Eur Urol Suppl 2009;8(4):377

1027Gene SIGnaTUReS OF pULMOnaRY RenaL ceLL caRcInOMa (Rcc) MeTaSTaSeS ReFLecT MeTaSTaSeS-FRee InTeRvaL and nUMBeR OF MeTaSTaSeS OF THe paTIenT

Wuttig D.1, Meinhardt M.2, Baier B.3, Fuessel S.1, Hoefling C.1, Toma M.I.2, Grimm M.O.1, Meye A.1, Rolle A.3, Wirth M.P.1

1Technical University of Dresden, Dept. of Urology, Dresden, Germany, 2Technical University Dresden, Dept. of Pathology, Dresden, Germany, 3Coswig Specialised Hospital, Thoracic and Vascular Surgery, Coswig, Germany

Introduction & Objectives: In RCC, where metastasis occurs in up to 60% of the patients, molecular prognosis markers are lacking. Knowing the molecular basis of metastatic spread would provide a promising tool for the identification of prognostic markers. We investigated a unique pool of clear cell RCC lung metastases (Mets) in order to identify expression signatures associated with two important prognostic factors in RCC, the metastases-free interval after nephrectomy (MSFI) and the number of Mets.

Material & Methods: Transcriptome-wide expression analysis of 20 cryo-preserved clear cell RCC lung Mets of 18 patients obtained during laser-based resection was performed (HG-U133 Plus 2.0 arrays, Affymetrix). The patients had undergone nephrectomy and pulmonary surgery (median follow-up after pulmonary surgery: 12 months) without any further immune therapy. RMAExpress software was used for data preprocessing and calculation of expression signals. Differentially expressed genes (fold change ≥1.8; t-test; false discovery rate 0.05) were identified using GenePattern software. Over-represented biological processes were defined using the Gene Ontology Mining Tool (Affymetrix). Prediction models were generated by weighted voting (WV) and k-nearest neighbouring (KNN; GeneCluster 2.0), based on the differentially expressed genes verified by two independent mathematical algorithms (GCOS, dCHIP).

Results: Comparing expression profiles of Mets derived from patients with ≤8 and ≥16 pulmonary Mets resulted in 135 differentially expressed genes. Based on the 35 genes verified with the two additional mathematical procedures (GCOS, dCHIP), we build an 11 gene signature (e.g. PBK, BCMP11, TITF1, GPR64) suitable for a correct leave-1-out cross validation (17 samples) and able to classify 3 independent samples correctly (Fisher test: p=0.001). Early (MSFI ≤9 month) and late Mets (MSFI ≥5 years) varied in 306 differentially expressed genes. Fifty-five of them were verified by additional algorithms (GCOS, dCHIP). Six of these (e.g. CD31, HSPG2, TSPAN7) were sufficient for a correct leave-1-out cross validation (11 samples; Fisher test: p=0.002) and classification of an additional sample into the appropriate group. Concerning biological processes (Gene Ontology), an activated cell division (e.g. PTTG1, BIRC5) was associated with a higher number of Mets, whereas differences of MSFI were related to a different metastatic potential of Mets (angiogenesis, cell migration; e.g. PDGFB, ETS1). Microarray results were completely confirmed by quantitative PCR (6/6 genes).

conclusions: Mets derived from patients with differing numbers of Mets or different MSFI are distinguishable based on their expression profiles. Further analyses will reveal, which of these features are already present in matched primary tumors and therefore, suitable for prognostic purposes.

1028pTen can FUncTIOn aS a TUMOUR SUppReSSOR In cLeaR ceLL RenaL caRcInOMa

Barod R.1, Domin J.2, Eccles S.3, O’ Brien T.S.4, Ashcroft M.1, Maxwell P.H.1

1University College London, Centre for Cellular Signalling & Molecular Genetics, London, United Kingdom, 2Imperial College London, Dept. of Medicine, London, United Kingdom, 3Institute of Cancer Research, Centre For Cancer Therapeutics, London, United Kingdom, 4Guy’s & St. Thomas’ NHS Foundation Trust, The Urology Centre, London, United Kingdom

Introduction & Objectives: PTEN is the 2nd most frequently inactivated tumour suppressor in human cancers and the PTEN gene is mutated in 10-33% of sporadic clear cell renal cell carcinomas (ccRCC). While its role has been well studied in other neoplasms, little is known about its function in ccRCC. Approximately 75% of patients with sporadic ccRCC have inactivation of the von Hippel Lindau (VHL) tumor suppressor. VHL loss leads to constitutive activation of hypoxia inducible factor (HIF). Activation of this transcription factor promotes angiogenesis, cell proliferation, growth and migration. The relatively aggressive 786-O ccRCC cell line is both PTEN and VHL deficient. Therefore, we asked whether PTEN functions as a tumour suppressor in ccRCC.

Material & Methods: Gain of function studies were performed on 786-O cells, which were retrovirally infected with PTEN, VHL or both to produce isogenic sublines. Cells were analysed for growth, migration, protein and mRNA expression. Tumour xenograft experiments were performed by injecting the isogenic 786-O cells into nude mice. Loss of function studies were performed by silencing PTEN in RCC4 and RCC10 cells. Pathological specimens of human ccRCC were analysed for PTEN expression.

Results: 1. PTEN expression is reduced in a subset of human ccRCC. 2. Restoration of PTEN in 786-O cells suppresses VEGF mRNA. 3. Silencing PTEN in RCC4 and RCC10 cells induces VEGF mRNA. 4. Restoration of PTEN in 786-O cells inhibits anchorage independent growth. 5. Restoration of PTEN in 786-O cells inhibits cell migration. 6. The effect of PTEN on VEGF, growth and migration is independent of VHL. 7. Restoration of PTEN in 786-O cells inhibits tumour growth as xenografts in nude mice.

conclusions: We have shown that PTEN is able to regulate growth, migration and VEGF levels in ccRCC cells. Additionally, restoration of PTEN inhibits tumour growth as xenografts in vivo. Collectively, our results show that PTEN is able to function as a tumour suppressor in ccRCC and that loss of PTEN may be a component of the multi-step process of ccRCC tumourigenesis. This makes PTEN a potential target for therapy in ccRCC.

1026pOLYMORpHIc deLeTIOnS OF GLUTaTHIOne-S-TRanSFeRaSe aFFecT THe RISK OF RenaL ceLL caRcInOMa and aGe aT dIaGnOSIS

De Martino M., Klatte T., Schatzl G., Remzi M., Waldert M., Kramer G., Marberger M.

Medical University of Vienna, Dept. of Urology, Vienna, Austria

Introduction & Objectives: Glutathione-S-transferases (GST) are cytosolic and transmembrane enzymes involved in detoxification of a wide range of endogenous and exogenous substrates, including products of oxidative stress and carcinogens. These enzymes are encoded by a total of 8 genes, of which complete genomic deletions of GST-M1 and GST-T1 have been associated with an increased risk of developing laryngeal, cervical and bladder cancer. The aim of this study was to investigate the association between GST-M1 and GST-T1 deletions and the risk of renal cell carcinoma (RCC).

Material & Methods: We performed a hospital-based, case-control study of 135 subjects, including 65 with histologically confirmed sporadic RCC and 70 age- and gender-matched controls without evidence or history of malignancy. Genomic DNA was extracted from whole blood using standard techniques. Multiplex polymerase chain reaction was performed to assess GST-M1 and GST-T1 genotypes. Associations with the risk of RCC and age at diagnosis were analyzed with chi-square tests, t-tests, and logistic regression.

Results: Cases and controls were similar in terms of age and gender distribution (p=0.919 and 0.879, respectively). Complete deletions (null genotype) of GST-M1 and GST-T1 were noted in 56% and 19% of the subjects, respectively. Among men, the GST-M1 null genotype carried a 2.61-fold increased odds of developing RCC compared with the controls (95% CI 1.07-6.36, p=0.035). In contrast, none of the GST-1 deletions affected the risk for RCC in women. Moreover, RCC patients showing the GST-T1 null genotype were significantly younger than their counterparts (mean age at diagnosis: 55.3 vs. 67.6 years, p=0.013).

conclusions: Polymorphic deletions of GST-M1 and GST-T1 increase the risk for RCC in men and impact age at diagnosis.

1025Gene expReSSIOn anaLYSIS In cLeaR ceLL RenaL ceLL caRcInOMa (ccRcc) FOR IdenTIFYInG pOTenTIaL MaRKeRS In pROGnOSIS and THeRapY cOnTROL

Maruschke M.1, Koczan D.2, Reuter D.2, Thiesen H.J.2, Hakenberg O.W.2

1University of Rostock, Dept. of Urology, Rostock, Germany, 2University of Rostock, Institute of Immunology, Rostock, Germany

Introduction & Objectives: The aim of the study was to establish aberrant genetic signatures of clear cell renal carcinoma (ccRCC) predisposing for metastatic spread.

Material & Methods: Fresh-frozen tumor tissue specimens of 29 patients with ccRCC were used, 15/29 patients with metastatic G3 ccRCC, and 14/29 from non-metastatic G1 ccRCC. Epidemiological, clinicopathological features (TNM stage, nuclear grade) and tumor specific survival data were obtained. Expression profiling was performed and candidate genes were discovered using the Human Genome U133 Plus 2.0 Array (Affymetrix Corp., Santa Clara, CA, USA), which interrogates 47.000 transcripts using 54.000 probesets in a genomewide manner. Each tumor tissue was compared to a non-neoplastic tissue sample from the same kidney. The primary probe level analysis was performed using the Affymetrix MAS 5 algorithm. Further analysis included an ER-scoring (event ratio scoring) and Gene Set Enrichment analysis (GSEA). For subsequent hierarchical clustering of the log-ratios of the fold-changes, the expression profile of each tumor sample was compared to the corresponding normal renal tissue sample.

Results: Hierarchical clustering provided some degree of separation of the two different tumour groups – non-metastatic G1 versus metastatic G3 - reflecting the fact that the genes under consideration were chosen for being different in these particular phenotypes, but did not prove sufficiently robust to provide a unanimous separation of the two tumor entities G1 and G3. To gain more specific gene selections, a voting scheme was applied to the complete dataset providing a number of 100 probesets from the total of 54000 probesets, thus, reducing technical and biological noise. In comparison of G1-tumors we found 158 up- and 340 down-regulated genes, in G3-tumors 124 up- and 412 down-regulated candidates were seen. Comparison of G1- and G3-tumors showed 24 upregulated and 49 down-regulated candidate genes.

conclusions: Sequences, which are normally expressed in healthy tissue and down-regulated in tumor tissue, decrease gradually from G1 to G3 or disappear completely in G3. Out of these gene sequences we verified the presence of several previously identified tumor associated genes and in addition several other genes: aldolase B, CUB domain-containing protein1, aquaporin 9, retinoic acid receptor responder, solute carrier transporter family 12A8, melanotransferrin / P97 (MFJ2), serum amyloid A1, interleukin 20, interleukin receptor 2, PTHLH, ceruloplasmin (ferrooxidase), laminin (LAMB3), which might signify the transition from G1 to G3 with metastatic behaviour.