PTTG1

1. Introduction and Objectives • Results of array analyses • early vs. late Mets • early Mets: PFS = 0 - 9 months, median = 1 month, n = 5 • late Mets: PFS = 60-156 months, median = 92.5 months, n = 6 • 306 differentially expressed genes (414 probe sets; Fig.1), 27 & 279 in early Mets biological processes: • motility/migration, angiogenesis, cell adhesion (e.g. KDR, PDGFRB, ETS1, PECAM1) = processes involved in metastatic spread → “activated” in “late Mets”  higher metastatic potential of late, or metachro- nous, Mets (Fig.2)? • prediction model: • 6/306 genes sufficient for prediction (WV, Tab.1) • → correct leave-1-out cross validation (p = 0.002) • → correct classification of 3/4 additional Mets (one late & two early Mets; one early Met misclassified) • Results of qPCR analyses •  validation of microarray results • six differentially expressed genes selected: • early vs. late Mets: GALNTL2, HSPG2, PECAM1 • few vs. multiple Mets: PTTG1, AGR3, CEACAM6  differential expression of the six genes in initial sample cohort confirmed (examples in Fig.4)  five genes (except CEACAM6) also differentially expressed when including additional samples (examples in Fig.4) • patients (pts) with renal cell carcinoma (RCC) have a high metastatic risk • molecular basis of particular characteristics of metastatic spread, like dormancy period or number of metastases (Mets) is largely unknown  molecular prognosis markers are lacking  aim of the study: investigation of an unique tool of pulmonary Mets of clear‑cell RCC in order to identify expression patterns associatedwith • multiple vs. few Mets • multiple Mets: no. of Mets = 16 - >80, median = 24, n = 7 • few Mets: no. of Mets = 1 - 8, median = 3, n = 10 • 135 differentially expressed genes (163 probe sets; Fig.3), 85 & 50 in “few Mets” biological processes: • cell division (e.g. PBK, BIRC5, PTTG1) → “activated” in “multiple Mets”  higher number of Mets might result from an increased growth potential prediction model: • 10/135 genes sufficient for prediction (KNN) • → correct leave-1-out cross validation (p <0.001) • → correct classification of 6/7 additional Mets • (four pts with few & two with multiple Mets; one pt with multiple Mets misclassified) • the progression-free survival after nephrectomy (PFS): early vs. late Mets • and the number of pulmonary Mets: multiple vs. few Mets • Material and Methods • Patients and tissue specimens • microarray analyses: 24 pulmonary Mets of 22 pts with clear-cell RCC obtained during 1318 nm-laser resection • validation by quantitative PCR (qPCR): nine additional pulmonary Mets • all pts had undergone nephrectomy, had clinically no other distant Mets before diagnosis of lung Mets, no immune- or immunchemotherapy (except one pt) and no other primary tumors Workflow early vs. late Mets: 9 additional Mets from pts with a median PFS (PFS=11–38 months) were classified based on a prediction model (KNN, WV) containing the 306 differentially expressed genes A snap-freezing of Mets in liquid nitrogen 8/9 Mets (PFS=11-38 months) 1/9 Mets (PFS=15 months) n-fold expression related to a fictitious samples with DCP=0 generation of cryo-sections (4µm) (tumor cell amount 70%) classified to classified to B early Mets (PFS ≤9 months) late Mets (PFS ≥5 years) since 8/9 Mets with a relatively short PFS (11-36 months) were classified to late Mets, the 306 differentially expressed genes might represent molecular differences of Mets developed synchronously or metachronouslyto the primary tumor RNA isolation (RNeasy Mini Kit, Qiagen; RNA integrity number 4.5 for array processing) Analyses of microarray data • pre-processing: global background correction, quantile normalization, calculation of expression signals & log2 transformation (RMAExpress 4.1) • genes defined to be differentially expressed if one or several probe sets representing these genes showed a fold-change ≥1.8 & a statistically significant difference (t-test, 100 permutations; false discovery rate, FDR, <5%) between average expression values of compared groups • identification of statistically significantly enriched biological processes (FDR <5%; fold-enrichment ≥2) in the list of the probe sets representing the differentially expressed genes compared to all probe sets on the array representing these processes using DAVID Bioinformatics Resources (http://david.abcc.ncifcrf.gov) • prediction models consist of minimal sets of genes fulfilling a correct leave-1-out cross validation (k-nearest neighbouring, KNN; weighted voting, WV; Fishers exact test); input for generation: differentially expressed genes verified by two independent mathematical algorithms (GCOS 1.4: FDR <5%; dCHIP 1.3: PM‑only, outlier detection, p <0.05) Fig.4: Gene expression quantified by qPCR for validation of expression patterns associated with a varying number of Mets (PTTG1) and PFS (GALNTL2). microarray processing (One-Cycle Target Labeling and Control Reagents Kit, HG‑U133 Plus 2.0 arrays, both Affymetrix) quantitative PCR  supposition when including functional relevance of differentially expressed genes : Late (metachronous) Mets might develop a higher metastatic potential right from the start of their growth, since the primary tumor does no longer exist as a source of metastatic spread. Gene expression values are shown for the initial sample cohorts (A) and for inclusion of additional samples (B). Fold changes (FC) = ratio of mean values “late vs. early Mets” or “multiple vs. few Mets” (array data: FC=2Dlog). Statistical significance of the expression differences measured by qPCR was determined by a two-sided heteroscedastic t-test based on the log2 transformed expression values. (Boxplots: median value, values within the 25. and 75. percentile, minimum and maximum expression value; n=number of Mets) validation Fig.1: Early vs. late Mets (complete linkage hierarchical clustering, Euclidean distance). Visualization of gene expression (heat map; columns and rows represent samples and probe sets, respectively) based on the 306 differentially expressed genes indicates the suitability of these genes to distinguish between the compared groups. Fig.3: Multiple vs. few Mets (complete linkage hierarchical clustering, Euclidean distance). Visualization of gene expression (heat map) based on the 135 differentially expressed genes indicates their suitability to distinguish between the compared groups. Fig.2: Classification of nine additional Mets to the „early“ or „late Mets“ groups indicates the existence of molecular differences between synchronously and metachronously developed Mets. PTTG1 GALNTL2 • Conclusion and Perspectives • Mets derived from pts with different PFS or differing numbers of Mets are distinguishable based on their expression profiles • present analyses are going to reveal, which of the identified features are already present in matched primary tumors and therefore, suitable for prognostic purposes (e.g. TSPAN7) • Quantitative PCR • cDNA synthesis: SuperScript II RNase H-Reverse Transcriptase (Invitrogen) & random hexamer primers • qPCR using TaqMan Gene Expression Assays (Applied Biosystems): BCMP11: Hs00411286_m1, CEACAM6: Hs00366002_m1, GALNTL2: Hs00365065_m1, HSPG2: Hs01078536_m1, PECAM1: Hs00169777_m1, PTTG1: Hs00851754_u1, reference gene TBP: Hs00427620_m1 •  calculation of transcript levels using the DDCP method [1] References: [1] Livak KJ & Schmittgen TD (2001) Methods 25:402. Acknowledgement:This study was kindly supported by the Dr.-Robert-Pfleger foundation. early Mets late Mets multiple Mets few Mets array results • Gene signatures of pulmonary renal cell carcinoma metastases reflect the progression-free interval and the number of metastases per patient Abstract no. 1027 D. Wuttig1, M. Meinhardt 2, M. Toma 2, S. Zastrow 1, C. Hoefling 1, B. Baier 3, S. Fuessel 1, M.-O. Grimm 1, A. Meye 1, A. Rolle 3 and M.P. Wirth 1 1 Department of Urology, 2 Institute of Pathology, Technical University of Dresden, Germany; 3 Department of Thoracic and Vascular Surgery, Center for Pneumology, Thoracic and Vascular Surgery, Coswig Specialised Hospital, Germany Tab.1: Six gene signature for the differentiation between late and early Mets. {Fold changes = 2Dlog} .