal protein that is phosphorylated by the mTOR-substrate, S6 kinase. Although larger set of samples should be tested to confirm this, phospho-S6 immunohistochemistry may prove useful to select tumors for alterations at any of these genes.

In conclusion, we have demonstrated that GA provides a means for ERBB2, MET, KIT and PDGFRA to be overexpressed in lung cancer and we propose with a diagram that indicate the steps for determining the oncogenic activation at GFRs in lung cancer (Fig. 3b). In this diagram, high protein or mRNA levels at these GFRs could be surrogate markers for tumor preselection. This should facilitate the stratification of lung cancer patients for treatment with targeted therapies.

Acknowledgments

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. References
  8. Supporting Information

The authors acknowledge the technical assistance of Albert Coll (Genes and Cancer Group) and the technical advice of their colleague Marga Nadal at IDIBELL.

References

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. References
  8. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. References
  8. Supporting Information

Additional Supporting Information may be found in the online version of this article.

FilenameFormatSizeDescription
ijc28090-sup-0001-suppinfo.docx27KSupporting Information Tables
ijc28090-sup-0002-suppinfo.pdf1823KSupporting Information Figures

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.