Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Frequent activation of EGFR in advanced chordomas

Frequent activation of EGFR in advanced chordomas Background: Chordomas are rare neoplasms, arising from notochordal remnants in the midline skeletal axis, for which the current treatment is limited to surgery and radiotherapy. Recent reports suggest that receptor tyrosine kinases (RTK) might be essential for the survival or proliferation of chordoma cells, providing a rationale for RTK targeted therapy. Nevertheless, the reported data are conflicting, most likely due to the assorted tumor specimens used for the studies and the heterogeneous methodological approaches. In the present study, we performed a comprehensive characterization of this rare entity using a wide range of assays in search for relevant therapeutic targets. Methods: Histopathological features of 42 chordoma specimens, 21 primary and 21 advanced, were assessed by immunohistochemistry and fluorescent in situ hybridization (FISH) using PDGFRB, CSF1R, and EGFR probes. Twenty- two of these cases, for which frozen material was available (nine primary and 13 advanced tumors), were selectively analyzed using the whole-genome 4.3 K TK-CGH-array, phospho-kinase antibody array or Western immunoblotting. The study was supplemented by direct sequencing of KIT, PDGFRB, CSF1R and EGFR. Results: We demonstrated that EGFR is frequently and the most significantly activated RTK in chordomas. Furthermore, concurrent to EGFR activation, the tumors commonly reveal co-activation of alternative RTK. The consistent activation of AKT, the frequent loss of the tumor suppressor PTEN allele, the recurrent activation of upstream RTK and of downstream effectors like p70S6K and mTOR, all indicate the PI3K/AKT pathway as an important mediator of transformation in chordomas. Conclusions: Given the complexity of the signaling in chordomas, combined treatment regimens targeting multiple RTK and downstream effectors are likely to be the most effective in these tumors. Personalized therapy with careful selection of the patients, based on the molecular profile of the specific tumor, is anticipated. Background soft tissue. Initial symptoms usually relate to local pro- Chordomas are rare tumors. With an incidence of about gression of the disease. Chordomas infrequently metas- 0.05/100000/year, they account for less than 5% of all tasize to lung, bone, soft tissue, lymph nodes and skin. primary malignant bone tumors. Mainly adults between On histology at low power magnification they show pro- 40 and 60 years are affected, but cases of children pre- minent lobules separated by fibrous septa. The tumors senting with chordoma were also rarely reported (5% of maybearranged in chords or sheets or maybefloating cases). These bone tumors arise from remnants of the singularly in the abundant myxoid matrix often present. fetal notochord, and hence occur along the midline, and The current treatment for chordoma is predominantly most often in the caudal spine or the base of the skull. surgery, followed by radiotherapy. Safe margins are They are slowly growing masses with the tendency to often difficult to obtain because of the anatomical loca- destroy the surrounding bone and to infiltrate adjacent tion of the tumors [1]. Unfortunately, standard che- motherapy was shown to be basically unsuccessful, * Correspondence: barbara.dewaele@uzleuven.be which causes serious problems for managing patients Department of Human Genetics, Catholic University of Leuven, University with locally recurrent or metastatic disease. Survival Hospitals, Leuven, Belgium Full list of author information is available at the end of the article © 2011 Dewaele et al; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Dewaele et al. Clinical Sarcoma Research 2011, 1:4 Page 2 of 16 http://www.clinicalsarcomaresearch.com/content/1/1/4 rates of 5 and 10 years are 68% and 40%, respectively addition, expression of the MET oncogene has been [2]. reported in chordoma [10]. Of note, the MET oncogene Cytogenetic studies in chordomas have revealed in is known to be expressed in various chondroid neo- general nearly diploid or rather hypodiploid karyotypes, plasms, normal articular cartilage and fetal notochord with a number of numerical and structural rearrange- [17,18]. Given their possible relationship to notochordal ments. Recurrent genetic events reported in chordoma development and chondroid differentiation, further include frequent losses of large parts of chromosomes 3, investigation is warranted to clarify the roles of these 4, 10 and 13 and the most commonly lost regions are and other RTK in chordomagenesis. 1p31-pter, 3p21-pter, 3q21-qter, 9p24-pter and 17q11- Activities of effectors more downstream in the main qter [3]. The most common gains affect the chromo- RTK pathways were also recently described. The ERK1/2, some 5q and the entire chromosomes 7 and 20 [4,5]. AKT and STAT3 activity was demonstrated in 18 (86%), Loss of heterozygosity at 1p36 was also found in familial 16 (76%) and 14 (67%) of cases, respectively, by immuno- chordomas, further supporting the hypothesis that an histochemistry performed on 21 chordomas [15]. important tumor suppressor might be located at the dis- Furthermore, analysis of 22 chordomas by Tamborini tal part of 1p [6]. Importantly, the CDKN2A tumor sup- and co-workers showed consistent ERK1/2 activation in pressor gene, which maps to 9p21.3, is reported to be all the cases, and activation of AKT in 20 (91%), mTOR lost in a high percentage (60%) of chordomas [7,8]. In in 18 (82%), and S6 in 16 (73%) of the tumors [8]. addition, loss of one copy of the PTEN tumor suppres- In the present study, we have performed a compre- sor gene (located on 10q23.31) was found in 37% (7/19) hensive molecular and biochemical analysis of 42 chor- of lesions, although no difference in PTEN expression domas, focusing on the role of RTK and their level was shown by Western blotting [8]. downstream signaling pathway in chordoma develop- In the literature, several RTK, specifically PDGFRA, ment, in primary tumors or their recurrent/metastatic PDGFRB, KIT, EGFR, MET and HER2, were reported to counterparts. be expressed in chordoma by immunohistochemistry [9-12]. Given that RTK could prove to be essential for Methods the survival or proliferation of chordoma tumor cells, Patients and histopathology targeting these RTK using antibodies or small molecule The present study included 31 patients [16 women and tyrosine kinase inhibitors (TKI) might offer new treat- 15 men; age range 18-84 (median 58 years)] (Table 1). ment options for chordoma patients. Interestingly, ima- In total, 42 tumor specimens from these patients were tinib was found to have antitumor activity in patients retrieved, of which 21 were annotated as primary with chordoma [13]. It was suggested that PDGFRB sig- tumors and 21 as recurrences or metastases (in the text naling might be implicated in the tumor growth, as ima- further referred to as advanced cases). The primary tinib-responding tumors were found to be chordomas originated from the spine (n = 9), the immunohistochemically positive for PDGFRB. Expres- sacrum (n = 10), the clivus (n = 1), and the cervix (n = sion of basic fibroblast growth factor (bFGF), transform- 1). Samples 10a and 10b represent primary samples ing growth factor alpha (TGF alpha) and fibronectin was from thesamepatient obtainedbyneedle biopsyand reported to correlate with an increased incidence of dis- subsequent surgical resection, respectively. Histopatho- ease recurrence in chordoma [14]. Moreover, clinical logical examination was performed on formalin fixed, response to imatinib in one case was accompanied by paraffin embedded tissue. Five μmsectionswereused the inhibition of PDGFRB as demonstrated by Western for routine hematoxylin and eosin (H&E) staining, and blot [13]. In recent reports, Tamborini and co-workers immunohistochemical staining was performed by the characterized 22 chordomas by immunoprecipitation avidin-biotin-peroxidase complex method, using the fol- and antibody arrays. The activation of PDGFRA, lowing monoclonal (mc) and polyclonal (pc) antibodies: PDGFRB, KIT, FLT3, CSF1R, EGFR, HER2, HER4, AXL Pankeratin (mc, dilution 1:200; Serotec, Oxford, UK), and DTK was reported in these studies [8,11]. Notably, Epithelial Membrane Antigen (EMA) (mc, 1:50; DAKO, PDGFRB activation was found in 95% (21/22) of cases. Glostrup, Denmark), Multikeratin (mc, dilution 1:10; The EGFR activation, mainly through EGFR/HER2 het- Novocastra, Newcastle Upon Tyne, UK), S-100 protein erodimer formation, was also suggested. Other groups (pc, 1:300; DAKO) and Vimentin (mc, dilution 1:500, found EGFR activation in three out of three and in DAKO). In addition, the EGFR (EGFR PharmDxTM, about 50% of chordomas evaluated by RTK antibody DAKO) and HER2/ERBB2 (HercepTestTM, DAKO) arrays and immunohistochemistry respectively [15,16]. staining kits were used. EGFR and HER2 protein expres- Partial response of metastatic chordoma to combined sion was reported as membranous brown staining of cetuximab/gefitinib treatment suggests that EGFR tar- neoplastic cells using a three-tier system ranging from geted treatment may benefit chordoma patients [9]. In “1+” to “3+”. Dewaele et al. Clinical Sarcoma Research 2011, 1:4 Page 3 of 16 http://www.clinicalsarcomaresearch.com/content/1/1/4 Table 1 Pathologic description of chordoma cases and results summary Cases Gender Age Tumor Site Immuno FISH Proteome TK Western TK status profiler array Mut aCGH EGFR HER2 EGFR HER2 PDGFRB/ PTEN P- P- EGFR PDGFRB CSF1R EGFR PDGFRB 1a F 56 P Spinal 1+ neg dis. dis. polys. nd nd nd nd nd nd nd 1b R Spinal neg nd dis. dis. polys. nd weak weak nd nd nd nd 2a M 33 R Spinal 1+ nd dis. dis. polys. nd nd nd nd nd nd nd 2b M 3+ nd dis. dis. polys. nd nd nd neg E E nd 3 F 43 R Spinal 3+ neg polys. dis. dis. dis. strong weak neg nd nd table 4a M 62 P Spinal neg nd dis. dis. dis. dis. nd nd nd nd nd nd 4b M 2+ nd polys. dis. polys. monos. interm weak neg E neg table 5 F 75 R Sacrum 3+ neg polys. polys. polys. polys. nd nd neg E/P E table 6 M 60 R Clivus 3+ nd dis. monos. dis. dis. strong interm. nd nd nd nd 7a F 62 P Sacrum neg nd polys. dis. dis. dis. nd nd nd nd nd nd 7b R Sacrum 2+ nd polys. monos. dis. monos. nd nd neg E/P E table 8 M 36 P Clivus neg nd dis. dis. dis. nd nd nd nd nd nd nd 9 M 52 R Coccyx 1+ 1+ monos. dis. dis. dis. nd nd neg E/P E table 10a F 41 P Spinal neg nd polys. dis. loss dis. nd nd nd nd nd nd 10b P Spinal neg nd polys. dis. loss monos. nd nd neg neg E/P table 10c R Spinal neg nd polys. dis. polys. nd nd nd nd nd nd nd 11 F 54 P Cervical 2+ nd polys. dis. dis. nd strong weak nd nd nd nd 12a M 55 P Sacrum 3+ neg l.l.amp. dis. dis. monos. nd nd neg nd nd nd 12b M 3+ 2+ h.l. dis. polys. nd strong weak nd E/P E nd amp. 13 M 80 R Coccyx 2+ nd polys. polys. polys. polys. interm weak neg nd nd table 14 F 60 R Sacrum 1+ nd dis. polys. dis. monos. nd nd neg E/P E table 15a F 73 P Spinal neg nd dis. dis. dis. monos. nd nd neg nd nd table 15b R Spinal neg nd polys. polys. polys. nd nd nd nd E E nd 16 M 84 R Sacrum 1+ neg dis. dis. dis. dis. nd nd neg nd nd table 17a F 58 P Sacrum 2+ 1+ polys. dis. dis. nd strong interm. neg nd nd table 17b R Sacrum 3+ neg polys. dis. dis. nd nd nd nd nd nd nd 18 F 57 P Sacrum 3+ neg polys. dis. dis. dis. strong weak nd nd nd nd 19 M 84 P Lumbal 1+ nd dis. monos. dis. nd nd nd nd nd nd nd 20 M 81 P Sacrum 3+ neg l.l.amp. dis. polys. nd strong weak nd nd nd nd 21 F 67 P Sacrum 1+ 1+ h.l. dis. dis. nd interm weak nd nd nd nd amp. 22 F 47 P Sacrum 1+ nd dis. dis. dis. nd weak weak nd nd nd nd 23 M 48 P Spinal nd nd dis. dis. dis. nd nd nd nd nd nd nd 24 F 60 R Clivus/ 3+ nd polys. polys. polys. polys. nd nd nd nd nd nd nc 25 F 60 R Sacrum neg nd monos. dis. nd nd nd nd nd nd nd nd 26 M 80 M nd nd polys. polys. polys. loss nd nd nd nd nd nd 27 M 48 P Sacrum 1+ nd dis. dis. nd nd nd nd nd nd nd nd 28 F 18 P Spinal nd nd dis. dis. nd nd nd nd nd nd nd nd 29 M 37 P Spinal 3+ neg dis. dis. nd nd nd nd nd nd nd nd Dewaele et al. Clinical Sarcoma Research 2011, 1:4 Page 4 of 16 http://www.clinicalsarcomaresearch.com/content/1/1/4 Table 1 Pathologic description of chordoma cases and results summary (Continued) 30a M 42 P Coccyx neg neg monos. dis. nd nd nd nd nd nd nd nd 30b 58 R Sacrum 2+ neg monos. dis. nd nd nd nd nd nd nd nd 31a F 60 P Sacrum neg 1+ polys. dis. nd nd nd nd nd nd nd nd 31b R Ilium neg nd polys. dis. nd nd nd nd nd nd nd nd TK Mut., Tyrosine Kinase Mutations; P, primary; R, recurrence; M, metastasis; nc, nasal cavity; neg, negative; l.l.amp., low level amplification; h.l.amp., high level amplification; polys., polysomy; dis., disomy; monos., monosomy; nd, not done; interm, intermediate; E, expressed; P, phosphorylated. Array-CGH (aCGH) analysis Fluorescence In Situ Hybridization (FISH) Array-CGH experiments were performed as previously Dual-color interphase FISH analysis was carried out on described on DNA extracted from 11 tumors (Table 2) 4 μm paraffin embedded tissue sections of 42 tumor [19]. For genomic profiling that included the evaluation biopsies. Sections were pretreated using the SPoT-Light of all 90 TK known in humans, the 4.3 K genomic DNA Tissue Pre-treatment Kit (Invitrogen, Life Technologies), tyrosine kinase array (TK-aCGH) was manufactured at according to the instructions of the manufacturer. FISH the Microarray Facility of the Flanders Interuniversity was performed as previously described [22]. Slides were Institute for Biotechnology, KULeuven [20]. In short, counterstained with 0.1 μM 4,6-diamidino-2-phenylin- the Sanger 1 Mb Clone Set containing 3527 BAC/PAC dole (DAPI) in an antifade solution for microscopy. clones was supplemented with 800 clones from 32 K For analysis of EGFR family members, FISH was per- CHORI BAC/PAC library, which specifically covers all formed using the locus specific identifier (LSI) EGFR- known human TK, and these two clone sets were SpectrumOrange(SO)/CEP7-SpectrumGreen(SG) and spotted together in duplicate on Code Linked Slides (AP PathVysion HER2-SO/CEP17-SG probes (Applied Bio- Biotech, US). The complete list of these clones is avail- systems/Ambion, Life Technologies, Carlsbad, CA, able upon request. The array-CGH data were statistically USA). For evaluation of PDGFRB/CSF1R copy numbers analyzed with aCGH-smooth, software especially and PDGFRB/CSF1R integrity, the SG-labeled bacterial designed for the analysis of heterogeneous samples [21]. artificial chromosome (BAC) RP11-21I20 (which maps Table 2 Gains and losses in chordoma using whole-genome 4.3 K TK-CGH-array Case 3 Case 4b Case 5 Case 7b Case 9 Case 10b Case 13 Case 14 Case 15a Case 16 Case 17a Gains 1q11-qter 5 2pter-p12 16q12.2- n.d. 7 13q31.2- 1q11-qter n.d. n.d. n.d. 7 7 q22.1 qter 2 8q11.21- 12pter- qter q24.23 10pter-p11 17q12.1- 20 qter Losses 3pter-p11.1 1pter-p11 1pter-p11.2 1pter-p13.1 2q21.1-qter 1 3pter- 3 1pter-p32.3 1pter- n.d. 8pter-p12 3 3p24.1-p13 3 3q11.2-q28 3pter-p12.1 p14.2 9pter-p11 1p22.3- p33.2 9 4 3q11.2- 4pter-p16.1 5q35.2-qter 9pter-p21 9 10 p21.3 3pter- 14 10 q13.31 6p22.3-p21.1 7pter-p22.1 10 14 14 1p21.2- p11.2 16q23.2- 11pter- 3q26.1- 9 8pter- 19p13.3- 17pter-p12 p13.2 22q12.1- q24.3 11p11 26.31 10 p11.21 p13.2 19p13.3- 2pter-p11.2 qter 13 3q28-qter 13q12.11- 11q12.2- 22q12.2- p13.2 2q31.2-qter 14 4p15.31- q12.13 q13.3 qter 6pter-p21.1 18 q21.21 16q12.1- 16pter- X 9 22 5pter-p15.2 q12.2 p12.1 10q11.23- Y 9pter-p21.1 16q22.3- 17pter- q24.2 9q34.11- q24.3 p11.2 18q11.2-q23 qter 17q12- 18q11.2- 19 11q12.2- q21.33 qter 21 q13.3 19p13.3- 20q11.21- 22 13q21.3- p13.11 qter q21.33 19q13.31- 22 13q33.1- qter qter n.d.: not detected. Dewaele et al. Clinical Sarcoma Research 2011, 1:4 Page 5 of 16 http://www.clinicalsarcomaresearch.com/content/1/1/4 centromeric to PDGFRB/5q33.1 and covers the adjacent described [22]. In short, tumor lysate aliquots containing CSF1R gene) and the SO-labeled RP11-368O19 (which 30 μg of protein were electrophoresed and blotted to maps telomeric to CSF1R and covers the PDGFRB gene) PVDF membranes (GE Healthcare, UK). Membranes DNA probes (both from Research Genetics, Huntsville, were blocked in PBS containing 5% blocking reagent AL, USA) were used. In addition, the PTEN copy num- (non-fat milk) and immunoblotted sequentially using bers were investigated using dual-color LSI PTEN/ rabbit antibodies against phospho-EGFR(Tyr1068) CEP10 probe (Applied Biosystems/Ambion). (Santa Cruz Biotechnology, Santa Cruz, CA, USA), total Hybridization signals were visualized using an epi- EGFR (Santa Cruz Biotechnology), phospho-PDGFRB fluorescence microscope (Leica DMRB, Wetzlar, Ger- (Y751) (mc; Cell Signaling, Beverly, MA, USA), total many) equipped with a cooled CCD camera and run by PDGFRB (mc; Cell Signaling), phospho-KIT(Tyr703) the ISIS digital image analysis system (MetaSystems, (mc; Invitrogen, Life Technologies), total KIT (pc; Altlussheim, Germany). One hundred nuclei were evalu- DAKO), phospho-ERK1/2 (Cell Signaling), total ERK1/2 ated for the number of red and green signals in different (Cell Signaling), phospho-AKT (Cell Signaling) and total areas corresponding to tumor tissues. AKT (Cell Signaling), diluted in 5% blocking reagent. FISH results were classified into five categories Total b-actin (Sigma Aldrich, St. Louis, MO, USA) was according to the percentage of tumor cells with a speci- used as a protein-loading and transfer control. The fic gene/CEP ratio and according to the gene copy num- HRP-conjugated anti-rabbit IgG (DAKO) were used at a ber per nucleus: 1) monosomy (1 signal from the gene dilution of 1:2000, and visualized with Enhanced Chemi- paralleled by one chromosome centromere signal) or luminescence (Thermo Scientific, Rockford, IL, USA). loss (a gene/CEP ratio of <0.6) in >40% of cells; 2) dis- omy (2 signals from the gene/CEP probes); 3) polysomy Receptor tyrosine kinases (RTK) activation profiling using (defined as > 2 gene signals per nucleus paralleled by antibody arrays similar increases in chromosome centromeric signals in The activation of RTK and their downstream signaling at least 10% of tumor cells); 4) low level gene amplifica- pathways were analyzed using the Proteome Profiler™ tion (gene/CEP ratio of > 2 in 10%-40% of tumor cells) Array kits (ARY001 and ARY003, R&D Systems, Min- or 5) high level gene amplification (presence of gene neapolis, MN, USA) in 12 fresh frozen chordoma tumor clusters or a gene/CEP ratio of > 2 in ≥40% of analyzed specimens. Assays were performed according to the cells). manufacturers’ protocol, and using 500 μgofprotein lysate per array. The images were captured and the level Mutation analysis of RTK activation was visualized with the FUJI mini- Mutational analysis was performed on genomic DNA LAS3000-plus imaging system (FUJIFILM, Tokyo, Japan) extracted from frozen tumor tissues (n = 13). The and densitometrically quantified with AIDA software sequence coding for the juxtamembrane and/or kinase (Raytest isotopenmessgeräte GmbH, Straubenhardt, Ger- domains of PDGFRA and PDGFRB (exons 12, 14 and many). The signal intensities of the probes and the local 18), KIT (exons 9, 11 and 17), CSF1R (exons 10 to 20) background of the probes were log transformed in and EGFR (exons 18 to 21) genes, were amplified by order to obtain a more symmetric distribution, and the polymerase chain reaction (PCR), using standard Taq differencebetween thesetwo resulted in alog trans- DNA polymerase (Roche Diagnostics, Basel, Switzerland) formed ratio (further referred to as log -intensity ratios). and the ABI PRISM 9700 (Applied Biosystems). Geno- For data normalization, within an array and within a mic sequences were obtained from online databases membrane the mean log -intensity ratio was calculated from the National Center for Biotechnology Information and then subtracted from the log -intensity ratio of each (NCBI), and specific primers for amplified fragments probe. Subsequently, the mean of the log -intensity were designed using the Primer3 software [23] (http:// ratios for eachkinasewithinanarray wascalculated. In frodo.wi.mit.edu/cgi-bin/primer3/primer3_www_slow. the statistical analysis, a linear mixed model was used cgi). Primers sequences are available upon request. The instead of a one-sample t-test per probe since the arrays PCR products were purified (QIAquick PCR Purification or membranes used to measure the probe intensities Kit, QIAGEN, Hilden, Germany) followed by direct bi- maydiffer. Thelinear mixedmodel hasthe log -inten- directional cycle sequencing using the ABI PRISM 3130 sity ratios as responses, the probes as fixed effects and XL Genetic Analyzer (Applied Biosystems, Foster City, the membrane as random effect per array [24,25]. The CA, USA). alpha level was set at 5%. As multiple testing correc- tions, the p-values from the tests for the different probes Western immunoblotting were adjusted to control the false discovery rate as Cell lysisoffrozentumors(n=9),SDS-PAGE, and described by Benjamini and Hochberg [26]. The ranking immunoblotting were carried out as previously of the probes was based on the adjusted p-values. All Dewaele et al. Clinical Sarcoma Research 2011, 1:4 Page 6 of 16 http://www.clinicalsarcomaresearch.com/content/1/1/4 analyses were performed with the statistical package tumors analyzed showed CNA by aCGH. CNA frequen- SAS (version 9.2), using the procedure PROC MIXED cies were calculated on these ten cases with CNA. for the linear mixed model. Losses were more common than gains, supporting pre- viousfindingsinchordoma[7].There wasamedian of Results one gain (range 0-5) and seven losses (range 0-14) per Histopathology and immunohistochemistry tumor. Genomic losses affecting five or more tumors (≥ All the chordomas in our cohort were reviewed and 50% of cases) were identified on chromosomes 1, 3, 9, classified as conventional chordomas by means of mor- 10, 19 and22(Table 2 and3,Figure2). Thesmallest phology and immunohistochemistry (IHC). They show common region of chromosome 3 deletion, covering prominent lobules separated by fibrous septa. The bands 3p24.1-p14.2, was lost in eight cases. Three tumor cells are arranged in cords or sheets or may be regions located on the short arm of chromosome 1, i.e. floating singularly in the abundant myxoid matrix often 1pter-p33.2, 1p22.3-p21.3 and 1p21.2-p13.2, were recur- present. The histologic hallmark is characterized by rently lost in six, five and five cases, respectively. Whole large tumor cells with abundant vacuolated cytoplasm, chromosome 9 loss was observed in four cases, and the referred to as physaliphorous cells [2]. The tumor cells region 9q34.11-qter, involving among others the TSC1 co-express keratin, EMA and S-100 protein. Of the 39 tumor suppressor gene, was lost in one additional case. chordomas tested by IHC for EGFR expression, 19 were Furthermore, the region 9pter-p21 was lost in three primary and 20 were advanced lesions. The EGFR extra cases of our cohort. Of note, homozygous deletion immunopositivity was found in 26 out of 39 cases of the chromosomal sub-band 9p21.3 (the region con- (67%), showing different levels of reactivity (Figure 1, taining the CDKN2A tumor suppressor gene) was found Table 1). Thus, 11 tumors presented with an intense in three of analyzes tumors. The entire chromosome 10 and diffuse cytoplasm membrane positivity in more than was lost in four cases and the region 10q11.23-q24.2, 10% of the cells (scored as “3+”), six cases showed encompassing the tumor suppressor PTEN, was lost in intense positive staining but in less than 10% of the cells another case. Losses that implicated chromosome 19, (scored as “2+”), and nine other cases were considered with the commonly deleted region 19p13.3-p13.2, were weakly and discontinuously stained in more than 10% of found in five cases. Total or partial chromosome 22 the cells (scored as “1+”). EGFR expression was more deletions, with the common region 22q12.2-qter, were frequently found in advanced tumors compared with recorded in six chordomas. The most common gain was primary tumors (80% versus 58%, respectively). In detail: the gain of the entire chromosome 7, observed in three 15 out of 20 advanced cases stained positive for EGFR chordoma cases (Table 2). Notably, the genes coding for versus 11 out of 19 primary cases. Additionally, when the EGFR, MET, LMTK2, EPHA1, EPHB4 and EPHB6 comparing the primary and the advanced stage within proteins are mapped on chromosome 7. No amplifica- patients, in cases 2, 4, 7, 17 and 30: stronger EGFR tions or rearrangements within the 90 known TK were staining was observed in the advanced in comparison detected in our cohort of chordomas. with the primary stage. Case 12 showed intense and dif- fuse (3+) staining in both the primary and the advanced FISH analysis stage. Case 1 was the only exception, showing stronger Thegenecopynumbers of the EGFR, HER2, CSF1R/ EGFR staining in the primary than in the advanced PDGFRB and PTEN were analyzed by FISH (Figure 3, stage. Cases 10, 15 and 31 stained negative for EGFR in Table 1). Sixteen out of 42 tumors analyzed revealed the primary stage and stayed negative upon progression. disomy for EGFR, while 16 (38%) cases displayed polyso- HER2 expression was tested in 16 cases, of which 11 mic cell clones. Two cases showed chromosome 7 polys- were negative, four displayed low level of staining inten- omy. Only a small fraction of tumors (four cases) sity and one case showed intense positive staining, albeit presented with EGFR amplification, and only in two in less than 10% of the cells. HER2 expression was cases at high level. Notably, four cases showed EGFR almost as frequent in primary as in advanced tumors loss. The gene copy number of HER2 was also analyzed (33% versus 29%, respectively). The HER2 immunoposi- in all cases, and six specimens revealed polysomy of tivity was associated with EGFR co-expression in all but HER2.Three casesshowed HER2 loss. Of note, half of one lesion, although the level of EGFR expression was the HER2 gains were not detectable by aCGH, probably heterogeneous. due to a low number of neoplastic cells in these speci- mens. Copy number gains of both, EGFR and HER2 aCGH study genes, correlated well with HER2 immuno-positivity by Using the whole genome 4.3 K TK-array, we studied IHC. Of the 34 cases analyzed, 13 tumors were polyso- copy number aberrations (CNA) in eleven cases for mic for CSF1R/PDGFRB and two revealed loss of which frozen tissue was available. Ten out of the 11 CSF1R/PDGFRB; the remaining presented disomy for Dewaele et al. Clinical Sarcoma Research 2011, 1:4 Page 7 of 16 http://www.clinicalsarcomaresearch.com/content/1/1/4 AB CD EF Figure 1 Histology and EGFR protein expression in chordomas. A and B/ Examples of histologic appearance of chordomas stained with hematoxylin and eosin (H&E). C - F/ Illustration of chordoma cases with heterogeneous type of positive EGFR immunostaining. F/ The typical physaliphorous cells with abundant vacuolated cytoplasm, showing EGFR membrane staining. Dewaele et al. Clinical Sarcoma Research 2011, 1:4 Page 8 of 16 http://www.clinicalsarcomaresearch.com/content/1/1/4 Table 3 Recurrent copy number losses in chordoma cases and examples are depicted in Figure 4. The probes are by aCGH ranked according to their false discovery rate (fdr) Regions lost in ≥ five cases adjusted p-value. The column “Estimate” shows the esti- mate mean log -intensity ratio for each RTK or kinase Chordoma cases (#) Cytogenetic Frequency Candidate 2 location genes over all experiments. The first three RTK-probes and 4b, 5, 7b, 10b, 15a, 16 1pter-p33.2 0.60 RUNX3 the first twelve kinase-probes in Table 4 have a log - 4b, 5, 7b, 10b, 15a 1p22.3-p21.3 0.50 intensity ratio significantly larger than zero at the alpha 3, 4b, 5, 7b, 10b, 13, 3p24.1-p14.2 0.80 RBM5, FHIT, level of 5%. Thus, the EPHB2, EGFR and macrophage- 14, 16 PTPRG stimulating protein receptor (MSPR) were found to be 4b, 5, 7b, 9, 14 3q11.2-q13.31 0.50 significantly activated in chordoma. Although present in 4b, 5, 7b, 9, 14 3q26.1-26.31 0.50 some of the analyzed specimens, activation of the 3, 5, 7b,10b, 13, 14, 9pter-p21 0.70 CDKN2A PDGFRB, FGFR3, CSF1R and ERBB4 was not statisti- 15a cally significant in our study. Strikingly, there was no 3, 5, 7b, 13, 15a 9q34.11-qter 0.50 TSC1 detectable activation of KIT or VEGF receptors. By ana- 4b, 7b, 10b, 14, 15a 10q11.23-q24.2 0.50 PTEN lyzing the signaling pathways (the profiles of 46 kinases 5, 7b, 10b, 14, 15a 19p13.3-p13.2 0.50 and protein substrates), AKT, RSK1/2/3, TP53, MSK1/2, 4b, 5, 9, 10b, 15a, 16 22q12.2-qter 0.60 CHEK2 YES, p38a, p70 S6K, CREB and SRC were the most fre- quently and strongest phosphorylated proteins in our these genes. The tumor suppressor PTEN was lost in cohort. Interestingly, SRC family members, as SRC and seven out of 18 analyzed tumors. YES, were recurrently activated in chordoma. Further- more, kinase-array revealed the activation of down- Mutation analysis stream effectors of both, the PI3K/AKT/mTOR and No activating mutations of EGFR,CSF1R,PDGFRB, RAS/RAF/MAPK pathways. PDGFRA or KIT in examined genes’ exons were found in any of the 13 analyzed cases (Table 1). Western immunoblotting The consistent protein expression of EGFR and RTK phosphorylation profiling using phospho-RTK and PDGFRB and the recurrent activation of EGFR were phospho-kinase antibody arrays confirmed by Western blotting (Figure 5). The expres- The results of the RTK- and kinase-analysis of 12 and sion status of EGFR in all cases was in agreement with 10 chordoma samples respectively are shown in Table 4 the results obtained by IHC (Table 1 Figure 5). Briefly, losses gains amplification 60% + 2p + 7 40% + 1q 20% 0% -20% -40% -60% 22q12.1-qter 9q34.11-qter 1pter-p33.2 19p13.3-p13.2 -80% 9pter-p21 3q11.2-q13.31 3p24.1-p14.2 -100% 1 3 5 7 9 11 13 15 17 19 21 2 4 6 8 10 12 14 16 18 20 22 Figure 2 Frequency (%) of gained and lost regions detected by 4.3K TK aCGH in chordomas. Gains are shown in grey, losses in blue and amplification in black. Important recurrent gains and losses are circled in red. No rearrangements or high level amplification of genes encoding TK were detected. Dewaele et al. Clinical Sarcoma Research 2011, 1:4 Page 9 of 16 http://www.clinicalsarcomaresearch.com/content/1/1/4 A B C D Figure 3 Representative examples of dual-color interphase FISH images on paraffin sections in chordomas.Detected by the co- hybridization of SpectrumOrange labeled EGFR DNA probe (red signals) and SpectrumGreen labeled chromosome 7 CEP probe (green signals). (A) Case 1a, showing EGFR disomy. (B) Case 10a reveals EGFR polysomy. (C) Case 20 shows low level amplification in < 10% of nuclei. Of note, this amplification is not detected by aCGH. (D) Case 12b, showing high level amplification of EGFR in > 40% of nuclei. cases 15b and 10b showing only faint EGFR staining on used as starting material for both experiments, which the Western blot were scored negative by immunostain- may bring about differences, as chordomas are proven ing. All other cases, presenting clear or intense EGFR to be heterogeneous lesions. By Western immunoblot, expression by Western, were immune-scored accord- PDGFRB was found to be expressed in all chordomas ingly as “1+”, “2+” or “3+”.Two specimenswereana- analyzed, although only one case (#10b) also presented lyzed in parallel by Western immunoblotting and RTK activated PDGFRB. KIT protein expression and low antibody array. The strong EGFR activation of case 12b level activation was found in three and two cases detected by Western was confirmed by RTK antibody respectively. array. In case 4b, EGFR was expressed but not activated by Western. However, intermediate activation of EGFR Discussion was disclosed for this lesion by RTK antibody array. Recent reports suggest that RTK might be essential for This apparent difference could be ascribed to the fact the survival or proliferation of chordoma tumor cells. that the antibody used for Western blot detects the Therefore, targeting RTK may offer new therapeutic phosphorylation status of just one EGFR tyrosine resi- options for chordoma treatment. Nevertheless, there are due(Y1068),while theantibodyarraydetects thephos- important discrepancies between the reported results, phorylation of all tyrosine residues on the EGFR which are most likely due to differences in the relative protein. Furthermore, different pieces of the tumor were sensitivities of the methods used or heterogeneity of the Dewaele et al. Clinical Sarcoma Research 2011, 1:4 Page 10 of 16 http://www.clinicalsarcomaresearch.com/content/1/1/4 Table 4 Significantly phosphorylated RTK and kinase sites in chordoma using Proteome Profiler arrays, ranked based on p-value Probe name Estimate Standard Error t-value Raw p-value fdr adjusted p-value Phospho-RTK EPHB2 0.1285 0.0263 4.9 6.6931E-07 2.8111E-05 EGFR 0.6762 0.1694 3.99 3.8547E-05 0.0008 MSPR 0.1241 0.0426 2.91 0.0019 0.0266 PDGFRB 0.0848 0.0334 2.54 0.0057 0.0600 FGFR3 0.1022 0.0484 2.11 0.0177 0.1487 CSF1R 0.0887 0.0445 1.99 0.0236 0.1652 ERBB4 0.0160 0.0289 1.78 0.0379 0.2272 Phosphorylated kinase site AKT (T308) 0.3117 0.0313 9.95 3.1253E-21 1.5001E-19 RSK 1/2/3 (S380) 0.1747 0.0212 8.25 1.2388E-15 2.9731E-14 TP53 (S46) 0.2394 0.0336 7.14 2.3075E-12 3.6920E-11 MSK 1/2 (S376/S360) 0.1557 0.0256 6.09 1.3564E-09 1.6277E-08 YES (Y426) 0.1639 0.0288 5.69 1.2512E-08 1.2012E-07 TP53 (S15) 0.2533 0.0469 5.41 5.5176E-08 4.4141E-07 p38a (T180/Y182) 0.2858 0.0625 4.57 3.2798E-06 2.2490E-05 p70 S6K (T421/S424) 0.1086 0.0242 4.49 4.6993E-06 2.8196E-05 CREB (S133) 0.3273 0.1018 3.21 0.0007 0.0038 RSK 1/2 (S221) 0.0707 0.0246 2.87 0.0022 0.0104 SRC (Y419) 0.0934 0.0349 2.68 0.0038 0.0158 TP53 (S392) 0.1237 0.0464 2.67 0.004 0.0158 TOR (S2448) 0.2407 0.1258 1.91 0.0284 0.105 JUN (S63) 0.0863 0.0533 1.62 0.053 0.1818 HSP27 (S78/S82) 0.1048 0.0691 1.52 0.0647 0.2016 eNOS (S1177) 0.2002 0.1331 1.50 0.0672 0.2016 STAT1 (Y701) 0.0465 0.0318 1.46 0.0725 0.2048 STAT5b (Y699) 0.0380 0.0286 1.33 0.0921 0.2457 LYN (Y397) 0.0351 0.0283 1.24 0.1079 0.2725 STAT6 (Y641) 0.0309 0.0284 1.09 0.1382 0.3317 STAT5A (Y699) 0.0656 0.0715 0.92 0.1791 0.4093 FYN (Y420) 0.0587 0.0768 0.76 0.2239 0.4884 STAT5A/B (Y699) 0.0168 0.0366 0.46 0.3229 0.6739 ERK1/2 (T202/Y204. T185/Y187) 0.0284 0.0708 0.40 0.3447 0.6894 * The probes written in bold have a log2-intensity ratio significantly larger than zero at the a-level of 5%. material analyzed. Moreover, the characterization of Interestinglythough, themostrecurrentcopynumber chordoma in most studies is rarely based on parallel gain, found in three out of ten cases, involved the entire multiple techniques. Our objective was to characterize chromosome 7. Gain of chromosome 7 is frequently this rare entity in search for relevant therapeutic targets reported in chordomas, and multiple genes that encode using a wide range of methodological approaches. TK are located on chromosome 7, including the EGFR Whole genome 4.3 K TK-array CGH revealed moder- [3,4,7,27-29]. Accordingly, copy number gains involving ately complex CNA across the genome in all but one the EGFR locus, were found by FISH in 22/42 (52%) of examined cases, with losses more common than gains. our cases. Polysomy of the EGFR/ERBB1 gene was pre- The CNA found in our cohort were in accordance with viously reported in a subset of chordomas, and the previously recognized imbalances in chordomas EGFR is an interesting target for therapy in chordoma [3,4,7,27-29]. No deletions or gains common to all sam- based on the availability of targeted molecular inhibitors ples were found, confirming that chordomas are geneti- [8,16]. Additionally, the status of the gene encoding cally heterogeneous tumors. HER2, a close family member and important dimeriza- Importantly, we did not identify any amplifications or tion partner of EGFR, was investigated. Copy number rearrangements involving genes coding for TK. gains of HER2 were identified in 6/42 (14%) of cases. Dewaele et al. Clinical Sarcoma Research 2011, 1:4 Page 11 of 16 http://www.clinicalsarcomaresearch.com/content/1/1/4 Case 18 9 10 Case 17a 9 10 1 13 Case 20 9 10 1 7 14 11 Case 21 9 10 Figure 4 Representative images from phospho-RTK (left panel) and phospho-kinase (right panel) arrays from chordoma cases 18, 17b, 20 and 21. The EGFR and EPHB2 TK are frequently activated and downstream RTK signaling intermediates are activated consistently in chordomas. Each kinase is spotted in duplicate. The pairs of dots in each corner are positive controls. Each pair of the most positive kinase dots is denoted by a numeral, with the identity of the corresponding kinases listed as follows: 1) EGFR, 2) CSF1R, 3) MSPR, 4) PDGFRB, 5) FGFR3, 6) EPHB2, 7) HER2, 8) TOR, 9) AKT, 10) TP53, 11) RSK1/2/3, 12) S6K, 13) CREB, 14) YES, 15) MSK1/2, 16) RSK1/2, 17) eNOS. Noteworthy, copy number gains of HER2 were exclu- Accordingly, loss of expression of the CDKN2A protein sively found in recurrent or metastatic cases in our in chordoma was also previously shown by immunos- cohort, further suggesting its possible association with taining [32]. Other recurrent losses, observed in the pre- poor clinical outcome. sent study by aCGH, involved regions carrying the Losses of large chromosomal regions are typically tumor suppressors PTEN/10q23.31, CHEK2/22q12.1 and found in chordoma. In this study, losses revealed by the transcription factor RUNX3/1p36.11, all previously aCGH predominantly involved chromosome 3; the smal- described in chordomas [7]. lest overlapping region of deletion, 3p24.1-p14.2, was In order to characterize the compendium of co-acti- lost in eight out of ten analyzed cases. This region con- vated RTK in chordoma, we used an antibody array that tains multiple genes, including RBM5, FHIT and allows the simultaneous characterization of the phos- PTPRG, but their involvement in chordoma pathogen- phorylation status of 42 different RTK. Most impor- esis has yet to be determined. Loss of the 9pter-p21 tantly, the EGFR kinase was consistently activated in all region, another frequent feature revealed by aCGH ana- 12 investigated cases. Furthermore, statistical analysis lysis, was found in seven out of ten tumors. Importantly, showed that EGFR activation was significant for chordo- in three cases the region was homozygous lost. The mas, based on the analysis of our cohort. The activation losses encompassed the tumor suppressor genes of EGFR in chordoma was previously shown by other CDKN2A and CDKN2B, which are frequently deleted in groups, although the reported frequencies of the EGFR many tumor types [30,31]. Correspondingly, Hallor and activation in chordoma vary significantly [8,16]. By RTK co-workers observed loss of the CDKN2A locus with an antibody array Tamborini and co-workers reported incidence of 70% in chordoma, and with an even higher EGFR, HER2 and HER4 activation in 6/7 (86%), 5/7 frequency considering just metastasizing lesions [7]. (71%) and 3/7 (43%) of cases, respectively [8]. However, Dewaele et al. Clinical Sarcoma Research 2011, 1:4 Page 12 of 16 http://www.clinicalsarcomaresearch.com/content/1/1/4 statistical analysis included a multiple testing correction. Chordoma cases The linear mixed model avoids the use of an arbitrarily 15b 4b 9 2b 10b 5 7b 12b 14 chosen cut-off that can lead to overestimation of the p-PDGFRB (Y751) 190 kDa activation of RTK and to uncertainty about the results. Notably, statistical analysis was never described by 190 kDa others in reports published so far in reference to RTK PDGFRB proteome profiling kits, thus the statistical significance of reported data is unknown. Importantly, we also found p-EGFR (Y1068) 175 kDa two other RTK: EPHB2 and MSPR, to be significantly activated in chordoma. The activation of EPHB2 was 175 kDa EGFR recently described in one chordoma study [16]. The role of EPHB2 in chordoma development and progression 145 kDa p_KIT (Y703) 125 kDa needs to be further evaluated. In general, EPHB2 func- tion depends on the tumor type and signaling context of 145 kDa KIT the neoplastic cell. The EPHB2 has a tumor suppressive 125 kDa role in colon carcinoma; in contrast, EPHB2 promotes Actin 42 kDa cell proliferation in adenomas and normal intestinal epithelium. Notably, it was recently shown in mice mod- Figure 5 Western immunoblot of nine chordoma cases.The els that the intrinsic kinase activity of EPHB2 conveys immunoblot confirms the frequent expression of EGFR and PDGFRB, mitogenic signals [33]. It is of interest that imatinib and frequent activation of EGFR, but not of PDGFRB and KIT mesylate is as an inhibitor of EPHB mitogenic signaling. proteins. Equal amounts of total protein extracts from nine tumors were separated on a gel, immunoblotted and then probed with the TheMSPR/RONtyrosinekinaseisamember of the indicated antibodies. MET family of RTK. MET expression was shown pre- viously in chordomas by several other groups, but MSPR expression and activation was only recently using immunoprecipitation assay, EGFR and HER2 were reported in all three investigated chordomas by Shalaby phosphorylated in respectively 17/22 (77%) and 6/14 andco-workers [16].Asitisthe case with itsbetter- (43%) of their cases [8]. Using the same RTK antibody known family member, MET, several lines of evidence array, Shalaby and colleagues recently showed activation suggest a role for RON in human cancer. Generally, of HER2, MSPR, EPHB2 and MER for the U-CH1 chor- RON overexpression is associated with poor clinical out- doma cell line and the three tested chordoma cases [16]. come and metastasis [34]. Foretinib, an oral multi-kinase inhibitor of MET, RON, AXL and VEGFR, is currently In our study, we found significant activation of EGFR, in phase I and II clinical testing [35]. HER2 and HER4 in respectively 12, one and one out of The multiple RTK co-activation is not a distinctive 12 cases, using the same antibody arrays. Interestingly, feature of chordomas, because similar patterns were the frequent activation of PDGFRB in chordomas [21/22 reported in other tumor types, such as colon adenocar- (95%) of cases] was described in the study by Tamborini cinomas, intimal sarcomas, glioblastomas or osteosarco- and collaborators [8]. In contrast, we found activation of mas [36-38]. Importantly, the simultaneous activation of PDGFRB only in five out of 12 (42%) chordomas, using multiple RTK provides the tumor cells with reduced thesameantibodyRTK arrays andusing thevalue of dependence on a single RTK for the maintenance of cri- the mean plus the standard deviation within an array as tical downstream signaling, and thus renders such the cut-off. However as indicated by statistical analysis, tumors refractory to single-agent RTK inhibition. PDGFRB activation was not significant in our cohort. This discrepancy might be attributable to the heteroge- The conflicting results on the frequency of EGFR, neity of chordoma tumors, the quality of the frozen HER2, PDGFRB expression and activation, and also tumor tissue used for the analysis, modifications of the copy number alterations in chordoma, might be due to technique and/or to subsequent dissimilar analysis of differences in sensitivity of the techniques used. In addi- the data. Thus, Tamborini and co-workers used high- tion, even if using the same technique, there are impor- concentrated (e.g. 2 mg/array) protein lysate per array in tant variations in methodology between different their study [8]. In contrast, we performed the experi- laboratories, with many confounding factors contribut- ments according to the manufacturers’ recommenda- ing to the inconsistencies, e.g. the different type and tions which indicate 500 μg of total protein as the source of the antibodies used in the immunohistochem- maximum amount to be used for each array. In addi- ical studies. When immunostaining is considered, it is tion, we have performed an extensive statistical analysis well known that the way of tissue fixation influences of thedatabyusing alinear mixedmodel.Our outcome [39]. Tumor specimens are frequently retrieved Dewaele et al. Clinical Sarcoma Research 2011, 1:4 Page 13 of 16 http://www.clinicalsarcomaresearch.com/content/1/1/4 from archives, and in case they are not preserved well, studies [8,42]. The discrepancy in the prevalence of the this maygiverisetofalse negative cases. Thelackof activated proteins between the reported results is most sensitivity of IHC to identify low expression levels of likely due to differences in the relative sensitivity and EGFR was comprehensively illustrated in colorectal can- specificity of the methods. This is well illustrated in a cer [40]. Similarly, chordoma immunostaining might study by Dobashi and co-workers, who found activated also show inconsistencies associated with these metho- mTOR in all five cases using immunohistochemistry, dological problems. Along this line, Weinberger and co- but only in one case using Western immunoblotting workers found EGFR and HER2 expression in respec- [44]. Nevertheless, the involvement of the AKT/mTOR tively 12 (100%) and seven (58%) out of 12 chordomas, pathway in chordoma is clear. Importantly, efficient using IHC on tissue micro-arrays (TMA), while Shalaby inhibition of the human chordoma cell line UCH-1 by and colleagues showed EGFR expression and activation PI-103, a dual PI3K and mTOR inhibitor, was recently in respectively 69% (79/114) and 50% (56/115) of chor- reported [43]. Notably, it was recently shown that AKT doma cases by the same technique, and while Fasig and activation persists in the UCH-1 chordoma cell line fol- co-authors reported EGFR activation in nine out of 21 lowing treatment with the EGFR inhibitor tyrphostin (43%) cases [12,15,16]. By conventional immunostaining, [16]. we have also found that EGFR and HER2 are expressed Furthermore, by kinase antibody arrays, we also found in chordomas, albeit in a lower fraction of cases 26/39 effectors of RAS/ERK1/2 signaling to be significantly (67%) and 5/16 (31%), respectively. In contrast to Wein- activated in chordoma, like ribosomal S6 kinases (RSK) berger and co-workers, however, we found more fre- 1/2/3, the CREB transcription factor and the chromatin quent EGFR expression in advanced (15/20, 75%) rather associated kinase p38. More downstream are the mito- than in primary (11/19, 58%) lesions. Again in contrast gen- and stress-activated protein kinases, MSK1 and the to Weinberger and co-workers, we did find a positive closely related isoform MSK2. These are nuclear kinases correlation between HER2 expression and EGFR expres- that are activated by the ERK1/2 and p38 MAPK signal- sion, which is in line with the HER2/EGFR heterodimers ing cascades [45]. Additionally, the SRC family mem- formation in chordomas reported by other groups [8,12]. bers, SRC and YES, were also activated. These pathways Moreover, we did not find a significant correlation were not extensively analyzed in chordoma by other between EGFR and HER2 gene status and their expres- groups, except for ERK1/2, which was described to be sion by immunostaining, this phenomenon was also consistently strongly phosphorylated in chordoma by described in colorectal cancer [40,41]. Tamborini and co-workers [8]. Nevertheless, these acti- The circuitry of intracellular signalling downstream of vated proteins are all confounding factors that might RTK is an area of dynamic investigations in many can- offer the tumors redundancy, making them less respon- cer types and advances in the characterization of this sive to upstream RTK and AKT pathway inhibition. signalling allows better selection of appropriate thera- Oncogenes often cooperate with additional mutations peutic agents. In the present study, we analyzed the acti- that disrupt tumor suppressor pathways. Phosphatase vation of important effectors of signalling downstream and tensin homologue deleted on chromosome ten of RTK. Using kinase antibody arrays, AKT was the (PTEN), is an important negative regulator of the AKT/ most frequent (found in nine out of ten cases analyzed) mTOR pathway, which when not expressed contributes and highest phosphorylated in chordomas. Similarly, to constitutive phosphorylation of AKT and activation Presneau and co-workers found AKT activation in 45 of downstream effectors. PTEN loss is also frequently out of 49 (92%) chordomas analysed by TMA, and Tam- found in chordomas. We observed loss of PTEN in five borini and colleagues in 21 out of 22 chordomas (95%) out of ten cases by aCGH, and in seven out of 18 (39%) using Western blotting [8,42]. The AKT protein trans- cases by FISH. Presneau and co-workers recently duces signals to several effector molecules, including revealed loss of PTEN protein expression in seven out TSC1/2. More specifically, AKT inhibits TSC1/2 and of 43 (16%) cases by IHC and semi-quantitative RT-PCR hereby relieves inhibition of mammalian target of rapa- [42]. Han and co-workers showed negative PTEN stain- mycin (mTOR), which functions downstream of TSC1/ ing by IHC in six out of ten sporadic chordoma [46]. 2. This occurs in part by phosphorylating two substrates, Just like in our cases, they did not find any correlation p70S6 kinase (S6K) and eukaryotic initiation factor 4E- between loss of PTEN and advanced disease. TSC1 is binding protein 1 (4E-BP1). Of note, p70S6K was acti- another critical tumor suppressor, implicated down- vated in five and mTOR in three of our ten chordoma stream in the PI3K/AKT and RAS/ERK pathways. In cases analyzed by kinase antibody arrays. These data are particular, upon growth factor activation, AKT, ERK and in accordance with previously published data [8,15,43]. p90 ribosomal S6 kinase 1 (RSK1) participate in TSC The phenomenon that p70S6K was activated in p- protein complex inhibition, hereby critically regulating mTOR negative chordomas was found in multiple cell growth and proliferation. Chordomas are reported Dewaele et al. Clinical Sarcoma Research 2011, 1:4 Page 14 of 16 http://www.clinicalsarcomaresearch.com/content/1/1/4 List of abbreviations in patients with tuberous sclerosis complex (TSC), an 4EBP1: eukaryotic translation initiation factor E4-binding protein 1; BAC: autosomal dominant disorder typified by hamartomas in bacterial artificial chromosome; CNA: copy number alterations; CSF1R: several organs, epilepsy, mental retardation and beha- colony-stimulating factor 1 receptor; DAPI: 4.6-diamidino-2-phenylindole; EGFR: epidermal growth factor receptor; ERK1/2: extracellular signal-regulated vioural problems. TSC is caused by germline mutations kinase; fdr: false discovery rate; HER2: v-ERBB2 Avian erythroblastic leukemia in the TSC1 or TSC2 genes and the loss of the corre- viral oncogene homolog 2; IHC: immunohistochemistry; IS: intimal sarcoma; sponding wild type allele. The chromosomal region MEK: mitogen-activated kinase kinase kinase 1; mTOR: mammalian target of rapamycin; NCBI: National Center for Biotechnology Information; PCR: 9q34.13, where the TSC1 gene is localized, is also fre- polymerase chain reaction; PDGFR: platelet derived growth factor receptor; quently lost in sporadic chordomas [7]. By aCGH, we PI3K: phosphatidyl inositol 3 kinase; PKB or AKT: protein kinase B; RTK: found loss of the region 9q34.11-qter, encompassing the receptor tyrosine kinase; S6K: ribosomal protein S6 kinase; SG: spectrum green; SO: spectrum orange; TK: tyrosine kinase; TKI: tyrosine kinase gene coding for TSC1, in five out of ten cases. Hallor inhibitors; TMA: tissue microarrays. and co-workers showed loss of this region in about 25% of 21 cases analyzed by aCGH. In contrast, Presneau Acknowledgements This work is supported by research grants from the EUROBONET consortium and co-workers found disomy for TSC1/2 by FISH in all (a network of excellence granted by the European Commission for studying of their 28 cases [42]. Generally, the consistent activa- the pathology and genetics of bone tumors), from the Fonds voor tion of AKT, the frequent activation of p70S6K and of Wetenschappelijk Onderzoek Vlaanderen (G.0589.09, MD-R), and by a Concerted Action Grant 2006/14 from the K.U.Leuven. mTOR, together with frequent loss of the TSC1 and PTEN genes, all suggest an important role for the PI3K/ Author details AKT pathway in chordoma. Department of Human Genetics, Catholic University of Leuven, University Hospitals, Leuven, Belgium. Department of Pathology, Catholic University of Leuven, University Hospitals, Leuven, Belgium. Laboratory of Experimental Conclusions Oncology, Department of General Medical Oncology, Catholic University of In summary, we found that EGFR is the strongest and Leuven, University Hospitals, Leuven, Belgium. I-BioStat, Catholic University of Leuven, Leuven, Belgium, and Hasselt University, Hasselt, Belgium. most frequently activated RTK in chordomas, and there- fore becomes a possible target for therapy. Lack of signifi- Authors’ contributions cant EGFR amplification and EGFR mutations suggests BD carried out the mutation analysis, participated in the aCGH data evaluation, Western immunoblotting analysis and antibody array analysis, activation by autocrine/paracrine ligand stimulation. and drafted the manuscript. FM carried out the histopathological PDGFRB is also activated in chordomas, but with a lower experiments and analysis and participated in the draft of the manuscript. GF frequency and/or to a lower level, which might not be participated in the antibody array experiments and analysis and histopathological analysis. MA performed the statistical analysis of the detectable by some current standard techniques. In the antibody arrays. VV carried out the FISH, aCGH, Western immunoblotting light of these findings, chordoma patients may benefit and antibody array experiments. AW performed the aCGH analysis and from treatment with multi-kinase inhibitors targeting both participated in the antibody array analysis. MDR participated in the design and coordination of the study and helped to draft the manuscript. RS EGFR and PDGFR. Furthermore, many other RTK are contributed tumor samples for this study, participated in the design of the activated in subsets of chordomas; these are likely to study and critically revised the manuscript. All authors read and approved increase treatment resistance in these tumors. These the final manuscript. results are currently only hypothesis-generating, and addi- Competing interests tional in vitro studies addressing the impact of inhibitors The authors declare that they have no competing interests. of RTK and their downstream effectors on chordoma Received: 25 January 2011 Accepted: 25 July 2011 tumor cells would be extremely useful in determining the Published: 25 July 2011 dominant and alternative RTKs in these tumors. As chor- domas are bone tumors, with a rigid, mineralized extracel- References lular matrix, ex-vivo studies on primary neoplastic 1. Stacchiotti S, Casali PG, Lo VS, Mariani L, Palassini E, Mercuri M, Alberghini M, Pilotti S, Zanella L, Gronchi A, Picci P: Chordoma of the chordoma cells will be difficult. Recent advances in com- mobile spine and sacrum: a retrospective analysis of a series of patients putational biology and network-based technologies gener- surgically treated at two referral centers. Ann Surg Oncol 2010, ating predictive models might be more of use [47]. 17:211-219. 2. Ferraresi V, Nuzzo C, Zoccali C, Marandino F, Vidiri A, Salducca N, Zeuli M, In conclusion, the consistent activation of AKT, the Giannarelli D, Cognetti F, Biagini R: Chordoma: clinical characteristics, recurrent activation of upstream EGFR and of down- management and prognosis of a case series of 25 patients. BMC Cancer stream effectors like p70S6K and mTOR, together with 2010, 10:22. 3. Mirra JM, Nelson SD, Della Rocca C, Mertens F: Chordoma. In World Health frequent loss of TSC1 and PTEN gene loci, all indicate Organization Classification of Tumours. Pathology and Genetics of Tumours of that the PI3K/AKT pathway is an important mediator of Soft Tissue and Bone. Volume 5.. 1 edition. Edited by: Fletcher CD, Unni KK, transformation in chordoma. Targeting this pathway is Mertens F. Lyon: IARC Press; 2002:315-317. 4. Scheil S, Bruderlein S, Liehr T, Starke H, Herms J, Schulte M, Möller P: likely to yield attractive data that will enlighten the Genome-wide analysis of sixteen chordomas by comparative genomic design of appropriate therapies. Individualized therapeu- hybridization and cytogenetics of the first human chordoma cell line, U- tic approaches depending on the genetic context of a CH1. Genes Chromosomes Cancer 2001, 32:203-211. 5. Tallini G, Dorfman H, Brys P, Dal Cin P, De Wever I, Fletcher CD, Jonson K, particular tumor are likely to be the most successful. Mandahl N, Mertens F, Mitelman F, Rosai J, Rydholm A, Samson I, Sciot R, Dewaele et al. Clinical Sarcoma Research 2011, 1:4 Page 15 of 16 http://www.clinicalsarcomaresearch.com/content/1/1/4 Van den Berghe H, Vanni R, Willén H: Correlation between 24. Chu T, Weir B, Wolfinger R: A systematic statistical linear modeling clinicopathological features and karyotype in 100 cartilaginous and approach to oligonucleotide array experiments. Mathematical Biosciences chordoid tumours. A report from the Chromosomes and Morphology 2002, 176:35-51. (CHAMP) Collaborative Study Group. J Pathol 2002, 196:194-203. 25. Wolfinger R, Gibson G, Wolfinger E, Bennett L, Hamadeh H, Bushel P, 6. Miozzo M, Dalpra L, Riva P, Volontà M, Macciardi F, Pericotti S, Tibiletti MG, Afshari C, Paules R: Assessing Gene Significance from cDNA Microarray Cerati M, Rohde K, Larizza L, Fuhrman Conti AM: A tumor suppressor locus Expression Data via Mixed Models. Journal of Computational Biology 2002, in familial and sporadic chordoma maps to 1p36. Int J Cancer 2000, 8:625-637. 87:68-72. 26. Benjamini Y, Hochberg Y: Controlling the False Discovery Rate: a Practical 7. Hallor KH, Staaf J, Jönsson G, Heidenblad M, Vult von Steyern F, Bauer HC, and Powerful Approach to Multiple Testing. Journal of the RSS: Series B Ijszenga M, Hogendoorn PC, Mandahl N, Szuhai K, Mertens F: Frequent (Statistical Methodology) 2010, 57:289-300. deletion of the CDKN2A locus in chordoma: analysis of chromosomal 27. Brandal P, Bjerkehagen B, Danielsen H, Heim S: Chromosome 7 imbalances using array comparative genomic hybridisation. Br J Cancer abnormalities are common in chordomas. Cancer Genet Cytogenet 2005, 2008, 98:434-442. 160:15-21. 8. Tamborini E, Virdis E, Negri T, Orsenigo M, Brich S, Conca E, Gronchi A, 28. Kuzniacka A, Mertens F, Strombeck B, Wiegant J, Mandahl N: Combined Stacchiotti S, Manenti G, Casali PG, Pierotti MA, Pilotti S: Analysis of binary ratio labeling fluorescence in situ hybridization analysis of receptor tyrosine kinases (RTKs) and downstream pathways in chordoma. Cancer Genet Cytogenet 2004, 151:178-181. chordoma. Neuro Oncol 2010, 12:776-89. 29. Sawyer JR, Husain M, Al-Mefty O: Identification of isochromosome 1q as a 9. Hof H, Welzel T, Debus J: Effectiveness of cetuximab/gefitinib in the recurring chromosome aberration in skull base chordomas: a new therapy of a sacral chordoma. Onkologie 2006, 29:572-574. marker for aggressive tumors? Neurosurg Focus 2001, 10:E6. 10. Naka T, Iwamoto Y, Shinohara N, Ushijima M, Chuman H, Tsuneyoshi M: 30. Li J, Rix U, Fang B, Bai Y, Edwards A, Colinge J, Bennett KL, Gao J, Song L, Expression of c-met proto-oncogene product (c-MET) in benign and Eschrich S, Superti-Furga G, Koomen J, Haura EB: A chemical and malignant bone tumors. Mod Pathol 1997, 10:832-838. phosphoproteomic characterization of dasatinib action in lung cancer. 11. Tamborini E, Miselli F, Negri T, Lagonigro MS, Staurengo S, Dagrada GP, Nat Chem Biol 2010, 6:291-299. Stacchiotti S, Pastore E, Gronchi A, Perrone F, Carbone A, Pierotti MA, 31. Olson LE, Soriano P: Increased PDGFRalpha activation disrupts connective Casali PG, Pilotti S: Molecular and biochemical analyses of platelet- tissue development and drives systemic fibrosis. Dev Cell 2009, derived growth factor receptor (PDGFR) B, PDGFRA, and KIT receptors in 16:303-313. chordomas. Clin Cancer Res 2006, 12:6920-6928. 32. Naka T, Boltze C, Kuester D, Schulz TO, Schneider-Stock R, Kellner A, 12. Weinberger PM, Yu Z, Kowalski D Joe J, Manger P, Psyrri A, Sasaki CT: Samii A, Herold C, Ostertag H, Roessner A: Alterations of G1-S checkpoint Differential expression of epidermal growth factor receptor, c-Met, and in chordoma: the prognostic impact of p53 overexpression. Cancer 2005, HER2/neu in chordoma compared with 17 other malignancies. Arch 104:1255-1263. Otolaryngol Head Neck Surg 2005, 131:707-711. 33. Genander M, Halford MM, Xu NJ, Eriksson M, Yu Z, Qiu Z, Martling A, 13. Casali PG, Messina A, Stacchiotti S, Tamborini E, Crippa F, Gronchi A, Greicius G, Thakar S, Catchpole T, Chumley MJ, Zdunek S, Wang C, Holm T, Orlandi R, Ripamonti C, Spreafico C, Bertieri R, Bertulli R, Colecchia M, Goff SP, Pettersson S, Pestell RG, Henkemeyer M, Frisén J: Dissociation of Fumagalli E, Greco A, Grosso F, Olmi P, Pierotti MA, Pilotti S: Imatinib EphB2 signaling pathways mediating progenitor cell proliferation and mesylate in chordoma. Cancer 2004, 101:2086-2097. tumor suppression. Cell 2009, 139:679-692. 14. Deniz ML, Kilic T, Almaata I, Kurtkaya O, Sav A, Pamir MN: Expression of 34. Wagh PK, Peace BE, Waltz SE: Met-related receptor tyrosine kinase Ron in growth factors and structural proteins in chordomas: basic fibroblast tumor growth and metastasis. Adv Cancer Res 2008, 100:1-33. growth factor, transforming growth factor alpha, and fibronectin are 35. Eder JP, Shapiro GI, Appleman L, Zhu AX, Miles D, Keer H, Cancilla B, Chu F, correlated with recurrence. Neurosurgery 2002, 51:753-760. Hitchcock-Bryan S, Sherman L, McCallum S, Heath EI, Boerner SA, 15. Fasig JH, Dupont WD, LaFleur BJ, Olson SJ, Cates JM: LoRusso PM: A phase I study of foretinib, a multi-targeted inhibitor of c- Immunohistochemical analysis of receptor tyrosine kinase signal Met and vascular endothelial growth factor receptor 2. Clin Cancer Res transduction activity in chordoma. Neuropathol Appl Neurobiol 2008, 2010, 16:3507-3516. 34:95-104. 36. Stommel JM, Kimmelman AC, Ying H, Nabioullin R, Ponugoti AH, 16. Shalaby A, Presneau N, Ye H, Halai D, Berisha F, Idowu B, Leithner A, Liegl B, Wiedemeyer R, Stegh AH, Bradner JE, Ligon KL, Brennan C, Chin L, Briggs TRW, Bacsi K, Kindblom LG, Athanasou N, Amary MF, DePinho RA: Coactivation of receptor tyrosine kinases affects the Hogendoorn PCW, Tirabosco R, Flanagan AM: The role of epidermal response of tumor cells to targeted therapies. Science 2007, 318:287-290. growth factor receptor in chordoma pathogenesis: a potential 37. Zhou Y, Li S, Hu YP, Wang J, Hauser J, Conway AN, Vinci MA, Humphrey L, therapeutic target. J Pathol 2011, 223:336-346. Zborowska E, Willson JK, Brattain MG: Blockade of EGFR and ErbB2 by the 17. Brand-Saberi B, Christ B: Evolution and development of distinct cell novel dual EGFR and ErbB2 tyrosine kinase inhibitor GW572016 lineages derived from somites. Curr Top Dev Biol 2000, 48:1-42. sensitizes human colon carcinoma GEO cells to apoptosis. Cancer Res 18. Ostroumov E, Hunter CJ: Identifying mechanisms for therapeutic 2006, 66:404-411. intervention in chordoma: c-Met oncoprotein. Spine (Phila Pa 1976) 2008, 38. Dewaele BM, Floris G, Finalet-Ferreiro J, Fletcher CD, Coindre JM, Guillou L, 33:2774-2780. Hogendoorn PC, Wozniak A, Vanspauwen V, Schöffski P, Marynen P, 19. Wozniak A, Sciot R, Guillou L, Pauwels P, Wasag B, Stul M, Vermeesch JR, Vandenberghe P, Sciot R, Debiec-Rychter M: Co-activated PDGFRA and Vandenberghe P, Limon J, Debiec-Rychter M: Array CGH analysis in EGFR are potential therapeutic targets in intimal sarcoma. Cancer Res primary gastrointestinal stromal tumors: cytogenetic profile correlates 2010, 70:7304-14. with anatomic site and tumor aggressiveness, irrespective of mutational 39. Atkins D, Reiffen KA, Tegtmeier CL, Winther H, Bonato MS, Storkel S: status. Genes Chromosomes Cancer 2007, 46:261-276. Immunohistochemical detection of EGFR in paraffin-embedded tumor 20. Lahortiga I, De Keersmaecker K, Van Vlierberghe P, Graux C, Cauwelier B, tissues: variation in staining intensity due to choice of fixative and Lambert F, Mentens N, Beverloo HB, Pieters R, Speleman F, Odero MD, storage time of tissue sections. J Histochem Cytochem 2004, 52:893-901. Bauters M, Froyen G, Marynen P, Vandenberghe P, Wlodarska I, Meijerink JP, 40. Chung KY, Shia J, Kemeny NE, Shah M, Schwartz GK, Tse A, Hamilton A, Cools J: Duplication of the MYB oncogene in T cell acute lymphoblastic Pan D, Schrag D, Schwartz L, Klimstra DS, Fridman D, Kelsen DP, Saltz LB: leukemia. Nat Genet 2007, 39:593-595. Cetuximab shows activity in colorectal cancer patients with tumors that 21. VU Micro-Array Data Analysis. [http://www.few.vu.nl/~vumarray/]. do not express the epidermal growth factor receptor by 22. Debiec-Rychter M, Wasag B, Stul M, De Wever I, Van Oosterom A, immunohistochemistry. J Clin Oncol 2005, 23:1803-1810. Hagemeijer A, Sciot R: Gastrointestinal stromal tumours (GISTs) negative 41. Moroni M, Veronese S, Benvenuti S, Marrapese G, Sartore-Bianchi A, Di for KIT (CD117 antigen) immunoreactivity. J Pathol 2004, 202:430-438. Nicolantonio F, Gambacorta M, Siena S, Bardelli A: Gene copy number for 23. Primer3: WWW primer tool. [http://frodo.wi.mit.edu/cgi-bin/primer3/ epidermal growth factor receptor (EGFR) and clinical response to primer3_www_slow.cgi]. antiEGFR treatment in colorectal cancer: a cohort study. Lancet Oncol 2005, 6:279-286. Dewaele et al. Clinical Sarcoma Research 2011, 1:4 Page 16 of 16 http://www.clinicalsarcomaresearch.com/content/1/1/4 42. Presneau N, Shalaby A, Idowu B, Gikas P, Cannon SR, Gout I, Diss T, Tirabosco R, Flanagan AM: Potential therapeutic targets for chordoma: PI3K/AKT/TSC1/TSC2/mTOR pathway. Br J Cancer 2009, 100:1406-1414. 43. Schwab J, Antonescu C, Boland P, Healey J, Rosenberg A, Nielsen P, Iafrate J, Delaney T, Yoon S, Choy E, Harmon D, Raskin K, Yang C, Mankin H, Springfield D, Hornicek F, Duan Z: Combination of PI3K/mTOR inhibition demonstrates efficacy in human chordoma. Anticancer Res 2009, 29:1867-1871. 44. Dobashi Y, Suzuki S, Sato E, Hamada Y, Yanagawa T, Ooi A: EGFR- dependent and independent activation of Akt/mTOR cascade in bone and soft tissue tumors. Mod Pathol 2009, 22:1328-1340. 45. Chiacchiera F, Simone C: Signal-dependent regulation of gene expression as a target for cancer treatment: inhibiting p38alpha in colorectal tumors. Cancer Lett 2008, 265:16-26. 46. Han S, Polizzano C, Nielsen GP, Hornicek FJ, Rosenberg AE, Ramesh V: Aberrant hyperactivation of akt and Mammalian target of rapamycin complex 1 signaling in sporadic chordomas. Clin Cancer Res 2009, 15:1940-1946. 47. Xu AM, Huang PH: Receptor tyrosine kinase coactivation networks in cancer. Cancer Res 2010, 70:3857-3860. doi:10.1186/2045-3329-1-4 Cite this article as: Dewaele et al.: Frequent activation of EGFR in advanced chordomas. Clinical Sarcoma Research 2011 1:4. Submit your next manuscript to BioMed Central and take full advantage of: • Convenient online submission • Thorough peer review • No space constraints or color figure charges • Immediate publication on acceptance • Inclusion in PubMed, CAS, Scopus and Google Scholar • Research which is freely available for redistribution Submit your manuscript at www.biomedcentral.com/submit http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Clinical Sarcoma Research Springer Journals

Frequent activation of EGFR in advanced chordomas

Download PDF
16 pages

Loading next page...
 
/lp/springer-journals/frequent-activation-of-egfr-in-advanced-chordomas-wu0WrrQb14

References (100)

Publisher
Springer Journals
Copyright
Copyright © 2011 by Dewaele et al; licensee BioMed Central Ltd.
Subject
Biomedicine; Cancer Research; Oncology; Surgical Oncology
eISSN
2045-3329
DOI
10.1186/2045-3329-1-4
pmid
22613809
Publisher site
See Article on Publisher Site

Abstract

Background: Chordomas are rare neoplasms, arising from notochordal remnants in the midline skeletal axis, for which the current treatment is limited to surgery and radiotherapy. Recent reports suggest that receptor tyrosine kinases (RTK) might be essential for the survival or proliferation of chordoma cells, providing a rationale for RTK targeted therapy. Nevertheless, the reported data are conflicting, most likely due to the assorted tumor specimens used for the studies and the heterogeneous methodological approaches. In the present study, we performed a comprehensive characterization of this rare entity using a wide range of assays in search for relevant therapeutic targets. Methods: Histopathological features of 42 chordoma specimens, 21 primary and 21 advanced, were assessed by immunohistochemistry and fluorescent in situ hybridization (FISH) using PDGFRB, CSF1R, and EGFR probes. Twenty- two of these cases, for which frozen material was available (nine primary and 13 advanced tumors), were selectively analyzed using the whole-genome 4.3 K TK-CGH-array, phospho-kinase antibody array or Western immunoblotting. The study was supplemented by direct sequencing of KIT, PDGFRB, CSF1R and EGFR. Results: We demonstrated that EGFR is frequently and the most significantly activated RTK in chordomas. Furthermore, concurrent to EGFR activation, the tumors commonly reveal co-activation of alternative RTK. The consistent activation of AKT, the frequent loss of the tumor suppressor PTEN allele, the recurrent activation of upstream RTK and of downstream effectors like p70S6K and mTOR, all indicate the PI3K/AKT pathway as an important mediator of transformation in chordomas. Conclusions: Given the complexity of the signaling in chordomas, combined treatment regimens targeting multiple RTK and downstream effectors are likely to be the most effective in these tumors. Personalized therapy with careful selection of the patients, based on the molecular profile of the specific tumor, is anticipated. Background soft tissue. Initial symptoms usually relate to local pro- Chordomas are rare tumors. With an incidence of about gression of the disease. Chordomas infrequently metas- 0.05/100000/year, they account for less than 5% of all tasize to lung, bone, soft tissue, lymph nodes and skin. primary malignant bone tumors. Mainly adults between On histology at low power magnification they show pro- 40 and 60 years are affected, but cases of children pre- minent lobules separated by fibrous septa. The tumors senting with chordoma were also rarely reported (5% of maybearranged in chords or sheets or maybefloating cases). These bone tumors arise from remnants of the singularly in the abundant myxoid matrix often present. fetal notochord, and hence occur along the midline, and The current treatment for chordoma is predominantly most often in the caudal spine or the base of the skull. surgery, followed by radiotherapy. Safe margins are They are slowly growing masses with the tendency to often difficult to obtain because of the anatomical loca- destroy the surrounding bone and to infiltrate adjacent tion of the tumors [1]. Unfortunately, standard che- motherapy was shown to be basically unsuccessful, * Correspondence: barbara.dewaele@uzleuven.be which causes serious problems for managing patients Department of Human Genetics, Catholic University of Leuven, University with locally recurrent or metastatic disease. Survival Hospitals, Leuven, Belgium Full list of author information is available at the end of the article © 2011 Dewaele et al; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Dewaele et al. Clinical Sarcoma Research 2011, 1:4 Page 2 of 16 http://www.clinicalsarcomaresearch.com/content/1/1/4 rates of 5 and 10 years are 68% and 40%, respectively addition, expression of the MET oncogene has been [2]. reported in chordoma [10]. Of note, the MET oncogene Cytogenetic studies in chordomas have revealed in is known to be expressed in various chondroid neo- general nearly diploid or rather hypodiploid karyotypes, plasms, normal articular cartilage and fetal notochord with a number of numerical and structural rearrange- [17,18]. Given their possible relationship to notochordal ments. Recurrent genetic events reported in chordoma development and chondroid differentiation, further include frequent losses of large parts of chromosomes 3, investigation is warranted to clarify the roles of these 4, 10 and 13 and the most commonly lost regions are and other RTK in chordomagenesis. 1p31-pter, 3p21-pter, 3q21-qter, 9p24-pter and 17q11- Activities of effectors more downstream in the main qter [3]. The most common gains affect the chromo- RTK pathways were also recently described. The ERK1/2, some 5q and the entire chromosomes 7 and 20 [4,5]. AKT and STAT3 activity was demonstrated in 18 (86%), Loss of heterozygosity at 1p36 was also found in familial 16 (76%) and 14 (67%) of cases, respectively, by immuno- chordomas, further supporting the hypothesis that an histochemistry performed on 21 chordomas [15]. important tumor suppressor might be located at the dis- Furthermore, analysis of 22 chordomas by Tamborini tal part of 1p [6]. Importantly, the CDKN2A tumor sup- and co-workers showed consistent ERK1/2 activation in pressor gene, which maps to 9p21.3, is reported to be all the cases, and activation of AKT in 20 (91%), mTOR lost in a high percentage (60%) of chordomas [7,8]. In in 18 (82%), and S6 in 16 (73%) of the tumors [8]. addition, loss of one copy of the PTEN tumor suppres- In the present study, we have performed a compre- sor gene (located on 10q23.31) was found in 37% (7/19) hensive molecular and biochemical analysis of 42 chor- of lesions, although no difference in PTEN expression domas, focusing on the role of RTK and their level was shown by Western blotting [8]. downstream signaling pathway in chordoma develop- In the literature, several RTK, specifically PDGFRA, ment, in primary tumors or their recurrent/metastatic PDGFRB, KIT, EGFR, MET and HER2, were reported to counterparts. be expressed in chordoma by immunohistochemistry [9-12]. Given that RTK could prove to be essential for Methods the survival or proliferation of chordoma tumor cells, Patients and histopathology targeting these RTK using antibodies or small molecule The present study included 31 patients [16 women and tyrosine kinase inhibitors (TKI) might offer new treat- 15 men; age range 18-84 (median 58 years)] (Table 1). ment options for chordoma patients. Interestingly, ima- In total, 42 tumor specimens from these patients were tinib was found to have antitumor activity in patients retrieved, of which 21 were annotated as primary with chordoma [13]. It was suggested that PDGFRB sig- tumors and 21 as recurrences or metastases (in the text naling might be implicated in the tumor growth, as ima- further referred to as advanced cases). The primary tinib-responding tumors were found to be chordomas originated from the spine (n = 9), the immunohistochemically positive for PDGFRB. Expres- sacrum (n = 10), the clivus (n = 1), and the cervix (n = sion of basic fibroblast growth factor (bFGF), transform- 1). Samples 10a and 10b represent primary samples ing growth factor alpha (TGF alpha) and fibronectin was from thesamepatient obtainedbyneedle biopsyand reported to correlate with an increased incidence of dis- subsequent surgical resection, respectively. Histopatho- ease recurrence in chordoma [14]. Moreover, clinical logical examination was performed on formalin fixed, response to imatinib in one case was accompanied by paraffin embedded tissue. Five μmsectionswereused the inhibition of PDGFRB as demonstrated by Western for routine hematoxylin and eosin (H&E) staining, and blot [13]. In recent reports, Tamborini and co-workers immunohistochemical staining was performed by the characterized 22 chordomas by immunoprecipitation avidin-biotin-peroxidase complex method, using the fol- and antibody arrays. The activation of PDGFRA, lowing monoclonal (mc) and polyclonal (pc) antibodies: PDGFRB, KIT, FLT3, CSF1R, EGFR, HER2, HER4, AXL Pankeratin (mc, dilution 1:200; Serotec, Oxford, UK), and DTK was reported in these studies [8,11]. Notably, Epithelial Membrane Antigen (EMA) (mc, 1:50; DAKO, PDGFRB activation was found in 95% (21/22) of cases. Glostrup, Denmark), Multikeratin (mc, dilution 1:10; The EGFR activation, mainly through EGFR/HER2 het- Novocastra, Newcastle Upon Tyne, UK), S-100 protein erodimer formation, was also suggested. Other groups (pc, 1:300; DAKO) and Vimentin (mc, dilution 1:500, found EGFR activation in three out of three and in DAKO). In addition, the EGFR (EGFR PharmDxTM, about 50% of chordomas evaluated by RTK antibody DAKO) and HER2/ERBB2 (HercepTestTM, DAKO) arrays and immunohistochemistry respectively [15,16]. staining kits were used. EGFR and HER2 protein expres- Partial response of metastatic chordoma to combined sion was reported as membranous brown staining of cetuximab/gefitinib treatment suggests that EGFR tar- neoplastic cells using a three-tier system ranging from geted treatment may benefit chordoma patients [9]. In “1+” to “3+”. Dewaele et al. Clinical Sarcoma Research 2011, 1:4 Page 3 of 16 http://www.clinicalsarcomaresearch.com/content/1/1/4 Table 1 Pathologic description of chordoma cases and results summary Cases Gender Age Tumor Site Immuno FISH Proteome TK Western TK status profiler array Mut aCGH EGFR HER2 EGFR HER2 PDGFRB/ PTEN P- P- EGFR PDGFRB CSF1R EGFR PDGFRB 1a F 56 P Spinal 1+ neg dis. dis. polys. nd nd nd nd nd nd nd 1b R Spinal neg nd dis. dis. polys. nd weak weak nd nd nd nd 2a M 33 R Spinal 1+ nd dis. dis. polys. nd nd nd nd nd nd nd 2b M 3+ nd dis. dis. polys. nd nd nd neg E E nd 3 F 43 R Spinal 3+ neg polys. dis. dis. dis. strong weak neg nd nd table 4a M 62 P Spinal neg nd dis. dis. dis. dis. nd nd nd nd nd nd 4b M 2+ nd polys. dis. polys. monos. interm weak neg E neg table 5 F 75 R Sacrum 3+ neg polys. polys. polys. polys. nd nd neg E/P E table 6 M 60 R Clivus 3+ nd dis. monos. dis. dis. strong interm. nd nd nd nd 7a F 62 P Sacrum neg nd polys. dis. dis. dis. nd nd nd nd nd nd 7b R Sacrum 2+ nd polys. monos. dis. monos. nd nd neg E/P E table 8 M 36 P Clivus neg nd dis. dis. dis. nd nd nd nd nd nd nd 9 M 52 R Coccyx 1+ 1+ monos. dis. dis. dis. nd nd neg E/P E table 10a F 41 P Spinal neg nd polys. dis. loss dis. nd nd nd nd nd nd 10b P Spinal neg nd polys. dis. loss monos. nd nd neg neg E/P table 10c R Spinal neg nd polys. dis. polys. nd nd nd nd nd nd nd 11 F 54 P Cervical 2+ nd polys. dis. dis. nd strong weak nd nd nd nd 12a M 55 P Sacrum 3+ neg l.l.amp. dis. dis. monos. nd nd neg nd nd nd 12b M 3+ 2+ h.l. dis. polys. nd strong weak nd E/P E nd amp. 13 M 80 R Coccyx 2+ nd polys. polys. polys. polys. interm weak neg nd nd table 14 F 60 R Sacrum 1+ nd dis. polys. dis. monos. nd nd neg E/P E table 15a F 73 P Spinal neg nd dis. dis. dis. monos. nd nd neg nd nd table 15b R Spinal neg nd polys. polys. polys. nd nd nd nd E E nd 16 M 84 R Sacrum 1+ neg dis. dis. dis. dis. nd nd neg nd nd table 17a F 58 P Sacrum 2+ 1+ polys. dis. dis. nd strong interm. neg nd nd table 17b R Sacrum 3+ neg polys. dis. dis. nd nd nd nd nd nd nd 18 F 57 P Sacrum 3+ neg polys. dis. dis. dis. strong weak nd nd nd nd 19 M 84 P Lumbal 1+ nd dis. monos. dis. nd nd nd nd nd nd nd 20 M 81 P Sacrum 3+ neg l.l.amp. dis. polys. nd strong weak nd nd nd nd 21 F 67 P Sacrum 1+ 1+ h.l. dis. dis. nd interm weak nd nd nd nd amp. 22 F 47 P Sacrum 1+ nd dis. dis. dis. nd weak weak nd nd nd nd 23 M 48 P Spinal nd nd dis. dis. dis. nd nd nd nd nd nd nd 24 F 60 R Clivus/ 3+ nd polys. polys. polys. polys. nd nd nd nd nd nd nc 25 F 60 R Sacrum neg nd monos. dis. nd nd nd nd nd nd nd nd 26 M 80 M nd nd polys. polys. polys. loss nd nd nd nd nd nd 27 M 48 P Sacrum 1+ nd dis. dis. nd nd nd nd nd nd nd nd 28 F 18 P Spinal nd nd dis. dis. nd nd nd nd nd nd nd nd 29 M 37 P Spinal 3+ neg dis. dis. nd nd nd nd nd nd nd nd Dewaele et al. Clinical Sarcoma Research 2011, 1:4 Page 4 of 16 http://www.clinicalsarcomaresearch.com/content/1/1/4 Table 1 Pathologic description of chordoma cases and results summary (Continued) 30a M 42 P Coccyx neg neg monos. dis. nd nd nd nd nd nd nd nd 30b 58 R Sacrum 2+ neg monos. dis. nd nd nd nd nd nd nd nd 31a F 60 P Sacrum neg 1+ polys. dis. nd nd nd nd nd nd nd nd 31b R Ilium neg nd polys. dis. nd nd nd nd nd nd nd nd TK Mut., Tyrosine Kinase Mutations; P, primary; R, recurrence; M, metastasis; nc, nasal cavity; neg, negative; l.l.amp., low level amplification; h.l.amp., high level amplification; polys., polysomy; dis., disomy; monos., monosomy; nd, not done; interm, intermediate; E, expressed; P, phosphorylated. Array-CGH (aCGH) analysis Fluorescence In Situ Hybridization (FISH) Array-CGH experiments were performed as previously Dual-color interphase FISH analysis was carried out on described on DNA extracted from 11 tumors (Table 2) 4 μm paraffin embedded tissue sections of 42 tumor [19]. For genomic profiling that included the evaluation biopsies. Sections were pretreated using the SPoT-Light of all 90 TK known in humans, the 4.3 K genomic DNA Tissue Pre-treatment Kit (Invitrogen, Life Technologies), tyrosine kinase array (TK-aCGH) was manufactured at according to the instructions of the manufacturer. FISH the Microarray Facility of the Flanders Interuniversity was performed as previously described [22]. Slides were Institute for Biotechnology, KULeuven [20]. In short, counterstained with 0.1 μM 4,6-diamidino-2-phenylin- the Sanger 1 Mb Clone Set containing 3527 BAC/PAC dole (DAPI) in an antifade solution for microscopy. clones was supplemented with 800 clones from 32 K For analysis of EGFR family members, FISH was per- CHORI BAC/PAC library, which specifically covers all formed using the locus specific identifier (LSI) EGFR- known human TK, and these two clone sets were SpectrumOrange(SO)/CEP7-SpectrumGreen(SG) and spotted together in duplicate on Code Linked Slides (AP PathVysion HER2-SO/CEP17-SG probes (Applied Bio- Biotech, US). The complete list of these clones is avail- systems/Ambion, Life Technologies, Carlsbad, CA, able upon request. The array-CGH data were statistically USA). For evaluation of PDGFRB/CSF1R copy numbers analyzed with aCGH-smooth, software especially and PDGFRB/CSF1R integrity, the SG-labeled bacterial designed for the analysis of heterogeneous samples [21]. artificial chromosome (BAC) RP11-21I20 (which maps Table 2 Gains and losses in chordoma using whole-genome 4.3 K TK-CGH-array Case 3 Case 4b Case 5 Case 7b Case 9 Case 10b Case 13 Case 14 Case 15a Case 16 Case 17a Gains 1q11-qter 5 2pter-p12 16q12.2- n.d. 7 13q31.2- 1q11-qter n.d. n.d. n.d. 7 7 q22.1 qter 2 8q11.21- 12pter- qter q24.23 10pter-p11 17q12.1- 20 qter Losses 3pter-p11.1 1pter-p11 1pter-p11.2 1pter-p13.1 2q21.1-qter 1 3pter- 3 1pter-p32.3 1pter- n.d. 8pter-p12 3 3p24.1-p13 3 3q11.2-q28 3pter-p12.1 p14.2 9pter-p11 1p22.3- p33.2 9 4 3q11.2- 4pter-p16.1 5q35.2-qter 9pter-p21 9 10 p21.3 3pter- 14 10 q13.31 6p22.3-p21.1 7pter-p22.1 10 14 14 1p21.2- p11.2 16q23.2- 11pter- 3q26.1- 9 8pter- 19p13.3- 17pter-p12 p13.2 22q12.1- q24.3 11p11 26.31 10 p11.21 p13.2 19p13.3- 2pter-p11.2 qter 13 3q28-qter 13q12.11- 11q12.2- 22q12.2- p13.2 2q31.2-qter 14 4p15.31- q12.13 q13.3 qter 6pter-p21.1 18 q21.21 16q12.1- 16pter- X 9 22 5pter-p15.2 q12.2 p12.1 10q11.23- Y 9pter-p21.1 16q22.3- 17pter- q24.2 9q34.11- q24.3 p11.2 18q11.2-q23 qter 17q12- 18q11.2- 19 11q12.2- q21.33 qter 21 q13.3 19p13.3- 20q11.21- 22 13q21.3- p13.11 qter q21.33 19q13.31- 22 13q33.1- qter qter n.d.: not detected. Dewaele et al. Clinical Sarcoma Research 2011, 1:4 Page 5 of 16 http://www.clinicalsarcomaresearch.com/content/1/1/4 centromeric to PDGFRB/5q33.1 and covers the adjacent described [22]. In short, tumor lysate aliquots containing CSF1R gene) and the SO-labeled RP11-368O19 (which 30 μg of protein were electrophoresed and blotted to maps telomeric to CSF1R and covers the PDGFRB gene) PVDF membranes (GE Healthcare, UK). Membranes DNA probes (both from Research Genetics, Huntsville, were blocked in PBS containing 5% blocking reagent AL, USA) were used. In addition, the PTEN copy num- (non-fat milk) and immunoblotted sequentially using bers were investigated using dual-color LSI PTEN/ rabbit antibodies against phospho-EGFR(Tyr1068) CEP10 probe (Applied Biosystems/Ambion). (Santa Cruz Biotechnology, Santa Cruz, CA, USA), total Hybridization signals were visualized using an epi- EGFR (Santa Cruz Biotechnology), phospho-PDGFRB fluorescence microscope (Leica DMRB, Wetzlar, Ger- (Y751) (mc; Cell Signaling, Beverly, MA, USA), total many) equipped with a cooled CCD camera and run by PDGFRB (mc; Cell Signaling), phospho-KIT(Tyr703) the ISIS digital image analysis system (MetaSystems, (mc; Invitrogen, Life Technologies), total KIT (pc; Altlussheim, Germany). One hundred nuclei were evalu- DAKO), phospho-ERK1/2 (Cell Signaling), total ERK1/2 ated for the number of red and green signals in different (Cell Signaling), phospho-AKT (Cell Signaling) and total areas corresponding to tumor tissues. AKT (Cell Signaling), diluted in 5% blocking reagent. FISH results were classified into five categories Total b-actin (Sigma Aldrich, St. Louis, MO, USA) was according to the percentage of tumor cells with a speci- used as a protein-loading and transfer control. The fic gene/CEP ratio and according to the gene copy num- HRP-conjugated anti-rabbit IgG (DAKO) were used at a ber per nucleus: 1) monosomy (1 signal from the gene dilution of 1:2000, and visualized with Enhanced Chemi- paralleled by one chromosome centromere signal) or luminescence (Thermo Scientific, Rockford, IL, USA). loss (a gene/CEP ratio of <0.6) in >40% of cells; 2) dis- omy (2 signals from the gene/CEP probes); 3) polysomy Receptor tyrosine kinases (RTK) activation profiling using (defined as > 2 gene signals per nucleus paralleled by antibody arrays similar increases in chromosome centromeric signals in The activation of RTK and their downstream signaling at least 10% of tumor cells); 4) low level gene amplifica- pathways were analyzed using the Proteome Profiler™ tion (gene/CEP ratio of > 2 in 10%-40% of tumor cells) Array kits (ARY001 and ARY003, R&D Systems, Min- or 5) high level gene amplification (presence of gene neapolis, MN, USA) in 12 fresh frozen chordoma tumor clusters or a gene/CEP ratio of > 2 in ≥40% of analyzed specimens. Assays were performed according to the cells). manufacturers’ protocol, and using 500 μgofprotein lysate per array. The images were captured and the level Mutation analysis of RTK activation was visualized with the FUJI mini- Mutational analysis was performed on genomic DNA LAS3000-plus imaging system (FUJIFILM, Tokyo, Japan) extracted from frozen tumor tissues (n = 13). The and densitometrically quantified with AIDA software sequence coding for the juxtamembrane and/or kinase (Raytest isotopenmessgeräte GmbH, Straubenhardt, Ger- domains of PDGFRA and PDGFRB (exons 12, 14 and many). The signal intensities of the probes and the local 18), KIT (exons 9, 11 and 17), CSF1R (exons 10 to 20) background of the probes were log transformed in and EGFR (exons 18 to 21) genes, were amplified by order to obtain a more symmetric distribution, and the polymerase chain reaction (PCR), using standard Taq differencebetween thesetwo resulted in alog trans- DNA polymerase (Roche Diagnostics, Basel, Switzerland) formed ratio (further referred to as log -intensity ratios). and the ABI PRISM 9700 (Applied Biosystems). Geno- For data normalization, within an array and within a mic sequences were obtained from online databases membrane the mean log -intensity ratio was calculated from the National Center for Biotechnology Information and then subtracted from the log -intensity ratio of each (NCBI), and specific primers for amplified fragments probe. Subsequently, the mean of the log -intensity were designed using the Primer3 software [23] (http:// ratios for eachkinasewithinanarray wascalculated. In frodo.wi.mit.edu/cgi-bin/primer3/primer3_www_slow. the statistical analysis, a linear mixed model was used cgi). Primers sequences are available upon request. The instead of a one-sample t-test per probe since the arrays PCR products were purified (QIAquick PCR Purification or membranes used to measure the probe intensities Kit, QIAGEN, Hilden, Germany) followed by direct bi- maydiffer. Thelinear mixedmodel hasthe log -inten- directional cycle sequencing using the ABI PRISM 3130 sity ratios as responses, the probes as fixed effects and XL Genetic Analyzer (Applied Biosystems, Foster City, the membrane as random effect per array [24,25]. The CA, USA). alpha level was set at 5%. As multiple testing correc- tions, the p-values from the tests for the different probes Western immunoblotting were adjusted to control the false discovery rate as Cell lysisoffrozentumors(n=9),SDS-PAGE, and described by Benjamini and Hochberg [26]. The ranking immunoblotting were carried out as previously of the probes was based on the adjusted p-values. All Dewaele et al. Clinical Sarcoma Research 2011, 1:4 Page 6 of 16 http://www.clinicalsarcomaresearch.com/content/1/1/4 analyses were performed with the statistical package tumors analyzed showed CNA by aCGH. CNA frequen- SAS (version 9.2), using the procedure PROC MIXED cies were calculated on these ten cases with CNA. for the linear mixed model. Losses were more common than gains, supporting pre- viousfindingsinchordoma[7].There wasamedian of Results one gain (range 0-5) and seven losses (range 0-14) per Histopathology and immunohistochemistry tumor. Genomic losses affecting five or more tumors (≥ All the chordomas in our cohort were reviewed and 50% of cases) were identified on chromosomes 1, 3, 9, classified as conventional chordomas by means of mor- 10, 19 and22(Table 2 and3,Figure2). Thesmallest phology and immunohistochemistry (IHC). They show common region of chromosome 3 deletion, covering prominent lobules separated by fibrous septa. The bands 3p24.1-p14.2, was lost in eight cases. Three tumor cells are arranged in cords or sheets or may be regions located on the short arm of chromosome 1, i.e. floating singularly in the abundant myxoid matrix often 1pter-p33.2, 1p22.3-p21.3 and 1p21.2-p13.2, were recur- present. The histologic hallmark is characterized by rently lost in six, five and five cases, respectively. Whole large tumor cells with abundant vacuolated cytoplasm, chromosome 9 loss was observed in four cases, and the referred to as physaliphorous cells [2]. The tumor cells region 9q34.11-qter, involving among others the TSC1 co-express keratin, EMA and S-100 protein. Of the 39 tumor suppressor gene, was lost in one additional case. chordomas tested by IHC for EGFR expression, 19 were Furthermore, the region 9pter-p21 was lost in three primary and 20 were advanced lesions. The EGFR extra cases of our cohort. Of note, homozygous deletion immunopositivity was found in 26 out of 39 cases of the chromosomal sub-band 9p21.3 (the region con- (67%), showing different levels of reactivity (Figure 1, taining the CDKN2A tumor suppressor gene) was found Table 1). Thus, 11 tumors presented with an intense in three of analyzes tumors. The entire chromosome 10 and diffuse cytoplasm membrane positivity in more than was lost in four cases and the region 10q11.23-q24.2, 10% of the cells (scored as “3+”), six cases showed encompassing the tumor suppressor PTEN, was lost in intense positive staining but in less than 10% of the cells another case. Losses that implicated chromosome 19, (scored as “2+”), and nine other cases were considered with the commonly deleted region 19p13.3-p13.2, were weakly and discontinuously stained in more than 10% of found in five cases. Total or partial chromosome 22 the cells (scored as “1+”). EGFR expression was more deletions, with the common region 22q12.2-qter, were frequently found in advanced tumors compared with recorded in six chordomas. The most common gain was primary tumors (80% versus 58%, respectively). In detail: the gain of the entire chromosome 7, observed in three 15 out of 20 advanced cases stained positive for EGFR chordoma cases (Table 2). Notably, the genes coding for versus 11 out of 19 primary cases. Additionally, when the EGFR, MET, LMTK2, EPHA1, EPHB4 and EPHB6 comparing the primary and the advanced stage within proteins are mapped on chromosome 7. No amplifica- patients, in cases 2, 4, 7, 17 and 30: stronger EGFR tions or rearrangements within the 90 known TK were staining was observed in the advanced in comparison detected in our cohort of chordomas. with the primary stage. Case 12 showed intense and dif- fuse (3+) staining in both the primary and the advanced FISH analysis stage. Case 1 was the only exception, showing stronger Thegenecopynumbers of the EGFR, HER2, CSF1R/ EGFR staining in the primary than in the advanced PDGFRB and PTEN were analyzed by FISH (Figure 3, stage. Cases 10, 15 and 31 stained negative for EGFR in Table 1). Sixteen out of 42 tumors analyzed revealed the primary stage and stayed negative upon progression. disomy for EGFR, while 16 (38%) cases displayed polyso- HER2 expression was tested in 16 cases, of which 11 mic cell clones. Two cases showed chromosome 7 polys- were negative, four displayed low level of staining inten- omy. Only a small fraction of tumors (four cases) sity and one case showed intense positive staining, albeit presented with EGFR amplification, and only in two in less than 10% of the cells. HER2 expression was cases at high level. Notably, four cases showed EGFR almost as frequent in primary as in advanced tumors loss. The gene copy number of HER2 was also analyzed (33% versus 29%, respectively). The HER2 immunoposi- in all cases, and six specimens revealed polysomy of tivity was associated with EGFR co-expression in all but HER2.Three casesshowed HER2 loss. Of note, half of one lesion, although the level of EGFR expression was the HER2 gains were not detectable by aCGH, probably heterogeneous. due to a low number of neoplastic cells in these speci- mens. Copy number gains of both, EGFR and HER2 aCGH study genes, correlated well with HER2 immuno-positivity by Using the whole genome 4.3 K TK-array, we studied IHC. Of the 34 cases analyzed, 13 tumors were polyso- copy number aberrations (CNA) in eleven cases for mic for CSF1R/PDGFRB and two revealed loss of which frozen tissue was available. Ten out of the 11 CSF1R/PDGFRB; the remaining presented disomy for Dewaele et al. Clinical Sarcoma Research 2011, 1:4 Page 7 of 16 http://www.clinicalsarcomaresearch.com/content/1/1/4 AB CD EF Figure 1 Histology and EGFR protein expression in chordomas. A and B/ Examples of histologic appearance of chordomas stained with hematoxylin and eosin (H&E). C - F/ Illustration of chordoma cases with heterogeneous type of positive EGFR immunostaining. F/ The typical physaliphorous cells with abundant vacuolated cytoplasm, showing EGFR membrane staining. Dewaele et al. Clinical Sarcoma Research 2011, 1:4 Page 8 of 16 http://www.clinicalsarcomaresearch.com/content/1/1/4 Table 3 Recurrent copy number losses in chordoma cases and examples are depicted in Figure 4. The probes are by aCGH ranked according to their false discovery rate (fdr) Regions lost in ≥ five cases adjusted p-value. The column “Estimate” shows the esti- mate mean log -intensity ratio for each RTK or kinase Chordoma cases (#) Cytogenetic Frequency Candidate 2 location genes over all experiments. The first three RTK-probes and 4b, 5, 7b, 10b, 15a, 16 1pter-p33.2 0.60 RUNX3 the first twelve kinase-probes in Table 4 have a log - 4b, 5, 7b, 10b, 15a 1p22.3-p21.3 0.50 intensity ratio significantly larger than zero at the alpha 3, 4b, 5, 7b, 10b, 13, 3p24.1-p14.2 0.80 RBM5, FHIT, level of 5%. Thus, the EPHB2, EGFR and macrophage- 14, 16 PTPRG stimulating protein receptor (MSPR) were found to be 4b, 5, 7b, 9, 14 3q11.2-q13.31 0.50 significantly activated in chordoma. Although present in 4b, 5, 7b, 9, 14 3q26.1-26.31 0.50 some of the analyzed specimens, activation of the 3, 5, 7b,10b, 13, 14, 9pter-p21 0.70 CDKN2A PDGFRB, FGFR3, CSF1R and ERBB4 was not statisti- 15a cally significant in our study. Strikingly, there was no 3, 5, 7b, 13, 15a 9q34.11-qter 0.50 TSC1 detectable activation of KIT or VEGF receptors. By ana- 4b, 7b, 10b, 14, 15a 10q11.23-q24.2 0.50 PTEN lyzing the signaling pathways (the profiles of 46 kinases 5, 7b, 10b, 14, 15a 19p13.3-p13.2 0.50 and protein substrates), AKT, RSK1/2/3, TP53, MSK1/2, 4b, 5, 9, 10b, 15a, 16 22q12.2-qter 0.60 CHEK2 YES, p38a, p70 S6K, CREB and SRC were the most fre- quently and strongest phosphorylated proteins in our these genes. The tumor suppressor PTEN was lost in cohort. Interestingly, SRC family members, as SRC and seven out of 18 analyzed tumors. YES, were recurrently activated in chordoma. Further- more, kinase-array revealed the activation of down- Mutation analysis stream effectors of both, the PI3K/AKT/mTOR and No activating mutations of EGFR,CSF1R,PDGFRB, RAS/RAF/MAPK pathways. PDGFRA or KIT in examined genes’ exons were found in any of the 13 analyzed cases (Table 1). Western immunoblotting The consistent protein expression of EGFR and RTK phosphorylation profiling using phospho-RTK and PDGFRB and the recurrent activation of EGFR were phospho-kinase antibody arrays confirmed by Western blotting (Figure 5). The expres- The results of the RTK- and kinase-analysis of 12 and sion status of EGFR in all cases was in agreement with 10 chordoma samples respectively are shown in Table 4 the results obtained by IHC (Table 1 Figure 5). Briefly, losses gains amplification 60% + 2p + 7 40% + 1q 20% 0% -20% -40% -60% 22q12.1-qter 9q34.11-qter 1pter-p33.2 19p13.3-p13.2 -80% 9pter-p21 3q11.2-q13.31 3p24.1-p14.2 -100% 1 3 5 7 9 11 13 15 17 19 21 2 4 6 8 10 12 14 16 18 20 22 Figure 2 Frequency (%) of gained and lost regions detected by 4.3K TK aCGH in chordomas. Gains are shown in grey, losses in blue and amplification in black. Important recurrent gains and losses are circled in red. No rearrangements or high level amplification of genes encoding TK were detected. Dewaele et al. Clinical Sarcoma Research 2011, 1:4 Page 9 of 16 http://www.clinicalsarcomaresearch.com/content/1/1/4 A B C D Figure 3 Representative examples of dual-color interphase FISH images on paraffin sections in chordomas.Detected by the co- hybridization of SpectrumOrange labeled EGFR DNA probe (red signals) and SpectrumGreen labeled chromosome 7 CEP probe (green signals). (A) Case 1a, showing EGFR disomy. (B) Case 10a reveals EGFR polysomy. (C) Case 20 shows low level amplification in < 10% of nuclei. Of note, this amplification is not detected by aCGH. (D) Case 12b, showing high level amplification of EGFR in > 40% of nuclei. cases 15b and 10b showing only faint EGFR staining on used as starting material for both experiments, which the Western blot were scored negative by immunostain- may bring about differences, as chordomas are proven ing. All other cases, presenting clear or intense EGFR to be heterogeneous lesions. By Western immunoblot, expression by Western, were immune-scored accord- PDGFRB was found to be expressed in all chordomas ingly as “1+”, “2+” or “3+”.Two specimenswereana- analyzed, although only one case (#10b) also presented lyzed in parallel by Western immunoblotting and RTK activated PDGFRB. KIT protein expression and low antibody array. The strong EGFR activation of case 12b level activation was found in three and two cases detected by Western was confirmed by RTK antibody respectively. array. In case 4b, EGFR was expressed but not activated by Western. However, intermediate activation of EGFR Discussion was disclosed for this lesion by RTK antibody array. Recent reports suggest that RTK might be essential for This apparent difference could be ascribed to the fact the survival or proliferation of chordoma tumor cells. that the antibody used for Western blot detects the Therefore, targeting RTK may offer new therapeutic phosphorylation status of just one EGFR tyrosine resi- options for chordoma treatment. Nevertheless, there are due(Y1068),while theantibodyarraydetects thephos- important discrepancies between the reported results, phorylation of all tyrosine residues on the EGFR which are most likely due to differences in the relative protein. Furthermore, different pieces of the tumor were sensitivities of the methods used or heterogeneity of the Dewaele et al. Clinical Sarcoma Research 2011, 1:4 Page 10 of 16 http://www.clinicalsarcomaresearch.com/content/1/1/4 Table 4 Significantly phosphorylated RTK and kinase sites in chordoma using Proteome Profiler arrays, ranked based on p-value Probe name Estimate Standard Error t-value Raw p-value fdr adjusted p-value Phospho-RTK EPHB2 0.1285 0.0263 4.9 6.6931E-07 2.8111E-05 EGFR 0.6762 0.1694 3.99 3.8547E-05 0.0008 MSPR 0.1241 0.0426 2.91 0.0019 0.0266 PDGFRB 0.0848 0.0334 2.54 0.0057 0.0600 FGFR3 0.1022 0.0484 2.11 0.0177 0.1487 CSF1R 0.0887 0.0445 1.99 0.0236 0.1652 ERBB4 0.0160 0.0289 1.78 0.0379 0.2272 Phosphorylated kinase site AKT (T308) 0.3117 0.0313 9.95 3.1253E-21 1.5001E-19 RSK 1/2/3 (S380) 0.1747 0.0212 8.25 1.2388E-15 2.9731E-14 TP53 (S46) 0.2394 0.0336 7.14 2.3075E-12 3.6920E-11 MSK 1/2 (S376/S360) 0.1557 0.0256 6.09 1.3564E-09 1.6277E-08 YES (Y426) 0.1639 0.0288 5.69 1.2512E-08 1.2012E-07 TP53 (S15) 0.2533 0.0469 5.41 5.5176E-08 4.4141E-07 p38a (T180/Y182) 0.2858 0.0625 4.57 3.2798E-06 2.2490E-05 p70 S6K (T421/S424) 0.1086 0.0242 4.49 4.6993E-06 2.8196E-05 CREB (S133) 0.3273 0.1018 3.21 0.0007 0.0038 RSK 1/2 (S221) 0.0707 0.0246 2.87 0.0022 0.0104 SRC (Y419) 0.0934 0.0349 2.68 0.0038 0.0158 TP53 (S392) 0.1237 0.0464 2.67 0.004 0.0158 TOR (S2448) 0.2407 0.1258 1.91 0.0284 0.105 JUN (S63) 0.0863 0.0533 1.62 0.053 0.1818 HSP27 (S78/S82) 0.1048 0.0691 1.52 0.0647 0.2016 eNOS (S1177) 0.2002 0.1331 1.50 0.0672 0.2016 STAT1 (Y701) 0.0465 0.0318 1.46 0.0725 0.2048 STAT5b (Y699) 0.0380 0.0286 1.33 0.0921 0.2457 LYN (Y397) 0.0351 0.0283 1.24 0.1079 0.2725 STAT6 (Y641) 0.0309 0.0284 1.09 0.1382 0.3317 STAT5A (Y699) 0.0656 0.0715 0.92 0.1791 0.4093 FYN (Y420) 0.0587 0.0768 0.76 0.2239 0.4884 STAT5A/B (Y699) 0.0168 0.0366 0.46 0.3229 0.6739 ERK1/2 (T202/Y204. T185/Y187) 0.0284 0.0708 0.40 0.3447 0.6894 * The probes written in bold have a log2-intensity ratio significantly larger than zero at the a-level of 5%. material analyzed. Moreover, the characterization of Interestinglythough, themostrecurrentcopynumber chordoma in most studies is rarely based on parallel gain, found in three out of ten cases, involved the entire multiple techniques. Our objective was to characterize chromosome 7. Gain of chromosome 7 is frequently this rare entity in search for relevant therapeutic targets reported in chordomas, and multiple genes that encode using a wide range of methodological approaches. TK are located on chromosome 7, including the EGFR Whole genome 4.3 K TK-array CGH revealed moder- [3,4,7,27-29]. Accordingly, copy number gains involving ately complex CNA across the genome in all but one the EGFR locus, were found by FISH in 22/42 (52%) of examined cases, with losses more common than gains. our cases. Polysomy of the EGFR/ERBB1 gene was pre- The CNA found in our cohort were in accordance with viously reported in a subset of chordomas, and the previously recognized imbalances in chordomas EGFR is an interesting target for therapy in chordoma [3,4,7,27-29]. No deletions or gains common to all sam- based on the availability of targeted molecular inhibitors ples were found, confirming that chordomas are geneti- [8,16]. Additionally, the status of the gene encoding cally heterogeneous tumors. HER2, a close family member and important dimeriza- Importantly, we did not identify any amplifications or tion partner of EGFR, was investigated. Copy number rearrangements involving genes coding for TK. gains of HER2 were identified in 6/42 (14%) of cases. Dewaele et al. Clinical Sarcoma Research 2011, 1:4 Page 11 of 16 http://www.clinicalsarcomaresearch.com/content/1/1/4 Case 18 9 10 Case 17a 9 10 1 13 Case 20 9 10 1 7 14 11 Case 21 9 10 Figure 4 Representative images from phospho-RTK (left panel) and phospho-kinase (right panel) arrays from chordoma cases 18, 17b, 20 and 21. The EGFR and EPHB2 TK are frequently activated and downstream RTK signaling intermediates are activated consistently in chordomas. Each kinase is spotted in duplicate. The pairs of dots in each corner are positive controls. Each pair of the most positive kinase dots is denoted by a numeral, with the identity of the corresponding kinases listed as follows: 1) EGFR, 2) CSF1R, 3) MSPR, 4) PDGFRB, 5) FGFR3, 6) EPHB2, 7) HER2, 8) TOR, 9) AKT, 10) TP53, 11) RSK1/2/3, 12) S6K, 13) CREB, 14) YES, 15) MSK1/2, 16) RSK1/2, 17) eNOS. Noteworthy, copy number gains of HER2 were exclu- Accordingly, loss of expression of the CDKN2A protein sively found in recurrent or metastatic cases in our in chordoma was also previously shown by immunos- cohort, further suggesting its possible association with taining [32]. Other recurrent losses, observed in the pre- poor clinical outcome. sent study by aCGH, involved regions carrying the Losses of large chromosomal regions are typically tumor suppressors PTEN/10q23.31, CHEK2/22q12.1 and found in chordoma. In this study, losses revealed by the transcription factor RUNX3/1p36.11, all previously aCGH predominantly involved chromosome 3; the smal- described in chordomas [7]. lest overlapping region of deletion, 3p24.1-p14.2, was In order to characterize the compendium of co-acti- lost in eight out of ten analyzed cases. This region con- vated RTK in chordoma, we used an antibody array that tains multiple genes, including RBM5, FHIT and allows the simultaneous characterization of the phos- PTPRG, but their involvement in chordoma pathogen- phorylation status of 42 different RTK. Most impor- esis has yet to be determined. Loss of the 9pter-p21 tantly, the EGFR kinase was consistently activated in all region, another frequent feature revealed by aCGH ana- 12 investigated cases. Furthermore, statistical analysis lysis, was found in seven out of ten tumors. Importantly, showed that EGFR activation was significant for chordo- in three cases the region was homozygous lost. The mas, based on the analysis of our cohort. The activation losses encompassed the tumor suppressor genes of EGFR in chordoma was previously shown by other CDKN2A and CDKN2B, which are frequently deleted in groups, although the reported frequencies of the EGFR many tumor types [30,31]. Correspondingly, Hallor and activation in chordoma vary significantly [8,16]. By RTK co-workers observed loss of the CDKN2A locus with an antibody array Tamborini and co-workers reported incidence of 70% in chordoma, and with an even higher EGFR, HER2 and HER4 activation in 6/7 (86%), 5/7 frequency considering just metastasizing lesions [7]. (71%) and 3/7 (43%) of cases, respectively [8]. However, Dewaele et al. Clinical Sarcoma Research 2011, 1:4 Page 12 of 16 http://www.clinicalsarcomaresearch.com/content/1/1/4 statistical analysis included a multiple testing correction. Chordoma cases The linear mixed model avoids the use of an arbitrarily 15b 4b 9 2b 10b 5 7b 12b 14 chosen cut-off that can lead to overestimation of the p-PDGFRB (Y751) 190 kDa activation of RTK and to uncertainty about the results. Notably, statistical analysis was never described by 190 kDa others in reports published so far in reference to RTK PDGFRB proteome profiling kits, thus the statistical significance of reported data is unknown. Importantly, we also found p-EGFR (Y1068) 175 kDa two other RTK: EPHB2 and MSPR, to be significantly activated in chordoma. The activation of EPHB2 was 175 kDa EGFR recently described in one chordoma study [16]. The role of EPHB2 in chordoma development and progression 145 kDa p_KIT (Y703) 125 kDa needs to be further evaluated. In general, EPHB2 func- tion depends on the tumor type and signaling context of 145 kDa KIT the neoplastic cell. The EPHB2 has a tumor suppressive 125 kDa role in colon carcinoma; in contrast, EPHB2 promotes Actin 42 kDa cell proliferation in adenomas and normal intestinal epithelium. Notably, it was recently shown in mice mod- Figure 5 Western immunoblot of nine chordoma cases.The els that the intrinsic kinase activity of EPHB2 conveys immunoblot confirms the frequent expression of EGFR and PDGFRB, mitogenic signals [33]. It is of interest that imatinib and frequent activation of EGFR, but not of PDGFRB and KIT mesylate is as an inhibitor of EPHB mitogenic signaling. proteins. Equal amounts of total protein extracts from nine tumors were separated on a gel, immunoblotted and then probed with the TheMSPR/RONtyrosinekinaseisamember of the indicated antibodies. MET family of RTK. MET expression was shown pre- viously in chordomas by several other groups, but MSPR expression and activation was only recently using immunoprecipitation assay, EGFR and HER2 were reported in all three investigated chordomas by Shalaby phosphorylated in respectively 17/22 (77%) and 6/14 andco-workers [16].Asitisthe case with itsbetter- (43%) of their cases [8]. Using the same RTK antibody known family member, MET, several lines of evidence array, Shalaby and colleagues recently showed activation suggest a role for RON in human cancer. Generally, of HER2, MSPR, EPHB2 and MER for the U-CH1 chor- RON overexpression is associated with poor clinical out- doma cell line and the three tested chordoma cases [16]. come and metastasis [34]. Foretinib, an oral multi-kinase inhibitor of MET, RON, AXL and VEGFR, is currently In our study, we found significant activation of EGFR, in phase I and II clinical testing [35]. HER2 and HER4 in respectively 12, one and one out of The multiple RTK co-activation is not a distinctive 12 cases, using the same antibody arrays. Interestingly, feature of chordomas, because similar patterns were the frequent activation of PDGFRB in chordomas [21/22 reported in other tumor types, such as colon adenocar- (95%) of cases] was described in the study by Tamborini cinomas, intimal sarcomas, glioblastomas or osteosarco- and collaborators [8]. In contrast, we found activation of mas [36-38]. Importantly, the simultaneous activation of PDGFRB only in five out of 12 (42%) chordomas, using multiple RTK provides the tumor cells with reduced thesameantibodyRTK arrays andusing thevalue of dependence on a single RTK for the maintenance of cri- the mean plus the standard deviation within an array as tical downstream signaling, and thus renders such the cut-off. However as indicated by statistical analysis, tumors refractory to single-agent RTK inhibition. PDGFRB activation was not significant in our cohort. This discrepancy might be attributable to the heteroge- The conflicting results on the frequency of EGFR, neity of chordoma tumors, the quality of the frozen HER2, PDGFRB expression and activation, and also tumor tissue used for the analysis, modifications of the copy number alterations in chordoma, might be due to technique and/or to subsequent dissimilar analysis of differences in sensitivity of the techniques used. In addi- the data. Thus, Tamborini and co-workers used high- tion, even if using the same technique, there are impor- concentrated (e.g. 2 mg/array) protein lysate per array in tant variations in methodology between different their study [8]. In contrast, we performed the experi- laboratories, with many confounding factors contribut- ments according to the manufacturers’ recommenda- ing to the inconsistencies, e.g. the different type and tions which indicate 500 μg of total protein as the source of the antibodies used in the immunohistochem- maximum amount to be used for each array. In addi- ical studies. When immunostaining is considered, it is tion, we have performed an extensive statistical analysis well known that the way of tissue fixation influences of thedatabyusing alinear mixedmodel.Our outcome [39]. Tumor specimens are frequently retrieved Dewaele et al. Clinical Sarcoma Research 2011, 1:4 Page 13 of 16 http://www.clinicalsarcomaresearch.com/content/1/1/4 from archives, and in case they are not preserved well, studies [8,42]. The discrepancy in the prevalence of the this maygiverisetofalse negative cases. Thelackof activated proteins between the reported results is most sensitivity of IHC to identify low expression levels of likely due to differences in the relative sensitivity and EGFR was comprehensively illustrated in colorectal can- specificity of the methods. This is well illustrated in a cer [40]. Similarly, chordoma immunostaining might study by Dobashi and co-workers, who found activated also show inconsistencies associated with these metho- mTOR in all five cases using immunohistochemistry, dological problems. Along this line, Weinberger and co- but only in one case using Western immunoblotting workers found EGFR and HER2 expression in respec- [44]. Nevertheless, the involvement of the AKT/mTOR tively 12 (100%) and seven (58%) out of 12 chordomas, pathway in chordoma is clear. Importantly, efficient using IHC on tissue micro-arrays (TMA), while Shalaby inhibition of the human chordoma cell line UCH-1 by and colleagues showed EGFR expression and activation PI-103, a dual PI3K and mTOR inhibitor, was recently in respectively 69% (79/114) and 50% (56/115) of chor- reported [43]. Notably, it was recently shown that AKT doma cases by the same technique, and while Fasig and activation persists in the UCH-1 chordoma cell line fol- co-authors reported EGFR activation in nine out of 21 lowing treatment with the EGFR inhibitor tyrphostin (43%) cases [12,15,16]. By conventional immunostaining, [16]. we have also found that EGFR and HER2 are expressed Furthermore, by kinase antibody arrays, we also found in chordomas, albeit in a lower fraction of cases 26/39 effectors of RAS/ERK1/2 signaling to be significantly (67%) and 5/16 (31%), respectively. In contrast to Wein- activated in chordoma, like ribosomal S6 kinases (RSK) berger and co-workers, however, we found more fre- 1/2/3, the CREB transcription factor and the chromatin quent EGFR expression in advanced (15/20, 75%) rather associated kinase p38. More downstream are the mito- than in primary (11/19, 58%) lesions. Again in contrast gen- and stress-activated protein kinases, MSK1 and the to Weinberger and co-workers, we did find a positive closely related isoform MSK2. These are nuclear kinases correlation between HER2 expression and EGFR expres- that are activated by the ERK1/2 and p38 MAPK signal- sion, which is in line with the HER2/EGFR heterodimers ing cascades [45]. Additionally, the SRC family mem- formation in chordomas reported by other groups [8,12]. bers, SRC and YES, were also activated. These pathways Moreover, we did not find a significant correlation were not extensively analyzed in chordoma by other between EGFR and HER2 gene status and their expres- groups, except for ERK1/2, which was described to be sion by immunostaining, this phenomenon was also consistently strongly phosphorylated in chordoma by described in colorectal cancer [40,41]. Tamborini and co-workers [8]. Nevertheless, these acti- The circuitry of intracellular signalling downstream of vated proteins are all confounding factors that might RTK is an area of dynamic investigations in many can- offer the tumors redundancy, making them less respon- cer types and advances in the characterization of this sive to upstream RTK and AKT pathway inhibition. signalling allows better selection of appropriate thera- Oncogenes often cooperate with additional mutations peutic agents. In the present study, we analyzed the acti- that disrupt tumor suppressor pathways. Phosphatase vation of important effectors of signalling downstream and tensin homologue deleted on chromosome ten of RTK. Using kinase antibody arrays, AKT was the (PTEN), is an important negative regulator of the AKT/ most frequent (found in nine out of ten cases analyzed) mTOR pathway, which when not expressed contributes and highest phosphorylated in chordomas. Similarly, to constitutive phosphorylation of AKT and activation Presneau and co-workers found AKT activation in 45 of downstream effectors. PTEN loss is also frequently out of 49 (92%) chordomas analysed by TMA, and Tam- found in chordomas. We observed loss of PTEN in five borini and colleagues in 21 out of 22 chordomas (95%) out of ten cases by aCGH, and in seven out of 18 (39%) using Western blotting [8,42]. The AKT protein trans- cases by FISH. Presneau and co-workers recently duces signals to several effector molecules, including revealed loss of PTEN protein expression in seven out TSC1/2. More specifically, AKT inhibits TSC1/2 and of 43 (16%) cases by IHC and semi-quantitative RT-PCR hereby relieves inhibition of mammalian target of rapa- [42]. Han and co-workers showed negative PTEN stain- mycin (mTOR), which functions downstream of TSC1/ ing by IHC in six out of ten sporadic chordoma [46]. 2. This occurs in part by phosphorylating two substrates, Just like in our cases, they did not find any correlation p70S6 kinase (S6K) and eukaryotic initiation factor 4E- between loss of PTEN and advanced disease. TSC1 is binding protein 1 (4E-BP1). Of note, p70S6K was acti- another critical tumor suppressor, implicated down- vated in five and mTOR in three of our ten chordoma stream in the PI3K/AKT and RAS/ERK pathways. In cases analyzed by kinase antibody arrays. These data are particular, upon growth factor activation, AKT, ERK and in accordance with previously published data [8,15,43]. p90 ribosomal S6 kinase 1 (RSK1) participate in TSC The phenomenon that p70S6K was activated in p- protein complex inhibition, hereby critically regulating mTOR negative chordomas was found in multiple cell growth and proliferation. Chordomas are reported Dewaele et al. Clinical Sarcoma Research 2011, 1:4 Page 14 of 16 http://www.clinicalsarcomaresearch.com/content/1/1/4 List of abbreviations in patients with tuberous sclerosis complex (TSC), an 4EBP1: eukaryotic translation initiation factor E4-binding protein 1; BAC: autosomal dominant disorder typified by hamartomas in bacterial artificial chromosome; CNA: copy number alterations; CSF1R: several organs, epilepsy, mental retardation and beha- colony-stimulating factor 1 receptor; DAPI: 4.6-diamidino-2-phenylindole; EGFR: epidermal growth factor receptor; ERK1/2: extracellular signal-regulated vioural problems. TSC is caused by germline mutations kinase; fdr: false discovery rate; HER2: v-ERBB2 Avian erythroblastic leukemia in the TSC1 or TSC2 genes and the loss of the corre- viral oncogene homolog 2; IHC: immunohistochemistry; IS: intimal sarcoma; sponding wild type allele. The chromosomal region MEK: mitogen-activated kinase kinase kinase 1; mTOR: mammalian target of rapamycin; NCBI: National Center for Biotechnology Information; PCR: 9q34.13, where the TSC1 gene is localized, is also fre- polymerase chain reaction; PDGFR: platelet derived growth factor receptor; quently lost in sporadic chordomas [7]. By aCGH, we PI3K: phosphatidyl inositol 3 kinase; PKB or AKT: protein kinase B; RTK: found loss of the region 9q34.11-qter, encompassing the receptor tyrosine kinase; S6K: ribosomal protein S6 kinase; SG: spectrum green; SO: spectrum orange; TK: tyrosine kinase; TKI: tyrosine kinase gene coding for TSC1, in five out of ten cases. Hallor inhibitors; TMA: tissue microarrays. and co-workers showed loss of this region in about 25% of 21 cases analyzed by aCGH. In contrast, Presneau Acknowledgements This work is supported by research grants from the EUROBONET consortium and co-workers found disomy for TSC1/2 by FISH in all (a network of excellence granted by the European Commission for studying of their 28 cases [42]. Generally, the consistent activa- the pathology and genetics of bone tumors), from the Fonds voor tion of AKT, the frequent activation of p70S6K and of Wetenschappelijk Onderzoek Vlaanderen (G.0589.09, MD-R), and by a Concerted Action Grant 2006/14 from the K.U.Leuven. mTOR, together with frequent loss of the TSC1 and PTEN genes, all suggest an important role for the PI3K/ Author details AKT pathway in chordoma. Department of Human Genetics, Catholic University of Leuven, University Hospitals, Leuven, Belgium. Department of Pathology, Catholic University of Leuven, University Hospitals, Leuven, Belgium. Laboratory of Experimental Conclusions Oncology, Department of General Medical Oncology, Catholic University of In summary, we found that EGFR is the strongest and Leuven, University Hospitals, Leuven, Belgium. I-BioStat, Catholic University of Leuven, Leuven, Belgium, and Hasselt University, Hasselt, Belgium. most frequently activated RTK in chordomas, and there- fore becomes a possible target for therapy. Lack of signifi- Authors’ contributions cant EGFR amplification and EGFR mutations suggests BD carried out the mutation analysis, participated in the aCGH data evaluation, Western immunoblotting analysis and antibody array analysis, activation by autocrine/paracrine ligand stimulation. and drafted the manuscript. FM carried out the histopathological PDGFRB is also activated in chordomas, but with a lower experiments and analysis and participated in the draft of the manuscript. GF frequency and/or to a lower level, which might not be participated in the antibody array experiments and analysis and histopathological analysis. MA performed the statistical analysis of the detectable by some current standard techniques. In the antibody arrays. VV carried out the FISH, aCGH, Western immunoblotting light of these findings, chordoma patients may benefit and antibody array experiments. AW performed the aCGH analysis and from treatment with multi-kinase inhibitors targeting both participated in the antibody array analysis. MDR participated in the design and coordination of the study and helped to draft the manuscript. RS EGFR and PDGFR. Furthermore, many other RTK are contributed tumor samples for this study, participated in the design of the activated in subsets of chordomas; these are likely to study and critically revised the manuscript. All authors read and approved increase treatment resistance in these tumors. These the final manuscript. results are currently only hypothesis-generating, and addi- Competing interests tional in vitro studies addressing the impact of inhibitors The authors declare that they have no competing interests. of RTK and their downstream effectors on chordoma Received: 25 January 2011 Accepted: 25 July 2011 tumor cells would be extremely useful in determining the Published: 25 July 2011 dominant and alternative RTKs in these tumors. As chor- domas are bone tumors, with a rigid, mineralized extracel- References lular matrix, ex-vivo studies on primary neoplastic 1. Stacchiotti S, Casali PG, Lo VS, Mariani L, Palassini E, Mercuri M, Alberghini M, Pilotti S, Zanella L, Gronchi A, Picci P: Chordoma of the chordoma cells will be difficult. Recent advances in com- mobile spine and sacrum: a retrospective analysis of a series of patients putational biology and network-based technologies gener- surgically treated at two referral centers. Ann Surg Oncol 2010, ating predictive models might be more of use [47]. 17:211-219. 2. Ferraresi V, Nuzzo C, Zoccali C, Marandino F, Vidiri A, Salducca N, Zeuli M, In conclusion, the consistent activation of AKT, the Giannarelli D, Cognetti F, Biagini R: Chordoma: clinical characteristics, recurrent activation of upstream EGFR and of down- management and prognosis of a case series of 25 patients. BMC Cancer stream effectors like p70S6K and mTOR, together with 2010, 10:22. 3. Mirra JM, Nelson SD, Della Rocca C, Mertens F: Chordoma. In World Health frequent loss of TSC1 and PTEN gene loci, all indicate Organization Classification of Tumours. Pathology and Genetics of Tumours of that the PI3K/AKT pathway is an important mediator of Soft Tissue and Bone. Volume 5.. 1 edition. Edited by: Fletcher CD, Unni KK, transformation in chordoma. Targeting this pathway is Mertens F. Lyon: IARC Press; 2002:315-317. 4. Scheil S, Bruderlein S, Liehr T, Starke H, Herms J, Schulte M, Möller P: likely to yield attractive data that will enlighten the Genome-wide analysis of sixteen chordomas by comparative genomic design of appropriate therapies. Individualized therapeu- hybridization and cytogenetics of the first human chordoma cell line, U- tic approaches depending on the genetic context of a CH1. Genes Chromosomes Cancer 2001, 32:203-211. 5. Tallini G, Dorfman H, Brys P, Dal Cin P, De Wever I, Fletcher CD, Jonson K, particular tumor are likely to be the most successful. Mandahl N, Mertens F, Mitelman F, Rosai J, Rydholm A, Samson I, Sciot R, Dewaele et al. Clinical Sarcoma Research 2011, 1:4 Page 15 of 16 http://www.clinicalsarcomaresearch.com/content/1/1/4 Van den Berghe H, Vanni R, Willén H: Correlation between 24. Chu T, Weir B, Wolfinger R: A systematic statistical linear modeling clinicopathological features and karyotype in 100 cartilaginous and approach to oligonucleotide array experiments. Mathematical Biosciences chordoid tumours. A report from the Chromosomes and Morphology 2002, 176:35-51. (CHAMP) Collaborative Study Group. J Pathol 2002, 196:194-203. 25. Wolfinger R, Gibson G, Wolfinger E, Bennett L, Hamadeh H, Bushel P, 6. Miozzo M, Dalpra L, Riva P, Volontà M, Macciardi F, Pericotti S, Tibiletti MG, Afshari C, Paules R: Assessing Gene Significance from cDNA Microarray Cerati M, Rohde K, Larizza L, Fuhrman Conti AM: A tumor suppressor locus Expression Data via Mixed Models. Journal of Computational Biology 2002, in familial and sporadic chordoma maps to 1p36. Int J Cancer 2000, 8:625-637. 87:68-72. 26. Benjamini Y, Hochberg Y: Controlling the False Discovery Rate: a Practical 7. Hallor KH, Staaf J, Jönsson G, Heidenblad M, Vult von Steyern F, Bauer HC, and Powerful Approach to Multiple Testing. Journal of the RSS: Series B Ijszenga M, Hogendoorn PC, Mandahl N, Szuhai K, Mertens F: Frequent (Statistical Methodology) 2010, 57:289-300. deletion of the CDKN2A locus in chordoma: analysis of chromosomal 27. Brandal P, Bjerkehagen B, Danielsen H, Heim S: Chromosome 7 imbalances using array comparative genomic hybridisation. Br J Cancer abnormalities are common in chordomas. Cancer Genet Cytogenet 2005, 2008, 98:434-442. 160:15-21. 8. Tamborini E, Virdis E, Negri T, Orsenigo M, Brich S, Conca E, Gronchi A, 28. Kuzniacka A, Mertens F, Strombeck B, Wiegant J, Mandahl N: Combined Stacchiotti S, Manenti G, Casali PG, Pierotti MA, Pilotti S: Analysis of binary ratio labeling fluorescence in situ hybridization analysis of receptor tyrosine kinases (RTKs) and downstream pathways in chordoma. Cancer Genet Cytogenet 2004, 151:178-181. chordoma. Neuro Oncol 2010, 12:776-89. 29. Sawyer JR, Husain M, Al-Mefty O: Identification of isochromosome 1q as a 9. Hof H, Welzel T, Debus J: Effectiveness of cetuximab/gefitinib in the recurring chromosome aberration in skull base chordomas: a new therapy of a sacral chordoma. Onkologie 2006, 29:572-574. marker for aggressive tumors? Neurosurg Focus 2001, 10:E6. 10. Naka T, Iwamoto Y, Shinohara N, Ushijima M, Chuman H, Tsuneyoshi M: 30. Li J, Rix U, Fang B, Bai Y, Edwards A, Colinge J, Bennett KL, Gao J, Song L, Expression of c-met proto-oncogene product (c-MET) in benign and Eschrich S, Superti-Furga G, Koomen J, Haura EB: A chemical and malignant bone tumors. Mod Pathol 1997, 10:832-838. phosphoproteomic characterization of dasatinib action in lung cancer. 11. Tamborini E, Miselli F, Negri T, Lagonigro MS, Staurengo S, Dagrada GP, Nat Chem Biol 2010, 6:291-299. Stacchiotti S, Pastore E, Gronchi A, Perrone F, Carbone A, Pierotti MA, 31. Olson LE, Soriano P: Increased PDGFRalpha activation disrupts connective Casali PG, Pilotti S: Molecular and biochemical analyses of platelet- tissue development and drives systemic fibrosis. Dev Cell 2009, derived growth factor receptor (PDGFR) B, PDGFRA, and KIT receptors in 16:303-313. chordomas. Clin Cancer Res 2006, 12:6920-6928. 32. Naka T, Boltze C, Kuester D, Schulz TO, Schneider-Stock R, Kellner A, 12. Weinberger PM, Yu Z, Kowalski D Joe J, Manger P, Psyrri A, Sasaki CT: Samii A, Herold C, Ostertag H, Roessner A: Alterations of G1-S checkpoint Differential expression of epidermal growth factor receptor, c-Met, and in chordoma: the prognostic impact of p53 overexpression. Cancer 2005, HER2/neu in chordoma compared with 17 other malignancies. Arch 104:1255-1263. Otolaryngol Head Neck Surg 2005, 131:707-711. 33. Genander M, Halford MM, Xu NJ, Eriksson M, Yu Z, Qiu Z, Martling A, 13. Casali PG, Messina A, Stacchiotti S, Tamborini E, Crippa F, Gronchi A, Greicius G, Thakar S, Catchpole T, Chumley MJ, Zdunek S, Wang C, Holm T, Orlandi R, Ripamonti C, Spreafico C, Bertieri R, Bertulli R, Colecchia M, Goff SP, Pettersson S, Pestell RG, Henkemeyer M, Frisén J: Dissociation of Fumagalli E, Greco A, Grosso F, Olmi P, Pierotti MA, Pilotti S: Imatinib EphB2 signaling pathways mediating progenitor cell proliferation and mesylate in chordoma. Cancer 2004, 101:2086-2097. tumor suppression. Cell 2009, 139:679-692. 14. Deniz ML, Kilic T, Almaata I, Kurtkaya O, Sav A, Pamir MN: Expression of 34. Wagh PK, Peace BE, Waltz SE: Met-related receptor tyrosine kinase Ron in growth factors and structural proteins in chordomas: basic fibroblast tumor growth and metastasis. Adv Cancer Res 2008, 100:1-33. growth factor, transforming growth factor alpha, and fibronectin are 35. Eder JP, Shapiro GI, Appleman L, Zhu AX, Miles D, Keer H, Cancilla B, Chu F, correlated with recurrence. Neurosurgery 2002, 51:753-760. Hitchcock-Bryan S, Sherman L, McCallum S, Heath EI, Boerner SA, 15. Fasig JH, Dupont WD, LaFleur BJ, Olson SJ, Cates JM: LoRusso PM: A phase I study of foretinib, a multi-targeted inhibitor of c- Immunohistochemical analysis of receptor tyrosine kinase signal Met and vascular endothelial growth factor receptor 2. Clin Cancer Res transduction activity in chordoma. Neuropathol Appl Neurobiol 2008, 2010, 16:3507-3516. 34:95-104. 36. Stommel JM, Kimmelman AC, Ying H, Nabioullin R, Ponugoti AH, 16. Shalaby A, Presneau N, Ye H, Halai D, Berisha F, Idowu B, Leithner A, Liegl B, Wiedemeyer R, Stegh AH, Bradner JE, Ligon KL, Brennan C, Chin L, Briggs TRW, Bacsi K, Kindblom LG, Athanasou N, Amary MF, DePinho RA: Coactivation of receptor tyrosine kinases affects the Hogendoorn PCW, Tirabosco R, Flanagan AM: The role of epidermal response of tumor cells to targeted therapies. Science 2007, 318:287-290. growth factor receptor in chordoma pathogenesis: a potential 37. Zhou Y, Li S, Hu YP, Wang J, Hauser J, Conway AN, Vinci MA, Humphrey L, therapeutic target. J Pathol 2011, 223:336-346. Zborowska E, Willson JK, Brattain MG: Blockade of EGFR and ErbB2 by the 17. Brand-Saberi B, Christ B: Evolution and development of distinct cell novel dual EGFR and ErbB2 tyrosine kinase inhibitor GW572016 lineages derived from somites. Curr Top Dev Biol 2000, 48:1-42. sensitizes human colon carcinoma GEO cells to apoptosis. Cancer Res 18. Ostroumov E, Hunter CJ: Identifying mechanisms for therapeutic 2006, 66:404-411. intervention in chordoma: c-Met oncoprotein. Spine (Phila Pa 1976) 2008, 38. Dewaele BM, Floris G, Finalet-Ferreiro J, Fletcher CD, Coindre JM, Guillou L, 33:2774-2780. Hogendoorn PC, Wozniak A, Vanspauwen V, Schöffski P, Marynen P, 19. Wozniak A, Sciot R, Guillou L, Pauwels P, Wasag B, Stul M, Vermeesch JR, Vandenberghe P, Sciot R, Debiec-Rychter M: Co-activated PDGFRA and Vandenberghe P, Limon J, Debiec-Rychter M: Array CGH analysis in EGFR are potential therapeutic targets in intimal sarcoma. Cancer Res primary gastrointestinal stromal tumors: cytogenetic profile correlates 2010, 70:7304-14. with anatomic site and tumor aggressiveness, irrespective of mutational 39. Atkins D, Reiffen KA, Tegtmeier CL, Winther H, Bonato MS, Storkel S: status. Genes Chromosomes Cancer 2007, 46:261-276. Immunohistochemical detection of EGFR in paraffin-embedded tumor 20. Lahortiga I, De Keersmaecker K, Van Vlierberghe P, Graux C, Cauwelier B, tissues: variation in staining intensity due to choice of fixative and Lambert F, Mentens N, Beverloo HB, Pieters R, Speleman F, Odero MD, storage time of tissue sections. J Histochem Cytochem 2004, 52:893-901. Bauters M, Froyen G, Marynen P, Vandenberghe P, Wlodarska I, Meijerink JP, 40. Chung KY, Shia J, Kemeny NE, Shah M, Schwartz GK, Tse A, Hamilton A, Cools J: Duplication of the MYB oncogene in T cell acute lymphoblastic Pan D, Schrag D, Schwartz L, Klimstra DS, Fridman D, Kelsen DP, Saltz LB: leukemia. Nat Genet 2007, 39:593-595. Cetuximab shows activity in colorectal cancer patients with tumors that 21. VU Micro-Array Data Analysis. [http://www.few.vu.nl/~vumarray/]. do not express the epidermal growth factor receptor by 22. Debiec-Rychter M, Wasag B, Stul M, De Wever I, Van Oosterom A, immunohistochemistry. J Clin Oncol 2005, 23:1803-1810. Hagemeijer A, Sciot R: Gastrointestinal stromal tumours (GISTs) negative 41. Moroni M, Veronese S, Benvenuti S, Marrapese G, Sartore-Bianchi A, Di for KIT (CD117 antigen) immunoreactivity. J Pathol 2004, 202:430-438. Nicolantonio F, Gambacorta M, Siena S, Bardelli A: Gene copy number for 23. Primer3: WWW primer tool. [http://frodo.wi.mit.edu/cgi-bin/primer3/ epidermal growth factor receptor (EGFR) and clinical response to primer3_www_slow.cgi]. antiEGFR treatment in colorectal cancer: a cohort study. Lancet Oncol 2005, 6:279-286. Dewaele et al. Clinical Sarcoma Research 2011, 1:4 Page 16 of 16 http://www.clinicalsarcomaresearch.com/content/1/1/4 42. Presneau N, Shalaby A, Idowu B, Gikas P, Cannon SR, Gout I, Diss T, Tirabosco R, Flanagan AM: Potential therapeutic targets for chordoma: PI3K/AKT/TSC1/TSC2/mTOR pathway. Br J Cancer 2009, 100:1406-1414. 43. Schwab J, Antonescu C, Boland P, Healey J, Rosenberg A, Nielsen P, Iafrate J, Delaney T, Yoon S, Choy E, Harmon D, Raskin K, Yang C, Mankin H, Springfield D, Hornicek F, Duan Z: Combination of PI3K/mTOR inhibition demonstrates efficacy in human chordoma. Anticancer Res 2009, 29:1867-1871. 44. Dobashi Y, Suzuki S, Sato E, Hamada Y, Yanagawa T, Ooi A: EGFR- dependent and independent activation of Akt/mTOR cascade in bone and soft tissue tumors. Mod Pathol 2009, 22:1328-1340. 45. Chiacchiera F, Simone C: Signal-dependent regulation of gene expression as a target for cancer treatment: inhibiting p38alpha in colorectal tumors. Cancer Lett 2008, 265:16-26. 46. Han S, Polizzano C, Nielsen GP, Hornicek FJ, Rosenberg AE, Ramesh V: Aberrant hyperactivation of akt and Mammalian target of rapamycin complex 1 signaling in sporadic chordomas. Clin Cancer Res 2009, 15:1940-1946. 47. Xu AM, Huang PH: Receptor tyrosine kinase coactivation networks in cancer. Cancer Res 2010, 70:3857-3860. doi:10.1186/2045-3329-1-4 Cite this article as: Dewaele et al.: Frequent activation of EGFR in advanced chordomas. Clinical Sarcoma Research 2011 1:4. Submit your next manuscript to BioMed Central and take full advantage of: • Convenient online submission • Thorough peer review • No space constraints or color figure charges • Immediate publication on acceptance • Inclusion in PubMed, CAS, Scopus and Google Scholar • Research which is freely available for redistribution Submit your manuscript at www.biomedcentral.com/submit

Journal

Clinical Sarcoma Research – Springer Journals

Published: Jul 25, 2011

There are no references for this article.