Wnt5a Expression Is Associated With the Tumor Proliferation and the Stromal Vascular Endothelial Growth Factor〞An Expression in Non每Small-Ce
http://www.100md.com
▲還散笫雖悝◎
the Departments of Second Surgery and Pathology and Host Defense, Faculty of Medicine, Kagawa University, Kagawa, Japan
ABSTRACT
PURPOSE: The Wnt gene family encodes the multifunctional signaling glycoproteins. We performed the present study to investigate the clinical significance of Wnt5a expression in non每small-cell lung cancer (NSCLC).
PATIENTS AND METHODS: One hundred twenty-three patients with NSCLC who had undergone resection were investigated. Real-time quantitative reverse transcriptase polymerase chain reaction was performed to evaluate the Wnt5a gene expression. Immunohistochemistry was performed to investigate the Wnt5a protein expression, the Ki-67 proliferation index, tumor angiogenesis, and the expression of beta-catenin and vascular endothelial growth factor-A (VEGF-A).
RESULTS: Wnt5a gene expression in squamous cell carcinoma was significantly higher than that in adenocarcinoma (P < .0001). There was a significant correlation between the normalized Wnt5a gene expression ratio and the intratumoral Wnt5a protein expression (r = 0.729; P < .0001). The intratumoral Wnt5a expression was significantly correlated with the Ki-67 proliferation index (r = 0.708; P < .0001). In contrast, no correlation was observed between the intratumoral Wnt5a expression and tumor angiogenesis. Furthermore, the intratumoral Wnt5a expression was significantly correlated with the stromal expression of beta-catenin (r = 0.729; P < .0001) and VEGF-A (r = 0.661; P < .0001). In addition, the stromal VEGF-A expression was also correlated with Ki-67 proliferation (r = 0.627; P < .0001). Cox regression analyses demonstrated Wnt5a status to be a significant prognostic factor for NSCLC patients (P = .0193), especially for patients with squamous cell carcinomas (P = .0491).
CONCLUSION: The present study revealed that an overexpression of Wnt5a could produce more aggressive NSCLC, especially in squamous cell carcinomas, during tumor progression.
INTRODUCTION
Non每small-cell lung cancer (NSCLC) is one of the most common human cancers with a poor prognosis. Surgical resections play a major role in managing patients with stage I and II NSCLCs, and surgery may also sometimes be used for patients with stage III NSCLCs.1 In addition, various concomitant treatments with chemotherapy and radiotherapy have been assessed for improving the outcome of patients with NSCLC.2 However, the 3-year survival rate is 35% to 55% in patients with stage II NSCLC and 28% in patients with stage III NSCLC.3 Therefore, it is considered important to clarify the mechanism of tumor biology to improve the clinical outcome of patients with NSCLC.
Recent studies on molecular biology in human cancers have revealed that many molecules affect various biologic behaviors of malignant tumors. The activation of oncogenes or the inactivation of tumor suppressor genes could initially cause carcinogenesis.4 During the subsequent tumor progression, many molecular markers associated with angiogenesis5 or metastasis6 could produce more aggressive malignant tumors. Furthermore, recent studies have indicated that the tumor-stromal interaction also plays an important role in tumor progression.7
Among the many molecular markers associated with tumor progression, the Wnt gene family has been shown to encode the multifunctional signaling glycoproteins that are involved in the regulation of a wide variety of normal and pathologic processes, including embriogenesis, differentiation, and tumorigenesis.8-10 We recently found that the transfection of the metastatic suppressor gene MRP-1/CD9 could downregulate the Wnt5a expression in tumor cell lines.11 In fact, many previous clinical studies have demonstrated that the Wnt5a expression is frequently upregulated in various human cancers, including gastric cancers, esophageal cancers, pancreatic cancers, and breast cancers.12-15
As a result, the overexpression of Wnt5a might affect tumor biology during tumor progression. Thus understanding the biologic mechanisms of Wnt5a could lead to the development of new strategies for the treatment of cancer patients. However, the biologic functions of Wnt5a in human cancers are still unclear. Therefore, we conducted the present study to investigate the clinical significance of Wnt5a expression in NSCLC. Because Wnt5a is considered to be multifunctional, we also evaluated the tumor proliferation rate using the Ki-67 index,16 tumor angiogenesis using CD34 staining,17 and the expression of beta-catenin9 and vascular endothelial growth factor-A (VEGF-A), a potent growth factor18 and one of the target genes of the Wnt family.19
PATIENTS AND METHODS
Clinical Characteristics of Patients
July 1999 to December 2002, consecutive patients with NSCLC who underwent surgery at the Second Department of Surgery of Kagawa University (Kagawa, Japan) were studied. This study was approved by the institutional review board of Kagawa University (14-7, a clinical study of biologic markers in NSCLCs). TNM staging designations were made according to the postsurgical pathologic international staging system.20 The lymph node status was pathologically evaluated using specimens resected by either thoracotomy or mediastinoscopy. In total, 123 patients with lung cancer up to stage IIIB, which included 67 patients with adenocarcinomas, 50 patients with squamous cell carcinomas, and six patients with large-cell carcinomas, were investigated (Table 1). The patients＊ clinical records and histopathologic diagnoses were fully documented. This report includes follow-up data as of December 28, 2004. The median follow-up period for all patients was 41.1 ㊣ 17.3 months.
Regarding methods of surgical resection, a pneumonectomy was performed in 13 patients with stage II to III NSCLCs. A lobectomy was performed in 101 patients: 65 patients with stage I NSCLC, 16 patients with stage II NSCLC, and 20 patients with stage III NSCLC. A segmentectomy was performed in five patients with stage I NSCLC, and a wedge resection was performed in four patients with stage I NSCLC. Systemic chemotherapy using mitomycin, vinblastine, and cisplatin was performed in all patients with stage II to III NSCLCs: neoadjuvant chemotherapy in 27 patients, and postoperative adjuvant chemotherapy in 22 patients with nodal metastases. Radiation therapy was performed in 19 patients: eight patients with T3 or T4 status and 11 patients with mediastinal lymph node metastases.
Real-Time Quantitative Reverse Transcriptase Polymerase Chain Reaction for Wnt5a Gene Expression
Total cellular RNA was extracted from frozen tissue specimens by the acid guanidinium thiocyanate procedure. First-strand cDNA synthesis was performed with 5 米g of total RNA using a cDNA synthesis kit (Pharmacia, Piscataway, NJ) according to the manufacturer＊s protocol. To quantify Wnt5a gene expression, real-time quantitative polymerase chain reaction (PCR) was performed with the ABI PRISM 7700 Sequence Detection System using TaqMan Gene Expression Assays and TaqMan Universal PCR Master Mix (Applied Biosystems, Foster City, CA). Glyceraldehyde 3-phophate dehydrogenase was used as an endogenous control gene to normalize the transcript content in each sample. The PCR cycling conditions for all samples were as follows: 50∼C, 2 minutes for AmpErase uracil-N-glycosylase activation; 95∼C, 10 minutes for AmpliTaq Gold activation; and 40 cycles for the melting (95∼C, 15 seconds) and annealing/extension (60∼C, 1 minute) steps. Real-time PCR assays were conducted in triplicate for each sample, and each PCR experiment included three nontemplate control wells. A standard curve for serial dilutions of cDNA of the Wnt5a-positive human lung cancer cell line A-549 was similarly generated.11 The comparative threshold cycle method (Applied Biosystems) was used to calculate the Wnt5a gene expression ratio in each sample relative to the value observed in the control A549 cells, using glyceraldehyde 3-phophate dehydrogenase as a control for normalization among samples.
Immunohistochemical Assays
We performed immunohistochemical studies to evaluate protein expression of Wnt5a, beta-catenin, and VEGF-A in tumor cells and stromal cells, tumor proliferation rate using the Ki-67 index, and tumor angiogenesis using anti-CD34 staining. The following antibodies were used, along with isotype antibodies as negative controls: a goat polyclonal antibody for Wnt5a (C-16; Santa Cruz Biotechnology, Santa Cruz, CA) diluted at 1:100, a mouse monoclonal antibody for beta-catenin (C19220, Transduction Laboratories, Lexington, KY) diluted at 1:200, a rabbit polyclonal antibody for VEGF-A (A-20, Santa Cruz Biotechnology) diluted at 1:200, a mouse monoclonal antibody for the Ki-67 antigen (MIB-1; DAKO, Glostrup, Denmark) diluted at 1:40, and a mouse monoclonal antibody for CD34 (NU-4A1; Nichirei Corporation, Tokyo, Japan) diluted at 1:10. Formalin-fixed paraffin-embedded tissue was cut into 4-米m sections and mounted on poly-L-lysine-coated slides. Sections were deparaffinized and rehydrated. The slides were then heated in a microwave for 10 minutes in a 10-米mol/L citrate buffer solution at pH 6.0 and cooled to room temperature for 20 minutes. After quenching the endogenous peroxidase activity with 0.3% H2O2 (in absolute methanol) for 30 minutes, the sections were treated for 2 hours at room temperature with 5% bovine serum albumin to block nonspecific staining. Duplicate sections were incubated overnight with the primary specific antibodies detecting Wnt5a, beta-catenin, VEGF-A, Ki-67, and CD34. Slides were then incubated for 1 hour with biotinylated antigoat immunoglobulin G (IgG; Vector Laboratories, Burlingame, CA) for Wnt5a; biotinylated antimouse IgG (Vector Laboratories) for beta-catenin, Ki-67, and CD34; and biotinylated antirabbit IgG (Vector Laboratories) for VEGF-A. The sections were incubated with the avidin-biotin-peroxidase complex (Vector Laboratories) for 1 hour, and antibody binding was visualized with 3,3'-diaminobenzidine tetrahydrochloride. Lastly, the sections were lightly counterstained with Mayer＊s hematoxylin. Sections of resected lung tumors known to express Wnt5a or VEGF-A were used as positive controls for immunohistochemical staining. Normal bronchial epithelium and normal mucosal glands within the tumor section were used as positive controls for the staining of beta-catenin.
All of the immunostained sections were reviewed by two authors (C.H. and M.U.) who had no knowledge of the patients＊ clinical status. Cases with discrepancies were jointly re-evaluated, and a consensus was reached. For protein expression of Wnt5a, beta-catenin, and VEGF-A, in cases with multiple areas of low intensity that occurred during evaluation of immunostaining, five areas were selected at random and scored. Also, one random field was selected in sections where all staining appeared intense. At least 200 cells were scored per x40 field about tumor cells or stromal cells, respectively. Regarding intratumoral beta-catenin expression, we used the classification of staining patterns as reported previously21: (1) a membranous pattern, if immunoreactivity was present solely at the cell membranes; (2) a membranous-cytoplasmic pattern, if immunoreactivity was also present in the cytoplasm; (3) a cytoplasmic pattern, if immunoreactivity was chiefly present in the cytoplasm and in less than 20% of the nuclei; and (4) a cytoplasmic-nuclear pattern, if immunoreactivity was present in the cytoplasm and concomitantly in more than 20% of the nuclei. Regarding stromal beta-catenin expression, stromal cells with cytoplasmic beta-catenin expression and/or nuclear beta-catenin expression were classified as beta-catenin每positive stromal cells. The percentage of carcinoma cells with positive staining for Ki-67 in a given specimen was scored as the Ki-67 proliferation index.16 For microvessel quantification, the three most highly vascularized areas detected by CD34 immunostaining were initially selected under the x40 field, and a x200 field (0.785 mm2 per field) was used to count vessels in each of these areas. The average of three x200 field counts was recorded as the intratumoral microvessel density (IMD).17
Statistical Methods
Because the distributions of nine values, including the normalized Wnt5a gene expression ratio (P = .1415), the percentages of Wnt5a-positve tumor cells (P = .5670), Wnt5a-positive stromal cells (P = .2159), tumor cells with cytoplasmic and/or nuclear staining of beta-catenin (P = .5194), beta-catenin每positive stromal cells (P = .4598), VEGF-A每positive tumor cells (P = .6639), and VEGF-A每positive stromal cells (P = .5597), the Ki-67 proliferation index (P = .8081), and the IMD (P = .4072), all showed normal distributions (Kolmogorov-Smirnov analysis), the statistical significances regarding these values were assessed by the t test, analysis of variance with Bonferroni/Dunn test, or Pearson＊s correlation coefficient. In addition, because the intratumoral Wnt5a protein expression cutoff line of 30% demonstrated the most significance in relation to the Ki-67 proliferation index and the stromal beta-catenin expression, the sample was classified as a Wnt5a-positive tumor when the percentage of Wnt5a-positive tumor cells was more than 30%. Because the stromal VEGF-A expression cutoff line of 30% demonstrated the most significance in relation to the Ki-67 proliferation index, the sample was classified as a stromal VEGF-A每positive tumor when the percentage of VEGF-A每positive stromal cells was more than 30%. Overall survival was defined as the time from treatment initiation (surgical resection, chemotherapy, or radiation) to the date of death from any cause. The Kaplan-Meier method was used to estimate the probability of overall survival as a function of time, and differences in the survival of subgroups of patients were compared using Mantel＊s log-rank test. A multivariate analysis was performed using the Cox proportional hazards regression model to study the effects of different variables on survival. All P values were based on two-tailed statistical analysis, and a P value of less than .05 was considered statistically significant.
RESULTS
Wnt5a Gene Expression in NSCLCs
First, we studied Wnt5a gene expression in 10 noncancerous samples to evaluate the Wnt5a expression in normal lung tissues. The normalized Wnt5a gene expression ratio was low in normal lung tissues (mean, 0.1548 ㊣ 0.0358). In contrast, the normalized Wnt5a gene expression ratio varied greatly among the 123 tumor tissues we studied (mean, 2.464 ㊣ 3.907). Regarding tumor histology, the normalized Wnt5a gene expression ratio was 0.774 ㊣ 0.944 in adenocarcinomas and 4.873 ㊣ 5.153 in squamous cell carcinomas (Fig 1). Wnt5a gene expression in squamous cell carcinomas was significantly higher than that in adenocarcinomas (P < .0001).
Wnt5a Protein Expression in Tumor Cells and Stromal Cells
We next performed immunohistochemistry to evaluate the Wnt5a protein expression in tumor cells and stromal cells, respectively. The Wnt5a protein expression appeared in the form of a cytoplasmic staining pattern (Fig 2A). Regarding the intratumoral Wnt5a protein expression, the percentage of Wnt5a-positive tumor cells varied greatly among the 123 NSCLCs (mean, 41.5% ㊣ 29.6%). In addition, a significant correlation was seen between the normalized Wnt5a gene expression ratio and the percentage of Wnt5a-positive tumor cells evaluated by immunohistochemistry (r = 0.729; P < .0001).
In contrast, regarding the stromal Wnt5a protein expression, the percentage of Wnt5a-positive stromal cells was low (mean, 5.3% ㊣ 4.7%). In addition, no correlation was observed between the normalized Wnt5a gene expression ratio and the percentage of Wnt5a-positive stromal cells (r = 0.095). Furthermore, the Wnt5a protein expression was also low in the normal alveolar epithelium.
We used the classification of the Wnt5a status according to the intratumoral Wnt5a protein expression because this value was considered to be appropriate to evaluate the tumor-stromal interaction. Seventy-one carcinomas (57.7%) were Wnt5a-positive, and 52 carcinomas (42.3%) were Wnt5a-negative (Table 2). Regarding tumor histology, Wnt5a-positive tumors in squamous cell carcinomas were significantly more common than those in adenocarcinomas (78.0% v 41.8%; P < .0001). Regarding the clinicopathologic parameters, no significant difference in the Wnt5a status was observed in relation to tumor status, nodal status, pathologic stage, or tumor differentiation.
Tumor Proliferation and Angiogenesis of NSCLCs in Relation to Wnt5a Status
To investigate the biologic function of Wnt5a in NSCLCs, we evaluated the Ki-67 proliferation index and the IMD in relation to the Wnt5a status (Fig 2B). Regarding tumor proliferation, the percentage of Wnt5a-positive tumor cells was significantly correlated with the Ki-67 proliferation index (r = 0.708; P < .0001). The Ki-67 proliferation index was significantly higher in Wnt5a-positive tumors than in Wnt5a-negative tumors (55.1% ㊣ 15.6% v 31.7% ㊣ 12.1%; P < .0001). However, concerning tumor angiogenesis, no correlation was observed between the percentage of Wnt5a-positive tumor cells and the IMD (r = 0.034). The IMD was 101.3 ㊣ 31.2 in Wnt5a-negative tumors and 114.5 ㊣ 28.8 in Wnt5a-positive tumors.
The Intratumoral Beta-Catenin Expression and the Stromal Beta-Catenin Expression in Relation to the Wnt5a Status
We next studied intratumoral beta-catenin expression and stromal beta-catenin expression in relation to the Wnt5a status (Fig 2). In tumor cells, beta-catenin generally exhibited homogenous staining with four expression patterns.21 Regarding the expression patterns of intratumoral beta-catenin in relation to the Wnt5a status, among 71 Wnt5a-positive tumors, 15 carcinomas (21.1%) had a membranous pattern, 24 carcinomas (33.8%) had a membranous-cytoplasmic pattern, nine carcinomas (12.7%) had a cytoplasmic pattern, and 23 carcinomas (32.4%) had a cytoplasmic-nuclear pattern. Among 52 Wnt5a-negative tumors, 18 carcinomas (34.6%) had a membranous pattern, 17 carcinomas (32.7%) had a membranous-cytoplasmic pattern, six carcinomas (11.5%) had a cytoplasmic pattern, and 11 carcinomas (21.2%) had a cytoplasmic-nuclear pattern. There was no significant difference in the expression patterns of intratumoral beta-catenin according to the Wnt5a status. Furthermore, no correlation was observed between the percentage of Wnt5a-positive tumor cells and the percentage of tumor cells with cytoplasmic and/or nuclear staining of beta-catenin (r = 0.032).
In contrast, the stromal beta-catenin expression appeared in the form of a cytoplasmic staining pattern with varied nuclear staining (Fig 2C). The percentage of beta-catenin每positive stromal cells also varied greatly among the 123 NSCLCs (mean, 25.3% ㊣ 25.4%). Furthermore, a significant correlation was seen between the percentage of Wnt5a-positive tumor cells and the percentage of beta-catenin每positive stromal cells (r = 0.729; P < .0001). The percentage of beta-catenin每positive stromal cells was significantly higher in Wnt5a-positive tumors than in Wnt5a-negative tumors (36.0% ㊣ 25.7% v 10.8% ㊣ 8.5%; P < .0001).
Intratumoral VEGF-A Expression and Stromal VEGF-A Expression in Relation to Wnt5a Status and Stromal Beta-Catenin Status
We next studied intratumoral VEGF-A expression and stromal VEGF-A expression in relation to Wnt5a status and the stromal beta-catenin status (Fig 2D). Regarding the intratumoral VEGF-A expression, no correlation was observed between the percentage of Wnt5a-positive tumor cells and the percentage of VEGF-A每positive tumor cells (r = 0.095). The percentage of VEGF-A每positive tumor cells was 34.1% ㊣ 14.5% in Wnt5a-negative tumors and 35.1% ㊣ 14.6% in Wnt5a-positive tumors.
However, regarding the stromal VEGF-A expression, the percentage of VEGF-A-positive stromal cells was significantly correlated with the percentage of Wnt5a-positive tumor cells (r = 0.661; P < .0001). The percentage of VEGF-A每positive stromal cells was significantly higher in Wnt5a-positive tumors than in Wnt5a-negative tumors (37.6% ㊣ 15.6% v 22.7% ㊣ 10.8%; P = .0005). Furthermore, the percentage of VEGF-A每positive stromal cells was also significantly correlated with the percentage of beta-catenin每positive stromal cells (r = 0.644; P < .0001).
Tumor Proliferation and Angiogenesis of NSCLCs in Relation to Stromal VEGF-A Status
To investigate the biologic function of stromal VEGF-A expression in NSCLCs, we evaluated Ki-67 proliferation index and the IMD in relation to stromal VEGF-A status. Among the 123 NSCLCs we studied, 50 carcinomas (40.7%) were stromal VEGF-A每positive, and 73 carcinomas (59.3%) were stromal VEGF-A每negative. Regarding tumor proliferation, the Ki-67 proliferation index was significantly correlated with the percentage of VEGF-A每positive stromal cells (r = 0.627; P < .0001). The Ki-67 proliferation index was significantly higher in stromal VEGF-A每positive tumors than in stromal VEGF-A每negative tumors (62.6% ㊣ 14.6% v 33.3% ㊣ 13.0%; P < .0001). In contrast, concerning tumor angiogenesis, no correlation was observed between the percentage of VEGF-A每positive stromal cells and the IMD (r = 0.096). The IMD was 104.2 ㊣ 27.6 in stromal VEGF-A每negative tumors and 110.7 ㊣ 34.1 in stromal VEGF-A每positive tumors.
Overall Survival of NSCLC Patients in Relation to Wnt5a Status
Regarding Wnt5a status, the 3-year survival was 75.5% in patients with Wnt5a-negative NSCLCs and 58.1% in patients with Wnt5a-positive NSCLCs. The overall survival was significantly lower in patients with Wnt5a-positive NSCLCs than in those with Wnt5a-negative NSCLCs (P = .0299; Fig 3A). Especially among patients with stage II to III NSCLC, the 3-year survival rate was significantly lower in patients with Wnt5a-positive tumors than in those with Wnt5a-negative tumors (30.2% v 69.2%; P = .0214; Fig 3B). In contrast, no difference was observed in the survival of patients with stage I NSCLC according to Wnt5a status (Fig 3C). Regarding tumor histology, the 3-year survival rate was significantly lower in patients with Wnt5a-positive squamous cell carcinomas than in those with Wnt5a-negative squamous cell carcinomas (50.1% v 90.9%; P = .0282; Fig 3D). On the other hand, no difference was observed in survival of patients with adenocarcinomas according to Wnt5a status (Fig 3E).
As shown in Table 3, a multivariate analysis using the Cox proportional hazards regression model demonstrated that Wnt5a status (hazard ratio, 2.451; P = .0193) and the pathologic stage (hazard ratio, 2.917; P < .0001) were significant prognostic factors for NSCLC patients. Especially, the Wnt5a status (hazard ratio, 3.514; P = .0491) and the pathologic stage (hazard ratio, 2.785; P < .0001) were also significant prognostic factors for patients with squamous cell carcinomas.
DISCUSSION
The Wnt family is a large family of homologous but distinct genes, which have been highly conserved across species in evolutions. The Wnt genes encode secreted cysteine-rich proteins with multidirectional biologic functions via autocrine or paracrine routes.9 Previous experimental studies using cell lines have demonstrated that Wnt5a is also an important regulator of many biologic functions, such as cell proliferation,22-24 differentiation,25,26 and cell motility.27 Furthermore, the Wnt5a expression is frequently upregulated in various human cancers.12-15 As a result, Wnt5a is considered to be involved in tumorigenesis.
However, the clinical significance of Wnt5a expression in human cancers is still unclear and also rather controversial. Wnt5a was initially reported to belong to nontransforming members of Wnts, according to their ability to transform C57MG mammary epithelial cells.28 However, many studies have reported that Wnt5a also can induce cell proliferation in several cell lines and tissues.22-24 In addition, a recent study on breast phyllodes tumors revealed that a Wnt5a overexpression in epithelial cells induced a stromal proliferation, but not an epithelial proliferation.29 These observations support the notion that effects of Wnt signaling are cell type specific. The autocrine and paracrine use of Wnt proteins, together with expression profiles of the Frizzled (Frz) family of Wnt receptors, will ultimately determine the cellular response.23
Therefore, to clarify the clinical significance of the multifunctional Wnt5a in NSCLCs, we conducted the present study on Wnt5a expression in relation to tumor proliferation, tumor angiogenesis, and the expression of beta-catenin9 and VEGF-A, a potent growth factor and one of the target genes of the canonical Wnt/beta-catenin pathway.18,19 Regarding Wnt5a expression, both the Wnt5a gene expression evaluated by reverse transcriptase PCR and the intratumoral Wnt5a protein expression evaluated by immunohistochemistry varied greatly, and a significant correlation was observed between them. In contrast, the Wnt5a protein expression was low in both stromal cells and normal lung tissues. We conducted the present study using the classification according to the intratumoral Wnt5a protein expression evaluated by immunohistochemistry because this value was considered to be appropriate for our investigative purposes to evaluate the tumor-stromal interaction in NSCLCs.
Interestingly, the present study revealed that the intratumoral Wnt5a overexpression was associated with the stromal expression of beta-catenin and VEGF-A, not with the intratumoral expression. This result suggested the existence of a tumor-stromal interaction. Previous studies also reported that Wnt5a was involved in the tumor-stromal interaction in human cancers, including breast phyllodes tumors and colorectal cancers.29,30 He et al31 demonstrated that Wnt5a may also play a role in the canonical Wnt/beta-catenin signaling pathway. The Wnt5a protein can act via Frz-5 receptor to initiate an intracellular pathway leading to the accumulation of beta-catenin.
Beta-catenin is a pivotal component of the Wnt signaling pathway and E-cadherin-associated homotypic cell adhesion.32 The present study revealed the beta-catenin expression levels to differ in tumor cells and in stromal cells, respectively. The beta-catenin expression is membranous in normal epithelium. In contrast, beta-catenin demonstrated four expression patterns in tumor cells. Previous studies reported that the beta-catenin expression in the cytoplasm and/or nuclear could be considered to be an indication for its aberrant expression.21,33 The activation of the Wnt signaling pathway by the stabilization of beta-catenin has been shown to play an important role in the development of colorectal carcinoma, which is mainly caused by inactivating mutations of the adenomatous polyposis coli gene or by activating mutations of the beta-catenin gene.34 However, previous clinical studies have revealed these mutations to be rare in NSCLCs.21,35 Therefore, the aberrant expression of beta-catenin in the cytoplasm and/or the nucleus among NSCLCs might be functional.21 The present study found that intratumoral Wnt5a expression was not associated with the intratumoral beta-catenin expression. This result might be due to the fact that tumor cells do not have a Frz receptor for Wnt5a, as suggested previously.29 Thus the intratumoral aberrant expression of beta-catenin among NSCLCs might be caused by other mechanisms, such as the overexpression of other members of the Wnt family or a reduced E-cadherin expression.21,36
In contrast, the stromal beta-catenin expression appeared in the form of a cytoplasmic staining pattern with varied nuclear staining. The present study revealed that the intratumoral Wnt5a expression was significantly correlated with the stromal beta-catenin expression. Furthermore, the intratumoral Wnt5a expression was also significantly correlated with the stromal VEGF-A expression. Although an immunohistochemistry assay does not precisely represent the source of VEGF-A, a known secreted protein, a significant correlation was seen between the stromal beta-catenin expression and the stromal VEGF-A expression in the present study. This result suggested that stromal cells with a positive beta-catenin expression could produce the VEGF-A expression. Therefore, the Wnt5a protein secreted by tumor cells could activate the canonical Wnt/beta-catenin pathway in stromal cells with a receptor for Wnt5a, thereby inducing the stromal VEGF-A expression. Subsequently, the stromal VEGF-A expression could affect tumor growth via the VEGF-receptor of tumor cells. Further studies including the Frz receptor and the VEGF-receptor should be performed to clarify this mechanism.
Concerning the regulation of Wnt5a expression, both gene amplification and gene rearrangement have been reported to occur infrequently in human cancers.37 In contrast, experimental studies using cell lines demonstrated the Wnt5a expression to be regulated by various molecules, including hepatocyte growth factor, protein kinase C activity, and MRP-1/CD9.11,38,39 Our pilot study found that there was no significant association between a reduced expression of MRP-1/CD9 and Wnt5a overexpression in NSCLCs (data not shown). Therefore, the Wnt5a expression might be secondarily regulated in response to a range of changes in these many biologic molecules during tumor progression.
Consequently, the present study revealed that Wnt5a expression in squamous cell carcinomas was significantly higher than that in adenocarcinomas. Regarding tumor biology, an overexpression of Wnt5a was associated with the Ki-67 proliferation index. Although the Ki-67 proliferation index was relatively high in the present study, this result might be due to epidermiologic difference. In addition, an overexpression of Wnt5a was associated with a poor prognosis in NSCLC patients, especially in stage II to III (locally advanced) NSCLCs. Our recent study has also shown that biologic markers associated with tumor cell proliferation could be prognostic factors in stage II to III NSCLCs.40 Furthermore, an overexpression of Wnt5a was associated with a poor prognosis, especially in squamous cell carcinomas. To our knowledge, the present study is the first clinical report demonstrating the clinical significance of intratumoral Wnt5a expression in NSCLCs. In particular, an overexpression of Wnt5a could affect the progression of squamous cell carcinoma of the lung.
In conclusion, the present study demonstrated that an overexpression of Wnt5a was associated with the tumor cell proliferation in NSCLCs. Furthermore, the intratumoral Wnt5a overexpression was also found to be associated with expression of beta-catenin and VEGF-A in stromal cells, which suggests the existence of a tumor-stromal interaction. In addition, the overexpression of Wnt5a is considered to be a significant prognostic factor for poor prognosis in NSCLC patients, especially in squamous cell carcinomas. Therefore, new strategies, such as the RNA inhibition of Wnt5a, secreted Frizzled-related proteins, and bevacizumab (Avastin; Genentech, Inc, South San Francisco, CA), may be potentially effective treatments for patients with Wnt5a-positive NSCLCs.41-43
Authors' Disclosures of Potential Conflicts of Interest
The authors indicated no potential conflicts of interest.
NOTES
Authors' disclosures of potential conflicts of interest are found at the end of this article.
REFERENCES
Cersosimo RJ: Lung cancer: A review. Am J Health Syst Pharm 59:611-642, 2002
Arriagada R, Bergman B, Dunant A, et al: Cisplatin-based adjuvant chemotherapy in patients with completely resected non-small-cell lung cancer. N Engl J Med 350:351-360, 2004
Martin J, Ginsberg RJ, Venkatraman ES, et al: Long-term results of combined-modality therapy in resectable non-small-cell lung cancer. J Clin Oncol 20:1989-1995, 2002
Weir B, Zhao X, Meyerson M: Somatic alterations in the human cancer genome. Cancer Cell 6:433-438, 2004
Imoto H, Osaki T, Taga S, et al: Vascular endothelial growth factor expression in non-small-cell lung cancer: Prognostic significance in squamous cell carcinoma. J Thorac Cardiovasc Surg 115:1007-1014, 1998
Adachi M, Taki T, Konishi T, et al: Novel staging protocol for non-small-cell lung cancers according to MRP-1/CD9 and KAI1/CD82 gene expression. J Clin Oncol 16:1397-1406, 1998
Kataoka H, Tanaka H, Nagaike K, et al: Role of cancer cell-stroma interaction in invasive growth of cancer cells. Hum Cell 16:1-14, 2003
Herzlinger D, Qiao J, Cohen D, et al: Induction of kidney epithelial morphogenesis by cells expressing Wnt-1. Dev Biol 166:815-818, 1994
Dale TC: Signal transduction by the Wnt family of ligands. Biochem J 329:209-223, 1998
You Z, Saims D, Chen S, et al: Wnt signaling promotes oncogenic transformation by inhibiting c-Myc-induced apoptosis. J Cell Biol 157:429-440, 2002
Huang C, Liu D, Masuya D, et al: MRP-1/CD9 gene transduction downregulates Wnt signal pathways. Oncogene 23:7475-7483, 2004
Saitoh T, Mine T, Katoh M: Frequent up-regulation of WNT5A mRNA in primary gastric cancer. Int J Mol Med 9:515-519, 2002
Saitoh T, Katoh M: Expression and regulation of Wnt5A and Wnt5B in human cancer: Up-regulation of Wnt5A by TNF in MKN45 cells and up-regulation of Wnt5B by beta-estradiol in MCF-7 cells. Int J Mol Med 10:345-349, 2002
Lejeune S, Huguet EL, Hamby A, et al: Wnt5a cloning, expression, and up-regulation in human primary breast cancers. Clin Cancer Res 1:215-222, 1995
Saldanha G, Ghura V, Potter L, et al: Nuclear beta-catenin in basal cell carcinoma correlates with increased proliferation. Br J Dermatol 151:157-164, 2004
Scagliotti GV, Micela M, Gubetta L, et al: Prognostic significance of Ki67 labeling in resected non small cell lung cancer. Eur J Cancer 29A:363-365, 1993
Masuya D, Huang C, Liu D, et al: The intratumoral expression of vascular endothelial growth factor and interleukin-8 associated with angiogenesis in nonsmall cell lung carcinoma patients. Cancer 92:2628-2638, 2001
Dvorak HF, Brown LF, Detmar M, et al: Vascular permeability factor/vascular endothelial growth factor, microvascular hyperpermeability, and angiogenesis. Am J Pathol 146:1029-1039, 1995
Zhang X, Gaspard JP, Chung DC: Regulation of vascular endothelial growth factor by the Wnt and K-ras pathways in colonic neoplasia. Cancer Res 61:6050-6054, 2001
Mountain CF: Revisions in the international system for staging lung cancer. Chest 111:1710-1717, 1997
Kotsinas A, Evangelou K, Zacharatos P, et al: Proliferation, but not apoptosis, is associated with distinct beta-catenin expression patterns in non-small-cell lung carcinomas. Am J Pathol 161:1619-1634, 2002
Yamaguchi TP, Bradley A, McMahon AP, et al: A Wnt5a pathway underlies outgrowth of multiple structures in the vertebrate embryo. Development 126:1211-1223, 1999
Murdoch B, Chadwick K, Martin M, et al: Wnt-5A augments repopulating capacity and primitive hematopoietic development of human blood stem cells in vivo. Proc Natl Acad Sci U S A 100:3422-3427, 2003
Yang Y, Topol L, Lee H, et al: Wnt5a and Wnt5b exhibit distinct activities in coordinating chondrocyte proliferation and differentiation. Development 130:1003-1015, 2003
Taki M, Kamata N, Yokoyama K, et al: Down-regulation of Wnt-4 and up-regulation of Wnt-5a expression by epithelial-mesenchymal transition in human squamous carcinoma cells. Cancer Sci 94:593-597, 2003
Li C, Xiao J, Hormi K, et al: Wnt5a participates in distal lung morphogenesis. Dev Biol 248:68-81, 2002
Weeraratna AT, Jiang Y, Hostetter G, et al: Wnt5a signaling directly affects cell motility and invasion of metastatic melanoma. Cancer Cell 1:279-288, 2002
Shimizu H, Julius MA, Giarre M, et al: Transformation by Wnt family proteins correlates with regulation of beta-catenin. Cell Growth Differ 8:1349-1358, 1997
Sawyer EJ, Hanby AM, Rowan AJ, et al: The Wnt pathway, epithelial-stromal interactions, and malignant progression in phyllodes tumours. J Pathol 196:437-444, 2002
Smith K, Bui TD, Poulsom R, et al: Up-regulation of macrophage Wnt gene expression in adenoma-carcinoma progression of human colorectal cancer. Br J Cancer 81:496-502, 1999
He X, Saint-Jeannet J, Wang Y, et al: A member of the frizzled protein family mediating axis induction by Wnt-5A. Science 275:1652-1654, 1997
Ben-Ze'ev A, Shtutman M, Zhurinsky J: The integration of cell adhesion with gene expression: The role of beta-catenin. Exp Cell Res 261:75-82, 2000
Pirinen RT, Hirvikoski P, Johansson RT, et al: Reduced expression of alpha-catenin, beta-catenin, and gamma-catenin is associated with high cell proliferative activity and poor differentiation in non-small cell lung cancer. J Clin Pathol 54:391-395, 2001
Bienz M, Clevers H: Linking colorectal cancer to Wnt signaling. Cell 103:311-320, 2000
Ueda M, Gemmill RM, West J, et al: Mutations of the beta- and gamma-catenin genes are uncommon in human lung, breast, kidney, cervical and ovarian carcinomas. Br J Cancer 85:64-68, 2001
Kase S, Sugio K, Yamazaki K, et al: Expression of E-cadherin and beta-catenin in human non-small cell lung cancer and the clinical significance. Clin Cancer Res 6:4789-4796, 2000
Iozzo RV, Eichstetter I, Danielson KG: Aberrant expression of the growth factor Wnt-5A in human malignancy. Cancer Res 55:3495-3499, 1995
Huguet EL, Smith K, Bicknell R, et al: Regulation of Wnt5a mRNA expression in human mammary epithelial cells by cell shape, confluence, and hepatocyte growth factor. J Biol Chem 270:12851-12856, 1995
Jonsson M, Smith K, Harris AL: Regulation of Wnt5a expression in human mammary cells by protein kinase C activity and the cytoskeleton. Br J Cancer 78:430-438, 1998
Huang C, Liu D, Masuya D, et al: Clinical application of biological markers for treatments of resectable non-small-cell lung cancers. Br J Cancer 92:1231-1239, 2005
You L, He B, Uematsu K, et al: Inhibition of Wnt-1 signaling induces apoptosis in beta-catenin-deficient mesothelioma cells. Cancer Res 64:3474-3478, 2004
He B, Lee AY, Dadfarmay S, et al: Secreted Frizzled-related protein 4 is silenced by hypermethylation and induces apoptosis in beta-catenin-deficient human mesothelioma cells. Cancer Res 65:743-748, 2005
Johnson DH, Fehrenbacher L, Novotny WF, et al: Randomized phase II trial comparing bevacizumab plus carboplatin and paclitaxel with carboplatin and paclitaxel alone in previously untreated locally advanced or metastatic non-small-cell lung cancer. J Clin Oncol 22:2184-2191, 2004(Cheng-long Huang, Dage Li)
ABSTRACT
PURPOSE: The Wnt gene family encodes the multifunctional signaling glycoproteins. We performed the present study to investigate the clinical significance of Wnt5a expression in non每small-cell lung cancer (NSCLC).
PATIENTS AND METHODS: One hundred twenty-three patients with NSCLC who had undergone resection were investigated. Real-time quantitative reverse transcriptase polymerase chain reaction was performed to evaluate the Wnt5a gene expression. Immunohistochemistry was performed to investigate the Wnt5a protein expression, the Ki-67 proliferation index, tumor angiogenesis, and the expression of beta-catenin and vascular endothelial growth factor-A (VEGF-A).
RESULTS: Wnt5a gene expression in squamous cell carcinoma was significantly higher than that in adenocarcinoma (P < .0001). There was a significant correlation between the normalized Wnt5a gene expression ratio and the intratumoral Wnt5a protein expression (r = 0.729; P < .0001). The intratumoral Wnt5a expression was significantly correlated with the Ki-67 proliferation index (r = 0.708; P < .0001). In contrast, no correlation was observed between the intratumoral Wnt5a expression and tumor angiogenesis. Furthermore, the intratumoral Wnt5a expression was significantly correlated with the stromal expression of beta-catenin (r = 0.729; P < .0001) and VEGF-A (r = 0.661; P < .0001). In addition, the stromal VEGF-A expression was also correlated with Ki-67 proliferation (r = 0.627; P < .0001). Cox regression analyses demonstrated Wnt5a status to be a significant prognostic factor for NSCLC patients (P = .0193), especially for patients with squamous cell carcinomas (P = .0491).
CONCLUSION: The present study revealed that an overexpression of Wnt5a could produce more aggressive NSCLC, especially in squamous cell carcinomas, during tumor progression.
INTRODUCTION
Non每small-cell lung cancer (NSCLC) is one of the most common human cancers with a poor prognosis. Surgical resections play a major role in managing patients with stage I and II NSCLCs, and surgery may also sometimes be used for patients with stage III NSCLCs.1 In addition, various concomitant treatments with chemotherapy and radiotherapy have been assessed for improving the outcome of patients with NSCLC.2 However, the 3-year survival rate is 35% to 55% in patients with stage II NSCLC and 28% in patients with stage III NSCLC.3 Therefore, it is considered important to clarify the mechanism of tumor biology to improve the clinical outcome of patients with NSCLC.
Recent studies on molecular biology in human cancers have revealed that many molecules affect various biologic behaviors of malignant tumors. The activation of oncogenes or the inactivation of tumor suppressor genes could initially cause carcinogenesis.4 During the subsequent tumor progression, many molecular markers associated with angiogenesis5 or metastasis6 could produce more aggressive malignant tumors. Furthermore, recent studies have indicated that the tumor-stromal interaction also plays an important role in tumor progression.7
Among the many molecular markers associated with tumor progression, the Wnt gene family has been shown to encode the multifunctional signaling glycoproteins that are involved in the regulation of a wide variety of normal and pathologic processes, including embriogenesis, differentiation, and tumorigenesis.8-10 We recently found that the transfection of the metastatic suppressor gene MRP-1/CD9 could downregulate the Wnt5a expression in tumor cell lines.11 In fact, many previous clinical studies have demonstrated that the Wnt5a expression is frequently upregulated in various human cancers, including gastric cancers, esophageal cancers, pancreatic cancers, and breast cancers.12-15
As a result, the overexpression of Wnt5a might affect tumor biology during tumor progression. Thus understanding the biologic mechanisms of Wnt5a could lead to the development of new strategies for the treatment of cancer patients. However, the biologic functions of Wnt5a in human cancers are still unclear. Therefore, we conducted the present study to investigate the clinical significance of Wnt5a expression in NSCLC. Because Wnt5a is considered to be multifunctional, we also evaluated the tumor proliferation rate using the Ki-67 index,16 tumor angiogenesis using CD34 staining,17 and the expression of beta-catenin9 and vascular endothelial growth factor-A (VEGF-A), a potent growth factor18 and one of the target genes of the Wnt family.19
PATIENTS AND METHODS
Clinical Characteristics of Patients
July 1999 to December 2002, consecutive patients with NSCLC who underwent surgery at the Second Department of Surgery of Kagawa University (Kagawa, Japan) were studied. This study was approved by the institutional review board of Kagawa University (14-7, a clinical study of biologic markers in NSCLCs). TNM staging designations were made according to the postsurgical pathologic international staging system.20 The lymph node status was pathologically evaluated using specimens resected by either thoracotomy or mediastinoscopy. In total, 123 patients with lung cancer up to stage IIIB, which included 67 patients with adenocarcinomas, 50 patients with squamous cell carcinomas, and six patients with large-cell carcinomas, were investigated (Table 1). The patients＊ clinical records and histopathologic diagnoses were fully documented. This report includes follow-up data as of December 28, 2004. The median follow-up period for all patients was 41.1 ㊣ 17.3 months.
Regarding methods of surgical resection, a pneumonectomy was performed in 13 patients with stage II to III NSCLCs. A lobectomy was performed in 101 patients: 65 patients with stage I NSCLC, 16 patients with stage II NSCLC, and 20 patients with stage III NSCLC. A segmentectomy was performed in five patients with stage I NSCLC, and a wedge resection was performed in four patients with stage I NSCLC. Systemic chemotherapy using mitomycin, vinblastine, and cisplatin was performed in all patients with stage II to III NSCLCs: neoadjuvant chemotherapy in 27 patients, and postoperative adjuvant chemotherapy in 22 patients with nodal metastases. Radiation therapy was performed in 19 patients: eight patients with T3 or T4 status and 11 patients with mediastinal lymph node metastases.
Real-Time Quantitative Reverse Transcriptase Polymerase Chain Reaction for Wnt5a Gene Expression
Total cellular RNA was extracted from frozen tissue specimens by the acid guanidinium thiocyanate procedure. First-strand cDNA synthesis was performed with 5 米g of total RNA using a cDNA synthesis kit (Pharmacia, Piscataway, NJ) according to the manufacturer＊s protocol. To quantify Wnt5a gene expression, real-time quantitative polymerase chain reaction (PCR) was performed with the ABI PRISM 7700 Sequence Detection System using TaqMan Gene Expression Assays and TaqMan Universal PCR Master Mix (Applied Biosystems, Foster City, CA). Glyceraldehyde 3-phophate dehydrogenase was used as an endogenous control gene to normalize the transcript content in each sample. The PCR cycling conditions for all samples were as follows: 50∼C, 2 minutes for AmpErase uracil-N-glycosylase activation; 95∼C, 10 minutes for AmpliTaq Gold activation; and 40 cycles for the melting (95∼C, 15 seconds) and annealing/extension (60∼C, 1 minute) steps. Real-time PCR assays were conducted in triplicate for each sample, and each PCR experiment included three nontemplate control wells. A standard curve for serial dilutions of cDNA of the Wnt5a-positive human lung cancer cell line A-549 was similarly generated.11 The comparative threshold cycle method (Applied Biosystems) was used to calculate the Wnt5a gene expression ratio in each sample relative to the value observed in the control A549 cells, using glyceraldehyde 3-phophate dehydrogenase as a control for normalization among samples.
Immunohistochemical Assays
We performed immunohistochemical studies to evaluate protein expression of Wnt5a, beta-catenin, and VEGF-A in tumor cells and stromal cells, tumor proliferation rate using the Ki-67 index, and tumor angiogenesis using anti-CD34 staining. The following antibodies were used, along with isotype antibodies as negative controls: a goat polyclonal antibody for Wnt5a (C-16; Santa Cruz Biotechnology, Santa Cruz, CA) diluted at 1:100, a mouse monoclonal antibody for beta-catenin (C19220, Transduction Laboratories, Lexington, KY) diluted at 1:200, a rabbit polyclonal antibody for VEGF-A (A-20, Santa Cruz Biotechnology) diluted at 1:200, a mouse monoclonal antibody for the Ki-67 antigen (MIB-1; DAKO, Glostrup, Denmark) diluted at 1:40, and a mouse monoclonal antibody for CD34 (NU-4A1; Nichirei Corporation, Tokyo, Japan) diluted at 1:10. Formalin-fixed paraffin-embedded tissue was cut into 4-米m sections and mounted on poly-L-lysine-coated slides. Sections were deparaffinized and rehydrated. The slides were then heated in a microwave for 10 minutes in a 10-米mol/L citrate buffer solution at pH 6.0 and cooled to room temperature for 20 minutes. After quenching the endogenous peroxidase activity with 0.3% H2O2 (in absolute methanol) for 30 minutes, the sections were treated for 2 hours at room temperature with 5% bovine serum albumin to block nonspecific staining. Duplicate sections were incubated overnight with the primary specific antibodies detecting Wnt5a, beta-catenin, VEGF-A, Ki-67, and CD34. Slides were then incubated for 1 hour with biotinylated antigoat immunoglobulin G (IgG; Vector Laboratories, Burlingame, CA) for Wnt5a; biotinylated antimouse IgG (Vector Laboratories) for beta-catenin, Ki-67, and CD34; and biotinylated antirabbit IgG (Vector Laboratories) for VEGF-A. The sections were incubated with the avidin-biotin-peroxidase complex (Vector Laboratories) for 1 hour, and antibody binding was visualized with 3,3'-diaminobenzidine tetrahydrochloride. Lastly, the sections were lightly counterstained with Mayer＊s hematoxylin. Sections of resected lung tumors known to express Wnt5a or VEGF-A were used as positive controls for immunohistochemical staining. Normal bronchial epithelium and normal mucosal glands within the tumor section were used as positive controls for the staining of beta-catenin.
All of the immunostained sections were reviewed by two authors (C.H. and M.U.) who had no knowledge of the patients＊ clinical status. Cases with discrepancies were jointly re-evaluated, and a consensus was reached. For protein expression of Wnt5a, beta-catenin, and VEGF-A, in cases with multiple areas of low intensity that occurred during evaluation of immunostaining, five areas were selected at random and scored. Also, one random field was selected in sections where all staining appeared intense. At least 200 cells were scored per x40 field about tumor cells or stromal cells, respectively. Regarding intratumoral beta-catenin expression, we used the classification of staining patterns as reported previously21: (1) a membranous pattern, if immunoreactivity was present solely at the cell membranes; (2) a membranous-cytoplasmic pattern, if immunoreactivity was also present in the cytoplasm; (3) a cytoplasmic pattern, if immunoreactivity was chiefly present in the cytoplasm and in less than 20% of the nuclei; and (4) a cytoplasmic-nuclear pattern, if immunoreactivity was present in the cytoplasm and concomitantly in more than 20% of the nuclei. Regarding stromal beta-catenin expression, stromal cells with cytoplasmic beta-catenin expression and/or nuclear beta-catenin expression were classified as beta-catenin每positive stromal cells. The percentage of carcinoma cells with positive staining for Ki-67 in a given specimen was scored as the Ki-67 proliferation index.16 For microvessel quantification, the three most highly vascularized areas detected by CD34 immunostaining were initially selected under the x40 field, and a x200 field (0.785 mm2 per field) was used to count vessels in each of these areas. The average of three x200 field counts was recorded as the intratumoral microvessel density (IMD).17
Statistical Methods
Because the distributions of nine values, including the normalized Wnt5a gene expression ratio (P = .1415), the percentages of Wnt5a-positve tumor cells (P = .5670), Wnt5a-positive stromal cells (P = .2159), tumor cells with cytoplasmic and/or nuclear staining of beta-catenin (P = .5194), beta-catenin每positive stromal cells (P = .4598), VEGF-A每positive tumor cells (P = .6639), and VEGF-A每positive stromal cells (P = .5597), the Ki-67 proliferation index (P = .8081), and the IMD (P = .4072), all showed normal distributions (Kolmogorov-Smirnov analysis), the statistical significances regarding these values were assessed by the t test, analysis of variance with Bonferroni/Dunn test, or Pearson＊s correlation coefficient. In addition, because the intratumoral Wnt5a protein expression cutoff line of 30% demonstrated the most significance in relation to the Ki-67 proliferation index and the stromal beta-catenin expression, the sample was classified as a Wnt5a-positive tumor when the percentage of Wnt5a-positive tumor cells was more than 30%. Because the stromal VEGF-A expression cutoff line of 30% demonstrated the most significance in relation to the Ki-67 proliferation index, the sample was classified as a stromal VEGF-A每positive tumor when the percentage of VEGF-A每positive stromal cells was more than 30%. Overall survival was defined as the time from treatment initiation (surgical resection, chemotherapy, or radiation) to the date of death from any cause. The Kaplan-Meier method was used to estimate the probability of overall survival as a function of time, and differences in the survival of subgroups of patients were compared using Mantel＊s log-rank test. A multivariate analysis was performed using the Cox proportional hazards regression model to study the effects of different variables on survival. All P values were based on two-tailed statistical analysis, and a P value of less than .05 was considered statistically significant.
RESULTS
Wnt5a Gene Expression in NSCLCs
First, we studied Wnt5a gene expression in 10 noncancerous samples to evaluate the Wnt5a expression in normal lung tissues. The normalized Wnt5a gene expression ratio was low in normal lung tissues (mean, 0.1548 ㊣ 0.0358). In contrast, the normalized Wnt5a gene expression ratio varied greatly among the 123 tumor tissues we studied (mean, 2.464 ㊣ 3.907). Regarding tumor histology, the normalized Wnt5a gene expression ratio was 0.774 ㊣ 0.944 in adenocarcinomas and 4.873 ㊣ 5.153 in squamous cell carcinomas (Fig 1). Wnt5a gene expression in squamous cell carcinomas was significantly higher than that in adenocarcinomas (P < .0001).
Wnt5a Protein Expression in Tumor Cells and Stromal Cells
We next performed immunohistochemistry to evaluate the Wnt5a protein expression in tumor cells and stromal cells, respectively. The Wnt5a protein expression appeared in the form of a cytoplasmic staining pattern (Fig 2A). Regarding the intratumoral Wnt5a protein expression, the percentage of Wnt5a-positive tumor cells varied greatly among the 123 NSCLCs (mean, 41.5% ㊣ 29.6%). In addition, a significant correlation was seen between the normalized Wnt5a gene expression ratio and the percentage of Wnt5a-positive tumor cells evaluated by immunohistochemistry (r = 0.729; P < .0001).
In contrast, regarding the stromal Wnt5a protein expression, the percentage of Wnt5a-positive stromal cells was low (mean, 5.3% ㊣ 4.7%). In addition, no correlation was observed between the normalized Wnt5a gene expression ratio and the percentage of Wnt5a-positive stromal cells (r = 0.095). Furthermore, the Wnt5a protein expression was also low in the normal alveolar epithelium.
We used the classification of the Wnt5a status according to the intratumoral Wnt5a protein expression because this value was considered to be appropriate to evaluate the tumor-stromal interaction. Seventy-one carcinomas (57.7%) were Wnt5a-positive, and 52 carcinomas (42.3%) were Wnt5a-negative (Table 2). Regarding tumor histology, Wnt5a-positive tumors in squamous cell carcinomas were significantly more common than those in adenocarcinomas (78.0% v 41.8%; P < .0001). Regarding the clinicopathologic parameters, no significant difference in the Wnt5a status was observed in relation to tumor status, nodal status, pathologic stage, or tumor differentiation.
Tumor Proliferation and Angiogenesis of NSCLCs in Relation to Wnt5a Status
To investigate the biologic function of Wnt5a in NSCLCs, we evaluated the Ki-67 proliferation index and the IMD in relation to the Wnt5a status (Fig 2B). Regarding tumor proliferation, the percentage of Wnt5a-positive tumor cells was significantly correlated with the Ki-67 proliferation index (r = 0.708; P < .0001). The Ki-67 proliferation index was significantly higher in Wnt5a-positive tumors than in Wnt5a-negative tumors (55.1% ㊣ 15.6% v 31.7% ㊣ 12.1%; P < .0001). However, concerning tumor angiogenesis, no correlation was observed between the percentage of Wnt5a-positive tumor cells and the IMD (r = 0.034). The IMD was 101.3 ㊣ 31.2 in Wnt5a-negative tumors and 114.5 ㊣ 28.8 in Wnt5a-positive tumors.
The Intratumoral Beta-Catenin Expression and the Stromal Beta-Catenin Expression in Relation to the Wnt5a Status
We next studied intratumoral beta-catenin expression and stromal beta-catenin expression in relation to the Wnt5a status (Fig 2). In tumor cells, beta-catenin generally exhibited homogenous staining with four expression patterns.21 Regarding the expression patterns of intratumoral beta-catenin in relation to the Wnt5a status, among 71 Wnt5a-positive tumors, 15 carcinomas (21.1%) had a membranous pattern, 24 carcinomas (33.8%) had a membranous-cytoplasmic pattern, nine carcinomas (12.7%) had a cytoplasmic pattern, and 23 carcinomas (32.4%) had a cytoplasmic-nuclear pattern. Among 52 Wnt5a-negative tumors, 18 carcinomas (34.6%) had a membranous pattern, 17 carcinomas (32.7%) had a membranous-cytoplasmic pattern, six carcinomas (11.5%) had a cytoplasmic pattern, and 11 carcinomas (21.2%) had a cytoplasmic-nuclear pattern. There was no significant difference in the expression patterns of intratumoral beta-catenin according to the Wnt5a status. Furthermore, no correlation was observed between the percentage of Wnt5a-positive tumor cells and the percentage of tumor cells with cytoplasmic and/or nuclear staining of beta-catenin (r = 0.032).
In contrast, the stromal beta-catenin expression appeared in the form of a cytoplasmic staining pattern with varied nuclear staining (Fig 2C). The percentage of beta-catenin每positive stromal cells also varied greatly among the 123 NSCLCs (mean, 25.3% ㊣ 25.4%). Furthermore, a significant correlation was seen between the percentage of Wnt5a-positive tumor cells and the percentage of beta-catenin每positive stromal cells (r = 0.729; P < .0001). The percentage of beta-catenin每positive stromal cells was significantly higher in Wnt5a-positive tumors than in Wnt5a-negative tumors (36.0% ㊣ 25.7% v 10.8% ㊣ 8.5%; P < .0001).
Intratumoral VEGF-A Expression and Stromal VEGF-A Expression in Relation to Wnt5a Status and Stromal Beta-Catenin Status
We next studied intratumoral VEGF-A expression and stromal VEGF-A expression in relation to Wnt5a status and the stromal beta-catenin status (Fig 2D). Regarding the intratumoral VEGF-A expression, no correlation was observed between the percentage of Wnt5a-positive tumor cells and the percentage of VEGF-A每positive tumor cells (r = 0.095). The percentage of VEGF-A每positive tumor cells was 34.1% ㊣ 14.5% in Wnt5a-negative tumors and 35.1% ㊣ 14.6% in Wnt5a-positive tumors.
However, regarding the stromal VEGF-A expression, the percentage of VEGF-A-positive stromal cells was significantly correlated with the percentage of Wnt5a-positive tumor cells (r = 0.661; P < .0001). The percentage of VEGF-A每positive stromal cells was significantly higher in Wnt5a-positive tumors than in Wnt5a-negative tumors (37.6% ㊣ 15.6% v 22.7% ㊣ 10.8%; P = .0005). Furthermore, the percentage of VEGF-A每positive stromal cells was also significantly correlated with the percentage of beta-catenin每positive stromal cells (r = 0.644; P < .0001).
Tumor Proliferation and Angiogenesis of NSCLCs in Relation to Stromal VEGF-A Status
To investigate the biologic function of stromal VEGF-A expression in NSCLCs, we evaluated Ki-67 proliferation index and the IMD in relation to stromal VEGF-A status. Among the 123 NSCLCs we studied, 50 carcinomas (40.7%) were stromal VEGF-A每positive, and 73 carcinomas (59.3%) were stromal VEGF-A每negative. Regarding tumor proliferation, the Ki-67 proliferation index was significantly correlated with the percentage of VEGF-A每positive stromal cells (r = 0.627; P < .0001). The Ki-67 proliferation index was significantly higher in stromal VEGF-A每positive tumors than in stromal VEGF-A每negative tumors (62.6% ㊣ 14.6% v 33.3% ㊣ 13.0%; P < .0001). In contrast, concerning tumor angiogenesis, no correlation was observed between the percentage of VEGF-A每positive stromal cells and the IMD (r = 0.096). The IMD was 104.2 ㊣ 27.6 in stromal VEGF-A每negative tumors and 110.7 ㊣ 34.1 in stromal VEGF-A每positive tumors.
Overall Survival of NSCLC Patients in Relation to Wnt5a Status
Regarding Wnt5a status, the 3-year survival was 75.5% in patients with Wnt5a-negative NSCLCs and 58.1% in patients with Wnt5a-positive NSCLCs. The overall survival was significantly lower in patients with Wnt5a-positive NSCLCs than in those with Wnt5a-negative NSCLCs (P = .0299; Fig 3A). Especially among patients with stage II to III NSCLC, the 3-year survival rate was significantly lower in patients with Wnt5a-positive tumors than in those with Wnt5a-negative tumors (30.2% v 69.2%; P = .0214; Fig 3B). In contrast, no difference was observed in the survival of patients with stage I NSCLC according to Wnt5a status (Fig 3C). Regarding tumor histology, the 3-year survival rate was significantly lower in patients with Wnt5a-positive squamous cell carcinomas than in those with Wnt5a-negative squamous cell carcinomas (50.1% v 90.9%; P = .0282; Fig 3D). On the other hand, no difference was observed in survival of patients with adenocarcinomas according to Wnt5a status (Fig 3E).
As shown in Table 3, a multivariate analysis using the Cox proportional hazards regression model demonstrated that Wnt5a status (hazard ratio, 2.451; P = .0193) and the pathologic stage (hazard ratio, 2.917; P < .0001) were significant prognostic factors for NSCLC patients. Especially, the Wnt5a status (hazard ratio, 3.514; P = .0491) and the pathologic stage (hazard ratio, 2.785; P < .0001) were also significant prognostic factors for patients with squamous cell carcinomas.
DISCUSSION
The Wnt family is a large family of homologous but distinct genes, which have been highly conserved across species in evolutions. The Wnt genes encode secreted cysteine-rich proteins with multidirectional biologic functions via autocrine or paracrine routes.9 Previous experimental studies using cell lines have demonstrated that Wnt5a is also an important regulator of many biologic functions, such as cell proliferation,22-24 differentiation,25,26 and cell motility.27 Furthermore, the Wnt5a expression is frequently upregulated in various human cancers.12-15 As a result, Wnt5a is considered to be involved in tumorigenesis.
However, the clinical significance of Wnt5a expression in human cancers is still unclear and also rather controversial. Wnt5a was initially reported to belong to nontransforming members of Wnts, according to their ability to transform C57MG mammary epithelial cells.28 However, many studies have reported that Wnt5a also can induce cell proliferation in several cell lines and tissues.22-24 In addition, a recent study on breast phyllodes tumors revealed that a Wnt5a overexpression in epithelial cells induced a stromal proliferation, but not an epithelial proliferation.29 These observations support the notion that effects of Wnt signaling are cell type specific. The autocrine and paracrine use of Wnt proteins, together with expression profiles of the Frizzled (Frz) family of Wnt receptors, will ultimately determine the cellular response.23
Therefore, to clarify the clinical significance of the multifunctional Wnt5a in NSCLCs, we conducted the present study on Wnt5a expression in relation to tumor proliferation, tumor angiogenesis, and the expression of beta-catenin9 and VEGF-A, a potent growth factor and one of the target genes of the canonical Wnt/beta-catenin pathway.18,19 Regarding Wnt5a expression, both the Wnt5a gene expression evaluated by reverse transcriptase PCR and the intratumoral Wnt5a protein expression evaluated by immunohistochemistry varied greatly, and a significant correlation was observed between them. In contrast, the Wnt5a protein expression was low in both stromal cells and normal lung tissues. We conducted the present study using the classification according to the intratumoral Wnt5a protein expression evaluated by immunohistochemistry because this value was considered to be appropriate for our investigative purposes to evaluate the tumor-stromal interaction in NSCLCs.
Interestingly, the present study revealed that the intratumoral Wnt5a overexpression was associated with the stromal expression of beta-catenin and VEGF-A, not with the intratumoral expression. This result suggested the existence of a tumor-stromal interaction. Previous studies also reported that Wnt5a was involved in the tumor-stromal interaction in human cancers, including breast phyllodes tumors and colorectal cancers.29,30 He et al31 demonstrated that Wnt5a may also play a role in the canonical Wnt/beta-catenin signaling pathway. The Wnt5a protein can act via Frz-5 receptor to initiate an intracellular pathway leading to the accumulation of beta-catenin.
Beta-catenin is a pivotal component of the Wnt signaling pathway and E-cadherin-associated homotypic cell adhesion.32 The present study revealed the beta-catenin expression levels to differ in tumor cells and in stromal cells, respectively. The beta-catenin expression is membranous in normal epithelium. In contrast, beta-catenin demonstrated four expression patterns in tumor cells. Previous studies reported that the beta-catenin expression in the cytoplasm and/or nuclear could be considered to be an indication for its aberrant expression.21,33 The activation of the Wnt signaling pathway by the stabilization of beta-catenin has been shown to play an important role in the development of colorectal carcinoma, which is mainly caused by inactivating mutations of the adenomatous polyposis coli gene or by activating mutations of the beta-catenin gene.34 However, previous clinical studies have revealed these mutations to be rare in NSCLCs.21,35 Therefore, the aberrant expression of beta-catenin in the cytoplasm and/or the nucleus among NSCLCs might be functional.21 The present study found that intratumoral Wnt5a expression was not associated with the intratumoral beta-catenin expression. This result might be due to the fact that tumor cells do not have a Frz receptor for Wnt5a, as suggested previously.29 Thus the intratumoral aberrant expression of beta-catenin among NSCLCs might be caused by other mechanisms, such as the overexpression of other members of the Wnt family or a reduced E-cadherin expression.21,36
In contrast, the stromal beta-catenin expression appeared in the form of a cytoplasmic staining pattern with varied nuclear staining. The present study revealed that the intratumoral Wnt5a expression was significantly correlated with the stromal beta-catenin expression. Furthermore, the intratumoral Wnt5a expression was also significantly correlated with the stromal VEGF-A expression. Although an immunohistochemistry assay does not precisely represent the source of VEGF-A, a known secreted protein, a significant correlation was seen between the stromal beta-catenin expression and the stromal VEGF-A expression in the present study. This result suggested that stromal cells with a positive beta-catenin expression could produce the VEGF-A expression. Therefore, the Wnt5a protein secreted by tumor cells could activate the canonical Wnt/beta-catenin pathway in stromal cells with a receptor for Wnt5a, thereby inducing the stromal VEGF-A expression. Subsequently, the stromal VEGF-A expression could affect tumor growth via the VEGF-receptor of tumor cells. Further studies including the Frz receptor and the VEGF-receptor should be performed to clarify this mechanism.
Concerning the regulation of Wnt5a expression, both gene amplification and gene rearrangement have been reported to occur infrequently in human cancers.37 In contrast, experimental studies using cell lines demonstrated the Wnt5a expression to be regulated by various molecules, including hepatocyte growth factor, protein kinase C activity, and MRP-1/CD9.11,38,39 Our pilot study found that there was no significant association between a reduced expression of MRP-1/CD9 and Wnt5a overexpression in NSCLCs (data not shown). Therefore, the Wnt5a expression might be secondarily regulated in response to a range of changes in these many biologic molecules during tumor progression.
Consequently, the present study revealed that Wnt5a expression in squamous cell carcinomas was significantly higher than that in adenocarcinomas. Regarding tumor biology, an overexpression of Wnt5a was associated with the Ki-67 proliferation index. Although the Ki-67 proliferation index was relatively high in the present study, this result might be due to epidermiologic difference. In addition, an overexpression of Wnt5a was associated with a poor prognosis in NSCLC patients, especially in stage II to III (locally advanced) NSCLCs. Our recent study has also shown that biologic markers associated with tumor cell proliferation could be prognostic factors in stage II to III NSCLCs.40 Furthermore, an overexpression of Wnt5a was associated with a poor prognosis, especially in squamous cell carcinomas. To our knowledge, the present study is the first clinical report demonstrating the clinical significance of intratumoral Wnt5a expression in NSCLCs. In particular, an overexpression of Wnt5a could affect the progression of squamous cell carcinoma of the lung.
In conclusion, the present study demonstrated that an overexpression of Wnt5a was associated with the tumor cell proliferation in NSCLCs. Furthermore, the intratumoral Wnt5a overexpression was also found to be associated with expression of beta-catenin and VEGF-A in stromal cells, which suggests the existence of a tumor-stromal interaction. In addition, the overexpression of Wnt5a is considered to be a significant prognostic factor for poor prognosis in NSCLC patients, especially in squamous cell carcinomas. Therefore, new strategies, such as the RNA inhibition of Wnt5a, secreted Frizzled-related proteins, and bevacizumab (Avastin; Genentech, Inc, South San Francisco, CA), may be potentially effective treatments for patients with Wnt5a-positive NSCLCs.41-43
Authors' Disclosures of Potential Conflicts of Interest
The authors indicated no potential conflicts of interest.
NOTES
Authors' disclosures of potential conflicts of interest are found at the end of this article.
REFERENCES
Cersosimo RJ: Lung cancer: A review. Am J Health Syst Pharm 59:611-642, 2002
Arriagada R, Bergman B, Dunant A, et al: Cisplatin-based adjuvant chemotherapy in patients with completely resected non-small-cell lung cancer. N Engl J Med 350:351-360, 2004
Martin J, Ginsberg RJ, Venkatraman ES, et al: Long-term results of combined-modality therapy in resectable non-small-cell lung cancer. J Clin Oncol 20:1989-1995, 2002
Weir B, Zhao X, Meyerson M: Somatic alterations in the human cancer genome. Cancer Cell 6:433-438, 2004
Imoto H, Osaki T, Taga S, et al: Vascular endothelial growth factor expression in non-small-cell lung cancer: Prognostic significance in squamous cell carcinoma. J Thorac Cardiovasc Surg 115:1007-1014, 1998
Adachi M, Taki T, Konishi T, et al: Novel staging protocol for non-small-cell lung cancers according to MRP-1/CD9 and KAI1/CD82 gene expression. J Clin Oncol 16:1397-1406, 1998
Kataoka H, Tanaka H, Nagaike K, et al: Role of cancer cell-stroma interaction in invasive growth of cancer cells. Hum Cell 16:1-14, 2003
Herzlinger D, Qiao J, Cohen D, et al: Induction of kidney epithelial morphogenesis by cells expressing Wnt-1. Dev Biol 166:815-818, 1994
Dale TC: Signal transduction by the Wnt family of ligands. Biochem J 329:209-223, 1998
You Z, Saims D, Chen S, et al: Wnt signaling promotes oncogenic transformation by inhibiting c-Myc-induced apoptosis. J Cell Biol 157:429-440, 2002
Huang C, Liu D, Masuya D, et al: MRP-1/CD9 gene transduction downregulates Wnt signal pathways. Oncogene 23:7475-7483, 2004
Saitoh T, Mine T, Katoh M: Frequent up-regulation of WNT5A mRNA in primary gastric cancer. Int J Mol Med 9:515-519, 2002
Saitoh T, Katoh M: Expression and regulation of Wnt5A and Wnt5B in human cancer: Up-regulation of Wnt5A by TNF in MKN45 cells and up-regulation of Wnt5B by beta-estradiol in MCF-7 cells. Int J Mol Med 10:345-349, 2002
Lejeune S, Huguet EL, Hamby A, et al: Wnt5a cloning, expression, and up-regulation in human primary breast cancers. Clin Cancer Res 1:215-222, 1995
Saldanha G, Ghura V, Potter L, et al: Nuclear beta-catenin in basal cell carcinoma correlates with increased proliferation. Br J Dermatol 151:157-164, 2004
Scagliotti GV, Micela M, Gubetta L, et al: Prognostic significance of Ki67 labeling in resected non small cell lung cancer. Eur J Cancer 29A:363-365, 1993
Masuya D, Huang C, Liu D, et al: The intratumoral expression of vascular endothelial growth factor and interleukin-8 associated with angiogenesis in nonsmall cell lung carcinoma patients. Cancer 92:2628-2638, 2001
Dvorak HF, Brown LF, Detmar M, et al: Vascular permeability factor/vascular endothelial growth factor, microvascular hyperpermeability, and angiogenesis. Am J Pathol 146:1029-1039, 1995
Zhang X, Gaspard JP, Chung DC: Regulation of vascular endothelial growth factor by the Wnt and K-ras pathways in colonic neoplasia. Cancer Res 61:6050-6054, 2001
Mountain CF: Revisions in the international system for staging lung cancer. Chest 111:1710-1717, 1997
Kotsinas A, Evangelou K, Zacharatos P, et al: Proliferation, but not apoptosis, is associated with distinct beta-catenin expression patterns in non-small-cell lung carcinomas. Am J Pathol 161:1619-1634, 2002
Yamaguchi TP, Bradley A, McMahon AP, et al: A Wnt5a pathway underlies outgrowth of multiple structures in the vertebrate embryo. Development 126:1211-1223, 1999
Murdoch B, Chadwick K, Martin M, et al: Wnt-5A augments repopulating capacity and primitive hematopoietic development of human blood stem cells in vivo. Proc Natl Acad Sci U S A 100:3422-3427, 2003
Yang Y, Topol L, Lee H, et al: Wnt5a and Wnt5b exhibit distinct activities in coordinating chondrocyte proliferation and differentiation. Development 130:1003-1015, 2003
Taki M, Kamata N, Yokoyama K, et al: Down-regulation of Wnt-4 and up-regulation of Wnt-5a expression by epithelial-mesenchymal transition in human squamous carcinoma cells. Cancer Sci 94:593-597, 2003
Li C, Xiao J, Hormi K, et al: Wnt5a participates in distal lung morphogenesis. Dev Biol 248:68-81, 2002
Weeraratna AT, Jiang Y, Hostetter G, et al: Wnt5a signaling directly affects cell motility and invasion of metastatic melanoma. Cancer Cell 1:279-288, 2002
Shimizu H, Julius MA, Giarre M, et al: Transformation by Wnt family proteins correlates with regulation of beta-catenin. Cell Growth Differ 8:1349-1358, 1997
Sawyer EJ, Hanby AM, Rowan AJ, et al: The Wnt pathway, epithelial-stromal interactions, and malignant progression in phyllodes tumours. J Pathol 196:437-444, 2002
Smith K, Bui TD, Poulsom R, et al: Up-regulation of macrophage Wnt gene expression in adenoma-carcinoma progression of human colorectal cancer. Br J Cancer 81:496-502, 1999
He X, Saint-Jeannet J, Wang Y, et al: A member of the frizzled protein family mediating axis induction by Wnt-5A. Science 275:1652-1654, 1997
Ben-Ze'ev A, Shtutman M, Zhurinsky J: The integration of cell adhesion with gene expression: The role of beta-catenin. Exp Cell Res 261:75-82, 2000
Pirinen RT, Hirvikoski P, Johansson RT, et al: Reduced expression of alpha-catenin, beta-catenin, and gamma-catenin is associated with high cell proliferative activity and poor differentiation in non-small cell lung cancer. J Clin Pathol 54:391-395, 2001
Bienz M, Clevers H: Linking colorectal cancer to Wnt signaling. Cell 103:311-320, 2000
Ueda M, Gemmill RM, West J, et al: Mutations of the beta- and gamma-catenin genes are uncommon in human lung, breast, kidney, cervical and ovarian carcinomas. Br J Cancer 85:64-68, 2001
Kase S, Sugio K, Yamazaki K, et al: Expression of E-cadherin and beta-catenin in human non-small cell lung cancer and the clinical significance. Clin Cancer Res 6:4789-4796, 2000
Iozzo RV, Eichstetter I, Danielson KG: Aberrant expression of the growth factor Wnt-5A in human malignancy. Cancer Res 55:3495-3499, 1995
Huguet EL, Smith K, Bicknell R, et al: Regulation of Wnt5a mRNA expression in human mammary epithelial cells by cell shape, confluence, and hepatocyte growth factor. J Biol Chem 270:12851-12856, 1995
Jonsson M, Smith K, Harris AL: Regulation of Wnt5a expression in human mammary cells by protein kinase C activity and the cytoskeleton. Br J Cancer 78:430-438, 1998
Huang C, Liu D, Masuya D, et al: Clinical application of biological markers for treatments of resectable non-small-cell lung cancers. Br J Cancer 92:1231-1239, 2005
You L, He B, Uematsu K, et al: Inhibition of Wnt-1 signaling induces apoptosis in beta-catenin-deficient mesothelioma cells. Cancer Res 64:3474-3478, 2004
He B, Lee AY, Dadfarmay S, et al: Secreted Frizzled-related protein 4 is silenced by hypermethylation and induces apoptosis in beta-catenin-deficient human mesothelioma cells. Cancer Res 65:743-748, 2005
Johnson DH, Fehrenbacher L, Novotny WF, et al: Randomized phase II trial comparing bevacizumab plus carboplatin and paclitaxel with carboplatin and paclitaxel alone in previously untreated locally advanced or metastatic non-small-cell lung cancer. J Clin Oncol 22:2184-2191, 2004(Cheng-long Huang, Dage Li)