Increased VEGFR-2 Gene Copy Is Associated with Chemoresistance and Shorter Survival in Patients with Non–Small-Cell Lung Carcinoma Who Receive Adjuvant Chemotherapy Free

Tumor growth is critically dependent on neovascularization (1). The ligand VEGF is an endothelial cell-specific mitogen known to be a highly potent and specific mediator of angiogenesis, and has 2 identified tyrosine kinase receptors, VEGF receptor-1 (VEGFR-1) and VEGF receptor-2 (VEGFR-2 or kinase insert domain receptor; KDR; refs. 2–5). The VEGFR-2 coded by the gene KDR (located in 4q12) is the predominant mediator of VEGF-stimulated endothelial cell functions, including cell migration, proliferation, survival, and enhancement of vascular permeability (6, 7). VEGFR-2 exhibits robust protein tyrosine kinase activity in response to the VEGF ligand (3).

In human epithelial tumors, including lung, VEGFR-2 has shown to be expressed in malignant cells as well as in the endothelial cell of tumor vasculature (8–11). In non–small-cell lung carcinoma (NSCLC), VEGFR-2 has been found to be overexpressed in malignant cells of tumor tissues, and associated with a poor outcome (8–12). The mechanism and biological impact of VEGFR-2 overexpression of NSCLC cells, however, is not known. Recent work from our group and others has shown that tumor cell expression of VEGFR-1 may drive tumor cell invasiveness (13, 14) and promote hypoxia-independent upregulation of hypoxia inducible factor-1α (HIF-1α), but it is not known whether VEGFR-2 signaling directly impacts the tumor cell phenotype in NSCLC.

Recently, a relatively high frequency (9%) of mutation and amplification of KDR has been detected in lung adenocarcinoma histology (15); however, the presence of these abnormalities in squamous cell carcinomas of the lung is unknown. In addition, there is no data available on the correlation of KDR abnormalities with tumor and patients' characteristics in lung cancer, including outcome and response to therapy.

The objective of this study was to characterize the molecular abnormalities of VEGFR-2 in epithelial malignant cells of NSCLC major histology types, adenocarcinoma and squamous cell carcinoma, and correlate with patients' clinical characteristics. We studied KDR copy number gain (CNG), mutation, and genetic variations in malignant cells of surgically resected NSCLC tumor tissues and correlated results with pathologic features in NSCLC patients' tumors and with their platinum adjuvant treatments and outcomes. In addition, using a series of NSCLC cell lines and tissue specimens, we investigated molecular mechanisms associated with KDR CNG in resistance to platinum, particularly the potential role of HIF-1α, a key regulator of angiogenesis in malignant tumors (16).

We obtained archived frozen and formalin-fixed and paraffin-embedded (FFPE) tissues from NSCLC patients who were surgically resected with curative intent from the Lung Cancer Specialized Program of Research Excellence (SPORE) tissue bank at The University of Texas MD Anderson Cancer Center (Houston, Texas). The tissue banking and the study were approved by the Institutional Review Board. We randomly selected 248 NSCLC specimens (159 adenocarcinomas and 89 squamous cell carcinomas) to test KDR abnormalities. Detailed clinical and pathologic information of the cases is presented in Supplementary Table S1.

We utilized 2 methodologies to test KDR CNG in NSCLC tumor specimens: real-time quantitative PCR (qPCR) and fluorescence in situ hybridization (FISH). To enrich for malignant cell content for qPCR analysis, tumor tissues were manually microdissected from optimal cutting temperature compound-embedded frozen tissue sections for subsequent DNA extraction. Tumor DNA was extracted using Pico Pure DNA Extraction Kit (Arcturus) according to the manufacturer's instructions. DNA samples with proportions of microdissected tumor cell greater than 70% were qualified for qPCR analysis. KDR gene copy number was detected by qPCR using the ABI 7300 real time PCR system (Applied Biosystems). The primers used to amplify KDR were KF-GACACACCCTCAGGCTCTTG and KR-ACTTTTCACCGCCTGTTCTC. Each PCR was carried out using Power SYBR Green PCR Master Mix (Applied Biosystems) at 50°C for 2 minutes and 95°C for 10 minutes followed by 40 cycles at 95°C for 15 seconds and 60°C for 1 minute. β-Actin was introduced as the endogenous reference gene, and TaqMan Control Human Genomic DNA (Applied Biosystems) was amplified as a standard control for calibration. All sample and standard DNA reactions were set in triplicate to gauge reaction accuracy. The target gene copy number was quantified using the comparative C_t method. Gene copy number of greater than 4 was considered as CNG, as previously reported (17).

KDR copy number analysis in NSCLC malignant tumor cells was also carried out using a dual-color FISH assay developed by 1 of the coauthors (M.V-G.). The KDR probe was prepared from the BAC clone RP11-21A18 obtained from CHORI. The FISH assay was conducted as we have previously published (18). Copy number analysis was done in approximately 50 nuclei per tumor in at least 4 areas. Greater than 2 gene copies per cell on average were considered as CNG.

All NSCLC cell lines were authenticated by DNA fingerprinting. Whole genome single nucleotide polymorphism (SNP) array profiling was carried out in 75 NSCLC cell lines using the Illumina Human1M-Duo DNA Analysis BeadChip (Illumina, Inc.). Prior to analysis, SNP data were normalized to the regional baseline copy number to account for aneuploidy. For VEGFR-2 reverse phase protein array (RPPA) analysis conducted in 63 NSCLC cell lines, protein lysate was collected from subconfluent cultures after 24 hours growth in media with 10% FBS and assayed by RPPA as previously described (19, 20). Cisplatin and carboplatin sensitivity was determined in triplicate by MTS (inner salt) assay for each cell line, and the concentration required for 50% growth inhibition (IC₅₀) was determined. For HIF-1α expression analysis, the cells were serum starved for 24 hours and stimulated with 50 ng/mL VEGF-A (R&D Systems). Cells were incubated in normoxia, and protein lysates were collected after 8 hours. HIF-1α ELISA (R&D Systems) was carried out according to the manufacturer's instructions (13).

Histology sections were incubated with primary antibodies against VEGFR-2 (dilution 1:50; Abcam) for 90 minutes, CD34 (dilution 1:100; Lab Vision) for 35 minutes, and HIF-1α (dilution 1:100; Novus Biologicals) for 65 minutes. Tissue sections were then incubated with the secondary antibody (EnVision Dual Link+; DAKO) for 30 minutes, after which diaminobenzidine chromogen was applied for 5 minutes.

Protein expression was quantified by immunohistochemistry using light microscopy with a 200× magnification by 2 observers (F.Y. and I.W.). Tissue samples were analyzed for VEGFR-2 expression in the cytoplasm and membrane of malignant cells and for HIF-1α in the nucleus. We used a 4-value intensity score (0, 1+, 2+, and 3+) and the percentage (0% to 100%) of the extent of reactivity. The final score was obtained by multiplying the intensity and extent-of-reactivity values (range, 0–300). Microvascular density (MVD) was assessed by Ariol 2.0 Image System (Ariol, Genetix) using the criteria of Weidner and colleagues (21).

We transfected NSCLC cells with 3 KDR gene-specific siRNA duplexes and control siRNA (OriGene Technology), at a final concentration of 10 nmol/L using Lipofectamine RNAiMAX (Invitrogen) according to the manufacturer's instructions. To verify the knockdown efficiency, mRNA and protein of transfected cells were collected for real-time reverse transcripatase PCR (RT-PCR) and Western blot analyses. The assessment of in vitro resistance to cisplatin and carboplatin was determined by the MTS assay. NSCLC cell lines were seeded in octuplicate at a density of 2,000 per well in 96-well plates. The following day, cells were treated with cisplatin and carboplatin at various concentrations ranging from 0 to 120 μmol/L for cisplatin and 0 to 200 μmol/L for carboplatin. After 72 hours of drugs exposure, 20 μL of MTS solution were added per well. Cells were incubated for 1 to 4 hours at 37°C and read at a wavelength of 490 nm. The cell migration assay using NSCLC cell lines was carried out as previously reported (13).

For KDR mutation and SNP genotyping analysis in NSCLC cell lines, we examined exons 7, 11, 21, 26, 27, and 30, using PCR-based sequencing and intron-based PCR primers as detailed in the Supplementary Table S2.

Demographic and clinical information were compared by using the χ² or Fisher exact tests for category variables, and Wilcoxon rank-sum or Kruskal-Wallis tests for continuous variables. The distributions of overall survival (OS) and recurrence-free survival (RFS) were estimated by the Kaplan–Meier method and compared between groups using the log-rank test. Cox proportional hazard models were used for regression analyses of survival data and conducted on OS defined as time from surgery to death or last contact, and on RFS defined as time from surgery to recurrence or last contact. Follow-up time was censored at 5 years. For the correlation analysis of KDR CNG in NSCLC cell lines using the whole genome SNP arrays data with cisplatin sensitivity, we used the Wilcoxon rank sum test. The NSCLC cell lines RPPA data were quantified using the SuperCurve method which detects changes in protein level as previously reported (22).

In epithelial malignant NSCLC cells microdissected from tumor tissues, KDR CNG was detected in 45 (32%) of 139 tumors examined. Similar frequency of KDR CNG was found in adenocarcinoma (26/85, 31%) and squamous cell carcinoma (19/54, 35%) histologies (P = 0.572). The range of increased KDR copy numbers was from 4 to 11 gene copies. None of 15 normal tissue samples adjacent to the NSCLC tested showed KDR CNG. To confirm KDR CNG results by qPCR, 20 tumor specimens with KDR CNG by qPCR were examined by FISH. KDR copy gains in the malignant cells were confirmed by FISH in all 20 NSCLC specimens detected by qPCR (Fig. 1A).

To assess the immunohistochemical (IHC) expression of VEGFR-2 in NSCLC malignant cells and the MVD (CD34) in lung tumor tissue stroma, we selected 52 lung tumor specimens with whole histologic sections from FFPE tissues. Of these, 26 cases had KDR CNG and 26 cases did not. VEGFR-2 protein expression was present both in the cytoplasm and membrane of malignant cells as well as in vessel endothelial cells (Fig. 1B).

Levels of VEGFR-2 expression in cytoplasm and in membrane were associated with KDR CNG in malignant cells of NSCLC. Tumors with KDR CNG showed significantly higher cytoplasmic (P = 0.013) and membrane (P = 0.009) VEGFR-2 protein expression in the malignant cells (Fig. 1C), and higher MVD (P = 0.018) and larger vessel areas (P = 0.033) in the tumor stroma than cases without KDR CNG (Fig. 2A and B).

When we correlated KDR CNG with patients' clinicopathologic features, we did not find correlation with tumor histology, smoking status, and tumor stage. In the multivariate analysis after adjusting for stage and adjuvant therapy, KDR CNG was associated with poor OS (HR = 4.0; 95% CI: 1.76–9.07; P = 0.001) and shortened RFS (HR = 1.83, 95% CI: 1.02–3.29; P = 0.044) in 115 NSCLC patients who underwent surgical resection. Strikingly, KDR CNG was associated with a significantly worse OS (HR = 5.16, 95% CI: 1.75–15.2; P = 0.003) in NSCLC patients receiving platinum adjuvant therapy, but not in patients without adjuvant therapy (P = 0.349; Fig. 3 and Table 1). These data suggest that KDR CNG in malignant cells may represent a predictive marker of worse outcome in patients with surgically resected NSCLC treated with platinum-based adjuvant chemotherapy.

We also investigated and examined the impact of neoadjuvant chemotherapy on KDR CNGs. The platinum neoadjuvant-treated tumors (33%, 8/24) had similar frequency of KDR CNGs than cases without neoadjuvant therapy (32%, 37/115).

The association detected between KDR CNG and worse outcome in patients treated with platinum adjuvant therapy prompted us to examine the correlation between KDR gain and VEGFR-2 protein levels in NSCLC cell lines with in vitro resistance to platinum drugs. KDR CNG was assessed by SNP array analysis in 75 NSCLC cell lines. Cell lines with KDR copy gains of 6 to 9 copies or 10 or more copies above the regional baseline copy number were identified. Nineteen (25%) cell lines showed KDR CNG defined as 6 or more copies. Of these, 3 (4%) cell lines contained high-level gains (≥10 copies), and 16 (21%) had CNG between 6 to 9. Of interest, cisplatin sensitivity in cell lines with 6 or more KDR copies showed significantly more resistance to cisplatin (P = 0.0179; Fig. 4A).

Then, we correlated the expression of VEGFR-2 protein in a panel of 63 untreated NSCLC cell lines by RPPA with each cell line's sensitivity to cisplatin or carboplatin. We found that higher VEGFR-2 expression levels were significantly associated with resistance to both cisplatin (Fig. 4B) and carboplatin (data not shown) by Pearson correlation. The correlation coefficient (r) between VEGFR-2 expression and the concentration of cisplatin and carboplatin required to inhibit cell growth by 50% (IC₅₀) were 0.346 (P = 0.005) and 0.319 (P = 0.011), respectively.

To investigate the role of KDR CNG and VEGFR-2 overexpression in resistance to both cisplatin and carboplatin, we utilized siRNA to knock down KDR expression in H23 and H461 NSCLC cell lines, which contain 6 to 9 KDR gene copies, and as control A549 NSCLC cell line with normal KDR copy number. In both cell lines, siRNA targeting KDR significantly decreased KDR mRNA expression by real-time RT-PCR, and VEGFR-2 expression by Western blot, compared with control cells transfected with scrambled siRNA and nontransfected cells (P < 0.05; Fig. 4C). The in vitro sensitivity of H23 and H461 cells to cisplatin (Fig. 4D) or carboplatin (data not shown) treatment was increased in siKDR transfected cells compared with control siRNA-transfected or untransfected cells, suggesting that VEGFR-2 is contributing to chemoresistance in this model. This phenomenon was not observed in cell A549 with normal KDR copy number.

In addition, we found that knockdown of reduction of VEGFR-2 expression induced by siKDR transfection significantly inhibited the migration of H23 and H461 cells compared with siRNA control-transfected or untransfected cells (Fig. 4E and F).

The observations that KDR CNGs were associated with increased angiogenesis, chemoresistance, and migration suggested that VEGFR-2 may be impacting the HIF-1α pathway, which is known to impact each of these cellular properties (13, 14). To investigate this further, we evaluated HIF-1α levels by ELISA in a panel of NSCLC cell lines with a range of KDR copy numbers and expression of VEGFR-2. HIF-1α levels were higher in cell lines with KDR CNG, and significantly (P = 0.02) higher in cells with 6 to 9 gene copies, compared with cells with no CNG (Fig. 5A). In H23 cells which have KDR CNG, stimulation with 50 ng/mL VEGF-A for 8 hours induced a rise in HIF-1α expression. Furthermore, knockdown of KDR with siRNA significantly (P = 0.01) reduced HIF-1α levels (Fig. 5B). This phenomenon was not detected in cell lines A549 with normal KDR copy number. These data indicated that VEGFR-2 can regulate HIF-1α in a ligand-dependent, but hypoxia-independent, manner in NSCLC cells.

We next investigated the potential association between KDR CNG and HIF-1α in NSCLC clinical specimens. Similar to the results in the NSCLC cell lines, tumor tissue specimens with KDR CNG (n = 25) showed a significantly (P = 0.037) higher expression of nuclear HIF-1α expression by immunohistochemistry than tumors without CNG (n = 22; Fig. 5C and D).

To investigate whether alterations in the KDR gene other than CNGs may impact NSCLC tumors, we assessed the KDR gene for mutations and SNPs. For KDR mutation analysis in NSCLC cell lines, we examined 6 KDR exons (7, 11, 21, 26, 27 and 30) they showed to be mutant in adenocarcinoma tumors in a study published by Ding and colleagues (15). In 37 tested NSCLC cell lines, we found only 2 mutations in the KDR gene, an intronic T + 2A exon 11 mutation in HCC2450 and a CGT946CAT point mutation in exon 21 in HCC2279. No mutation affecting exons 11 and 21 was detected in 200 NSCLC tissues specimens examined.

In addition, 3 KDR SNPs (889G/A, 1416A/T, and -37A/G) were genotyped in DNA extracted from 200 NSCLC tumors (Supplementary Table S3), and correlated with patients clinicopathologic features, including outcome. We did not find correlation between the SNP genotypes distribution and OS or RFS of all NSCLC patients examined. In adenocarcinoma patients both KDR 1416 AT/TT (HR = 0.45, 95% CI: 0.2–0.99; P = 0.048) and -37AG/GG (HR = 0.43, 95% CI: 0.2–0.92; P = 0.031) variant genotypes were associated with a favorable OS in the multivariate analysis after adjusting for tumor stage and neoadjuvant therapy (Supplementary Fig. S1, and Supplementary Table S4).

Furthermore, among NSCLC patients with the KDR 889 GA/AA variant genotypes, those who received platinum neoadjuvant and/or adjuvant chemotherapy showed a significantly better OS (HR = 0.22, 95% CI: 0.05–0.94; P = 0.041) than patients who did not receive chemotherapy in the multivariate analysis after adjusting for histology and tumor stage. However, no survival benefit was found in NSCLC patients with KDR 889 GG wild genotype (HR = 1.23, 95% CI: 0.64–2.35; P = 0.538).

Our study represents the first report in lung cancer showing a high frequency of KDR CNG (32%) in both major histology types of NSCLC, adenocarcinoma, and squamous cell carcinoma, by qPCR and confirmed in a subset of cases by FISH. Notably, KDR CNG predicted worse OS in patients who received platinum adjuvant therapy but not in untreated patients. In NSCLC cell lines, we found that KDR CNGs were significantly associated with in vitro resistance to platinum chemotherapy, as well as increased levels of nuclear HIF-1α. Furthermore, KDR knockdown experiments using siRNA reduced platinum resistance, cell migration, and HIF-1α levels in cells bearing KDR CNGs, providing evidence for direct involvement of KDR. Our findings suggest that tumor cell KDR CNGs may promote a more malignant phenotype including increased chemoresistance, angiogenesis, and HIF-1α levels.

In our study, tumors with KDR CNG in the malignant cells showed significantly higher VEGFR-2 protein expression in the cytoplasm and membrane of those cells, as well as higher MVD and larger vessel areas in the tumor stroma, compared with tumors lacking the KDR CNG. One possible explanation for this association is that tumor cell VEGFR-2 binds circulating VEGF, increasing local concentrations of the ligand which in turn increases angiogenesis through effects on tumor endothelium. Another possible explanation is that VEGFR-2 overexpressing lung cancer cells may express increased levels of VEGF and other proangiogenic factors via upregulation of HIF-1α, which in turn could promote autocrine or paracrine signaling that further increases expression. These mechanisms are not mutually exclusive and merit further investigation. Our finding of correlations between KDR CNG and higher expression of HIF-1α in NSCLC cell lines and tumor specimens support the latter hypothesis. It has been showed that activation of several receptor tyrosine kinases (RTK; RET, VEGFR-1, epidermel growth factor receptor, and platelet-derived growth factor receptor) increases HIF-1α levels in a cell-specific manner in tumors (13, 23, 24); therefore, our data represent the first evidence suggesting that VEGFR-2 may be another RTK that plays a role in increasing the levels of HIF-1α expression in cancer.

A provocative finding of this study is this first report that KDR CNG in malignant cells predicted a worse outcome of NSCLC patients receiving platinum adjuvant chemotherapy after surgical resection with curative intent, but was not predictive in patients without adjuvant therapy. These findings suggest that KDR CNG may represent a potential biomarker for predicting resistance to adjuvant platinum-based chemotherapy in NSCLC patients. It is also noteworthy that VEGFR-2 knockdown reduced chemoresistance and cell migration, and lowered HIF-1α levels, using in vitro NSCLC models. One potential implication of these findings is that VEGFR-2 blockade may sensitize tumors bearing KDR CNGs to chemotherapy directly through effects of the tumor cells themselves, in addition to its effect on tumor endothelial cells. KDR CNGs may therefore identify a group of NSCLC patients that would receive greater relative benefit from combinations of VEGF pathway inhibitors with chemotherapy than patients lacking KDR CNGs. Further prospective studies with larger patient cohorts are needed to assess the role of KDR CNG in NSCLC tumors and outcome of NSCLC patients treated with platinum-based chemotherapy in both surgically resected and advanced metastatic tumor settings, and to determine whether KDR CNGs are predictive of either chemoresistance or benefit for VEGF inhibitor benefit/chemotherapy combinations compared with chemotherapy alone.

Our finding that KDR CNG by SNP array and higher levels of VEGFR-2 expression by RPPA in a large series of NSCLC cell lines correlated significantly with in vitro resistance to platinum dugs (cisplatin for KDR CNG, and cisplatin and carboplatin for VEGFR-2 expression) provides support to our clinical observation. The increased sensitivity of the NSCLC cell lines having KDR CNG to in vitro treatment with cisplatin or carboplatin after inhibition of KDR mRNA and protein expressions further supports the concept that KDR CNG may promote platinum resistance in NSCLC. Although the exact mechanism needs to be elucidated, we postulate that the increased expression of HIF-1α induced by KDR CNG, and subsequent VEGFR-2 expression, in malignant NSCLC cells may explain increased platinum resistance in NSCLC. Interestingly, HIF-1α has been previously associated to chemoresistance in NSCLC (25, 26) and other solid tumor types (27, 28).

The finding that inhibition of KDR and VEGFR-2 expression resulted in decreased NSCLC cell migration points out another new interesting role of VEGFR-2 in NSCLC malignant cells. It has been established that, among other functions, VEGFR-2 is an important mediator of VEGF-stimulated endothelial cell migration (29, 30). We have also observed that that HIF-1α mediates migration driven by another RTK, EGFR, in NSCLC, independent of hypoxia (31).

In our study, the variant genotypes of KDR SNPs 1416 (AT/TT) and -37(AG/GG) associated with a favorable OS in the multivariate analysis. Ours is the first report showing association between KDR SNP genotypes and prognosis in lung cancer. In breast cancer patients, the KDR SNP 1416 A/T genotypic variant was associated with the expression of progesterone receptors, and its presence suggested a better prognosis for carriers of the T allele (32). Questions remain about the functional roles of the KDR SNPs responsible for the associations with outcome of NSCLC patients, particularly in adenocarcinoma patients, found in our study.

In summary, our findings indicate that KDR CNG was frequently detected in NSCLC tumors and associated with platinum resistance in vivo and in vitro, and may be a useful biomarker for identifying patients at high risk for recurrence after adjuvant therapy, a group that may benefit from VEGFR-2 blockade.

No potential conflicts of interest were disclosed.

This study was supported by grants from the Department of Defense (W81XWH-07-1-0306 to J.D. Minna, J.V. Heymach, and I.I. Wistuba), the Specialized Program of Research Excellence in Lung Cancer (P50CA70907 to J.D. Minna, J.V. Heymach, and I.I. Wistuba; P50CA58187 to M. Varella-Garcia), and the National Cancer Institute (Cancer Center Support Grant CA-16672).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.