Esophageal cancer is an aggressive human malignancy with increasing incidence in the developed world. VEGF-C makes crucial contributions to esophageal cancer progression that are not well understood. Here, we report the discovery of regulatory relationship in esophageal cancers between the expression of VEGF-C and cortactin (CTTN), a regulator of the cortical actin cytoskeleton. Upregulation of CTTN expression by VEGF-C enhanced the invasive properties of esophageal squamous cell carcinoma in vitro and tumor metastasis in vivo. Mechanistic investigations showed that VEGF-C increased CTTN expression by downregulating Dicer-mediated maturation of miR326, thereby relieving the suppressive effect of miR326 on CTTN expression. Clinically, expression of Dicer and miR326 correlated with poor prognosis in patients with esophageal cancer. Our findings offer insights into how VEGF-C enhances the robust invasive and metastatic properties of esophageal cancer, which has potential implications for the development of new biomarkers or therapies in this setting. Cancer Res; 74(21); 6280–90. ©2014 AACR.
Esophageal cancer is the fifth and seventh leading cause of cancer-related death worldwide in males and females, respectively (1). Esophageal squamous cell carcinoma (ESCC) is the most frequent cause death of esophageal cancer in most Asian countries, an area often described as the Asian esophageal cancer belt (2). Although treatment and perioperative management have evolved in recent years with dramatic advances in diagnosis, operative methods, and combined chemoradiotherapy, the prognosis of patients with ESCC is not ideal. Only a small subset of patients (20%–30%) exhibit a 5-year survival rate after surgery (3, 4). Thus, understanding the detailed molecular mechanisms involved in ESCC progression is crucial for the development of novel therapeutic strategies.
The VEGF gene family, which encodes seven polypeptide growth factors, VEGF-A, -B, -C, -D, -E, and -F, and placental growth factor (PlGF), plays important roles in tumor development and metastasis due to its strong mitogenic, angiogenic, and lymphangiogenic properties (5, 6). VEGF-C and its receptors are expressed in a substantial fraction of human cancers, including lung, breast, colon, stomach, prostate, and esophageal cancers (6). VEGF-C promotes lymphangiogenesis, angiogenesis, and tumor metastasis by binding to its receptors, VEGFR-3 (also called Flt-4) and VEGFR-2 (Flk-1; ref. 7). The VEGF-C/VEGFR-3 axis exerts diverse biologic effects on cancer cells that facilitate tumor progression, such as increased cell survival, proliferation, invasion, migration, and resistance to chemotherapy (8–10). In patients with ESCC, serum VEGF-C levels are upregulated and are considered a reliable diagnostic biomarker for ESCC (11). However, the underlying molecular mechanism involved in VEGF-C–mediated ESCC tumor progression is unclear. Previous studies suggest that VEGF-C is an important target for anticancer research and that interruption of VEGF-C–mediated signal transduction may be a viable cancer therapeutic strategy.
MicroRNAs (miRNA) are conserved, endogenous, small, noncoding RNA molecules of 21 to 23 nucleotides that function as posttranscriptional gene regulators (12). Posttranscriptional repression mediated by miRNAs can occur through the inhibition of protein synthesis, the degradation of target mRNAs, and the translocation of target mRNAs into cytoplasmic processing bodies (13). Deregulated biogenesis of miRNAs has been widely observed in cancer (14). The role of Dicer, one of the key factors involved in miRNA biogenesis, has been identified as a tumor suppressor. Dicer is often downregulated in cancer and exerts its function by regulating specific miRNAs (15–18).
In the current work, we have identified an oncogenic role of VEGF-C through Dicer suppression in ESCC, which attenuates miR326 maturation and further enhances cortactin (CTTN) expression. This study provides a mechanistic demonstration of the prometastatic function of VEGF-C in esophageal cancer.
Materials and Methods
Antibodies and chemicals
Recombinant human VEGF-C protein (rhVEGF-C) was purchased from R&D Systems. Western blotting and immunohistochemistry (IHC) were performed with the use of the following antibodies: CTTN (Santa Cruz Biotechnology); Dicer1 and Ago2 (Abcam); VEGF-C (Invitrogen); VEGFR-3 (Abgent); p-VEGFR-3 (Y1063/1068, Cell Applications); and β-actin (Millipore). Actinomycin D and cycloheximide were purchased from Sigma.
Cell culture and transfection
Human ESCC cell lines TE-1, TE-3, TE-7, TE-8, and TE-12 were kindly provided by Dr. Mien-Chie Hung (MD Anderson Cancer Center, Houston, TX). TE1/pcDNA6 and TE1/VEGF-C cells were transfected with pcDNA6 and pcDNA6/VEGF-C vectors using Lipofectamine LTX (Invitrogen). Forty-eight hours after transfection, cells were trypsinized and plated in DMEM/F12 with 10% FBS and blasticidin S (5 μg/mL; Invitrogen). TE-8/shVEGF-C/CTTN cells were transfected with a pcDNA6/CTTN vector using electroporation (BTX ECM630). CTTN protein expression in TE-8 cells was determined by Western blotting. Blasticidin S–resistant clones (TE1/pcDNA6 and TE1/VEGF-C) were selected and kept in 4 μg/mL of blasticidin S and then expanded for further studies. All cell lines were maintained in DMEM/F12 (1:1) supplemented with 10% FBS (Gibco), 100 units/mL penicillin, and 100 μg/mL streptomycin (Hyclone) at 37°C in a 5% CO2 humidified atmosphere.
Patients and specimens
Esophageal squamous carcinoma tissues were obtained from a total of 118 consecutive patients who underwent surgical resection at the China Medical University Hospital (Taichung, Taiwan) from January 12, 1995, to December 31, 2008. Patients who had previous history of cancers or had been treated with neoadjuvant chemotherapy and/or radiation therapy were not included in this study. Only patients who provided esophageal squamous carcinoma specimens were included in this study. Paraffin-embedded, formalin-fixed surgical specimens were collected for IHC staining for VEGF-C, CTTN, and Dicer. Written informed consent was obtained from all patients. Tumor size, local invasion, and lymph node metastasis were determined at pathologic examination. All cases were staged according to the cancer staging manual of the American Joint Committee on Cancer, and the histological cancer type was classified according to World Health Organization (19). Follow-up data were obtained from the patients' medical charts and from our tumor registry service. The survival time of patients was calculated from the date of surgery to the date of death. Tissue preparation and analysis were approved by the Institutional Review Boards of China Medical University Hospital.
Formalin-fixed, paraffin-embedded tissue section (5 μm) were dewaxed in xylene and hydrated in graded concentrations of ethanol. Endogenous peroxidase was blocked with 3% hydrogen peroxide followed by normal goat serum (10%) blocking for 1 hour. Sections were next incubated at 4°C for 12 hours with primary anti-human CTTN antibody (1:800 dilution), anti-human VEGF-C (1:100 dilution), and anti-human Dicer1 (1:100 dilution). Sections were rinsed with PBS and incubated for 1 hour with biotinylated secondary antibody followed by 30-minute incubation at room temperature with horseradish peroxidase–conjugated streptavidin. Immunostaining was developed with 3,3′-diaminobenzidine tetrahydrochloride (DAB) for 3 minutes and counterstained with hematoxylin.
miRNA inhibition and overexpression
The mature miRNA sequences were obtained from the Sanger Center miRNA Registry (http://microrna.sanger.ac.uk/sequences/). The sequences of the top strands (5′) and bottom strands (3′) for miR326 were: 5′-TCGAGCCTCTGGGCCCTTCCTCCAGGTTTTGGCCACTGACTGACCTGGAGGAGGCCCAGAGG-3′ (top) and 5′-GGCCCCTCTGGGCCTCCTCCAGGTCAGTCAGTGGCCAAAACCTGGAGGAAGGGCCCAGAGGC-3′ (bottom). The miRNA double strands were ligated with the pLemiR vector with restriction enzyme XhoI + NotI (Open Biosystem). The construct was sequenced. Subsequently, the miRNA expression vector or control vector, pCMV-deltaR8.91, and pCMV-VSV-G vector were transfected into 293T cells using PEI reagent (Sigma) according to the manufacturer's instructions to produce lentivirus. Lentiviruses were used to infect TE-1 cells to overexpress miR326. Anti-miR, miRNA inhibitors (Ambion) are single-stranded chemically modified oligonucleotides designed to inhibit endogenous miRNAs. For in vitro miRNA inhibition studies, TE-8/shCtrl and TE-8/shVEGF-C cells were transfected with anti-miR inhibitors (150 nmol/L) based on the manufacturer's instructions. Forty-eight hours after transfection, cells were used for migration assays, invasion assays, and Western blot assays.
Pri-miRNA and pre-miRNA detection
Typically, 1 μg of total RNA was used in reverse transcription. A 10 μL reaction was assembled using 10 μmol/L of the antisense primer and a primer for the internal control (GAPDH). The reaction was heated to 80°C for 5 minutes to denature the RNA, followed by 5-minute incubation at 65°C to anneal the primers. The reactions were cooled to room temperature and the remaining reagents [5 × buffer, dNTPs, dithiothreitol (DTT), RNase inhibitor, M-MLV reverse transcriptase] were added as specified in the protocol. Real-time PCR was done by Roche LightCycler 480 Real-Time PCR System. PCR reactions contained 0.5 μmol/L of each sense and antisense primer, 1 × LightCycler SYBR Green I Master Mix, and 2 μL of cDNA. Primary precursor microRNA was detected with TaqMan Pri-miRNA Assays (Applies Biosystems).
The 1-kb 3′ untranslated region (UTR) of human CTTN gene (GenBank accession no. NM_005231) was amplified from TE-8 cDNA and cloned into the HindIII/SpeI site of pMIR-REPORT Luciferase vector (Ambion). The primers for CTTN 3′UTR were forward: 5′-ACTAGTTGGAGCTGCGGCAGTAGG-3′ and reverse: 5′-AAGCTTGGAGGAGGGTTCCTGGC-3′. The segment (base pairs: 2,370–2,376) of the CTTN 3′UTR containing the mutated miR326 target sequence (CCCAGAG to GGGTCTC) was also cloned into the pMIR-REPORT Luciferase vector using a QuickChange II Site-Directed Mutagenesis Kit (Stratagene). The primers for the mutant CTTN 3′UTR were forward: 5′-TTAACCCGGAGCTAAGTCAGGGTCTCCACAGGAGCTGCCATGTCA-3′ and reverse: 5′- TGACATGGCAGCTCCTGTGGAGACCCTGACTTAGCTCCGGGTTAA-3′. Cells (50% confluent in 24-well plates) were transfected with indicated plasmids by PEI reagent (Sigma). Firefly luciferase reporter gene construct and miRNA expression vectors were cotransfected. Cell extracts were prepared 48 hours after transfection, and luciferase activity was measured by the Luciferase Reporter Assay System (Promega).
Transwell migration and invasion assays
Transwell migration and invasion assays were performed as described (20). Briefly, ESCC cells (2.5 × 104) with serum-free DMEM/F12 medium were seeded in the upper well (8-μm pore size, Corning) for migration assay or Matrigel-coated membrane for invasion assay, following addition of 10% FBS DMEM/F12 medium in the lower well (40×, three random fields per well).
All animal work was done in accordance with a protocol approved by the China Medical University Institutional Animal Care and Use Committee (IACUC No. 97-2-H and 99-9-N). Female and male SCID mice (supplied by LASCO), age matched 4 to 6 weeks old, were used for a lung colonization metastasis model. For experimental lung metastasis assays, 5 × 106 viable cells were resuspended in 0.1 mL of PBS and introduced into the circulation via tail vein injection. Forty-two days after implantation, mice were sacrificed and lungs were excised, fixed in 4% formaldehyde solution, and processed for hematoxylin and eosin (H&E) staining. The number of metastatic nodules on the surfaces of lung was counted under a stereomicroscope.
Data are analyzed as the mean ± SE. Two-tailed Student t test was used to analyze the means of the 2 different groups. Survival was analyzed by the Kaplan–Meier method, and the log-rank test was used to test the difference in relapse time and survival between patients. Difference with a P value of less than 0.05 was considered statistically significant.
VEGF-C facilitates the migration/invasion of ESCC cells by upregulating CTTN expression
To investigate the relationship between VEGF-C expression and cell migration/invasion in ESCC, we analyzed the mRNA and protein expression levels of VEGF-C in ESCC cell lines (TE-1, TE-12, TE-3, TE-7, and TE-8). As shown in Fig. 1A, VEGF-C mRNA is highly expressed in TE-3, TE-7, and TE-8 cells, and VEGFR-3, a VEGF-C receptor, as well as Nrp1/Nrp2, VEGF co-receptors, are generally expressed among all tested ESCC cell lines (Fig. 1A, top and Supplementary Fig. S1). Western blotting and ELISA revealed that highly invasive TE-8 cells expressed the highest protein level of VEGF-C (Fig. 1A, middle and bottom), and VEGF-C, not VEGF-A, activated phosphorylation of VEGFR-3 in TE-1 cells (Supplementary Fig. S2). A positive correlation between VEGF-C expression and migration/invasion abilities was observed in ESCC cells (Fig. 1B).
CTTN has been reported to be an oncogene in human cancers and is involved in the regulation of VEGF-mediated angiogenesis activities, including cell migration and tubular formation (21, 22). We examined the protein expression of CTTN in ESCC cells by Western blotting (Fig. 1C, top) and observed a positive correlation between VEGF-C and CTTN among the esophageal cancer cell lines (Fig. 1C, bottom). These results prompted us to investigate the role of CTTN in VEGF-C–mediated oncogenic effects. To study the functional role of CTTN in VEGF-C–mediated cell motility, we established VEGF-C–expressing TE-1 cells and examined the effects of VEGF-C expression on CTTN expression and cell migration. The CTTN expression and cell migration distances were increased in TE-1/VEGF-C cells by more than 2-fold (Fig. 1D, lane 2 vs. lane 1) compared with the TE-1/vector cells. Knockdown of CTTN significantly abolished VEGF-C–enhanced cell migration in TE-1 cells (Fig. 1D, bottom). We introduced shRNA targeting VEGF-C into TE-8 cells and observed that downregulation of CTTN suppressed cell migration/invasion in TE-8/shVEGF-C cells, and restoring CTTN expression in TE-8/shVEGF-C cells recovered the migration/invasive abilities (Fig. 1E). These results indicate that CTTN is a crucial and functional downstream molecule of VEGF-C. Moreover, we observed a significant positive correlation between CTTN and VEGF-C expression in tumors from patients with ESCC (Fig. 1F and Table 1), which provides further support of VEGF-C–induced CTTN expression in human esophageal tumors. Together, these results suggest a crucial role of CTTN in VEGF-C–mediated cell invasion and tumor progression in ESCC.
|.||Low .||High .||Pa .||Correlation coefficientb .|
|.||Low .||High .||Pa .||Correlation coefficientb .|
aP value was determined by the Pearson χ2 test.
bCorrelation coefficient was calculated using phi coefficient.
VEGF-C induces CTTN expression by reducing miR326 expression
Although the protein expression of CTTN was dramatically induced by VEGF-C, no change was observed at its mRNA level (Supplementary Fig. S3), suggesting that a posttranscriptional regulation was involved in VEGF-C–induced CTTN expression. To investigate whether this regulation was dependent on the regulation of miRNAs, we first predicted several candidate miRNAs that may target CTTN (miR326, miR542-3p, and miR758) using the miRbase and TargetScan algorithms. The expression of miR542-3p and miR758 was not altered by VEGF-C knockdown in TE-8 cells, but the expression of miR326 was greatly increased (Fig. 2A). Consistent with knockdown of VEGFR-3 in TE-8 cells, TE-8/shVEGFR-3 cells also enhanced miR326 expression, compared with TE-8/shCtrl cells (Supplementary Fig. S4). Overexpression of VEGF-C dramatically reduced miR326 expression in TE-1 cells (Fig. 2B), and treatment with rhVEGF-C caused a dose-dependent reduction in miR326 but not miR758 expression (Fig. 2C). These results indicate that VEGF-C might enhance CTTN expression by downregulating miR326 expression. To further address this possible mechanism, we established a firefly luciferase reporter with the 3′UTR sequence of CTTN and mutated the seed region of the miR326 recognition site to disrupt the interaction between miR326 and CTTN (Fig. 2D, top). In contrast to the dose-dependent suppression of the wild-type 3′UTR (Fig. 2D, bottom; WT-3′UTR), miR326 had no effect on the seed region containing the mutated CTTN 3′UTR (Fig. 2D, bottom; MT-3′UTR), supporting the idea that CTTN is regulated by miR326 and that this regulation requires a direct interaction between the 3′UTR of CTTN and miR326. Upon restoring miR326 expression in TE-1/VEGF-C cells, the VEGF-C–mediated upregulation of CTTN was no longer observed (Fig. 2E). By antagonizing shVEGF-C–induced miR326 expression, we also observed a significant restoration of CTTN expression (Fig. 2F). In support of above results, we observed an inverse association between miR326 levels and VEGF-C expression in ESCC cell lines (Fig. 2G). These findings indicate that VEGF-C suppresses miR326 expression and subsequently upregulates CTTN expression in esophageal cancer.
Downregulation of miR326 is required for VEGF-C–suppressed ESCC metastasis
We further investigated the functional role of miR326 in VEGF-C–mediated metastasis. First, we observed that VEGF-C–induced cell migration was significantly diminished upon the expression of miR326 (Fig. 3A). Significant restoration of migration/invasion was also observed in anti–miR326-transfected TE-8/shVEGF-C cells (Fig. 3B). In addition, introducing a CTTN-expressing vector without a 3′UTR abrogated the effect of miR326 in mitigating VEGF-C–induced cell migration (Fig. 3C), suggesting that miR326-mediated CTTN downregulation plays an important role in VEGF-C–enhanced cell migration. We further restored the expression of miR326 into TE-1/VEGF-C cells and performed experimental lung metastasis by intravenous injection. Consistently, the VEGF-C–enhanced tumor metastasis was dramatically abolished by miR326 overexpression (Fig. 3D). Furthermore, higher levels of miR326 were correlated with better clinical outcomes in patients with ESCC (Fig. 3E). The inverse correlation between miR326 and VEGF-C was also observed in ESCC tumors (Table 2). These results demonstrate that VEGF-C downregulates miR326, which posttranscriptionally suppresses CTTN protein expression and inhibits the cancer metastasis of ESCC.
|.||Low .||High .||Pa .||Correlation coefficientb .|
|.||Low .||High .||Pa .||Correlation coefficientb .|
aP value was determined by the Pearson χ2 test.
bCorrelation coefficient was calculated using phi coefficient.
Dicer-dependent miR326 maturation is attenuated by VEGF-C
The expression of miRNA is controlled by factors regulating primary miRNA transcription, maturation, and degradation. Interestingly, we observed that only the mature form, but not primary and precursor forms, of miR326 was downregulated by VEGF-C (Fig. 4A). To investigate whether the change in miR326 is due to the regulation of the degradation pathway, we blocked the de novo synthesis of miR326 by treatment with actinomycin D. Under such a scenario, no significant difference in the degradation rate of miR326 was observed between TE-1 cells cultured in the presence and absence of rhVEGF-C (Fig. 4B). The results prompted us to identify the factors affecting miR326 maturation. In TE-1/VEGF-C cells, we observed the suppression of Dicer protein but not Ago2 (Fig. 4C, left). Consistent with these findings, we also observed that Dicer was dramatically induced by shVEGF-C in TE-8/shVEGF-C cells (Fig. 4C, right). However, the mRNA expression was not affected by VEGF-C, suggesting that Dicer is not regulated at a transcriptional level (Supplementary Fig. S5). Upon treatment with VEGF-C, the expression of Dicer was dramatically reduced (Fig. 4D). We also measured the degradation rate using cycloheximide and observed that VEGF-C significantly accelerated Dicer degradation (Fig. 4E). These results demonstrate a novel mechanism regulating the expression of Dicer in human cancer. Restoring the expression of Dicer abolished the effects of VEGF-C on miR326 suppression (Fig. 4F, bottom), CTTN upregulation (Fig. 4F, top), and cell migration/invasion (Fig. 5A), whereas knockdown of Dicer in TE-8/shVEGF-C cells rescued miR326 expression (Fig. 5B) and restored the migration ability and CTTN protein level (Fig. 5C). The above results suggest that VEGF-C–enhanced invasion is the result of the loss of Dicer-mediated CTTN suppression in esophageal cancer. Next, we investigated whether the Dicer-dependent CTTN regulation plays a clinical role in tumors. We observed an inverse correlation in the expression pattern between VEGF-C and Dicer expression in the tumor tissue of patients with ESCC (Fig. 5D and Table 2). In addition, higher levels of Dicer expression result in better clinical outcome in patients with ESCC (Fig. 5E). These results indicate that Dicer plays a key role in VEGF-C–inhibited miR326 maturation and subsequent CTTN regulation.
Recent studies have reported that VEGF-C not only induces lymphangiogenesis but also triggers oncogenic signaling pathways in an autocrine mechanism to promote tumor metastasis in several types of cancer (20, 23–25). Although the expression of VEGF-C has been observed in esophageal cancer (26–30), the mechanism by which VEGF-C promotes tumor progression remains to be fully elucidated. In present work, we demonstrate that VEGF-C promotes migration/invasion in ESCC cells. VEGF-C also enhances metastatic potential in animal experiments through the regulation of CTTN protein expression. From in vitro and in vivo experiments as well as clinical studies, our study demonstrates that VEGF-C downregulates Dicer-dependent miR326 maturation, enhancing the synthesis of CTTN protein to facilitate tumor metastasis. Our results elucidate a novel mechanism underlying VEGF-C–accelerated cancer progression (Fig. 5F).
In esophageal tumors, the overexpression of CTTN has been widely reported to promote cell migration and anoikis resistance after carcinogen induction (31). Elevated levels of CTTN in human cancers are more commonly associated with gene amplification, and little is known about its translational/transcriptional regulation. The only evidence of transcriptional regulation in esophageal cancer is based on the ability of calreticulin to induce CTTN activity through STAT3 signaling (32). Here, we identify VEGF-C as a novel regulator of CTTN expression in ESCC cells. Our findings connect the functional mechanisms that are controlled by VEGF-C and CTTN in esophageal cancer.
miRNAs are the newly identified, small noncoding RNAs belonging to a novel class of gene regulators that suppress gene expression by recognizing complementary sequences in the 3′UTRs of target mRNAs. Increasing data indicate that miRNAs such as miR145, 146a, 133a, 133b, 21, 205, and 373 might affect ESCC progression (33–36). Furthermore, it has been reported that miR326 overexpression could potentially cause multiple sclerosis by regulating helper T cell (TH)-17 cell differentiation (37). Other studies report that miR326 acts as a tumor-suppressive miRNA in brain tumors and may prevent chemoresistance in breast cancer by downregulating multidrug resistance–associated protein 1 (38, 39). Here, we identify miR326 as a tumor-suppressive miRNA involved in the regulation of VEGF-C–mediated CTTN expression and the subsequent invasion of ESCC cells. CTTN is well documented as an oncogenic protein in human malignancies; however, the regulation of CTTN by miRNA has not been reported in esophageal cancer. Indeed, only one miRNA, miR182, has been reported to suppress CTTN in lung adenocarcinoma cells (40). Through this study, for the first time, we identify the miRNA, miR326, that is able to regulate CTTN expression in both ESCC cells and clinical ESCC tumors. The observed correlations in esophageal tumors and the results obtained from in vitro/in vivo experiments support the following pathway: VEGF-C ―| miR326 ―| CTTN. Furthermore, the elevated expression of miR326 can predict better survival in patients, suggesting an important prognostic role of this miRNA in esophageal cancer. This mechanism demonstrates a new layer of CTTN regulation in esophageal cancer.
Dicer is an essential enzyme in miRNA biogenesis to its maturation, functionally active form able to regulate gene expression (41). Downregulation of Dicer has been shown to be associated with solid tumor progression in breast, ovarian, lung, and colorectal cancers (15, 18, 42–44), whereas upregulation of Dicer has been observed in prostate and oral cancer (45, 46). Although Sugito and colleagues reported that Dicer1 mRNA expression level did not exhibit significant associations with survival in patients with esophageal cancer (47), we observed that higher protein levels of Dicer in ESCC were associated with better clinical outcomes in patients with ESCC. Recent studies have reported that Dicer can be posttranscriptionally suppressed by miR103/107, resulting in the enhancement of metastasis in breast cancer (16), and the upregulation of Dicer expression by p53-related transcription factor (TAp63) reduces the numbers of metastatic tumors in transgenic mice (48). In addition, our finding reveals that the protein level but not the mRNA expression of Dicer was reduced by VEGF-C, suggesting that VEGF-C may not regulate Dicer expression at the transcriptional level in ESCC. Further investigation of VEGF-C–mediated Dicer expression is needed to clarify the detailed mechanisms of these effects.
Disclosure of Potential Conflicts of Interest
Y.-H. Yu is a director at Tainan Municipal An-Nan Hospital-China Medical University. No potential conflicts of interest were disclosed by the other authors.
Conception and design: P.-S. Chen, J. Chiou, Y.-H. Yu, M.-C. Hung, J.-L. Su
Development of methodology: C.-C. Hong, J. Chiou
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): C.-C. Hong, J. Chiou, C.-Y. Yang, Y.-W. Chang, Y.-H. Yu, N.-W. Hsu
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): C.-C. Hong, P.-S. Chen, J. Chiou, C.-F. Chiu, M.-C. Hung, J.-L. Su
Writing, review, and/or revision of the manuscript: C.-C. Hong, P.-S. Chen, J. Chiou, C.-F. Chiu, M.-C. Hung, J.-L. Su
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): C.-C. Hong, J. Chiou, S.-G. Shiah, N.-Y. Hsu, J.-L. Su
Study supervision: M. Hsiao, M.-C. Hung, S.-G. Shiah, J.-L. Su
This work was supported by grants from the National Science Council, Taiwan (NSC 102-2314-B-039-200, NSC 102-2314-B-038-028-MY3, NSC 101-2320-B-400-016-MY3); from Ministry of Health and Welfare, Taiwan (DOH 102-TD-C-111-004); from National Health Research Institutes, Taiwan (CA-102-PP-41, CA-103-PP-35); from China Medical University Hospital (DMR-101-014); and from China Medical University (CMU99-TC-22, CMU100-S-22).
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.