Reduced p21^WAF1/CIP1 via Alteration of p53-DDX3 Pathway Is Associated with Poor Relapse-Free Survival in Early-Stage Human Papillomavirus–Associated Lung Cancer

Lung cancer in persons who have never smoked (never-smokers) is now becoming increasingly apparent as approximately 25% of lung cancer cases worldwide are not attributable to cigarette smoking (1). In East Asia, where few women are smokers, lung cancer incidence rates are higher and more variable than in other geographic areas that have low numbers of female smokers (2). For example, more than 90% of Taiwanese women are lifetime never-smokers (3), but lung cancer has been the leading cause of their cancer deaths for the past 3 decades (4). Human papillomavirus (HPV) 16/18 infection has been shown to be associated with the development of lung cancer in female Taiwanese never-smokers (5). Therefore, identification of molecular markers for this disease is urgently needed to improve therapeutic strategies for never-smokers with lung cancer.

The cyclin-dependent kinase (CDK) inhibitor p21^WAF1/CIP1 (p21) is the target of the HPV E6 oncoprotein (6, 7). Induction of p21 by p53-dependent (8) or p53-independent pathways (9, 10) results in the inhibition of cyclin/CDK complexes that regulate cellular proliferation (11, 12). In humans, loss of p21 expression correlates with lung cancer that has poor prognosis (13–16). In mouse model studies, p21^−/− mice showed an increased incidence of spontaneous lung tumors compared with p21-sufficient mice (17). Therefore, p21 is considered to act as a tumor suppressor in lung tumorigenesis. In addition, real-time reverse-transcriptase (RT) PCR data have revealed that p21 expression decreases significantly in HPV 16/18 E6–positive lung tumors compared with E6-negative tumors (18). Thus, p21 appears to be a relevant target of E6 in HPV-associated lung tumors. However, the role of p21 in tumor progression and metastasis is unclear in HPV-associated lung tumorigenesis.

The human DEAD-box RNA helicase DDX3 may play a role in the regulation of gene expression via RNA metabolism including transcription, splicing, mRNA export, and translation (19). Recently, DEAD-box RNA helicases have been shown to participate in the development of certain viral-associated cancers. For example, inactivation of DDX3 by HBx and HCV core proteins may promote tumor growth via suppression of p21 transcription through the p53-independent pathway (20, 21). In addition, inactivation of DDX3 by phosphorylation at Thr204 by the cyclin B/cdc2 complex halts HeLa cells in the S phase through reduced cyclin A expression (22). However, the role of DDX3 in HPV-associated tumorigenesis remains to be elucidated, especially with respect to lung tumorigenesis.

In the present study, we identified HPV E6 regulation of DDX3 transcription via p53 inactivation. We constructed different lengths of the DDX3 promoter by 5′-deletion mutation for use in luciferase reporter assays in HPV 16 E6–positive TL-1 and E6-negative H1299 lung cancer cells. Chromatin immunoprecipitation (ChIP) assays and site-directed mutagenesis were performed to identify direct binding of p53 onto the putative p53 binding sites of the DDX3 promoter. We also used E6, p53, and p21 small RNA interference (RNAi) to confirm that DDX3 transcription is predominantly regulated by p53 and that p21 transcription was synergistically suppressed by the alteration of the p53-DDX3 pathway via E6. In addition, we observed that cell proliferation and colony formation were dependent on p21 expression and consistent findings were also observed in lung tumors. More importantly, reduction of p21 expression by the alteration of the p53-DDX3 pathway in lung tumors was associated with poor relapse-free survival (RFS) in lung cancer patients.

Lung tumor specimens were collected from 138 patients with primary lung cancer in the Department of Thoracic Surgery, Taichung Veterans General Hospital between 1998 and 2004 and patients were asked to submit a written informed consent approved by the Institutional Review Board. The tumor type and stage of each collected specimen were histologically determined according to the WHO classification system. Cancer relapse data were obtained by chart review and confirmed by thoracic surgeons. The A549, H1299, Ch27, and H460 lung cancer cell lines were maintained in DMEM (Dulbecco's modified Eagle's medium). The H1355, H441, H520, TL-1, TL-2, and TL-4 lung cancer cells and SiHa cervical cancer cells were maintained in RPMI-1640. C33A cervical cancer cells were maintained in MEM. The medium contained 10% FBS supplemented with penicillin (100 U/mL) and streptomycin (100 mg/mL). Cells were grown at 37°C in a humidified incubator at 5% CO₂.

The wild-type (WT) p53, p21-Luc, and HPV 16 E6 were kindly provided by Drs. J.L. Ko and J.H. Chang Tsai from the Institute of Medical and Molecular Toxicology at Chung Shan Medical University. The DDX3 overexpression plasmid was purchased from OriGene (OriGene Technologies). The p21 overexpression plasmid was purchased from Addgene (Addgene Technologies). The DDX3-Luc plasmid was constructed by inserting a 2,060 bps HindIII/KpnI fragment (spanning the promoter region −2,060/+1 and 5′-deleted −1,242/+1, −734/+1, −309/+1 related to the translation start site of the human DDX3 gene) into a HindIII/KpnI-treated pGL3 vector (Promega Corp.). DDX3 promoter–driven luciferase reporters containing multiple point mutations of the p53 sites (Mut1) DDX3-Luc, (Mut2/3) DDX3-Luc, and (Mut1/2/3) DDX3-Luc were generated using the QuickChange site-directed mutagenesis system (Stratagene). RNAi was performed by expression of small hairpin RNA (shRNA) to target p53, DDX3, and p21 mRNA in lung cancer cell lines. The shRNA template was constructed from 2 oligonucleotides with a complementary sequence in the loop region (Supplementary Table 1; ref. 23). The different concentrations of expression plasmids were transiently transfected into lung cancer cells (1 × 10⁶) using the Transfast reagent. After 48 hours, cells were harvested and whole-cell extracts were assayed in the following experiments.

The RNAi target sequences for HPV 16 E6 siRNA (E6si) have been previously verified (24, 25). The procedures and methods were as described previously (18).

Total RNA was extracted by homogenization in 1 mL TRIzol reagent, followed by chloroform extraction and isopropanol precipitation. A 3-μg sample of total RNA from lung tumor tissues was reverse transcribed using SuperScript II Reverse Transcriptase (Invitrogen Life Technologies) and oligo(dT)₁₅ primer. Primers used for real-time PCR analysis are listed in Supplementary Table 1. Reactions were performed as previously described (26). Since DDX3 is located on chromosome X and escapes X-inactivation in the female (27), the expression levels of DDX3 in male and female genders might be different. When at least a 2-fold reduction existed in the mRNA level of the tumor tissue compared with the normal tissue, DDX3 mRNA was defined as “low”. Conversely, the situation was defined as “high” (20).

For the luciferase reporter assay, appropriate numbers of cells were transfected with sufficient reporter plasmid, p21-Luc, DDX3-Luc or its derivatives, and either the control vector or the p53 and DDX3 expression plasmid. For normalization of transfection efficiency, β-gal was also cotransfected. Transfected cells were harvested at 48 hours posttransfection and a luciferase assay was performed according to the manufacturer's instructions. The luciferase activity was measured with an AutoLumat LB953 luminometer (Berthold) and normalized with the cotransfected β-gal activity.

ChIP analysis was performed as described in a previous report (28) with the following modifications: Immunoprecipitated DNA was precipitated with ethanol and resuspended in 20 μL ddH₂O (double distilled water). Samples were resuspended in 100 μL ddH₂O and diluted 1:100 before PCR analysis. PCR amplification of immunoprecipitated DNA was carried out with diluted aliquots, using the primers consisting of the oligonucleotides that encompass the promoter region of DDX3 and p21 (Supplementary Table 1). PCR products were separated on 2% agarose gels and analyzed using ethidium bromide staining. All ChIP assays were performed at least twice with similar results.

For the immunoprecipitation experiments, cells transfected with plasmids were harvested and cell lysates were prepared using immunoprecipitation lysis buffer [20 mmol/L Tris-Cl (pH 7.5), 150 mmol/L NaCl, 10% glycerol, and 1% Triton X-100]. Cell extracts (1.5 mg) were incubated with 40 μL of anti-Sp1-agarose affinity gel (Millipore). After extensive washing with immunoprecipitation lysis buffer, the immunoprecipitated proteins were analyzed by immunoblotting using specific antibodies against DDX3, Sp1, and p53 antibody (Dako).

Cell counting and colony formation assays were performed to assess cell proliferation rates as previously described (18).

Flow cytometric analysis was performed for cell knockdown by DDX3si or overexpression by DDX3 overexpression plasmid as described previously (18).

These assays were performed according to a previously report (29).

The immunohistochemical procedures for the tissue array were similar to those described in a previous report (18). The anti-rabbit DDX3 antibody was kindly provided by Professor Yan-Hwa Wu Lee (National Yang-Ming University; ref. 20) and the monoclonal anti-p21 and anti-Ki-67 antibody was purchased from Zymed Laboratories and Dako. Immunostaining results of HPV 16/18 E6 expression in lung tumors were obtained from a previous report (18, 26). Negative immunostaining of HPV 16/18 E6, p21, and Ki-67 was defined as less than 10% of tumor cells showing immunoreactivity in the nuclei, whereas a value greater than 10% was defined as positive immunostaining (18). The scoring of DDX3 immunostaining was performed as described previously (20).

Statistical analysis was performed using the SPSS statistical software program (Version 15.0 SPSS Inc.). The χ² test (2-tailed) was applied for statistical analysis. The associations between HPV 16/18 E6, DDX3, p21, and Ki-67 protein expression were analyzed by a χ² test. Survival plots were generated using the Kaplan–Meier method and differences between patient groups were determined by a log-rank test.

Our data indicated that DDX3 mRNA expression was significantly reduced in 73 of 138 (53%) of the tumor samples compared with their normal counterpart tissues (Table 1). However, the actual factor involved in DDX3 transcription remained unclear. Therefore, we used real-time PCR to study 10 lung cancer cell lines and 2 cervical cancer cell lines known to harbor p53 WT, p53 mutation, p53-null, or HPV 16 E6, to understand whether DDX3 mRNA expression could be related to p53 status. Interestingly, DDX3 mRNA levels in p53 WT lung cancer cells were higher than in E6-positive and p53-mutant or p53-null cancer cells, except for H1355 cells (Fig. 1A). Expression of HPV 16/18 E6 protein has been demonstrated in lung tumors and appears to be related to p53 inactivation (18). We next investigated whether the deregulation of DDX3 by E6 is mediated through p53-dependent pathway, E6 knocked down in TL-1 cells, and E6 overexpressed in A549 cells were performed. E6 knockdown gradually increased the p53 protein level, whereas forced expression of E6 gradually decreased the p53 level (Fig. 1B). The elevation and reduction of DDX3 expression was dependent on the restoration and degradation of p53 by E6 knockdown and E6 overexpression (Fig. 1B). This phenomenon was also observed in HPV 16 E6–positive TL-2 and HPV 16 E6–negative TL-4 cells (Supplementary Fig. 1). Decreased and increased DDX3 expression levels occurred in a dose-dependent manner in p53-knockdown or p53-overexpression cells (Fig. 1B). These results suggest that p53 may regulate DDX3 transcription.

The DDX3 promoter region located at −1,100 to −700 has been shown predominantly to activate DDX3 transcription in HeLa cells (30). In this study, 3 putative p53 binding sites in the promoter region were predicted by the software from the web site: http://motif.genome.jp/ (Fig. 1C). This region contained a sequence between −1,080/−1,070 (5′-GCGCGTGTCT-3′), −695/−685 (5′-TGGCCTGCCG-3′), and −283/−273 (5′-AGGCAGGACT-3′) and was homologous to the consensus p53-binding site 5′-NGRCWTGYCY-3′, where R is a purine and Y is a pyrimidine base. Four lengths of the DDX3 promoter sequence (−2,060 to +1, −1,242 to +1, −734 to +1, and −309 to +1) were constructed by PCR and deletion mutation (Fig. 1C) and were transfected into different cells for luciferase reporter activity analysis. As shown in Figure 1C, the reporter activity of the 4 lengths of the DDX3 promoter increased markedly in E6-knockdown TL-1 cells and in H1299 cells with p53 overexpression (Fig. 1C). The reporter activity of DDX3 (−1,242/+1)-Luc was similar to that of DDX3 (−2,060/+1)-Luc, revealing that the −1,242/+1 promoter region may be sufficient to regulate DDX3 transcription in E6-knockdown TL-1 cells. To delineate which of the 3 putative p53 binding sites were responsible for the DDX3 reporter activity, DDX3 promoter–directed reporter plasmids harboring the mutations of the p53 binding site at −1,080/−1,070 (Mut1), the other 2 p53 binding sites at −695/−685 and −283/−273 (Mut2/3), and all 3 p53 binding sites (Mut1/2/3; Supplementary Fig. 2) were introduced into E6-positive or E6-knockdown TL-1 cells and p53-positive or p53-negative H1299 cells for luciferase reporter assay. The reporter activity of Mut1, Mut2/3, and Mut1/2/3 reporter plasmids in E6-positive TL-1 cells was similar to that of the WT reporter plasmid. However, the reporter activity of Mut1 and Mut2/3 reporter plasmids in E6-knockdown TL-1 cells decreased significantly compared with that of the WT reporter plasmid (Fig. 1D). The reporter activity of Mut1/2/3 in E6-knockdown TL-1 cells was almost completely abolished and was similar to that of the WT reporter plasmid (Fig. 1D). ChIP analysis further showed that p53 binding ability on p53 binding sites was significantly modulated by E6 and p53 status (Fig. 1D). These results clearly indicate that DDX3 transcription is directly regulated by p53.

The transcription of p21 is predominantly transactivated by p53 via increased Sp1 binding affinity (31). A recent report indicated that p21 transcription was upregulated by DDX3 via the p53-independent pathway (21). In the present study, we consistently observed that DDX3 overexpression in TL-1 and H1299 cells significantly increased p21 reporter activity (Fig. 2A). Because DDX3 transcription is directly regulated by p53 in lung cells, we studied the role of DDX3 in p53 transactivation of the p21 promoter by conducting a series of transient transfection experiments in TL-1 and A549 cells. E6 knockdown in TL-1 cells resulted in a significant increase in p21 promoter activity (3.5-fold), whereas cotransfection with DDX3 knockdown decreased the reporter activity to 1.31-fold (Fig. 2B). When p53 was silenced in E6-knockdown TL-1 cells, the p21 reporter activity was significantly reduced to 0.33-fold of that of TL-1 NC (nonspecific RNAi control) cells. These results clearly indicate that DDX3 may play an important role in p21 transcription via a p53-dependent mechanism. Conversely, forced expression of DDX3 increased p21 reporter activity 5-fold in A549 cells, but p21 reporter activity only increased 2-fold in DDX3+p53si A549 cells, as compared with p53si A549 cells (Fig. 2B). These results clearly indicate that p21 activated by p53 is synergistically enhanced by DDX3.

We next examined whether the binding activity of Sp1 on the p21 promoter could be increased by DDX3, perhaps via an increase in interaction between Sp1 and p53. Immunoprecipitation and ChIP analysis indicated that the interaction between p53 and Sp1 and the binding affinity of Sp1 onto the p21 promoter was markedly increased in E6-knockdown TL-1 cells compared with TL-1 NC cells but changed slightly in DDX3-knockdown TL-1 cells when E6 was further silenced (Fig. 2C and D). Conversely, forced expression of DDX3 in A549 cells increased the interaction between p53 and Sp1 and the ability of Sp1 to bind onto the p21 promoter, but these factors were increased only slightly in p53-knockdown A549 cells (Fig. 2C and D). Thus, DDX3 appeared to enhance p53-activated p21 transcription via increased Sp1 binding onto the p21 promoter.

Because DDX3 synergistically enhances p53-activated p21 transcription, we next examined whether this situation could alter cell proliferation, colony formation, and cell invasion capability. Cell growth and colony formation were significantly inhibited by E6 knockdown in TL-1 cells, but no effect was seen in response to E6si in DDX3si, p53si, or p21si cells compared with NC cells (Fig. 3A and B). DDX3 significantly reduced cell proliferation and colony formation in A549 cells but only slightly retarded these processes in p53-knockdown A549 cells (Fig. 3C, left). In addition, p21 protein expression in H1299 cells evaluated by Western blotting was significantly increased by transfection of p53, DDX3, and p53 + DDX3 expression vectors. Among these, p53 + DDX3 transfection had the most inhibition on cell proliferation and colony formation followed by p53 or DDX3 transfection (Supplementary Fig. 3). Cell invasion ability was significantly inhibited by E6 knockdown in TL-1 cells; however, there was no difference by p21-knockdown + E6-knockdown in TL-1 cells as compared with E6-knockdown TL-1 cells (Fig. 3C, left). Similar observations were also seen in A549 cells (Fig. 3C, right). These results suggest that the alteration of DDX3 could predominantly inhibit p53-dependent cell growth arrest via the p21 pathway.

To understand whether DDX3 transcription could be influenced by E6 and p53 mutations in lung tumors, 138 lung tumors were examined by real-time PCR to determine their DDX3 mRNA level. Low DDX3 mRNA expression levels were more prevalent in E6-positive tumors than in E6-negative tumors (72% vs. 44%, P = 0.002; Table 1). This phenomenon was observed particularly in early-stage tumors (I + II; 73% vs. 37%, P = 0.002; Table 1) but not in late-stage tumors. However, the association between the DDX3 mRNA level and p53 mutation was marginal in all studied tumor groups, except in early- or late-stage tumors (Table 1). Low levels of DDX3 mRNA expression were more prevalent in E6-positive or p53 mutation tumors than in E6-negative and p53 WT tumors (66% vs. 37%, P = 0.001; Table 1), especially in early-stage tumors (64% vs. 29%, P = 0.003; Table 1). Expression of DDX3 mRNA in lung tumors was also consistent with DDX3 immunostaining results (P = 0.001; Table 1).

We next examined whether p21 expression could be affected by DDX3 in lung tumors by immunohistochemistry. Representative immunostaining results for DDX3 and p21 expression in lung tumors are shown in Figure 4A. DDX3 expression correlated positively with p21 expression in early-stage tumors (P = 0.001; Table 2). A high mitotic and Ki-67 proliferative index has been shown to be correlated with reduction of cell-cycle regulators (32–34) and the prognostic value of Ki-67 has been shown in non–small cell lung cancer (NSCLC; refs. 35–37). Ki-67 expression was also examined by immunohistochemistry and a representative immunostaining of Ki-67 in lung tumors was shown in Figure 4A. Due to sample availability, Ki-67 analysis was performed on samples from 78 patients. As shown in Table 2, p21 expression in lung tumors correlated negatively with Ki-67 expression (P = 0.001), particularly in early-stage tumors (P < 0.001). Notably, these observations from all studied patients were predominately shown in patients from nonsmokers (Supplementary Tables 2 and 3). These results suggest that DDX3 deregulated by E6 may promote tumor growth via decreased p21.

To verify whether decreased p21 expression could be associated with a poor RFS rate in patients with early-stage tumors, Kaplan–Meier analysis was performed. Of the 111 patients who were enrolled in this study, over a median follow-up period of 32.4 months, 44 patients relapsed (4 had local recurrence, 26 had distant metastasis, and 14 had local and distant metastasis) and 30 patients died from this disease. None of the patients received adjuvant treatment before surgical therapy. Patients with low p21 expression had a shorter median RFS than those with high p21 expression (26.2 months vs. 43.8 months, P = 0.029; Fig. 4B). However, the prognostic value of p21 expression was not seen in all patients (22.1 months vs. 35.4 months, P = 0.124; Fig. 4B), which suggests that decreased p21 expression may partially contribute to early-stage tumor progression. In addition, patients with p53 mutation or E6-positive/low DDX3/low p21 expression had a shorter median RFS compared with the late-stage group (23.5 months vs. 50.0 months, P = 0.001; Fig. 4C). Moreover, p53 mutation or E6-positive/low DDX3/low p21 expression correlated with a shorter median survival than seen in patients with low p21 expression (23.5 months vs. 26.2 months). These results suggest that reduction of p21 by p53-DDX3 pathway may contribute to tumor progression and a poor RFS in early-stage HPV-associated lung cancer.

Lung cancer is characterized by sequential accumulation of specific genetic and morphologic changes in the normal epithelial cells of the lung. The most important pathway involved in tumor malignancy is disruption of the tightly regulated cell-cycle regulators that control cell proliferation. In the present study, the reduction of p21 levels by the alteration of the p53-DDX3 pathway via E6 appeared to play a crucial role in early-stage HPV 16/18-associated lung tumorigenesis. Among tumors, DDX3 expression correlated negatively with E6 expression, particularly in early-stage tumors (stage I + II; Table 1), whereas p21 expression correlated positively with DDX3 expression (P = 0.001; Table 2). The latter finding was consistent with our previous report in which p21 mRNA expression levels in lung tumors were found to correlate negatively with HPV 16/18 E6 expression (18). In the present study, patients with early-stage tumors with low p21 expression had a worse RFS rate compared with patients with high p21 expression. In addition, p21 expression in lung tumors correlated negatively with the expression of the cell proliferation marker Ki-67. Therefore, we suggest that reduction of p21 in response to alteration of p53-DDX3 via E6 may further enhance tumor growth and recurrence in early-stage lung cancers associated with HPV infection.

p21 is frequently deregulated during human tumorigenesis and may act as a tumor suppressor that inhibits cell-cycle progression (38). Chao and colleagues (21) showed that DDX3 may inhibit hepatocellular cell growth via upregulated p21 expression. This suggestion is consistent with our present study, the expression of p21 was significantly altered by DDX3 overexpression in TL-1 cells, even when p53 was inactivated by E6 (Supplementary Fig. 4A). The doubling time, colony formation, and flow cytometric assay further showed that the doubling time of DDX3-knockdown TL-4 and A549 cells were reduced and concomitantly, the S-phase cell proportion and colony numbers of both cells was elevated compared with their control cells. Conversely, an increase of doubling time and decrease of S-phase population and colony numbers was observed in DDX3-overexpressed TL-1 cells (Supplementary Fig. 4A). In addition, DDX3 inhibited lung cancer cell growth via the reduction of p21 transcription (Supplementary Fig. 4B). These results suggest that p21 deregulated by p53-DDX3 pathway may increase lung cancer cell growth. Cell proliferation and colony formation were significantly altered by p21 expression in TL-1 and A549 cells (Fig. 4). We also observed a dose dependency in the inhibition of colony formation by exogenously expressed p21 plasmid in TL-1 cells (Supplementary Fig. 4C). However, cell invasion activity was not changed by a reduction in p21 expression (Fig. 4). Among lung tumors, p21 expression was correlated with expression of the cell proliferation marker Ki-67, further supporting an association of the reduction of p21 with tumor growth in early-stage tumors. Therefore, p21 deregulation by the p53-DDX3 pathway may predominantly increase the cell proliferation rate in vitro and tumor growth in vivo.

The prognostic value of p21 has been examined extensively in lung cancer, but results for prognosis based on p21 have been inconsistent (13–16, 39). In the present study, patients, and especially early-stage patients, with positive p21 expression had favorable prognosis. This observation was consistent with most previous reports (13–16) but was contrary to Dworakowska's report. Possibly, the few female cases and the few adenocarcinomas studied by Dworakowska and colleagues (2005; ref. 39) may account for this difference. In the present study, early-stage patients with negative p21 expression showed a more negative RFS rate when associated with a p53 mutation and/or the E6-positive condition, compared with other combinations (P < 0.001; Fig. 4C). This finding was consistent with previous reports indicating that lung cancer patients with p21 reduction due to alteration of the p53 pathway may have a poor prognosis (13).

In cell experiments, the reduction in DDX3 in response to E6 has been shown to result in a synergistic reduction in p21 expression and a promotion of cell proliferation (Fig. 4). Pyeon and colleagues (2009) showed that G₁, S, G₂, and early M phase cell-cycle inhibitors efficiently inhibited cell-cycle progression and prevented HPV infection (40). Soft agar assay further showed that the independent anchorage soft agar colony number in TL-1 cells was markedly reduced by E6si transfection but slightly changed by the combined treatment of DDX3si + E6si as compared wih TL-1 cells with DDX3si transfection (data not shown).Therefore, the alteration in the p53-DDX3 pathway may promote the initiation step of lung tumorigenesis in HPV-infected tissues. However, this conclusion needs to be further investigated in E6 transgenic mouse model.

An association between HPV 16/18 and lung cancer in never-smokers has been noted (5, 18). However, the molecular pathogenesis of HPV-associated lung cancer remains unclear. This study is the first report to indicate that DDX3 transcription is regulated by p53. Tumor progression appears to require E6-mediated inactivation of DDX3, degradation of p53, and synergistic suppression of p21 transcription. These factors contribute to a poor RFS in early-stage lung cancer associated with HPV infection. Therefore, we suggest that DDX3 could represent a promising molecular target for therapeutics of HPV-associated lung cancer.

The authors declare no conflict of interest.

This work was jointly supported by grants from the National Health Research Institute (NHRI96-TD-G-111-006; NHRI97-TD-G-111-006), Department of Health (DOH94-TD-G-111-017; DOH100-TD-C-111-005) and the National Science Council (NSC-96-2628-B-040-002-MY3) of Taiwan, ROC.

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.