Coactivated Platelet-Derived Growth Factor Receptor α and Epidermal Growth Factor Receptor Are Potential Therapeutic Targets in Intimal Sarcoma

It is well established that platelet-derived growth factors (PDGF) are involved in several pathologic settings, including tumor growth (1, 2). PDGFs bind to two receptor tyrosine kinases (RTK), PDGF receptor-α (PDGFRA) and PDGF receptor-β (PDGFRB), which activate the receptor kinases and initiate a number of signaling pathways, including the RAS/extracellular signal–regulated kinase 1/2 (ERK1/2), phosphoinositide-3-kinase/AKT and protein kinase C pathways. The cellular responses to PDGF signaling involve proliferation, survival, migration, and differentiation. In mouse models, elevated PDGFRA activation leads to connective tissue hyperplasia, progressive chronic fibrosis in many organs, and sarcomagenesis (3). In humans, aberrant PDGFR signaling has been associated with a number of malignancies (4). In sarcomas, several types of genetic alterations, including gene amplification, translocations, and activating mutations, result in ligand and/or receptor overexpression (5). These aberrations may lead to dysregulated activity of PDGFRA/B. The activating mutations of PDGFRA in gastrointestinal stromal tumors (GIST) and the overexpression of PDGFB due to a unique COL1A1-PDGFB fusion in dermatofibrosarcoma protuberans serve as examples (6, 7). Imatinib mesylate, the inhibitor of the tyrosine kinase activity of BCR-ABL, which also targets PDGFRA/B, proved to be very effective for the treatment of both conditions (8, 9). There are several reports on PDGFRA amplifications in solid tumors, such as glioblastomas (10, 11), anaplastic oligodendrogliomas (12), and esophageal squamous cell carcinomas (13).

Intimal sarcoma (IS) is a rare, lethal mesenchymal tumor arising in the large arteries, mainly the pulmonary artery and the aorta. Presumably, these sarcomas develop from the subendothelial cells of the blood vessel wall located in the intima (14). The defining feature is the predominant intraluminal growth with obstruction of the lumen of the vessel of origin and embolic tumor dissemination (15). On histology, IS usually displays features of a poorly differentiated sarcoma, composed of spindle cells, with varying degrees of atypia, mitotic activity, nuclear polymorphisms, and necrosis (16). Some tumors may show considerable heterogeneity in terms of extracellular matrix and tumor cell composition. Large myxoid regions or an epithelioid component can be observed, as well as areas of rhabdomyo-, angio-, or osteosarcomatous differentiation (17–21). Angiosarcoma and leiomyosarcoma are the most important entities that have to be included in the differential diagnosis. The clinical presentation of IS is often not specific and related to tumor emboli (15). Pulmonary IS behaves very aggressively on site, conferring lung infiltrations in 40% of the patients (22, 23). Aortic IS mostly arises in the abdominal aorta. Arterial embolic tumor dissemination might result in distant metastases involving the bone, peritoneum, liver, and mesenteric lymph nodes (15). The prognosis of IS is poor, with a mean survival time of about 12 months (24).

To date, the understanding of the histogenesis and molecular mechanisms leading to the development of IS remains fragmentary. In recent years, recurrent molecular genetic events associated with IS development have been reported based on the analysis of a few cases. By comparative genomic hybridization (CGH), recurrent amplification of the regions 4q12 and 12q13-q15 was described in a subset of IS (25). Amplification of the chromosomal region 12q13-q15 was reported in 75% of these tumors (14, 25). Multiple genes in this region, such as MDM2, SAS, CDK4, and GLI1, are frequently amplified in many types of human cancer (14, 26). Notably, amplification of the PDGFRA gene, which maps to chromosome 4q12, was reported in 5 of 8 (62.5%) pulmonary artery IS (25). In a more recent report, deregulated copy numbers of PDGFRA/PDGFRB, KIT, and EGFR were described in seven examined IS cases, and the activation of the PDGFRs and EGFR has been disclosed by biochemical assays in one of these IS cases (24).

Due to the rareness of IS, these lesions have been a neglected area and there is a lack of specific biomarkers. The therapy is currently limited to surgery (23). In this study, we aimed to identify potential novel diagnostic markers and molecular targets for the treatment of IS. To this aim, we have performed an integrated histopathologic, cytogenetic, molecular, and biochemical analysis of a large cohort of 21 cases with IS. In view of the clinical availability of several therapeutic agents against RTKs, we have focused on the relevance of activation of RTKs in the pathobiology of IS and on the inhibitory effect of RTK inhibitors on primary IS cells.

The present study included 21 patients (11 females and 10 males; age range, 27–74 years; median age, 52 years; Table 1). The tumors were located in the pulmonary artery trunk (n = 12), right or left pulmonary artery (n = 3), heart (n = 4), femoral artery (n = 1), and splenic artery (n = 1), and one presented as a femoral embolus from the aorta primary site.

Histopathologic examination was performed on formalin-fixed, paraffin-embedded tissues. Five-micrometer sections were used for routine H&E and immunohistochemical stainings (avidin-biotin-peroxidase complex method), using the following monoclonal (mc) and polyclonal (pc) antibodies: α-smooth muscle actin (α-SMA; mc, 1/100; DAKO), desmin (mc, 1/20; ICN), CD31 (mc, 1/50; DAKO), CD34 (mc, 1/10; Becton Dickinson), MDM2 (mc, 1/100; Invitrogen, Life Technologies), CD117 (pc, 1/250; DAKO), EGFR (EGFR PharmDx Kit, DAKO), and human epidermal growth factor receptor 2 (HER2; HercepTest, DAKO). EGFR and HER2 protein expression was reported as membranous brown staining of neoplastic cells using a three-tier system ranging from 1+ (weak intensity) to 3+ (strong intensity).

Karyotyping was performed on primary tumor cells from three cases after 7 days in culture.

Genomic DNA was isolated from eight frozen tumor tissues using the High Pure PCR Template Preparation Kit (Roche Diagnostics). The Agilent Human Genome Microarray Kit 244A (Agilent Technologies) was used to perform array-CGH (aCGH) analysis. Labeling, hybridization, washing, and scanning were carried out in accordance with the protocols provided by the manufacturer. The Agilent Feature Extraction software (version 10.5.1.1) was used to process the images obtained from the scanner (Agilent DNA microarray scanner, with up to 2-μm precision) and to generate a text file (FE file). Subsequently, the FE files were analyzed using the Agilent Genomic Workbench Standard Edition 5.0.14 software based on the HG18 genome build. The ADM-2 algorithm was chosen for reporting copy number aberrations (CNA). Regions with a log 2 ratio >2 were considered as amplified, and regions with a log 2 ratio <−1 were considered as homozygously deleted. The data were submitted to the Gene Expression Omnibus database (accession no. 15834535).

Fluorescence in situ hybridization (FISH) analysis was performed on all 21 cases as previously described (27). Pretreatment and enzyme digestion were done using the SPoT-Light Tissue Pretreatment Kit (Invitrogen). Chromosome 4 and PDGFRA copy numbers were determined by cohybridization of Spectrum Green (SG)–labeled centromer-specific probe CEP4 (Vysis, Inc., Abbott Laboratories) and Spectrum Orange (SO)–labeled bacterial artificial chromosome (BAC) DNA probe RP11-231C18, which maps to PDGFRA (Research Genetics). For evaluation of the integrity of PDGFRA, BAC probes centromeric (RP11-3H20-SO) and telomeric (RP11-24O10-SG) to the PDGFRA locus (both from Research Genetics) were used. The coamplification of PDGFRA and MDM2 or EGFR on a cellular level was determined by cohybridization of PDGFRA/RP11-24O10-SO and MDM2/RP11-1064P9-SG or PDGFRA/RP11-24O10-SG and EGFR/RP11-231C18-SO labeled probes. The BAC DNA was isolated and labeled with fluorochrome dyes using standard techniques. Detection was performed as previously described (27). In addition, the MDM2, EGFR, and HER2 gene copy numbers were investigated using the double-color locus-specific identifier (LSI) MDM2-SO/CEP12-SG (Applied Biosystems/Ambion, Life Technologies), LSI EGFR-SO/CEP7-SG (Vysis), and LSI HER2-SO/CEP17-SG (Vysis) probes, respectively, according to the manufacturers' recommendations. Slides were counterstained with 0.1 μmol/L 4,6-diaminido-2-phenylindole in an antifade solution and viewed under a fluorescence microscope equipped with an ISIS digital image analysis system (MetaSystems). Signal copy numbers were counted from 100 nonoverlapping nuclei from at least three different areas and the percentage of affected nuclei was recorded. Chromosome polysomy was defined as >4 gene signals per nucleus paralleled by similar increases in chromosome centromere signals in at least 10% of tumor cells. A gene/CEP ratio of >2 in at least 10% of tumor cells was defined as specific gene amplification.

Mutational analysis was performed on genomic DNA extracted from frozen tumor tissues (n = 6). The whole coding sequence of the PDGFRA gene and the sequences coding for the juxtamembrane and/or kinase domains of PDGFRB and EGFR were amplified. Specific primers for amplified fragments are listed in Supplementary Table S1. The PCR products were purified and bidirectionally sequenced using the ABI PRISM 3130 XL Genetic Analyzer (Applied Biosystems).

The expression and activation of RTKs and their downstream signaling pathways were analyzed using the Proteome Profiler Array kits (R&D Systems) according to the manufacturer's protocol and using 500 μg of protein lysate per array. The images were captured and the level of RTK activation was densitometrically quantified with AIDA software (TBC Software, Inc.). The intensity values of the probes and the local background of the probes were log 2 transformed to obtain a more symmetrical distribution, and the difference between these two resulted in a log 2 transformed ratio (further referred to as log 2 intensity ratios). Subsequently, the mean of the log 2 intensity ratios for each kinase within an array was calculated. Probes with a log 2 intensity ratio higher than the mean plus SD were considered as significantly activated.

Tumor specimens (n = 7) were snap-frozen at −80°C and used for Western immunoblotting. A primary GIST that harbored a PDGFRA-D842V mutation and the human GIST882 (28) and SQD9 squamous cell carcinoma cell lines (29) were incorporated as controls. Cell lysis, SDS-PAGE, and immunoblotting were carried out as previously described (30). Membranes were immunoblotted sequentially using rabbit antibodies against phospho-EGFR (Tyr845; R&D Systems), total EGFR (Santa Cruz Biotechnology), phospho-PDGFRA (Tyr754; Santa Cruz Biotechnology), total PDGFRA (Santa Cruz Biotechnology), phospho-ERK1/2 (Cell Signaling), total ERK1/2 (Invitrogen, Life Technologies), phospho-AKT (Cell Signaling), total AKT (Cell Signaling), and total actin (Sigma-Aldrich), diluted in 5% blocking reagent. The horseradish peroxidase–conjugated antirabbit IgGs were used for enhanced chemiluminescence visualization (Thermo Scientific).

Primary cells from a collagenase-disaggregated tumor specimen from case no. 1 were used for the cell cultures, as previously described (31). In short, cells were grown for 3 days in DMEM supplemented with 10% fetal bovine serum, MITO⁺ serum extender (used according to the manufacturer's recommendation; BD Biosciences), 30 μg/mL bovine pituitary extract (BD Biosciences), 0.1 mmol/L nonessential amino acids, and 1.0 mmol/L sodium pyruvate. Cells were exposed to imatinib, dasatinib, or vehicle alone for 2 hours at 37°C. After a wash in ice-cold PBS, cells were lysed and used for Western immunoblotting as described above.

The inhibitors imatinib (Glivec/Gleevec, Novartis) and dasatinib (BMS-354825, Bristol-Myers Squibb) were provided by MedImmune, Inc. The inhibitors were diluted in culture medium directly before use.

On histology, the tumors showed a variable morphology, ranging from spindle cell proliferations with variable cellularity to islands of high anaplastic pleomorphic tumor cells with atypical mitotic figures (Supplementary Fig. S1). Low-grade– and high-grade–appearing areas were simultaneously present in all tumors. The matrix was usually collagenous, but on occasion was very myxoid; in one case, osteoid production was observed. Immunohistochemistry revealed focal and weak expression of CD31, CD34, and desmin in 10% to 20% of cases (Table 1). About half of the tumors investigated were α-SMA positive. EGFR was expressed in all but one case, although the staining pattern was heterogeneous, from strongly diffuse (3+; n = 6) to intermediate (2+; n = 7) or low and focal (1+; n = 7). MDM2 staining was positive in 52% of examined cases. CD117 (KIT) expression was rare and occurred in only two tumors mainly as focal and weak staining.

All three IS tumors featured a polyploid range of chromosome number and complex structural rearrangements, with the presence of homogeneously staining regions within unidentified markers in two cases (nos. 1 and 4) and double minutes in the third case (no. 19; data not shown).

Using high-resolution oligonucleotide aCGH, we studied CNAs in eight cases for which frozen tissue was available. There was a median of 64 gains (range, 36–83) and 42 losses (range, 13–63) per tumor. CNAs affecting six or more tumors (≥75% of cases) were identified on chromosomes 3q, 4q, 7p, 8q, 9p, 10q, and 12q (Table 2; Fig. 1). Major segments of high-level amplification were localized within chromosomes 1, 3, 4, 6, 8, 9, 12, 16, 21, 22, and X (Supplementary Table S2). Six of eight tumors showed chromosome 4q12 amplification, with the common region (54.78–55.21 Mb) containing only the PDGFRA gene (Fig. 2). The region at 12q15 (68.32–68.45 Mb) was amplified in 50% of the cases; the common region of amplification encompassed the BEST3 gene (the encoded protein is a Ca-activated Cl channel candidate in the development of vascular smooth muscle cells; ref. 32). Three chromosomal regions were amplified in 37.5% of cases, that is, the chromosomal regions 12q14.1 (KUB3), 12q15 (67.43–67.79 Mb; containing MDM2, SLC35E3, and CPM genes), and 22q11.23 (GSTT1). Four additional regions, 8p11.23-p11.22, 12q13.3-q14, 12q14.2-q14.3, and Xp21.1, were amplified in two cases each. Seventeen other amplified regions, which harbor known cancer-related genes, were observed in single cases.

Homozygous deletions were found on chromosomes 3, 5, 6, 7, 9, 10, 12, 13, 18, 20, and 22 (Supplementary Table S2). The homozygous deletions at 9p21.3-p21.2 (21.89–25.65 Mb), covering the CDKN2A/CDKN2B loci, were found in three of eight tumors. In addition, loss of one copy of CDKN2A/CDKN2B loci was revealed in three additional cases. This finding indicates that loss of CDKN2A/CDKN2B, although not tumor type specific, may be important for IS development and/or progression. The three chromosomal sub-bands—3p26.2 (ITPR1), 3p14.1 (FOXP1), and 18q22.3 (FBXO15)—were homozygously lost in two cases each. Twelve other homozygous deletions were found in one case each, and some of these deletions harbor known putative tumor suppressor genes.

The eight tumors analyzed with 244A Agilent array and 13 additional paraffin-embedded IS were evaluated by FISH (Table 1; Fig. 3) using specific probes for PDGFRA, EGFR, MDM2, and HER2. The PDGFRA amplification was the most prevalent change, found in 17 of 21 (81%) cases (Fig. 3A). Notably, two tumors (nos. 2 and 7) that showed focal PDGFRA amplification (from 12% to 18% affected cells) by FISH did not reveal this amplification by the parallel aCGH analysis (Figs. 2 and 3B). This finding indicates that the FISH technique is more suitable than aCGH for detecting the focal CNA in the context of cytogenetically heterogeneous and polyploid tumors. Three other IS showed high-level polysomy of PDGFRA, and only one case did not reveal any CNA of PDGFRA. Importantly, none of the examined cases showed a rearrangement of PDGFRA.

The MDM2 gene was amplified in a large subset of tumors in our series (11 of 17 cases; 65%) and one case showed MDM2 gene polysomy.

In total, 16 (76%) tumors showed either EGFR amplification (n = 2) or EGFR aneuploidy (n = 14). Additionally, the status of the gene encoding HER2, a close family member and an important dimerization partner of EGFR, was investigated. There was polysomy of HER2 in one case. By immunohistochemistry, strong EGFR-positive staining tended to correlate with EGFR gene amplification and aneuploidy, suggesting that the amplified EGFR is transcribed and translated to protein.

In the subsequent step of FISH analysis, we have explored the colocalization of PDGFRA, EGFR, and MDM2 amplicons on a cellular level in three cases (nos. 1, 4, and 19). Coamplification of PDGFRA with MDM2 or EGFR was observed as partially overlapping large-clustered amplicons and/or as multiple, scattered signals (Fig. 3C and D). Interestingly, subpopulations of tumor cells with mutual or exclusive amplification of PDGFRA and MDM2 or EGFR coexisted in each of the examined specimens.

No activating mutations were identified in PDGFRA/B and EGFR in the six IS cases (nos. 1, 2, 3, 7, 9, and 10) analyzed.

All seven analyzed tumors revealed strongly coactivated PDGFRA and EGFR kinases (Fig. 4). Six cases concurrently showed phosphorylation of PDGFRB. Four cases concomitantly showed low-level phosphorylation of EPHB2, and three cases also had low-level phosphorylation of AXL. Strikingly, there was no detectable activation of KIT or vascular endothelial growth factor receptors, nor activation of the members of the SRC-family proteins. By analyzing the signaling pathways (the profiles of 46 kinases and protein substrates), CREB, eNOS, TP53, JUN, STAT5B, and AMPKα1 were found to be the most frequently and strongest phosphorylated proteins (Supplementary Table S3; Fig. 4).

The consistent protein expression and activation of PDGFRA and EGFR were confirmed by Western immunoblotting or immunohistochemistry (Supplementary Fig. S2; Table 1). By immunoblotting, both the AKT and ERK1/2 downstream signaling pathways were activated, the latter to a much stronger level, which is in line with PDGFR/EGFR–driven tumors (31, 33, 34).

We examined the consequences of PDGFRA inhibition with imatinib and dasatinib in primary IS cultures from case no. 1. Both inhibitors caused a dose-responsive decrease in PDGFRA phosphorylation (Fig. 5). Dasatinib completely blocked the phosphorylation of the target PDGFRA at 0.1 μmol/L. In contrast, 0.5 μmol/L imatinib only slightly affected PDGFRA activation, and 5.0 μmol/L imatinib was needed to completely abrogate PDGFRA phosphorylation. Hence, dasatinib inhibited PDGFRA at much lower (clinically achievable) concentrations. Using p-AKT and p-ERK1/2 proteins as molecular surrogates of downstream RTK signaling, we found that both the AKT and ERK1/2 pathways were activated in untreated IS primary cells. Importantly, although PDGFRA activation was completely abolished with 5 μmol/L imatinib, the downstream signaling pathways were still active. In contrast, complete inactivation of PDGFRA was associated with substantial inactivation of both ERK1/2 and AKT with 0.5 μmol/L dasatinib. This suggests that other dasatinib targets, next to PDGFRA, might be responsible for the inactivation of downstream ERK1/2 signaling by dasatinib.

Using a variety of methodologic approaches, we aimed for a comprehensive characterization of IS in search for relevant diagnostic biomarkers and therapeutic targets.

As expected, high-resolution 244A Agilent aCGH revealed complex CNA in all examined cases across the genome, including previously recognized oncogenes and tumor suppressor genes, as well as completely novel loci. Most importantly, high-level amplifications or copy number gains frequently involved chromosomal regions that contain the PDGFRA, EGFR, and/or MDM2 genes.

Accordingly, PDGFRA amplification and high-level polysomy by FISH were present in 81% and 15% of tumors, respectively. Remarkably, using high-resolution array, we were able to delineate the common region of 4q12 amplification to 54.78 to 55.21 Mb, containing only the PDGFRA locus. Thus, PDGFRA is the most likely target for amplification in IS. Our findings are comparable to the results reported previously on smaller IS cohorts (24, 25). To the best of our knowledge, no other sarcoma reported thus far presents consistent PDGFRA amplification (35). Thus, given its specificity and high frequency of occurrence, the PDGFRA amplification in IS might be considered as a molecular hallmark of the entity. Importantly, PDGFRA amplification, detected by FISH, is a valuable marker for IS diagnosis, distinguishing IS from the other undifferentiated sarcomas known to reside in similar anatomic sites.

The amplification of the 12q15 region, harboring the MDM2 locus, was another consistent and frequent feature revealed by aCGH. Amplification and consequent upregulation of the MDM2 oncogene represent a recurrent molecular feature of many sarcomas, such as de-differentiated liposarcomas (36), and it was also reported in IS (14, 24, 25). The extended FISH analysis revealed that 65% of tumors have MDM2 amplification in our cohort. MDM2 amplification was closely correlated with the protein expression of this oncogene, as evidenced by immunohistochemistry. Because the amplification or overexpression of MDM2 has been found in many cancers, inhibiting MDM2 and thereby reactivating TP53 in cancer cells is an emerging new therapeutic strategy. Nutlin-3 and MI-219 are MDM2 inhibitors with desirable pharmacologic properties and are in advanced preclinical development or early-phase clinical trials (37). Interestingly, cancer cell lines with MDM2 gene amplification were shown to be more sensitive to Nutlin-3 than cell lines lacking this amplification (38).

The gain of chromosome 7p, encompassing the EGFR locus, was accounted for in six of eight cases analyzed by aCGH in the present study. High-level EGFR aneuploidy (or rarely amplification) was confirmed by FISH (76% of cases). EGFR belongs to the ERBB family of RTKs, which recognize at least 11 known ligands (39, 40). There is ample evidence of amplified and activated forms of EGFR in many human cancers, and drugs that specifically target EGFR show promise in inhibiting the growth of certain tumor types (39). Sato and colleagues showed that 168 of 281 (60%) adult soft tissue sarcomas overexpressed EGFR (41); this overexpression significantly correlated with higher histologic grade, poorer survival, and chemoresistance. It is noteworthy that gefitinib, an EGFR inhibitor, has shown potential antitumor effects in several sarcoma cell lines when used in combination with irinotecan (42).

To define the compendium of coactivated RTKs in IS, we used an antibody array that allows simultaneous assessment of the phosphorylation status of 42 RTKs. Of utmost importance and coherent with the crucial biological role for the IS pathogenesis, the PDGFRA and EGFR kinases were consistently activated in all seven investigated cases of our study. The PDGFRB kinase was activated in a smaller subgroup of tumors. The coactivation of multiple RTKs is not a distinctive feature of IS because similar patterns were reported in other tumor types (43–46). Importantly, the concomitant activation of multiple RTKs serves to reduce dependence on a single RTK for the maintenance of critical downstream signaling and, thus, renders such tumors refractory to single-agent RTK inhibition.

The recurrent high-level amplification and strong activation of PDGFRA in absence of resistance mutations prompted us to investigate the effect of the PDGFRA inhibitors imatinib and dasatinib in IS. Both inhibitors were able to specifically inhibit PDGFRA kinase activity, but dasatinib showed higher potency. The stronger inhibitory capacities of dasatinib toward PDGFR in reference to imatinib were previously shown in rat A10 vascular smooth muscle cells (47). Additionally, dasatinib showed a more effective inhibition of downstream ERK1/2 and AKT signaling than did imatinib in primary IS tumor cells. It is likely that IS cells under PDGFR-targeted treatment are still able to maintain downstream signaling through the ERK1/2 and AKT pathways due to EGFR coactivation. By RTK array, the propagated ex vivo tumor cells lacked activation of SRC family members. Thus, the possibility that inhibition of SRC-family kinases could contribute significantly to the inhibitory effect of dasatinib on the PDGFR downstream signaling pathway is remote. Yet, the discrete SRC activation to levels not discernible by our technique could still exist in the tested cells, rendering them more prone to dasatinib inhibition. Interestingly, it was recently shown in lung cancer that dasatinib may reduce EGFR activation through the inhibition of SRC, which mediates the phosphorylation of the Y845 residue in the receptor (48–50).

It is essential to emphasize that the success of the therapy in IS might strongly depend on the genetic context of the tumor, which is complex, as indicated by the numerous CNAs of cancer-related genes identified by aCGH. The CDKN2A/CDKN2B losses were recurrent in IS in our cohort. This is in line with previous studies that have shown a synergistic effect on neoplasia induction when aberrant oncogenic PDGFRA signaling is combined with loss of the tumor suppressors CDKN2A/CDKN2B (3, 47). Moreover, the TP53 pathway seems to be frequently affected in IS (e.g., common MDM2/CDK4 amplifications), pointing to confounding factors that might attenuate the response to RTK inhibition.

Second, an important and hitherto unreported finding in our study is the concomitant existence of cells with exclusive amplification and cells with coamplification of PDGFRA, EGFR, or MDM2 within a single tumor, as detected by FISH. The mechanism leading to gene coamplification remains to be clarified. In breast cancer, the coamplification most likely arises by the selection of clones that have both foci amplified rather than as a direct result of prior fusions (51). In our study, the populations with coamplification and single amplifications seem to be able to coexist independently. It is equally important to keep in mind that selection of certain populations might happen once targeted treatment is started. This provides a rationale for multitargeted treatment of IS.

In conclusion, PDGFRA amplifications are common and consistently associated with the activation of this gene in IS. Activation of EGFR is concurrent to PDGFRA activation and may coexist with amplification and overexpression of MDM2. Our data provide a rationale for the targeted treatment of IS with specific PDGFRA, EGFR, and MDM2 inhibitors. Dasatinib may be more efficient than imatinib for the inhibition of activated PDGFRs, but the effective therapy for these tumors may require combined regimens targeting multiple RTKs and/or downstream pathway regulatory effectors.

No potential conflicts of interest were disclosed.

We thank Dr. Pascale Cervera (Hôpital Saint Antoine, Paris, France) for providing tumor tissue biopsy samples for this study.

Grant Support: Fonds voor Wetenschappelijk Onderzoek Vlaanderen grant G.0589.09 (M. Debiec-Rychter) and a Concerted Action Grant (2006/14) from the K.U. Leuven. P. Vandenberghe is a senior clinical investigator of FWO.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.