|
Advertisement |
|
|
||
 |
|
|||||
|
|||||
|
 |
||||||
|
||||||
|
||||||
|
Journal of Clinical Oncology, Vol 22, No 5 (March 1), 2004: pp. 811-819 © 2004 American Society of Clinical Oncology. DOI: 10.1200/JCO.2004.04.109 Prognostic Model of Pulmonary Adenocarcinoma by Expression Profiling of Eight Genes As Determined by Quantitative Real-Time Reverse Transcriptase Polymerase Chain ReactionFrom the Department of Thoracic Surgery, Department of Pathology and Molecular Diagnostics, Aichi Cancer Center Hospital; Division of Molecular Oncology, Aichi Cancer Center Research Institute, Aichi Cancer Center, Nagoya, Japan; and Department of Surgery I, Gunma University, School of Medicine, Maebashi, Japan. Address reprint requests to Tetsuya Mitsudomi, MD, PhD, Department of Thoracic Surgery, Aichi Cancer Center, Hospital, 1-1 Kanokoden, Chikusa-ku, Nagoya 464-8681, Japan; e-mail: mitsudom{at}aichi-cc.jp
PURPOSE: Recently, several expression-profiling experiments have shown that adenocarcinoma can be classified into subgroups that also reflect patient survival. In this study, we examined the expression patterns of 44 genes selected by these studies to test whether their expression patterns were relevant to prognosis in our cohort as well, and to create a prognostic model applicable to clinical practice. PATIENTS AND METHODS: Expression levels were determined in 85 adenocarcinoma patients by quantitative reverse transcriptase polymerase chain reaction. Cluster analysis was performed, and a prognostic model was created by the proportional hazards model using a stepwise method. RESULTS: Hierarchical clustering divided the cases into three major groups, and group B, comprising 21 cases, had significantly poor survival (P = .0297). Next, we tried to identify a smaller number of genes of particular predictive value, and eight genes (PTK7, CIT, SCNN1A, PGES, ERO1L, ZWINT, and two ESTs) were selected. We then calculated a risk index that was defined as a linear combination of gene expression values weighted by their estimated regression coefficients. The risk index was a significant independent prognostic factor (P = .0021) by multivariate analysis. Furthermore, the robustness of this model was confirmed using an independent set of 21 patients (P = .0085). CONCLUSION: By analyzing a reasonably small number of genes, patients with adenocarcinoma could be stratified according to their prognosis. The prognostic model could be applicable to future decisions concerning treatment.
In Japan, as in many Western countries, lung cancer is the leading cause of cancer-related death, claiming more than 50,000 lives annually, and the situation is worsening [1]. Approximately 30% of patients with non–small-cell lung carcinoma (NSCLC) have localized disease, and successful surgical management with long-term disease control is generally restricted to this group of early-stage patients. NSCLC is histopathologically and clinically distinct from small-cell lung carcinoma (SCLC), and is further subdivided into adenocarcinoma, squamous cell carcinoma, and large-cell carcinoma [2]. Although these types share common characteristics, they are thought to develop from at least partially different sets of genetic alterations [3]. Adenocarcinoma is currently the most predominant histological subtype of NSCLC in Japan as well as in the United States [4]. Although morphological features and clinical stage based on the tumor-node-metastasis system can roughly stratify patients for prognosis, it is often difficult to predict either which surgically managed patients are at risk for early relapse or which rare advanced-stage patients may experience prolonged survival [2]. To guide clinical decisions on the optimum treatment regimen, there is clearly a need to accurately identify patients at high risk for recurrent or metastatic disease. Therefore, search of the genetic lesions identified by recent advances in cancer molecular biology for those relevant to predicting patient prognosis is considered to be of great importance. Many molecular markers that predict patient survival independent of the tumor-node-metastasis staging system have been reported [5]. These include oncogenes (K-ras, Bcl2, Her2/neu, EGFR), tumor suppressor genes (p53, RB, p16, p27), cell cycle modulators (cyclins), molecules related to tumor invasion and metastasis (CD44, cathepsin B, matrix metalloproteinase), telomerase, molecules involved in tumor angiogenesis (vascular endothelial growth factor, vascular endothelial growth factor receptor) and cyclo-oxygenase 2 [5]. However, for the moment there is no single biomarker available that can be routinely used for prediction of prognosis of NSCLC. This may be quite reasonable considering that cancer is a complex multigene disease. Recently, cDNA microarray technologies that simultaneously analyze the expression tens of thousands of genes have been used to correlate gene-expression patterns in individuals with various clinical parameters, including morphologic features and tumor behavior. In 2001, groups from Stanford [6] and Boston [7] applied expression profiling technologies to lung cancer, and they both concluded that (1) clusters defined by gene expression patterns recapitulate morphological classification of the tumors into squamous, small-cell, large-cell, and adenocarcinoma, and that (2) adenocarcinoma can be classified into subgroups that reflect patient survival. However, these two groups selected quite different sets of genes that influenced patient survival [6,7]. Similar subsequent studies also identified sets of genes of prognostic interest; but again, the selected sets of genes were quite different among reports [8-10]. In this study, we used real-time reverse transcriptase polymerase chain reaction (RT-PCR) to examine expression of 44 genes reported previously [6,7] as putative prognostic markers in our cohort of pulmonary adenocarcinoma patients. Our objectives were (1) to confirm whether expression patterns of these 44 genes were relevant to prognosis in our cohort of patients, and (2) to create a prognostic model that could reasonably be used in routine clinical practice by further selecting a smaller number of genes.
Cell Lines and Patients RNAs derived from five SCLC and 15 NSCLC cell lines were used for optimization of PCR conditions. Details of their derivation and culture conditions have been described [11,12]. Adenocarcinoma samples were obtained from 85 consecutive patients who underwent pulmonary resection at the Aichi Cancer Center Hospital (Nagoya, Japan) from December 1995 through May 1998, after obtaining approval from the institutional review board, and patients' written informed consent. The patients were 44 males and 41 females, with age at diagnosis ranging from 32 to 84 years (median age, 62 years). Forty-eight patients had stage I disease, six had stage II, 30 had stage III, and one patient had stage IV disease. Twenty-four patients had poorly differentiated; 47, moderately differentiated; and 14, well-differentiated adenocarcinoma. Thirty-eight patients were smokers, with a median Brinkman index (number of cigarettes per day x years) of 855; the remaining 47 were never smokers. For validation of our prognostic model, we used an independent set of 21 patients with pulmonary adenocarcinoma who had undergone pulmonary resection from March 1994 through November 1995. These patients were 12 males and nine females, with an age at diagnosis ranging from 43 to 78 years (median age, 60 years), and 13 patients were smokers with a median Brinkman index of 750. Eight patients had stage I disease, one had stage II, 10 had stage III, and two had stage IV disease. Nineteen patients had moderately differentiated, and two had well-differentiated adenocarcinoma. All patients underwent potentially curative resection (84 of 85 and 21 of 21 underwent formal pulmonary lobectomy or more, with systematic ipsilateral mediastinal lymph node dissection). One remaining patient underwent partial resection due to poor pulmonary reserve. Stages were determined after pathologic evaluation of resected specimens according to the International System for Staging Lung Cancer, revised in 1997 [13]. Tumor samples were processed immediately after surgical removal. Tissue specimens were grossly examined by a surgical pathologist (Y.Y.), and a piece of the tumor tissue was carefully obtained so as to maximize tumor content. A half of the piece was snap frozen in liquid nitrogen, followed by storage at -80°C until use. The other half was fixed with prechilled acetone, and embedded in paraffin for evaluation of tumor contents and scored in one of four classes (eg, < 25%, 25% to 50%, 50% to 75%, and > 75%). Total RNA was isolated using the acid guanidinium isothiocyanate/cesium chloride procedure [14]. All samples used in this study were analyzed by RT-PCR of glyceraldehyde-3-phosphate dehydrogenase (GAPDH) gene (539 base pair [bp]) to ensure the integrity of RNA using the following primers: 5'-GTCAACGGATTTGGTCGTATT-3' and 5'-AGTCTTCTGGGTGGCAGTGAT-3'.
Genes Examined in the Present Study
Relative Quantification by Real-Time RT-PCR First-strand cDNAs were synthesized from total RNA using Superscript II (Invitrogen, Carlsbad, CA) and random hexamer primers (Roche Applied Science, Alameda, CA). Real-time quantitative PCR amplifications were performed by SYBR Green assay in an ABI PRISM 7900-HT (Applied Biosystems). The reactions were carried out in a 96-well plate in 25-µL reactions containing 2 x SYBR Green Master Mix (Applied Biosystems), 200 to 250 nmol/L each, forward and reverse primer, and a cDNA template corresponding to 20 ng total RNA. SYBR Green PCR conditions were 50°C for 2 minutes, 95°C for 10 minutes, followed by 95°C for 50 seconds, 57°C or 60°C for 50 seconds (except Ret/Ptc2, 68°C), and 72°C for 1 minute for 40 to 50 cycles. In the SYBR Green Master Mix, there is an internal passive dye, ROX, in addition to the SYBR Green dye. The increase in the fluorescence of SYBR Green against that of ROX was measured at the end of each cycle. In each 96-well reaction plate, six standard samples, diluted up to 1/1000 of cDNA of any lung cancer cell line selected in preliminary experiments of each gene, were run with unknown tumor samples. Finally, relative quantitative values of each sample were compared with those of their 18S ribosomal RNA (rRNA) genes (186 bp), since expression of 18S rRNA was more consistent than expression of ß-actin (275 bp) or GAPDH (225 bp) among 20 cell lines in our preliminary experiments (ie, standard deviations of these housekeeping gene expressions were 0.87 for ß-actin, 0.56 for GAPDH, and 0.25 for 18S rRNA when expression levels in the ACC-LC-319 cell line were set to 1.0).
Hierarchical Clustering
Data Analysis
Relative Quantification by Real-Time RT-PCR We have preliminarily examined reliability of this assay, including reproducibility using cell line samples. Four of the 48 genes (EST AA468094, DDC [dopa decarboxylase], HNF3A [hepatocyte nuclear factor 3  ], and KIAA0767) did not give consistent bands on agarose gels or identical melting curve of PCR products, and thus, subsequent analyses were performed using the remaining 44 genes. We next asked whether there was a good correlation between real-time RT-PCR assay and immunohistochemistry (IHC). Since we previously examined overlapping cohorts of tumors by IHC of thyroid transcription factor 1 (TITF1) [16], we compared the results obtained by RT-PCR with IHC. TITF1 expressions were highly in agreement with mRNA quantities of our samples (P < .0001, Spearman rank correlation coefficient).
Hierarchical Clustering
Identification of a Smaller Number of Survival-Related Genes and Creation of a Prognostic Model Although it is of interest that the above unsupervised cluster analysis of 44 genes was able to identify patients at higher risk, examination of 44 genes is too laborious in clinical practice. Furthermore, the hierarchical clustering method can only be applicable to a retrospective analysis of a cohort of patients and cannot be used to predict clinical outcome for any future patients. Therefore, we tried to identify a smaller number of genes relevant to patient prognosis, and to create a prognostic model that could be applied prospectively. For the selection of genes, a stepwise multivariate Cox proportional hazards model was used, and eight genes were selected as significant independent prognostic factors (Table 4) when the cutoff P value was set at .1. The genes were PTK7, CIT, SCNN1A (sodium channel, nonvoltage-gated 1 alpha), PGES (prostaglandin E synthase), ERO1L (ERO1-like), ZWINT (ZW10 interactor), and two ESTs. Coefficients for PTK7, CIT, ERO1L, and EST AA434256 were negative, while those for the other four genes were positive, suggesting that high expression of the former four genes was associated with good prognosis, and that high expression of the latter four genes was associated with poor prognosis. We then calculated risk indices (RIs) that were defined as a linear combination of gene expression values weighed by their estimated regression coefficients. When the cutoff of the RI was set at the 50th percentile, Kaplan-Meier analysis and a log-rank test showed that there was a large difference in survival between high- and low-risk patients (P < .0001, log-rank test; Fig 3A). There was a significant association between RIs and tumor stage or tumor differentiation. Seventy percent of patients with stage III and IV disease, and 40% of patients with stage I and II disease belonged to the high risk group (P = .0082,  2 test). In addition, the high-risk group contained 57% of patients with poorly to moderately differentiated carcinoma, and only 15% of those with well-differentiated carcinoma (P = .0052). The prognostic stratification was even more prominent when patients were divided into four groups by setting the cutoffs at the 25th, 50th, and 75th percentiles (Fig 3B).
We also created a model including conventional prognostic factors (sex, age, differentiation, and stage) as well as RI. The Cox proportional hazards model selected pathological stage (hazard ratio [HR], 2.915; P = .0069) and RI (HR, 5.017; P = .0021) as independent prognostic factors (Table 5).
Since these eight genes were selected and regression coefficients were calculated to explain prognosis of these learning set of 85 patients, it was important to validate this approach by applying the model to an independent set of patients. Therefore, quantitative RT-PCR was performed for these eight genes, and RI was calculated in a test set of 21 additional adenocarcinoma patients who had undergone surgery from 1994 through 1995—a period just before the learning set of cases was collected. The Kaplan-Meier curves stratified by RI are shown in Figure 4. The difference in overall survival between patients with low RI and high RI was also significant (P = .0273, log-rank test), validating this model. As with the learning set, multivariate analysis with the test cases also identified RI (HR, 6.562; P = .0085) and pathological stage (HR, 14.819; P = .0038) as independent prognostic factors (Table 6).
cDNA microarray technology is promising in that it can analyze thousands of unselected genes simultaneously, and thus, it gives a comprehensive view of gene expression characteristics of tumors in each patient. However, microarray technology is still developing, and it has potential technical variances that may compromise the reproducibility of results. These technical variances may be derived from variation in printing or processing of chips, hybridization or scanning, sample preparation, or probes [17]. In addition, cDNA chips are still very expensive for routine clinical use. As a complementary approach, we determined the expression levels of each gene by quantitative RT-PCR, which can provide more accurate and reproducible RNA quantification and requires smaller quantities of tumor tissue [18-20]. In the present study, we did not use microdissected tissue but used bulk of cancer tissue. This may raise argument that contamination of normal stromal cells may dilute molecular characteristic of tumors of interest. However, we have shown that the degree of tumor content did not significantly affect clustering. In a clinical setting, it is too laborious to perform microdissection for each patient. Furthermore, gene-expression signature might arise from both malignant and stromal elements in primary tumors, as suggested by Ramaswamy et al [10]. We first analyzed our expression data using unsupervised hierarchical clustering. Hierarchical clustering separates samples into subgroups of related expression patterns in an unbiased manner. Although the number of genes we examined was small, we were able to cluster adenocarcinoma into three groups, confirming the heterogeneity of this type of NSCLC as has been suggested by other researchers [6-8]. Patients in group B by our expression profiling had a significantly poor prognosis. Group B had more male elderly patients with smoking habits and larger tumors. However, there was no statistical difference in terms of pleural invasion or lymph node metastasis. Tumors in group B were characterized by lower expression of adhesion molecules, such as DPP4, ICAM1, and CEACAM1, and growth suppressors such as HPN [21] and DOC-1R [22]. ICAM1 is involved in intracellular signaling in a variety of physiological and pathological processes, including metastasis and tumor growth [23]. WFDC2, TITF1, and CIT were also low in expression in group B. WFDC2 (HE4), a protease inhibitor, is expressed in pulmonary epithelium and may be part of the host defense shield of the airways [24]. It was suggested to be a growth inhibitor [25], as similarly, its family member gene WFDC1 is a candidate tumor suppressor gene. TITF1 is implicated in the regulation of surfactant gene expression, and TITF1 expression is initiated at a very early stage of lung morphogenesis [26]. CIT is a serine/threonine kinase and rho effector suggested as a crucial regulator in cytokinesis [27]. SLC2A1 is also known as glucose transporter (GLUT1), and it was highly expressed in group B. Increased SLC2A1 expression is observed under conditions that induce greater dependence on glycolysis as an energy source, such as ischemia or hypoxia [28-30]. These data suggest that overexpression of SLC2A1 may play an important role in the survival tumor cells, especially in large tumors. In the literature, decreased expression of ICAM1 [31] and TITF1 [16], and increased expression of SLC2A1 [32], have been previously reported to be poor prognostic factors in NSCLC. For several genes, the prognostic impact appeared to be somewhat at odds with their function. Although protein tyrosine kinases (PTKs) are regarded as oncogenes, PTK7 was more strongly expressed in tumors with favorable prognosis, in agreement with the Stanford group [6]. It is also reported to be downregulated during melanoma development [33]. DUSP4 (MKP-2) is one of the dual specificity phosphatases, which are also known as MKPs (mitogen-activated protein kinase phosphatases). DUSP4 inactivates MAPKs (extracellular signal-regulated kinase and c-Jun N-terminal kinase) [34] and is thought to be a negative growth regulator. However, high expression of DUSP4 was associated with poor prognosis in the present study, in agreement with the Boston group [7] and with the Stanford group [6]. Indeed, DUSP4 was recently reported to be a possible component of a novel, transforming pathway [35], and to be highly expressed in hepatoma [36] and pancreatic tumor cell lines [37]. Moreover, it is reported that expression of MKP-1, highly homologous to DUSP4, is associated with a shorter progression-free survival in ovarian tumors [38]. Although the Boston group reported that adenocarcinoma with overexpression of neuroendocrine markers had the worst prognosis [7], our group C, which had highest expression of neuroendocrine markers, did not have poor prognosis. Using Cox proportional hazards model, we selected eight genes that would jointly predict patient prognosis. These were PTK7, CIT, SCNN1A, PGES, ERO1L, ZWINT, and two ESTs, and half of these genes also characterized hierarchical cluster group B. PGES has been shown to stimulate transcription, influence mitogenesis of normal human bone cells, and to promote growth and metastasis of tumors [39]. PGE2 is synthesized from PGH2, whose production is catalyzed by cyclo-oxygenase (COX), and inducible PGES is specifically overexpressed in NSCLC [40]. Overexpression of inducible form of COX (COX-2) is shown to be associated with poor prognosis in NSCLC patients [41]. ZWINT, human ZW10 interacting protein-1, plays an important role in normal centromere function [42]. SCNN1A is related to hypertension [43], and ERO1L is thought to be involved in oxidative protein folding in the endoplasmic reticulum [44]. Although the mechanisms by which some of these genes affect patient prognosis are not very clear, it seemed that whole set of expression pattern of these genes might have the important information for cancer prognosis. In conclusion, we were able to select eight genes that were more relevant to patients' survival from expression data obtained by real-time PCR. Both unsupervised clustering and a supervised method were useful to predict survival of patients with pulmonary adenocarcinoma. It is of special clinical interest that the risk index calculated from expression of a reasonably small number of genes may be useful in routine clinical practice.
The authors indicated no potential conflicts of interest.
We thank Dr Hamajima, professor of Nagoya University, for help in statistical approaches such as Cox proportional hazards regression modeling.
Supported in part by the Aichi Cancer Research Foundation. Authors’ disclosures of potential conflicts of interest are found at the end of this article.
1. Kuroishi T, Hirose K, Tajima K, et al: Cancer mortality in Japan, in Tominaga S, Oshima A (eds): Cancer mortality and morbidity statistics. Gann Monograph Cancer Research, No. 47, Tokyo, Japan, Japan Scientific Societies Press, 1999, pp 1-38 2. Vaporciyan AA, Nesbitt JC, Lee JS, et al: Cancer of the lung, in Bast RC, Kufe DW, Pollock RE, et al (eds): Cancer Medicine (ed 5). Hamilton, Canada, B.C. Decker Inc, 2000, pp 1227-1292 3. Balsara BR, Testa JR: Chromosomal imbalance in human lung cancer. Oncogene 21:6877-6883, 2002[CrossRef][Medline] 4. Fry WA, Phillips JL, Menck HR: Ten-year survey of lung cancer treatment and survival in hospitals in the United States: A national cancer data base report. Cancer 86:1867-1876, 1999[CrossRef][Medline]
5. Brundage MD, Davies D, Mackillop WJ: Prognostic factors in non-small cell lung cancer: A decade of progress. Chest 122:1037-1057, 2002
6. Garber ME, Troyanskaya OG, Schluens K, et al: Diversity of gene expression in adenocarcinoma of the lung. Proc Natl Acad Sci U S A 98:13784-13789, 2001
7. Bhattacharjee A, Richards WG, Staunton J, et al: Classification of human lung carcinomas by mRNA expression profiling reveals distinct adenocarcinoma subclasses. Proc Natl Acad Sci U S A 98:13790-13795, 2001 8. Beer DG, Kardia SL, Huang CC, et al: Gene-expression profiles predict survival of patients with lung adenocarcinoma. Nat Med 8:816-824, 2002[Medline]
9. Wigle DA, Jurisica I, Radulovich N, et al: Molecular profiling of non-small cell lung cancer and correlation with disease-free survival. Cancer Res 62:3005-3008, 2002 10. Ramaswamy S, Ross KN, Lander ES, et al: A molecular signature of metastasis in primary solid tumors. Nat Genet 33:49-54, 2003[CrossRef][Medline]
11. Takahashi T, Ueda R, Song X, et al: Two novel cell surface antigens on small cell lung carcinoma defined by mouse monoclonal antibodies NE-25 and PE-35. Cancer Res 46:4770-4775, 1986
12. Hida T, Ariyoshi Y, Kuwabara M, et al: Glutathione S-transferase pi levels in a panel of lung cancer cell lines and its relation to chemo-radiosensitivity. Jpn J Clin Oncol 23:14-19, 1993
13. Mountain CF: Revisions in the international system for staging lung cancer. Chest 111:1710-1717, 1997 14. Sambrook J, Fritsch EF, Maniatis T: Isolation of total RNA by extraction with strong denaturants, in Molecular Cloning: A Laboratory Manual, New York, NY, Cold Spring Harbor Laboratory Press, 1989, pp 7.18-7.25
15. Eisen MB, Spellman PT, Brown PO, et al: Cluster analysis and display of genome-wide expression patterns. Proc Natl Acad Sci U S A 95:14863-14868, 1998 16. Yatabe Y, Mitsudomi T, Takahashi T: TTF-1 expression in pulmonary adenocarcinomas. Am J Surg Pathol 26:767-773, 2002[CrossRef][Medline] 17. Miller LD, Long PM, Wong L, et al: Optimal gene expression analysis by microarrays. Cancer Cell 2:353-361, 2002[CrossRef][Medline]
18. Dhar AK, Roux MM, Klimpel KR: Detection and quantification of infectious hypodermal and hematopoietic necrosis virus and white spot virus in shrimp using real-time quantitative PCR and SYBR Green chemistry. J Clin Microbiol 39:2835-2845, 2001
19. Nazarenko I, Lowe B, Darfler M, et al: Multiplex quantitative PCR using self-quenched primers labeled with a single fluorophore. Nucleic Acids Res 30:e37, 2002
20. Riou P, Saffroy R, Comoy J, et al: Investigation in liver tissues and cell lines of the transcription of 13 genes mapping to the 16q24 region that are frequently deleted in hepatocellular carcinoma. Clin Cancer Res 8:3178-8316, 2002
21. Srikantan V, Valladares M, Rhim JS, et al: HEPSIN inhibits cell growth/invasion in prostate cancer cells. Cancer Res 62:6812-6816, 2002 22. Zhang X, Tsao H, Tsuji T, et al: Identification and mutation analysis of DOC-1R, a DOC-1 growth suppressor-related gene. Biochem Biophys Res Commun 255:59-63, 1999[CrossRef][Medline]
23. Yasuda M, Tanaka Y, Tamura M, et al: Stimulation of beta1 integrin down-regulates ICAM-1 expression and ICAM-1-dependent adhesion of lung cancer cells through focal adhesion kinase. Cancer Res 61:2022-2030, 2001 24. Bingle L, Singleton V, Bingle CD: The putative ovarian tumour marker gene HE4 (WFDC2), is expressed in normal tissues and undergoes complex alternative splicing to yield multiple protein isoforms. Oncogene 21:2768-2773, 2002[CrossRef][Medline]
25. Larsen M, Ressler SJ, Lu B, et al: Molecular cloning and expression of ps20 growth inhibitor: A novel WAP-type "four-disulfide core" domain protein expressed in smooth muscle. J Biol Chem 273:4574-4584, 1998 26. Minoo P, Su G, Drum H: Defects in traceoesophageal and lung morphogenesis in Nkx2.1(-/-) mouse embryos. Dev Biol 209:60-71, 1999[CrossRef][Medline]
27. Di Cunto F, Calautti E, Hsiao J, et al: Citron rho-interacting kinase, a novel tissue-specific ser/thr kinase encompassing the Rho-Rac-binding protein Citron. J Biol Chem 273:29706-29711, 1998 28. Kurata T, Oguri T, Isobe T, et al: Differential expression of facilitative glucose transporter (GLUT) genes in primary lung cancers and their liver metastases. Jpn J Cancer Res 90:1238-1243, 1999[CrossRef][Medline] 29. Merrall NW, Plevin R, Gould GW: Growth factors, mitogens, oncogenes and the regulation of glucose transport. Cell Signal 5:667-675, 1993[CrossRef][Medline]
30. Clavo A, Brown R, Wahl R: 2-fluoro-2-deoxy-D-glucose (FDG) uptake into human cancer cell lines is increased by hypoxia. J Nucl Med 36:1625-1632, 1995 31. Passlick B, Izbicki JR, Simmel S, et al: Expression of major histocompatibility class I and class II antigens and intracellular adhesion molecule-1 on operable non-small cell lung carcinomas: Frequency and prognostic significance. Eur J Cancer 30:376-381, 1994[CrossRef] 32. Younes M, Brown RW, Stephenson M, et al: Overexpression of Glut1 and Glut3 in stage I nonsmall cell lung carcinoma is associated with poor survival. Cancer 80:1046-1051, 1997[CrossRef][Medline] 33. Eastey DJ, Mitchell PJ, Patel K, et al: Loss of expression of receptor tyrosine kinase family genes PTK7 and SEK in metastatic melanoma. Int J Cancer 71:1061-1065, 1997[CrossRef][Medline]
34. Hirsch DD, Stork PJ: Mitogen-activated protein kinase phosphatases inactivate stress-activated protein kinase pathways in vivo. J Biol Chem 272:4568-4575, 1997 35. Fu SL, Waha A, Vogt PK: Identification and characterization of genes upregulated in cells transformed by v-Jun. Oncogene 19:3537-3545, 2000[CrossRef][Medline] 36. Yokoyama A, Karasaki H, Urushibara N, et al: The characteristic gene expressions of MAPK phosphatases 1 and 2 in hepatocarcinogenesis, rat ascites hepatoma cells, and regenerating rat liver. Biochem Biophys Res Commun 239:746-751, 1997[CrossRef][Medline] 37. Yip-Schneider MT, Lin A, Marshall MS: Pancreatic tumor cells with mutant K-ras suppress ERK activity by MEK-dependent induction of MAP kinase phosphatase-2. Biochem Biophys Res Commun 280:992-997, 2001[CrossRef][Medline] 38. Denkert C, Schmitt WD, Berger S, et al: Expression of mitogen-activated protein kinase phosphatase-1 (MKP-1) in primary human ovarian carcinoma. Int J Cancer 102:507-513, 2002[CrossRef][Medline]
39. Sales KJ, Katz AA, Howard B, et al: Cyclooxygenase-1 is up-regulated in cervical carcinomas: Autocrine/paracrine regulation of cyclooxygenase-2, prostaglandin E receptors, and angiogenic factors by cyclooxygenase-1. Cancer Res 62:424-432, 2002
40. Yoshimatsu K, Altorki NK, Golijanin D, et al: Inducible prostaglandin E synthase is overexpressed in non-small cell lung cancer. Clin Cancer Res 7:2669-2674, 2001
41. Achiwa H, Yatabe Y, Hida T, et al: Prognostic significance of elevated cyclooxygenase 2 expression in primary, resected lung adenocarcinomas. Clin Cancer Res 5:1001-1005, 1999 42. Starr DA, Saffery R, Li Z, et al: Hzwint-1, a novel human kinetochore component that interacts with HZW10. J Cell Sci 113:1939-1950, 2000[Abstract]
43. Iwai N, Baba S, Mannami T, et al: Association of a sodium channel alpha subunit promoter variant with blood pressure. J Am Soc Nephrol 13:80-85, 2002
44. Cabibbo A, Pagani M, Fabbri M, et al: ERO1-L, a human protein that favors disulfide bond formation in the endoplasmic reticulum. J Biol Chem 275:4827-4833, 2000 Submitted April 14, 2003; accepted October 3, 2003. This article has been cited by other articles:
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
TOP
ABSTRACT
INTRODUCTION
