Clinical Cancer Research Candidate Pathways, Whole Genome Scans: Reconciling Results, Looking into the Future Translational Cancer Medicine 2008: Cancer Clinical Trials and Personalized Medicine
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
Cancer Research Clinical Cancer Research
Cancer Epidemiology Biomarkers & Prevention Molecular Cancer Therapeutics
Molecular Cancer Research Cancer Prevention Research
Cancer Prevention Journals Portal Cancer Reviews Online
Annual Meeting Education Book Cell Growth & Differentiation
This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Haller, F.
Right arrow Articles by Füzesi, L.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Haller, F.
Right arrow Articles by Füzesi, L.
Clinical Cancer Research Vol. 11, 6589-6597, September 15, 2005
© 2005 American Association for Cancer Research

Imaging, Diagnosis, Prognosis

Prognostic Role of E2F1 and Members of the CDKN2A Network in Gastrointestinal Stromal Tumors

Florian Haller1, Bastian Gunawan1, Anja von Heydebreck4, Stefanie Schwager1, Hans-Jürgen Schulten1, Judith Wolf-Salgó1, Claus Langer2, Giuliano Ramadori3, Holger Sültmann5 and László Füzesi1

Authors' Affiliations: Departments of 1 Pathology, 2 Surgery, 3 Gastroenterology, University of Göttingen, Germany; 4 Department of Bio- and Chemoinformatics, Merck KGaA, Darmstadt, Germany; and 5 Department of Molecular Genome Analysis, German Cancer Research Center, Heidelberg, Germany

Requests for reprints: Florian Haller, Department of Pathology, University of Göttingen, Robert-Koch-Str. 40, D-37075 Göttingen, Germany. Phone: 49-551-396858; Fax: 49-551-398627; E-mail: florian.haller{at}med.uni-goettingen.de.


   Abstract
Top
 Abstract
Materials and Methods
 Results
 Discussion
 References
 
Purpose: The aim of the current study was to examine the prognostic relevance of the CDKN2A tumor suppressor pathway in gastrointestinal stromal tumors (GIST).

Experimental Design: We determined the mRNA expression of p1INK4A, p14ARF, CDK4, RB1, MDM2, TP53, and E2F1 by quantitative reverse transcription-PCR in 38 cases of GISTs and correlated the findings with clinicopathologic factors, including mutation analysis of KIT and PDGFRA.

Results: The k-means cluster analysis yielded three prognostic subgroups of GISTs with distinct mRNA expression patterns of the CDKN2A pathway. GISTs with low mRNA expression of the CDKN2A transcripts p16INK4A and p14ARF but high mRNA expression of CDK4, RB1, MDM2, TP53, and E2F1 were associated with aggressive clinical behavior and unfavorable prognosis, whereas GISTs with a low mRNA expression of CDK4, RB1, MDM2, TP53, and E2F1 were not. GISTs with a moderate to high mRNA expression of all examined genes also seemed to be associated with unfavorable prognosis. Regarding mutation analysis, we found significant differences in the KIT/PDGFRA genotype among the three clusters. Univariate analysis revealed high expression of E2F1 to be associated with mitotic count, proliferation rate, KIT mutation, and aggressive clinical behavior. These findings on mRNA level could be confirmed by immunohistochemistry.

Conclusion: Our findings implicate differential regulation schemes of the CDKN2A tumor suppressor pathway converging to up-regulation of E2F1 as the critical link to increased cell proliferation and adverse prognosis of GISTs.

Gastrointestinal stromal tumors (GIST) are the most common mesenchymal tumors of the gastrointestinal tract. Due to a limited value of most morphologic criteria for assessment of clinical outcome in GISTs, generally accepted prognostic variables are tumor size, mitotic count, proliferation index, and tumor site (1–3). Central to the tumorigenesis of GISTs are activating mutations in the proto-oncogene tyrosine-protein kinase Kit (KIT) or platelet-derived growth factor receptor {alpha} (PDGFRA; refs. 4, 5), which are regarded as alternative and mutually exclusive. However, in a subset of so-called wild-type GISTs, no mutation in these two genes can be found. Downstream signaling of constitutively activated oncogenic KIT in GISTs is under current investigation, with implication for distinct activation of AKT, MAPK, and STAT signal transduction pathways (6).

Recently, functional inactivation of the p16INK4A transcript at 9p21 via mutation, deletion, or promoter hypermethylation causing loss or down-regulation of the corresponding protein has been identified as an independent unfavorable prognostic factor in GISTs (7–10). p16INK4A is one of two alternate transcripts of the cyclin-dependent kinase inhibitor 2A (CDKN2A) gene. The other transcript, p14ARF, results from an alternative reading frame of the first exon (11). The CDKN2A gene, with its two transcripts p16INK4A and p14ARF, is an important tumor suppressor gene, connecting a complex network of genes (Fig. 1), and has a central role in the control of cell proliferation and apoptosis (12). The p14ARF gene product inhibits the mouse double minute 2 p53-binding protein (MDM2) from degrading the tumor protein p53 (TP53; ref. 13), which prevents uncontrolled proliferation of mutated cells (14). The p16INK4A gene product binds to the cyclin-dependent kinase 4 (CDK4), thus inhibiting CDK4 orylating the retinoblastoma-associated protein (RB1; ref. 15). Unphosphorylated RB1 binds and blocks the E2F transcription factor 1 (E2F1; ref. 16), whereas its functional inactivation via CDK-mediated phosphorylation disrupts its ability to suppress E2F1 (17, 18). When RB1 is phosphorylated, free E2F1 accumulates in the nucleus and initiates S-phase entry via transcription of several genes necessary for DNA synthesis and cell cycle progression (19). Depending on its status and interacting partners, E2F1 may influence the expression of >1,000 genes involved in proliferation, differentiation, and apoptosis (20, 21). Hence, E2F1 directs the final step of cell proliferation and occupies a central position in controlling the progression of cells from a quiescent state to proliferation. The release of free E2F1 in intact cells is strictly controlled by its interaction partner RB1 and multiple further negative feedback control mechanisms. Once released, free E2F1 has the ability to up-regulate its own mRNA expression via positive feedback (21). In human fibroblasts, a simple increased activity of E2F1 has been described to lead to replicative senescence or apoptosis, mainly via p14ARF transactivation and consecutive TP53 stabilization (22). Additionally, p14ARF directly inhibits E2F1 from initiating S-phase entry itself (23). Pathologic inactivation of RB1 with deregulation of E2F1 induces p16INK4A expression (24), foreclosing further CDK-mediated phosphorylation of intact RB1. However, these multiple negative feedback loop control mechanisms have been shown to be at least partially interrupted in many solid tumors (25). Functional or quantitative loss of p16INK4A as well as alterations of other genes from the CDKN2A network are believed to be involved in ineffective cell cycle control mechanisms, increased cell proliferation, and progression of a variety of solid tumors.



View larger version (21K):
[in this window]
[in a new window]
  		Fig. 1. CDKN2A tumor suppressor pathway. Multiple interactions among the seven genes examined in the current study. Pointed arrows, induction or activation; blunt arrows, inhibition.

 
The aim of the current study was to examine the role of the CDKN2A pathway in tumor progression of GISTs. We determined the mRNA expression of p16INK4A, p14ARF, CDK4, RB1, MDM2, TP53, and E2F1 by quantitative reverse transcription-PCR (qRT-PCR) and the immunohistochemical expression of p16INK4A, p14ARF, CDK4, and E2F1 in 38 cases of GIST. The prognostic relevance of the genetic findings was evaluated by a cluster analysis using the k-means algorithm, which allows evaluation of distinct mRNA expression patterns of the complete CDKN2A network, and comparison with clinicopathologic features including mutation analysis of KIT and PDGFRA as well as patient outcome.


   Materials and Methods
Top
 Abstract
 Materials and Methods
Results
 Discussion
 References
 
Tumor samples. For this study, snap-frozen as well as formalin-fixed and paraffin wax–embedded tumor samples from primary GISTs of 38 patients were recruited. None of the patients had received imatinib before surgery. Risk of clinically aggressive behavior was evaluated according to the consensus approach published by Fletcher et al. (2).

Mutation analysis of KIT and PDGFRA genes. For mutation analysis of KIT exons 9, 11, 13, and 17 as well as PDGFRA exons 12 and 18, genomic DNA was extracted from deparaffinized samples of tumor tissue or snap-frozen tissue using spin column purification (Qiagen, Hilden, Germany). PCR was done in 50 µL reaction mixtures containing 0.2 µg DNA, 10 nmol deoxynucleotide triphosphates, and 20 pmol of each primer, 75 nmol MgCl2, 1 unit Platinum Taq polymerase (Invitrogen Life Technologies, Karlsruhe, Germany), and 5 µL of 10x PCR buffer. Control reactions contained no genomic DNA template. The following primers were used: KIT exon 9 (26), KIT exon 11 (27), KIT exon 13 (forward/reverse; refs. 26, 28), KIT exon 17 (29), and PDGFRA exons 12 and 18 (5). All PCRs followed the protocol described elsewhere (27). Direct sequencing of purified PCR products was carried out at a sequencing facility (Seqlab GmbH, Göttingen, Germany).

RNA isolation and reverse transcription. After homogenization of the snap-frozen tissue samples with an Ultra Turrax T25 (IKA-Werke GmbH, Staufen, Germany), total RNA was isolated with TRIzol reagent (Invitrogen Life Technologies) according to the manufacturer's manual using ~ 50 mg frozen tissue per mL TRIzol. Total RNA concentration was quantified with the RNA 6000 Nano LabChip using the Agilent 2100 Bioanalyzer (Agilent Technologies, Palo Alto, CA). Only high-quality total RNA, as confirmed by high peaks for 18S and 28S rRNA, was used. First-strand cDNA was generated from 5 µg total RNA per sample using the SuperScript II RNase H-Reverse Transcriptase (Invitrogen Life Technologies), including 125 ng random hexamer primers (Invitrogen Life Technologies) and 40 units RNasIn RNase Inhibitor (Promega, Mannheim, Germany). Reverse transcription product was aliquoted in equal volumes and stored at –20°C.

Quantitative reverse transcription-PCR. Gene-specific primers (Table 1) were designed on different exons with a 60°C melting temperature and a length of 18 to 26 bp for PCR products with a length of 50 to 270 bp. PCR was run in 20 µL reactions in triplicates on an iCycler (Bio-Rad Laboratories GmbH, München, Germany) using the Eurogentec qPCR Core kit for SYBR Green I (Eurogentec, Seraing, Belgium) and gene-specific primers in a final concentration of 300 nmol/L. The temperature profile consisted of (a) an initial step of 95°C for 10 minutes for Taq activation, (b) 40 cycles of 95°C for 15 seconds and 60°C for 1 minute, and (c) a final melt curve analysis with a temperature ramp from 60°C to 95°C with a heating rate of 3°C/min. PCR efficiencies were calculated with a relative standard curve derived from a cDNA mixture (a 2-fold dilution series with seven measuring points in triplicates) and gave regression coefficients more than 0.95 and reproducible primer-specific efficiencies of 85% to 99%. Gene-specific amplification was confirmed by a single peak in melt curve analysis and a single band in high-resolution agarose gel electrophoresis (SeaKem LE agarose, BMA, Rockland, ME). No template controls (no cDNA in PCR) and genomic controls (no enzyme in reverse transcription reaction) were run for each gene to detect unspecific/genomic amplification or primer dimerization. Relative expression levels in relation to total RNA input were calculated from the relative standard curve as described in refs. 30, 31 and logarithmized to obtain approximately normally distributed data. Additionally, the reference genes 18S, HPRT1, and SDHA (32) could be confirmed to be equivalently expressed within most of the examined clinicopathologic variables as described previously (33). Therefore, two different and independent methods for interpretation of gene expression data were applicable, which both gave approximately similar results, including significance and fold change levels for differential gene expression.


View this table:
[in this window]
[in a new window]
  		Table 1. Gene-specific primers used for qRT-PCR

 
Immunohistochemistry. Immunohistochemical evaluation of formalin-fixed and paraffin wax–embedded tumor tissue sections was done using the following monoclonal primary antibodies: c-kit (1:400 dilution, DAKOCytomation, Hamburg, Germany), Ki-67 antigen (1:50 dilution, clone Mib1, DAKOCytomation), p16INK4A (1:100 dilution, JC-8, Quartett, Berlin, Germany), p14ARF (1:50 dilution, 14P03, Dianova, Hamburg, Germany), CDK4 (1:100 dilution, DCS-35, Chemicon, Temecula, CA), and E2F1 (1:200 dilution, KH95, Dianova). Before incubation with the primary antibody, the slides underwent heat treatment for antigen retrieval (15 minutes at 95°C in 0.1 mol sodium citrate buffer). All primary antibodies were incubated at 4°C overnight, except c-kit and Ki-67, which were incubated at room temperature for 30 minutes. Visualization of the specific primary antibody was done using the DAKO ChemMate Detection kit (DAKOCytomation) with NeoFuchsin as chromogen. The c-kit staining reactions were interpreted in the presence of tissue mast cells or interstitial cells of Cajal as internal positive control. Mitoses were counted in 50 high-power fields (HPF), whereas proliferation rate was estimated as the percentage of Ki-67-positive nuclei from tumor areas with the highest mitotic activity. For the evaluation of p16INK4A expression, staining of the nuclei or the cytoplasm in >20% of the tumor cells was evaluated as "p16INK4A positive" according to the proposal of Schneider-Stock et al. (7, 10). As external positive controls, a case of malignant melanoma and a case of squamous cell carcinoma of the cervix were used, which were both strongly positive for p16INK4A. For p14ARF, a three-scale semiquantitative scoring for the intensity of the cytoplasmic staining was applied (0, no staining; 1, weak staining; 2, strong staining) using a case of squamous cell carcinoma of the cervix as external positive control. In the reactions using the antibody against CDK4, a specific staining of mitotic cells was detectable. For E2F1, the percentage of tumor cells with a strong nuclear staining was estimated using a case of malignant melanoma as external positive control.

Statistics. Descriptive statistics, tests, and graphs were done with Statistica 6.0 (StatSoft, Hamburg, Germany) and the statistical software system R (34). Associations among the clinicopathologic, the molecular genetic, and the immunohistochemical variables were evaluated using the t test for independent samples (qRT-PCR), the two-sample Wilcoxon test (immunohistochemistry), and the Fisher's exact test in the case of categorical variables. To identify particular patterns of gene expression of all seven genes, a cluster analysis using the k-means algorithm was done. The k-means algorithm groups a given set of objects into a user-defined number k of clusters, such that the sum of within-cluster squared Euclidean distances is minimized (35). The average silhouette width was used as a criterion to determine the optimal number of clusters (36). Disease-free survival rates were plotted by the Kaplan-Meier method. Associations of patient and tumor variables with disease-free survival times were assessed with the log-rank test (in the case of categorical variables) and with univariate Cox proportional hazards models (qRT-PCR). For evaluation of correlation between qRT-PCR and immunohistochemistry, the one-sided Kendall's {tau} test was used.


   Results
Top
 Abstract
 Materials and Methods
 Results
Discussion
 References
 
Patients and follow-up. The clinicopathologic data of 38 patients (16 female, 22 male) with primary GIST are summarized in Table 2. The median age of patients at the time of operation was 64 years (mean, 64 ± 12 years; range, 39-82 years). The present series comprised 26 (68%) cases of gastric location, 9 (24%) of intestinal location (including 3 duodenal, 3 small intestinal, and 3 colorectal), and 3 (8%) mesenteric GISTs. Tumor size ranged from 1.7 to 30 cm (median, 7 cm; mean, 8.4 ± 5.8 cm). Proliferation rate ranged from 1% to 30% (median, 5%, mean, 9.1 ± 8.2%). Follow-up was available for 37 patients and ranged from 2 to 84 months with a median of 32 months (mean, 35 ± 22 months), during which 13 patients developed disease progression. Shorter disease-free survival was significantly associated with intestinal/mesenteric location (P = 2 x 10–3), tumor size >5 cm (P = 0.04), mitotic count >5/50 HPFs (P = 2 x 10–4), proliferation rate ≥10% (P = 2 x 10–5), and high-risk (P = 2 x 10–4).


View this table:
[in this window]
[in a new window]
  		Table 2. Clinicopathologic data in 38 cases of GIST

 
Mutation analysis. Mutation analysis was done of KIT exons 9, 11, 13, and 17 as well as PDGFRA exons 12 and 18 in all 38 GISTs. As expected in primary and untreated GISTs, we could not find more than one mutation in each GIST (Table 2), confirming the mutually exclusive character of KIT or PDGFRA mutation (except in case 38, which carried both a KIT exon 11 mutation and a KIT exon 9 mutation). We found 27 GISTs with mutation of KIT (71.1%; 25 tumors with exon 11 mutation and 2 tumors with exon 9 mutation), whereas 4 GISTs had a mutation of the PDGFRA gene (10.5%; all tumors had exon 18 mutations). In the remaining 7 (18.4%) GISTs, all showing immunohistochemical c-kit expression, we could not find a mutation in any of the examined exons; thus, these cases were considered wild-type GISTs (for both KIT and PDGFRA). As a tendency, the different types of mutation seemed to be associated with clinical behavior. None of the four GISTs with PDGFRA mutation and only one of three GISTs with internal tandem duplication in the 3'-juxtamembrane domain of KIT developed disease progression. Both GISTs with KIT exon 9 mutation were of high risk, and one of them was clinically progressive. However, the frequency of these mutations was too low to reach statistical significance in the present study.

Quantitative reverse transcription-PCR. Significant associations between genotype and clinicopathologic variables on the one hand and the mRNA expression of individual genes on the other hand are summarized in Table 3. GISTs with KIT mutation had higher mean expression of CDK4, RB1, MDM2, TP53, and E2F1 compared with GISTs with PDGFRA mutation (2.9- to 5.2-fold) or wild-type GISTs (2.2- to 2.7-fold). On the other hand, the mean expression of p16INK4A and p14ARF was significantly higher in wild-type GISTs than in GISTs with PDGFRA mutation (10.5- and 28.6-fold, respectively). Compared with gastric GISTs, intestinal tumors had higher mean expression of CDK4, RB1, MDM2, and TP53 (2.3- to 3.4-fold). Compared with GISTs with mitotic counts ≤5/50 HPFs, tumors with mitotic counts >5/50 HPFs had higher mean expression of CDK4 and E2F1 (2.1- and 4.0-fold, respectively). Similarly, high-risk GISTs revealed higher mean expression of CDK4 and E2F1 than non-high-risk tumors (1.9- to 2.7-fold, respectively). Compared with nonprogressive GISTs, tumors with disease progression showed higher mean expression of CDK4, RB1, MDM2, TP53, and E2F1 (1.7- and 3.7-fold). In univariate Cox models, the differences in disease-free survival were statistically significant for CDK4, RB1, MDM2, TP53, and E2F1.


View this table:
[in this window]
[in a new window]
  		Table 3. Associations of mRNA expression and clinicopathologic variables

 
Clustering of gastrointestinal stromal tumors based on mRNA expression. Based on the similarity of the mRNA expression patterns of p16INK4A, p14ARF, CDK4, RB1, MDM2, TP53, and E2F1, the k-means cluster algorithm grouped the 38 GISTs into three clusters A, B, and C containing 17, 12, and 9 cases, respectively (Fig. 2). GISTs from cluster A were characterized by a moderate to high mRNA expression of all seven analyzed genes p16INK4A, p14ARF, CDK4, RB1, MDM2, TP53, and E2F1. The GISTs of cluster B shared a low mRNA expression of p16INK4A and p14ARF combined with a high mRNA expression of the other five genes CDK4, RB1, MDM2, TP53, and E2F1. In contrast, GISTs of cluster C also showed a low mRNA expression of p16INK4A and p14ARF but together with a low mRNA expression of the other five genes CDK4, RB1, MDM2, TP53, and E2F1. There were significant differences in the distribution of tumors with KIT mutation, tumors with PDGFRA mutation, and tumors without mutation of KIT or PDGFRA among these three clusters (cluster A, 11/1/5; cluster B, 12/0/0; cluster C, 4/3/2; P = 0.01). As regards clinicopathologic variables, there was a trend for GISTs of cluster C to have lower mitotic counts and to be of low risk, but this trend did not reach statistical significance. However, there were significant differences in the distribution of gastric, intestinal, and mesenteric sites (cluster A, 12/3/2; cluster B, 5/6/1; cluster C, 9/0/0; P = 0.01) and the proportion of tumors showing disease progression (cluster A, 7 of 17; cluster B, 7 of 12; cluster C, none of 9; P = 0.01). The differences in disease-free survival were statistically significant (P = 0.03; Fig. 3).



View larger version (52K):
[in this window]
[in a new window]
  		Fig. 2. Clustering of 38 GISTs by the k-means algorithm. Left, the seven genes used for the cluster analysis. Depending on particular mRNA expression patterns of these genes, the individual tumor samples (bottom) were grouped into three clusters A, B, and C. Color-coded log2-transformed relative mRNA expression levels, with a categorical increase in color intensity from white representing 0 to dark red representing 10. Each row represents the mRNA expression of a separate gene over all tumor samples, and each column represents the mRNA expression of all genes in a separate tumor.

 


View larger version (10K):
[in this window]
[in a new window]
  		Fig. 3. Kaplan-Meier plot for disease-free survival of 38 GISTs. The cluster analysis based on the mRNA expression of seven members of the CDKN2A network yielded three different subgroups of GISTs with distinct disease-free survival (P = 0.03). None of the 9 GISTs from cluster C showed aggressive clinical behavior, whereas 7 of 17 GISTs from cluster A did. The GISTs assigned to cluster B had the most unfavorable outcome (7 of 12).

 
Immunohistochemistry. Twenty-five of 38 (66%) cases were considered negative for p16INK4A (Fig. 4A), and there was a significant association between loss of p16INK4A expression and high mitotic count (>5/50 HPFs; P = 0.03) or high risk (P = 0.04). Using the antibody against p14ARF, 32% cases showed strong, 55% cases weak, and 13% cases no cytoplasmic staining (Fig. 4E). For CDK4, there was a specific reactivity of mitotic figures (Fig. 4F), and similar to mitotic count, higher protein expression of CDK4 was significantly associated with shorter disease-free survival (P = 0.008). The estimated percentage of E2F1-positive cells (Fig. 4D) ranged from 2% to 70% (mean, 9%), with a significant positive correlation between E2F1 staining and mitotic count (P = 0.02).



View larger version (124K):
[in this window]
[in a new window]
  		Fig. 4. Immunohistochemical staining of GISTs for p16INK4A (A and B), E2F1 (C and D), p14ARF (E), and CDK4 (F). A, p16INK4A negative (<20% positive nuclear staining). Magnification, x 200. B, p16INK4A positive (>20% positive nuclear staining). Magnification, x 200. C, low fraction of cells with nuclear expression of E2F1. Magnification, x 200. D, high fraction of cells with nuclear expression of E2F1. Magnification, x 200. E, staining for p14ARF showing cytoplasmic staining in the single cell in the middle, and granular nucleic staining in a fraction of the surrounding cells. Magnification x 400. F, staining for CDK4 showing selective staining of mitotic figures. Magnification, x200.

 
Correlation between quantitative reverse transcription-PCR and immunohistochemistry. For p16INK4A, CDK4, and E2F1, there was a significant positive correlation between mRNA expression and immunohistochemical reactivity (P = 0.002, 0.04, and 0.0001, respectively). Comparing the three clusters derived from the k-means algorithm based on the mRNA expression with the semiquantitative immunohistochemical staining for p16INK4A and p14ARF, there also was a significant association detectable, in that clusters B and C (with low mean mRNA expression of p16INK4A and p14ARF) contained a significantly higher proportion of tumors that were immunohistochemically negative for p16INK4A (P = 0.01) and p14ARF (P = 0.03).


   Discussion
Top
 Abstract
 Materials and Methods
 Results
 Discussion
References
 
With respect to the close functional linkage of p16INK4A, p14ARF, CDK4, RB1, MDM2, TP53, and E2F1 combined in the CDKN2A tumor suppressor pathway, we used a cluster analysis based on the k-means algorithm to explore particular mRNA expression patterns of these genes in a series of 38 GISTs. This analysis yielded three subgroups of GISTs with distinct clinical behavior (Fig. 2). The GISTs assigned to the first cluster A shared a moderate to high expression of all seven analyzed genes and revealed aggressive clinical behavior (Fig. 3). Cluster B comprised GISTs with a low mRNA expression of the two CDKN2A transcripts p16INK4A and p14ARF combined with a very high expression of the other five members of the pathway, CDK4, RB1, MDM2, TP53, and E2F1. These tumors had a distinctly unfavorable prognosis. In contrast, the GISTs of cluster C with a low mRNA expression of the CDKN2A transcripts p16INK4A and p14ARF along with also low mRNA expression of the other five genes CDK4, RB1, MDM2, TP53, and E2F1 revealed no aggressive clinical behavior. These observations confirm previous reports on a possible involvement of p16INK4A in tumor progression of GISTs (7–9) and furthermore implicate a role for the downstream members of the CDKN2A tumor suppressor pathway, CDK4, RB1, MDM2, TP53, and E2F1.

In general, malignant behavior of GISTs is associated with higher cell proliferation, and most schemes to prognostication of GISTs, including the current consensus approach, use mitotic count as a key variable (1–3). Herein, we showed that a higher mRNA and protein expression of E2F1 in GISTs was significantly associated with higher mitotic counts and higher proliferation rates. This is consistent with the critical role of E2F1 for initiation of S-phase entry and consecutive mitosis (19, 37, 38). Furthermore, a high mRNA expression of E2F1 was significantly associated with aggressive clinical behavior and shorter disease-free survival. These results emphasize the crucial role of E2F1 in the control of cell proliferation and suggest E2F1 expression as a useful additional marker for prognostication of GISTs, as it has been proposed previously for non–small cell lung carcinomas (39).

Another member of the CDKN2A tumor suppressor pathway, CDK4, has the ability of supporting the proliferative effect of E2F1 via phosphorylation of the RB1 protein. This is of special importance, as the release of free E2F1 in intact cells is closely controlled by its interaction partner RB1. In that CDK4 inhibits RB1 from eliminating E2F1, free E2F1 has the ability to up-regulate its own mRNA expression via positive feedback, thus enabling an uncontrollable amplification of itself (21). Similar to E2F1, high mRNA expression of CDK4 was significantly correlated with high mitotic count and tumor progression. Immunohistochemistry revealed CDK4 solely in mitotic cells. Thus, CDK4 may critically contribute to the release of free E2F1 in GISTs. These observations further underline the close functional interconnection of the analyzed genes within the CDKN2A tumor suppressor pathway.

In normal cells, E2F1 is under negative feedback control of p14ARF transactivation (23, 40, 41) with consecutive TP53 stabilization (22) and induction of p16INK4A expression with CDK4 inactivation (24), respectively (Fig. 1). We observed the highest frequency of tumor progression in GISTs of cluster B with high expression of E2F1 and CDK4 but low expression of p16INK4A and p14ARF. The combined low mRNA expression of p16INK4A and p14ARF implicates a coregulation of these two alternative CDKN2A transcripts, which differ only by their first exon (11, 42), and may represent genetic alteration of their common gene CDKN2A (7–9). Thus, there is evidence for disrupted negative feedback control mechanisms for E2F1 with further uncontrolled cell growth in this subset of GISTs.

There is strong evidence that the genotype of KIT and PDGFRA in GISTs correlates with distinct regulation of downstream signaling cascades involving AKT, MAPK, and STAT signal transduction pathways (6), leading to distinguishable gene expression profiles in GISTs with KIT mutation, PDGFRA mutation, or wild-type GISTs (43–45). This may be the reason for the observed differences in clinical behavior. GISTs with KIT mutation have an adverse prognosis compared with wild-type GISTs (46, 47), whereas mutations of PDGFRA appear preferentially in gastric GISTs and are associated with favorable prognosis (48, 49). In this study, GISTs with KIT mutation had a significantly higher expression of CDK4, RB1, MDM2, TP53, and E2F1 compared with GISTs with PDGFRA mutation or wild-type GISTs. Furthermore, we found significant differences in the KIT/PDGFRA genotype among the three different clusters. All GISTs of the unfavorable cluster B carried a KIT exon 11 mutation, whereas GISTs of the more favorable cluster C had a disproportionate high incidence of PDGFRA mutation. From the four GISTs with KIT mutation in cluster C, two had the variant internal tandem duplication in the 3' end of KIT juxtamembrane domain, for which an association with favorable course has been reported (48). A recent investigation of KIT signaling found stronger AKT phosphorylation in KIT exon 11 mutant GISTs, which was correlated with a higher proliferation rate and a presumably more aggressive clinical behavior (6). Notably, AKT activation was shown to inhibit E2F1-induced apoptosis, suggesting a negative feedback loop involving E2F1 and AKT (50, 51). Thus, the GISTs with KIT exon 11 mutation assigned to cluster B may represent a subgroup of GISTs with up-regulated expression of E2F1, which may be partially due to interrupted negative feedback control of p16INK4A/p14ARF. Whether as a consequence of activated AKT in these KIT exon 11 mutated GISTs such up-regulation of E2F1 leads to increased cell proliferation instead of apoptosis remains to be determined.

Tumor site in GISTs is viewed as an independent prognostic variable, with tumors arising from the stomach generally having a better prognosis than those arising from the intestinum (3). This perception was further supported by a recent study, which revealed distinct gene expression profiles of gastric and intestinal GISTs (43). Indeed, we also found a significantly higher mRNA expression of CDK4, RB1, MDM2, and TP53 in intestinal GISTs, suggesting that GISTs of intestinal sites may be genetically distinct from gastric GISTs. Furthermore, all GISTs of the favorable cluster C were of gastric site, whereas the unfavorable cluster B contained a disproportionate high number of intestinal GISTs. Whether this observation reflects an alternate route of the CDKN2A network in intestinal GISTs, or represents a consequence of the differences in the distribution of mutation types among the three clusters, cannot be clearly distinguished. However, up-regulated expression of E2F1 may occur in GISTs of both gastric and intestinal sites and may carry prognostic significance for GISTs irrespective of site.

In conclusion, there is evidence for molecular alterations of the CDKN2A tumor suppressor pathway to contribute to tumor progression in GISTs. The common and critical link of alternate routes within this network is E2F1, which finally controls progression of a quiescent state to proliferation. We herein suggest that up-regulated expression of E2F1 is of prognostic relevance in GISTs, and determination of the E2F1 expression status may be a helpful prognosticator in predicting clinical behavior. Eventually, E2F1 or other genes within the CDKN2A pathway may provide potential targets for new molecular-based therapies in GISTs.


   Footnotes
 
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.

Received 2/11/05; revised 5/27/05; accepted 6/21/05.


   References
Top
 Abstract
 Materials and Methods
 Results
 Discussion
 References
 

  1. Franquemont DW. Differentiation and risk assessment of gastrointestinal stromal tumors. Am J Clin Pathol 1995;103:41–7.[Medline]
  2. Fletcher CD, Berman JJ, Corless C, et al. Diagnosis of gastrointestinal stromal tumors: a consensus approach. Hum Pathol 2002;33:459–65.[CrossRef][Medline]
  3. Miettinen M, El Rifai W, Sobin HL, Lasota J. Evaluation of malignancy and prognosis of gastrointestinal stromal tumors: a review. Hum Pathol 2002;33:478–83.[CrossRef][Medline]
  4. Hirota S, Isozaki K, Moriyama Y, et al. Gain-of-function mutations of c-kit in human gastrointestinal stromal tumors. Science 1998;279:577–80.[Abstract/Free Full Text]
  5. Heinrich MC, Corless CL, Duensing A, et al. PDGFRA activating mutations in gastrointestinal stromal tumors. Science 2003;299:708–10.[Abstract/Free Full Text]
  6. Duensing A, Medeiros F, McConarty B, et al. Mechanisms of oncogenic KIT signal transduction in primary gastrointestinal stromal tumors (GISTs). Oncogene 2004;23:3999–4006.[CrossRef][Medline]
  7. Schneider-Stock R, Boltze C, Lasota J, et al. High prognostic value of p16INK4 alterations in gastrointestinal stromal tumors. J Clin Oncol 2003;21:1688–97.[Abstract/Free Full Text]
  8. Ricci R, Arena V, Castri F, et al. Role of p16/INK4a in gastrointestinal stromal tumor progression. Am J Clin Pathol 2004;122:35–43.[CrossRef][Medline]
  9. Sabah M, Cummins R, Leader M, Kay E. Loss of heterozygosity of chromosome 9p and loss of p16INK4A expression are associated with malignant gastrointestinal stromal tumors. Mod Pathol 2004;17:1364–71.[CrossRef][Medline]
  10. Schneider-Stock R, Boltze C, Lasota J, et al. Loss of p16 protein defines high-risk patients with gastrointestinal stromal tumors: a tissue microarray study. Clin Cancer Res 2005;11:638–45.[Abstract/Free Full Text]
  11. Quelle DE, Zindy F, Ashmun RA, Sherr CJ. Alternative reading frames of the INK4a tumor suppressor gene encode two unrelated proteins capable of inducing cell cycle arrest. Cell 1995;83:993–1000.[CrossRef][Medline]
  12. Sherr CJ. The INK4a/ARF network in tumour suppression. Nat Rev Mol Cell Biol 2001;2:731–7.[CrossRef][Medline]
  13. Pomerantz J, Schreiber-Agus N, Liegeois NJ, et al. The Ink4a tumor suppressor gene product, p19Arf, interacts with MDM2 and neutralizes MDM2's inhibition of p53. Cell 1998;92:713–23.[CrossRef][Medline]
  14. Levine AJ. p53, the cellular gatekeeper for growth and division. Cell 1997;88:323–31.[CrossRef][Medline]
  15. Serrano M, Hannon GJ, Beach D. A new regulatory motif in cell-cycle control causing specific inhibition of cyclin D/CDK4. Nature 1993;366:704–7.[CrossRef][Medline]
  16. Weintraub SJ, Prater CA, Dean DC. Retinoblastoma protein switches the E2F site from positive to negative element. Nature 1992;358:259–61.[CrossRef][Medline]
  17. Weinberg RA. The retinoblastoma protein and cell cycle control. Cell 1995;81:323–30.[CrossRef][Medline]
  18. Sherr CJ. Cancer cell cycles. Science 1996;274:1672–7.[Abstract/Free Full Text]
  19. Lukas J, Petersen BO, Holm K, Bartek J, Helin K. Deregulated expression of E2F family members induces S-phase entry and overcomes p16INK4A-mediated growth suppression. Mol Cell Biol 1996;16:1047–57.[Abstract]
  20. Müller H, Bracken AP, Vernell R, et al. E2Fs regulate the expression of genes involved in differentiation, development, proliferation, and apoptosis. Genes Dev 2001;15:267–85.[Abstract/Free Full Text]
  21. Ishida S, Huang E, Zuzan H, et al. Role for E2F in control of both DNA replication and mitotic functions as revealed from DNA microarray analysis. Mol Cell Biol 2001;21:4684–99.[Abstract/Free Full Text]
  22. Dimri GP, Itahana K, Acosta M, Campisi J. Regulation of a senescence checkpoint response by the E2F1 transcription factor and p14(ARF) tumor suppressor. Mol Cell Biol 2000;20:273–85.[Abstract/Free Full Text]
  23. Eymin B, Karayan L, Seite P, et al. Human ARF binds E2F1 and inhibits its transcriptional activity. Oncogene 2001;20:1033–41.[CrossRef][Medline]
  24. Khleif SN, DeGregori J, Yee CL, et al. Inhibition of cyclin D-CDK4/CDK6 activity is associated with an E2F-mediated induction of cyclin kinase inhibitor activity. Proc Natl Acad Sci U S A 1996;93:4350–4.[Abstract/Free Full Text]
  25. Wyllie AH. E2F1 selects tumour cells for both life and death. J Pathol 2002;198:139–41.[CrossRef][Medline]
  26. Lux ML, Rubin BP, Biase TL, et al. KIT extracellular and kinase domain mutations in gastrointestinal stromal tumors. Am J Pathol 2000;156:791–5.[Abstract/Free Full Text]
  27. Wardelmann E, Losen I, Hans V, et al. Deletion of Trp-557 and Lys-558 in the juxtamembrane domain of the c-kit protooncogene is associated with metastatic behavior of gastrointestinal stromal tumors. Int J Cancer 2003;106:887–95.[CrossRef][Medline]
  28. Hou YY, Tan YS, Sun MH, et al. C-kit gene mutation in human gastrointestinal stromal tumors. World J Gastroenterol 2004;10:1310–4.[Medline]
  29. Rubin BP, Singer S, Tsao C, et al. KIT activation is a ubiquitous feature of gastrointestinal stromal tumors. Cancer Res 2001;61:8118–21.[Abstract/Free Full Text]
  30. Bustin SA. Absolute quantification of mRNA using real-time reverse transcription polymerase chain reaction assays. J Mol Endocrinol 2000;25:169–93.[Abstract]
  31. Pfaffl MW. A new mathematical model for relative quantification in real-time RT-PCR. Nucleic Acids Res 2001;29:45e.[Abstract/Free Full Text]
  32. Vandesompele J, De Preter K, Pattyn F, et al. Accurate normalization of real-time quantitative RT-PCR data by geometric averaging of multiple internal control genes. Genome Biol 2002;3:34–41.
  33. Haller F, Kulle B, Schwager S, et al. Equivalence test in quantitative reverse transcription polymerase chain reaction: confirmation of reference genes suitable for normalization. Anal Biochem 2004;335:1–9.[CrossRef][Medline]
  34. Ihaka R, Gentleman R. R. A language for data analysis and graphics. J Comp Graph Stat 1996;5:299–314.[CrossRef]
  35. Jain AK, Dubes RC. Algorithms for clustering data. Englewood Cliffs (NJ): Prentice Hall; 1988.
  36. Kaufman L, Rousseeuw PJ. Finding groups in data. New York: Wiley; 1990.
  37. Johnson DG, Schwarz JK, Cress WD, Nevins JR. Expression of transcription factor E2F1 induces quiescent cells to enter S phase. Nature 1993;365:349–52.[CrossRef][Medline]
  38. Wu L, Timmers C, Maiti B, et al. The E2F1-3 transcription factors are essential for cellular proliferation. Nature 2001;414:457–62.[CrossRef][Medline]
  39. Gorgoulis VG, Zacharatos P, Mariatos G, et al. Transcription factor E2F-1 acts as a growth-promoting factor and is associated with adverse prognosis in non-small cell lung carcinomas. J Pathol 2002;198:142–56.[CrossRef][Medline]
  40. Martelli F, Hamilton T, Silver DP, et al. p19ARF targets certain E2F species for degradation. Proc Natl Acad Sci U S A 2001;98:4455–60.[Abstract/Free Full Text]
  41. Mason SL, Loughran O, La Thangue NB. p14(ARF) regulates E2F activity. Oncogene 2002;21:4220–30.[CrossRef][Medline]
  42. Mao L, Merlo A, Bedi G, et al. A novel p16INK4A transcript. Cancer Res 1995;55:2995–7.[Abstract/Free Full Text]
  43. Antonescu CR, Viale A, Sarran L, et al. Gene expression in gastrointestinal stromal tumors is distinguished by KIT genotype and anatomic site. Clin Cancer Res 2004;10:3282–90.[Abstract/Free Full Text]
  44. Subramanian S, West RB, Corless CL, et al. Gastrointestinal stromal tumors (GISTs) with KIT and PDGFRA mutations have distinct gene expression profiles. Oncogene 2004;23:7780–90.[CrossRef][Medline]
  45. Kang HJ, Nam SW, Kim H, et al. Correlation of KIT and platelet-derived growth factor receptor {alpha} mutations with gene activation and expression profiles in gastrointestinal stromal tumors. Oncogene 2005;24:1066–74.[CrossRef][Medline]
  46. Singer S, Rubin BP, Lux ML, et al. Prognostic value of KIT mutation type, mitotic activity, and histologic subtype in gastrointestinal stromal tumors. J Clin Oncol 2002;20:3898–905.[Abstract/Free Full Text]
  47. Kim TW, Lee H, Kang YK, et al. Prognostic significance of c-kit mutation in localized gastrointestinal stromal tumors. Clin Cancer Res 2004;10:3076–81.[Abstract/Free Full Text]
  48. Lasota J, Dansonka-Mieszkowska A, Stachura T, et al. Gastrointestinal stromal tumors with internal tandem duplications in 3' end of KIT juxtamembrane domain occur predominantly in stomach and generally seem to have a favorable course. Mod Pathol 2003;16:1257–64.[CrossRef][Medline]
  49. Lasota J, Dansonka-Mieszkowska A, Sobin LH, Miettinen M. A great majority of GISTs with PDGFRA mutations represent gastric tumors of low or no malignant potential. Lab Invest 2004;84:874–83.[CrossRef][Medline]
  50. Hallstrom TC, Nevins JR. Specificity in the activation and control of transcription factor E2F-dependent apoptosis. Proc Natl Acad Sci U S A 2003;100:10848–53.[Abstract/Free Full Text]
  51. Chaussepied M, Ginsberg D. Transcriptional regulation of AKT activation by E2F. Mol Cell 2004;16:831–7.[CrossRef][Medline]



This article has been cited by other articles:


Home page
 Ann. Surg. Oncol.Home page
F. Haller, S. Detken, H.-J. Schulten, N. Happel, B. Gunawan, J. Kuhlgatz, and L. Fuzesi
Surgical Management After Neoadjuvant Imatinib Therapy in Gastrointestinal Stromal Tumours (GISTs) with Respect to Imatinib Resistance Caused by Secondary KIT Mutations
Ann. Surg. Oncol., February 1, 2007; 14(2): 526 - 532.
[Abstract] [Full Text] [PDF]

Home page
 J. Mol. Diagn.Home page
J. Lasota, B. Wasag, S. E. Steigen, J. Limon, and M. Miettinen
Improved Detection of KIT Exon 11 Duplications in Formalin-Fixed, Paraffin-Embedded Gastrointestinal Stromal Tumors
J. Mol. Diagn., February 1, 2007; 9(1): 89 - 94.
[Abstract] [Full Text] [PDF]

Home page
 J. Clin. Pathol.Home page
L Tornillo and L M Terracciano
An update on molecular genetics of gastrointestinal stromal tumours.
J. Clin. Pathol., June 1, 2006; 59(6): 557 - 563.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Haller, F.
Right arrow Articles by Füzesi, L.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Haller, F.
Right arrow Articles by Füzesi, L.

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
Cancer Research Clinical Cancer Research
Cancer Epidemiology Biomarkers & Prevention Molecular Cancer Therapeutics
Molecular Cancer Research Cancer Prevention Research
Cancer Prevention Journals Portal Cancer Reviews Online
Annual Meeting Education Book Cell Growth & Differentiation