Article Tools
ORIGINAL REPORTS
Sarcomas
Article Tools
Overcoming Resistance to Conventional Drugs in Ewing Sarcoma and Identification of Molecular Predictors of Outcome
Abstract
The improvement of Ewing sarcoma (EWS) therapy is currently linked to the discovery of strategies to select patients with poor and good prognosis and of modified treatment regimens. In this study, we analyzed the molecular factors governing EWS response to chemotherapy to identify genetic signatures to be used for risk-adapted therapy.
Microarray technology was used for profiling 30 primary tumors and seven metastases of patients who were classified according to event-free survival. For selected genes, real-time polymerase chain reaction was applied in 42 EWS primary tumors as validation assay. 3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide test was used to evaluate in vitro drug sensitivity.
We identified molecular signatures that reflect tumor resistance to chemotherapy. Annotation analysis was applied to reveal the biologic functions that critically influenced clinical outcome. The prognostic relevance of glutathione metabolism pathway was validated. The expression of MGST1, the microsomal glutathione S-transferase (GST), was found to clearly predict EWS prognosis. MGST1 expression was associated with doxorubicin chemosensitivity. This prompted us to assess the in vitro effectiveness of 6-(7-nitro-2,1,3-benzoxadiazol-4-ylthio)hexanol (NBDHEX), a new anticancer agent that efficiently inhibits GST enzymes. Six cell lines were found to be sensitive to this new drug.
Classification of EWS patients into high- and low-risk groups is feasible with restricted molecular signatures that may have practical value at diagnosis for selecting patients with EWS who are unresponsive to current treatments. Glutathione metabolism pathway emerged as one of the most significantly altered prognosis-associated pathway. NBDHEX is proposed as a new potential therapeutic possibility.
Ewing sarcoma (EWS) is the second most common malignant bone tumor. Despite its rareness (1/million total population), it has an important impact on the health system because it generally occurs in children and young adults.
Because of its aggressiveness, therapy for EWS requires either surgery and/or radiation therapy for local control, along with intensive chemotherapy to treat micrometastasis. These multimodal therapies dramatically improve survival of patients with localized EWS at diagnosis to 50% to 70%.1–3 Nevertheless, the increase in long-term survival is obtained at the price of high toxicity and the considerable risk of developing a second disease,4,5 two important life-threatening events considering the young age of patients with EWS. In addition, prognosis is still disappointingly low (approximately 20% overall survival) for the patients who present clinically detectable metastases at diagnosis,6 and a comparable poor outcome is shared by the significant group of patients who experience local and/or distant recurrent disease.7 Therefore, it is imperative to identify prognostic factors to detect chemotherapy-resistant tumors at diagnosis and to generate more individualized treatment regimens. However, at present, only clinical features, such as patient age, presence of clinically evident metastases at diagnosis, and poor response to neoadjuvant chemotherapy are widely accepted as prognostic indicators in EWS.8–11
Nowadays, techniques like microarray hybridization provide an opportunity to perform genome-wide analysis at different levels. Several recent articles, focusing on expression profiles of EWS cell lines, have given new and precious clues on molecular mechanisms related to the origin, development and progression of this disease.12–16 However, only two reports, so far, correlate genetic information with patient outcome.17,18 In this study, 30 EWS primary tumors, treated in a single Institution, were globally profiled for gene expression by using an Affymetrix GeneChip Expression Analysis platform (Affymetrix, Santa Clara, CA) to identify molecular signatures that clearly distinguish at diagnosis patients with EWS who do not respond to chemotherapy and quickly experience relapse (median, 12 months) from patients with a long-lasting favorable prognosis. Annotation analysis was applied to reveal the biologic functions that critically influenced the clinical outcome. Real-time polymerase chain reaction (PCR) validated the prognostic relevance of glutathione metabolism pathway, which emerged as one of the most significantly altered prognosis-associated pathways. The expression of MGST1, the membrane-bound microsomal glutathione S-transferase (GST), was found to clearly predict EWS prognosis and to be associated with doxorubicin chemoresistance. This prompted us to assess the in vitro effectiveness of 6-(7-nitro-2,1,3-benzoxadiazol-4-ylthio)hexanol (NBDHEX), a new anticancer agent that efficiently inhibits GST enzymes.19
Forty-two EWS frozen samples from primary lesions and seven EWS metastases (four lung and three bone metastasis) were collected from the tissue bank of the Laboratory of Oncologic Research, Rizzoli Institute, Bologna, Italy. A pathologist panel reviewed histology, and the presence of EWS/FLI or EWS/ERG translocation was confirmed by reverse-transcriptase (RT) PCR. Patients underwent three consecutive programs of multidrug chemotherapy20 and local treatments (surgery; surgery plus radiotherapy; radiotherapy only). After completion, patients were continuously observed and clinical data were updated (median follow-up, 7.4 years). Adverse events were defined as recurrence of the tumor at any site or death during remission, and event-free survival was calculated from the date of initial diagnosis. Clinical data are listed in Table 1. Sample sets were handled in a coded fashion. The ethical committee of the Rizzoli Institute approved the studies, and informed consent was obtained from all subjects involved.
|
| Characteristic | MicroArray (n = 30) | Real-Time PCR (n = 42) | ||
|---|---|---|---|---|
| No. | % | No. | % | |
| Sex | ||||
| Male | 19 | 63 | 25 | 60 |
| Female | 11 | 37 | 17 | 40 |
| Age, years | ||||
| ≤ 14 | 9 | 30 | 15 | 36 |
| > 14 | 21 | 70 | 27 | 64 |
| Histologic variant | ||||
| Ewing | 23 | 77 | 33 | 79 |
| PNET | 6 | 20 | 8 | 19 |
| Askin | 1 | 3 | 1 | 2 |
| Tumor site | ||||
| Femur | 11 | 37 | 14 | 33 |
| Humerus | 3 | 10 | 5 | 12 |
| Tibia | 3 | 10 | 4 | 10 |
| Fibula | 2 | 7 | 2 | 5 |
| Pelvis | 5 | 17 | 6 | 14 |
| Other | 6 | 20 | 11 | 26 |
| Location | ||||
| Extremity | 22 | 73 | 28 | 67 |
| Other | 8 | 27 | 14 | 33 |
| Surgery | ||||
| Resection | 19 | 63 | 30 | 71 |
| Amputation | 2 | 7 | 2 | 5 |
| Not done | 9 | 30 | 10 | 24 |
| Treatment | ||||
| Combined | 21 | 70 | 32 | 76 |
| Chemotherapy and radiotherapy | 9 | 30 | 10 | 24 |
| Radiotherapy | ||||
| Done | 14 | 47 | 19 | 46 |
| Not done | 16 | 53 | 22 | 54 |
| Histologic response | ||||
| Total necrosis | 3 | 14 | 4 | 14 |
| Nontotal necrosis | 18 | 86 | 28 | 86 |
| Fusion transcripts | ||||
| EWS/Fli1 type 1 | 18 | 60 | 24 | 57 |
| EWS/Fli1 non type 1 | 12 | 40 | 18 | 43 |
| Event-free survival outcome | ||||
| Relapsed | 17 | 57 | 23 | 55 |
| NED | 13 | 43 | 19 | 45 |
| Final outcome | ||||
| Dead | 14 | 47 | 18 | 43 |
| NED | 16 | 53 | 24 | 57 |
Abbreviations: PCR, polymerase chain reaction; PNET, primitive neuroectodermal tumor; EWS, Ewing sarcoma; NED, no evidence of disease.
Analysis of CD34+ cells and EWS clinical samples was performed according to the Affymetrix GeneChip Expression Analysis Technical Manual using HG-U133 Plus 2.0 Array. Affymetrix Microarray Analysis Suite 5.0 was used for chip-to-chip normalization and gene expression value determination. A complete description of the microarray analysis is provided as in the Appendix (online only). Microarray data are available at Gene Expression Omnibus21 public database with the access number GSE12102. The microarray data of two bone marrow–derived mesenchymal stem cell (BM-MSC) samples were taken from Gene Expression Omnibus with the accession number GSE2248/GSM38627, GSE7007/GSM161538.15,22
RNA extraction and RT-PCR of clinical samples and cell lines were performed as described previously.23 Predesigned TaqMan probes and primers sets for genes were chosen, and three replicates per gene were considered. Samples were analyzed using an ABI Prism 7900 Sequence Detection System (Applied Biosystems, Foster City, CA), according to manufacturer's instructions. Expression levels of target genes were normalized to that of glyceraldehyde 3-phosphate dehydrogenase, and the relative quantification analysis was performed on the basis of the ΔΔCT method. cDNA from human bone marrow CD34+ cells was used as calibrator for the comparative analysis.
The EWS cell lines TC-71, 6647, SK-N-MC, SK-ES-1, and RD-ES were obtained from the American Type Culture Collection (Rockville, MD). LAP35, IOR/BRZ, IOR/CLB, IOR/CAR, IOR/NGR, and IOR/RCH were obtained and characterized at the Laboratory of Oncologic Research, Rizzoli Institute.
Doxorubicin and vincristine were purchased from Sigma-Aldrich (Milan, Italy). NBDHEX was synthesized as reported by Ricci et al.19 Cells (2,500/well) were seeded in Iscove modified Dulbecco medium 10% fetal bovine serum, and after 24 hours, the medium was changed without (control) or with a different drug concentration. After 72 hours, cytotoxicity was evaluated by 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide assay (Roche Applied Science, Basel, Switzerland). Sensitivity to different drugs was expressed as drug concentration resulting in 50% inhibition of cell growth.
Global array processing and following analyses were performed using R Bioconductor24 and MATLAB (MathWorks Inc, Natick, MA) software packages. Data were rescaled within each array and between arrays through an affine transformation.25 Unsupervised biclustering (using Pearson correlation metrics) and principal component analysis (PCA) were performed. A one-way analysis of variance as well as quadratic discriminant analysis with leave-one-out cross-validation were performed on the data set. Biologic pathways were defined according to the Kyoto Encyclopedia of Genes and Genomes (KEGG)26 annotation, and the mapping between probes and pathways was accomplished by querying the KEGG Database via R software. Pathway significance analysis was calculated considering the hypergeometric distribution. Differences among means were analyzed using t test. Kaplan-Meier and log-rank methods were used, respectively, to draw and evaluate the significance of survival curves in patients with EWS. To define drug-drug interactions (in terms of synergism, additivity, or antagonism), the combination index of each two-drug treatment was calculated with the isobologram equation,27 by using CalcuSyn software (Biosoft, Stapleford, United Kingdom).
The study included 30 primary tumors from patients with localized EWS. As a first step, PCA was performed onto EWS samples to check for data set homogeneity. As normal control, we used data sets derived from three different stem cells, one of hematopoietic origin (CD34+) and two BM-MSCs. PCA was performed onto 22,277 probes, resulting from the intersection of the probes found in U133a (BM-MSC samples) and U133 Plus 2 (EWS samples and CD34+ cells) Affymetrix arrays. We noted the good homogeneity of the EWS data set compared with the reference samples (Fig 1) and confirmed that EWS form a quite homogeneous group of tumors.
Fig 1. Scatterplot of principal component (PC) analysis performed on 33 samples: 30 Ewing sarcoma samples, CD34+ cells (CD34+ SC), and two bone marrow–derived mesenchymal stem-cell (BM MSC) samples. NED, no evidence of disease; REL, relapse.
To identify genes related to treatment response, we then compared the gene expression profiles of 17 tumors from patients who had tumor progression within 3 years from diagnosis (median relapse time, 12 months; range, 3 to 33 months; defined as patients who experienced a relapse [REL]) with those obtained from 13 tumors from patients who did not experience recurrence (median follow-up, 89 months; range, 36 to 161 months; defined as no evidence of disease [NED]). Using relatively low stringency statistical criteria, we identified 2,092 probes with significant changes in expression levels between NED and REL tumors (P < .05; one-way analysis of variance test). These 2,092 probes sets represented 1,358 unique genes according to Affymetrix annotation database. Their expression clearly distinguished NED from REL tumors, either with PCA analysis (Fig 2A) or hierarchical unsupervised clustering (Fig 2B). Analysis using KEGG annotations and the hypergeometric-based method identified a number of pathways that were differentially expressed in NED versus REL tumors (Table 2). Twelve pathways were found to be overrepresented (P < .05). As expected for genes that should reflect a different tumor response to chemotherapy, seven pathways are related with metabolism (glutathione, cyanoamino acid, taurine and hypotaurine, arachidonic acid, pyruvate metabolism) or biosynthesis (monoterpenoid or fatty acid biosynthesis). The others are related to gap junction or to pathways related to nervous system development and functions.
Fig 2. Scatterplot of (A, C, E) principal component (PC) analysis and (B, D, F) unsupervised hierarchical clustering performed on different subsets of the no evidence of disease (NED)/relapse (REL) data set. (A, B) Analysis performed on 2,092 probes (P ≤ .05). (C, D) Analysis performed on 260 probes (P ≤ .05, fold change ≥ 2, average probe intensity ≥ 10). (E, F) Analysis performed on a subset of 35 probes.
|
| Pathway | P | Probe Sets in Pathway | Significant Probe Sets |
|---|---|---|---|
| Glutathione metabolism | .0000039 | 74 | 13 |
| Cyanoamino acid metabolism | .0003634 | 26 | 6 |
| Taurine and hypotaurine metabolism | .0005573 | 28 | 6 |
| Arachidonic acid metabolism | .0015845 | 116 | 12 |
| Olfactory transduction | .0026458 | 92 | 10 |
| Gap junction | .0096549 | 308 | 21 |
| Amyotrophic lateral sclerosis | .0108459 | 64 | 7 |
| Proteasome | .0139396 | 52 | 6 |
| Fatty acid biosynthesis | .0253844 | 17 | 3 |
| Pyruvate metabolism | .0258213 | 93 | 8 |
| Monoterpenoid biosynthesis | .0351119 | 8 | 2 |
| Renin-angiotensin system | .0396476 | 34 | 4 |
Abbreviation: KEGG, Kyoto Encyclopedia of Genes and Genomes.25
To obtain a molecular signature of potential value for clinical practice, we applied a more selective filter: a fold change ≥ 2 was required, and a minimum expression value (averaged over the whole data set) was required for each probe (Pi ≥ 10; Pi = mean expression of the i-th probe over the 30 samples of the data set). A total of 260 probe sets, representing 169 genes, were selected by this procedure (Figs 2C and 2D). Diagonal quadratic discriminant analysis with leave-one-out cross-validation was then performed over subsets of the 260-probe set: a core subset of 35 probes, representing 20 genes, was selected as the best probe set for classification, with 100% (30 of 30 samples) correct classification performance. Moreover, this restricted molecular signature was able to separate REL from NED tumors either by PCA analysis (Fig 2E) or by hierarchical clustering (Fig 2F).
We further tested whether the poor- and good-prognosis signatures that we identified can classify EWS metastasis to one of the prognostic groups or as a distinct group. We confirmed that the metastasis gene expression profile did not differ from the one observed for primary tumors. In contrast, all the metastases were classified in the poor-prognosis signature group with all the three signatures, with the only exception being one metastasis when the most restricted signature was applied (Fig 3).
Fig 3. Scatterplots of the first two principal components for unsupervised classification of primaries (no evidence of disease [NED] or relapse [REL]) and metastasis (MET) performed on the different signatures (Fig 2). (A) A total of 2,092 probes (P ≤ .05). (B) A total of 260 probes (P ≤ .05, fold change ≥ 2, average probe intensity ≥ 10). (C) Thirty-five probes selected with quadratic discriminant analysis–based method.
The pathway represented by the highest number of genes involved, compared with the total number of genes composing the pathway, is the one related to glutathione metabolism. Thirteen probe sets, representing seven genes, were differentially expressed in NED versus REL tumors. We thus validated the expression of these genes (GCLM, GCT1, GGTL3, ANPEP, GSTP1, GSTA3, MGST1) by real-time PCR in a larger series of 42 EWS primary tumors, and we evaluated whether their expression can predict clinical outcome. GSTA3 was found to be generally undetectable and was excluded from further analysis. For the other six genes, we calculated median expression value, and patients were stratified as high expressors (H) or low expressors (L) relative to the median value. Only MGST1 was found to be significantly related with EFS (Fig 4A), with low expressors showing a better prognosis, in keeping with the microarray data. Upregulation of ANPEP in long-term survivors was also in line with the microarray data, but this did not achieve statistical significance (data not shown). Of note, although all the other genes showed significant correlations among them (r ≥ 0.6; P < .001, Spearman correlation test), MGST1 did not, indicating a different role. We also evaluated the expression of MGST1 and GSTP1 in a panel of EWS cell lines (CD34+ mRNA was used as calibrator; Fig 4B) and confirmed the absence of correlation between the expression of the two enzymes. However, the relative mRNA expression of MGST1, but not of GSTP1, was correlated with the level of sensitivity of EWS cell lines to doxorubicin (r = 0.98, P = .002, Spearman correlation test), which is one of the most important drugs in the treatment of EWS (Fig 4C). This prompted us to assess the in vitro effectiveness of NBDHEX, a new anticancer agent that efficiently inhibits GST enzymes. Six EWS cell lines were found to be sensitive to this new drug (Fig 4C), including those with higher MGST1 levels, indicating that NBDHEX may be considered for the treatment of patients with EWS that is refractory to other conventional drugs. Additive effects were observed when NBDHEX is associated with DXR (Figs 4D and E; 0.90 ≤ combination index ≤ 1.10, according to Chou et al27).
Fig 4. (A) Prognostic impact of MGST1 on event-free survival (EFS) of 42 patients with primary Ewing sarcoma (EWS) according to Kaplan-Meier curves. (B) GSTP1 and MGST1 expression in EWS cell lines. (C) In vitro sensitivity of EWS cell lines to doxorubicin (DXR), vincristine (VCR), and 6-(7-nitro-2,1,3-benzoxadiazol-4-ylthio)hexanol (NBDHEX). (D) Isobologram analysis of combination of NBDHEX and DXR in two EWS cell lines. Combination index values are represented by points above, on, or below the lines, indicating antagonism, additivity, or synergy, respectively. (E) Sensitivity to NBDHEX in combination with DXR. (*), P < .05; (**), P < .001. A dose of NBDHEX that gives 20% to 30% growth inhibition was combined with increasing doses of DXR. Cum., cumulative; H, high expressors; L, low expressors; OS, overall survival; IC50, drug concentration resulting in 50% inhibition of cell growth; ED50, effective dose resulting in 50% growth inhibition; ED75, effective dose resulting in 75% growth inhibition; ED90, effective dose resulting in 90% growth inhibition.
The study reports the determination of patients with high-risk EWS and outcome prediction using high-resolution expression array profiling. We identified gene expression signatures that distinguished patients with quickly progressing EWS (median: 12 months from diagnosis) from those with a favorable prognosis of long-term disease-free survival (median, 89 months). None of the conventional clinicopathologic parameters (age, sex, site, response to treatment) were found to be associated with prognosis in our series of patients, indicating a superior statistical power of the genetic analysis that we performed. We propose here three different molecular signatures defined by progressive restrictive statistical criteria. The limited number of genes in the second and third signatures has a potential practical value because it can be easily managed also for routine purposes in pathology laboratories. The last signature includes only 20 genes and was particularly statistically robust. We offered these signatures to the scientific communities for cross-validation and meta-analysis, which are indispensable tools for a rare tumor such as EWS. Of note, all the three signatures that we propose were able to classify EWS lung and bone metastasis in the poor-prognosis subgroup, indicating that themetastasis profile is already present at diagnosis and is indistinguishable from the primary tumors that are unresponsive to therapy and develop metastasis earlier. Our data are in keeping with observations reported in other tumors28–30 and confirm that metastatic capacity is an inherent feature of EWS and not a late acquired feature.
Although many studies have been published in recent years indicating the ability of gene-expression microarray analysis to predict outcome in different tumors, only two articles directly correlated gene expression profile of primary EWS tumors with clinical outcome.17,18 Of these two, the study by Ohali et al17 used arrays with a more limited resolution and was performed only on 14 primary tumors. Nevertheless, the biologic information that could be drawn from these studies is highly compatible with that obtained in this study. Network analysis of our signature clearly indicated a role for Rac-1 gene in discriminating patients with poor or good prognosis. Researchers have recently found that the small GTPase Rac1 is a crucial component of the canonical Wnt signaling pathway,31 which was reported to be significantly modulated also in the study of Schaefer et al.18 Integrins and molecules involved in cell-cell or cell-extracellular matrix adhesion also emerged in all the three studies as central genes, the differential expression of which is associated with different clinical outcome. In this study, we focused our attention on the pathway related to glutathione metabolism and detoxification. We reasoned that because we are distinguishing patients who experience relapse early (during or soon after chemotherapy) from patients with favorable prognosis, we are in the process of identifying genes and pathways that are mainly related to drug unresponsiveness. Conventional markers of drug resistance, such as the multidrug resistance MDR-1 or MRPs genes, have never been found as significant predictors of EWS prognosis.11,32 Consistently, we did not observe these genes as differently expressed, but rather genes related to cellular glutathione-related detoxification system. GSTs are a family of detoxification enzymes that catalyze the conjugation of glutathione to a wide variety of endogenous and exogenous compounds. GSTs have been implicated in the development of resistance toward chemotherapeutic agents (see review in Townsend and Tew33). They act via direct detoxification as well as inhibitors of the MAP/Ras kinase pathway, which also has been reported to be a critical pathway for EWS.17,18 Quantitative PCR analysis of seven genes of this pathway indicated that MGST1, the microsomal GST, is a reliable predictor of prognosis. The gene was present with five different probe sets in the first and second signatures that we proposed here, indicating the soundness of the results, and low expression of MGST1 was found to be significantly associated with better prognosis. Of note, the expression of MGST1 was not correlated with that of the other components of the glutathione metabolism pathway, and particularly with other GSTs. The contribution of MGST1 to cellular drug resistance was only recently reported.34 Here, we confirm the role of this enzyme in cellular resistance to anticancer drugs by observing a strong correlation between the level of expression of MGST1 in EWS cell lines and their level of sensitivity to doxorubicin, one of the leading drugs in the treatment of EWS. Although further studies are necessary to establish the functional mechanisms of this still poorly understood enzyme in drug resistance, we propose this enzyme as a crucial mediator of sensitivity to EWS therapy.
By looking for drugs targeting GSTs, we focused on the new anticancer agent NBDHEX, which proved to be active in several human experimental tumors and was not extruded from tumor cells by multidrug resistance protein pumps.19,35,36 Although NBDHEX is still under preclinical in vivo evaluation, the low concentrations, which are necessary to exert cytotoxic effects on tumor cells, together with the low toxicity exhibited in mice and rats, have supported its potential clinical use.36 We showed that targeting GST isoenzymes by NBDHEX was effective on human EWS cell lines, independently from the relative level of expression of MGST1, thus demonstrating a new potentially interesting approach.
Taken together, our data show that classification of patients with EWS into high- and low-risk subgroups is feasible with restricted molecular signatures that can be practically and easily applied at diagnosis for selecting patients with EWS who are unresponsive to current treatments. Metastasis was classified in the same group of primaries with poor prognosis when the molecular signatures were applied. Glutathione-metabolism pathway was identified as a major pathway regulating EWS chemoresistance. For patients expressing high levels of MGST1 and showing worse prognosis, the new drug NBDHEX might be considered.
Supported by the European Project PROTHETS (Grant No. 503036), the Italian Association for Cancer Research (grant to K.S.), the Italian Ministry of Health (Alliance against Cancer to P.P.), and the Italian Ministry of University and Research (Grant No. PRIN 2007 to M.M.).
Authors' disclosures of potential conflicts of interest and author contributions are found at the end of this article.
The author(s) indicated no potential conflicts of interest.
Conception and design: Katia Scotlandi, Sakari Knuutila, Piero Picci
Provision of study materials or patients: Mario Mercuri, Anna Maria Caccuri
Collection and assembly of data: Katia Scotlandi, Maria Cristina Manara, Filippo Nardi, Lara Cantiani, Massimo Serra, Sakari Knuutila, Piero Picci
Data analysis and interpretation: Katia Scotlandi, Daniel Remondini, Gastone Castellani, Mirko Francesconi, Massimo Serra
Manuscript writing: Katia Scotlandi, Daniel Remondini
Final approval of manuscript: Katia Scotlandi, Daniel Remondini, Gastone Castellani, Maria Cristina Manara, Filippo Nardi, Lara Cantiani, Mirko Francesconi, Mario Mercuri, Anna Maria Caccuri, Massimo Serra, Sakari Knuutila, Piero Picci
| 1. | G Bacci, C Forni, A Longhi , etal: Long-term outcome for patients with non-metastatic Ewing's sarcoma treated with adjuvant and neoadjuvant chemotherapies. 402 patients treated at Rizzoli between 1972 and 1992 Eur J Cancer 40: 73– 83,2004 Crossref, Medline, Google Scholar |
| 2. | HE Grier, MD Krailo, NJ Tarbell , etal: Addition of ifosfamide and etoposide to standard chemotherapy for Ewing's sarcoma and primitive neuroectodermal tumor of bone N Engl J Med 348: 694– 701,2003 Crossref, Medline, Google Scholar |
| 3. | EA Kolb, BH Kushner, R Gorlick , etal: Long-term event-free survival after intensive chemotherapy for Ewing's family of tumors in children and young adults J Clin Oncol 21: 3423– 3430,2003 Link, Google Scholar |
| 4. | M Paulussen, S Ahrens, M Lehnert , etal: Second malignancies after Ewing tumor treatment in 690 patients from a cooperative German/Austrian/Dutch study Ann Oncol 12: 1619– 1630,2001 Crossref, Medline, Google Scholar |
| 5. | G Bacci, A Longhi, E Barbieri , etal: Second malignancy in 597 patients with Ewing sarcoma of bone treated at a single institution with adjuvant and neoadjuvant chemotherapy between 1972 and 1999 J Pediatr Hematol Oncol 27: 517– 520,2005 Crossref, Medline, Google Scholar |
| 6. | M Paulussen, S Ahrens, AW Craft , etal: Ewing's tumors with primary lung metastases: Survival analysis of 114 (European Intergroup) Cooperative Ewing's Sarcoma Studies patients J Clin Oncol 16: 3044– 3052,1998 Link, Google Scholar |
| 7. | G Bacci, S Ferrari, A Longhi , etal: Therapy and survival after recurrence of Ewing's tumors: The Rizzoli experience in 195 patients treated with adjuvant and neoadjuvant chemotherapy from 1979 to 1997 Ann Oncol 14: 1654– 1659,2003 Crossref, Medline, Google Scholar |
| 8. | M Bernstein, H Kovar, M Paulussen , etal: Ewing's sarcoma family of tumors: Current management Oncologist 11: 503– 519,2006 Crossref, Medline, Google Scholar |
| 9. | G Bacci, A Longhi, S Ferrari , etal: Prognostic factors in non-metastatic Ewing's sarcoma tumor of bone: An analysis of 579 patients treated at a single institution with adjuvant or neoadjuvant chemotherapy between 1972 and 1998 Acta Oncol 45: 469– 475,2006 Crossref, Medline, Google Scholar |
| 10. | SJ Cotterill, S Ahrens, M Paulussen , etal: Prognostic factors in Ewing's tumor of bone: Analysis of 975 patients from the European Intergroup Cooperative Ewing's Sarcoma Study Group J Clin Oncol 18: 3108– 3114,2000 Link, Google Scholar |
| 11. | RD Riley, SA Burchill, KR Abrams , etal: A systematic review and evaluation of the use of tumour markers in paediatric oncology: Ewing's sarcoma and neuroblastoma Health Technol Assess 7: 1– 162,2003 Crossref, Medline, Google Scholar |
| 12. | R Smith, LA Owen, DJ Trem , etal: Expression profiling of EWS/FLI identifies NKX2.2 as a critical target gene in Ewing's sarcoma Cancer Cell 9: 405– 416,2006 Crossref, Medline, Google Scholar |
| 13. | E Bandrés, R Malumbres, A Escalada , etal: Gene expression profile of ewing sarcoma cell lines differing in their EWS-FLI1 fusion type J Pediatr Hematol Oncol 27: 537– 542,2005 Crossref, Medline, Google Scholar |
| 14. | MS Staege, C Hutter, I Neumann , etal: DNA microarrays reveal relationship of Ewing family tumors to both endothelial and fetal neural crest-derived cells and define novel targets Cancer Res 64: 8213– 8221,2004 Crossref, Medline, Google Scholar |
| 15. | F Tirode, K Laud-Duval, A Prieur , etal: Mesenchymal stem cell features of Ewing tumors Cancer Cell 11: 421– 429,2007 Crossref, Medline, Google Scholar |
| 16. | JD Hancock, SL Lessnick: A transcriptional profiling meta-analysis reveals a core EWS-FLI gene expression signature Cell Cycle 7: 250– 256,2008 Crossref, Medline, Google Scholar |
| 17. | A Ohali, S Avigad, R Zaizov , etal: Prediction of high risk Ewing's sarcoma by gene expression profiling Oncogene 23: 8997– 9006,2004 Crossref, Medline, Google Scholar |
| 18. | KL Schaefer, M Eisenacher, Y Braun , etal: Microarray analysis of Ewing's sarcoma family of tumours reveals characteristic gene expression signatures associated with metastasis and resistance to chemotherapy Eur J Cancer 44: 699– 709,2008 Crossref, Medline, Google Scholar |
| 19. | G Ricci, F De Maria, G Antonini , etal: 7-Nitro-2,1,3-benzoxadiazole derivatives, a new class of suicide inhibitors for glutathione S-transferases. Mechanism of action of potential anticancer drugs J Biol Chem 280: 26397– 26405,2005 Crossref, Medline, Google Scholar |
| 20. | G Bacci, A Balladelli, C Forni , etal: Adjuvant and neoadjuvant chemotherapy for Ewing sarcoma family tumors in patients aged between 40 and 60: Report of 35 cases and comparison of results with 586 younger patients treated with the same protocols in the same years Cancer 109: 780– 786,2007 Crossref, Medline, Google Scholar |
| 21. | Gene Expression Omnibus (GEO) National Center for Biotechnology Information http://www.ncbi.nlm.nih.gov/geo/ Google Scholar |
| 22. | T Barberi, LM Willis, ND Socci , etal: Derivation of multipotent mesenchymal precursors from human embryonic stem cells PLoS Med 2: e161,2005 Crossref, Medline, Google Scholar |
| 23. | B Perbal, M Zuntini, D Zambelli , etal: Prognostic value of CCN3 in osteosarcoma Clin Cancer Res 14: 701– 709,2008 Crossref, Medline, Google Scholar |
| 24. | RC Gentleman, VJ Carey, DM Bates , etal: Bioconductor: Open software development for computational biology and bioinformatics Genome Biol 5: R80,2004 Crossref, Medline, Google Scholar |
| 25. | H Bengtsson, O Hossjer: Methodological study of affine transformations of gene expression data with proposed robust non-parametric multi-dimensional normalization method BMC Bioinformatics 7: 100,2006 Crossref, Medline, Google Scholar |
| 26. | M Francesconi, D Remondini, N Neretti , etal: Reconstructing networks of pathways via significance analysis of their intersections BMC Bioinformatics 9: S9,2008 suppl 4 Crossref, Medline, Google Scholar |
| 27. | TC Chou, RJ Motzer, Y Tong , etal: Computerized quantitation of synergism and antagonism of taxol, topotecan, and cisplatin against human teratocarcinoma cell growth: A rational approach to clinical protocol design J Natl Cancer Inst 86: 1517– 1524,1994 Crossref, Medline, Google Scholar |
| 28. | LJ van 't Veer, H Dai, MJ van de Vijver , etal: Expression profiling predicts outcome in breast cancer Breast Cancer Res 5: 57– 58,2003 Crossref, Medline, Google Scholar |
| 29. | MJ van de Vijver, YD He, LJ van't Veer , etal: A gene-expression signature as a predictor of survival in breast cancer N Engl J Med 347: 1999– 2009,2002 Crossref, Medline, Google Scholar |
| 30. | S Ramaswamy, KN Ross, ES Lander , etal: A molecular signature of metastasis in primary solid tumors Nat Genet 33: 49– 54,2003 Crossref, Medline, Google Scholar |
| 31. | X Wu, X Tu, KS Joeng , etal: Rac1 activation controls nuclear localization of beta-catenin during canonical Wnt signaling Cell 133: 340– 353,2008 Crossref, Medline, Google Scholar |
| 32. | T Perri, M Fogel, S Mor , etal: Effect of P-glycoprotein expression on outcome in the Ewing family of tumors Pediatr Hematol Oncol 18: 325– 334,2001 Crossref, Medline, Google Scholar |
| 33. | DM Townsend, KD Tew: The role of glutathione-S-transferase in anti-cancer drug resistance Oncogene 22: 7369– 7375,2003 Crossref, Medline, Google Scholar |
| 34. | K Johansson, K Ahlen, R Rinaldi , etal: Microsomal glutathione transferase 1 in anticancer drug resistance Carcinogenesis 28: 465– 470,2007 Crossref, Medline, Google Scholar |
| 35. | P Turella, C Cerella, G Filomeni , etal: Proapoptotic activity of new glutathione S-transferase inhibitors Cancer Res 65: 3751– 3761,2005 Crossref, Medline, Google Scholar |
| 36. | P Turella, G Filomeni, ML Dupuis , etal: A strong glutathione S-transferase inhibitor overcomes the P-glycoprotein-mediated resistance in tumor cells: 6-(7-Nitro-2,1,3-benzoxadiazol-4-ylthio)hexanol (NBDHEX) triggers a caspase-dependent apoptosis in MDR1-expressing leukemia cells J Biol Chem 281: 23725– 23732,2006 Crossref, Medline, Google Scholar |
Acknowledgment
We thank Giovanna Magagnoli, MSC, for the excellent work in taking care of the tumor tissue bank, Cristina Ferrari, MSC, for the active participation in collecting follow-up data, and Cristina Ghinelli for editing.
Microarray analysis was performed on mRNA samples extracted from a total of 30 Ewing sarcoma (EWS) specimens from primary lesions. Fresh-frozen samples (stored at −80°C) were collected from the archives of the Laboratory of Oncologic Research, Rizzoli Institute, Bologna, Italy, and are related to patients treated at the Rizzoli Institute. Histology was reviewed by a pathologist panel, and the presence of EWS/FLI or EWS/ERG translocation was confirmed by reverse transcriptase polymerase chain reaction (RT-PCR). Patients were treated with three consecutive programs of multidrug chemotherapy20 and with local treatments (surgery; surgery plus radiotherapy; radiotherapy only). After completion, patients were continuously observed and clinical data were updated (median follow-up, 7.4 years). Adverse events were defined as recurrence of the tumor at any site or death during remission, and event-free survival was calculated from the date of initial diagnosis. Clinical data are listed in Table 1. Thirteen patients never showed evidence of disease after treatment, whereas 17 patients suffered a relapse of disease after treatment. Of these, 14 patients died. Sample set was handled in a coded fashion. The ethical committee of the Rizzoli Institute approved the studies, and informed consent was obtained from all subjects involved.
Total RNA was extracted by TRIzol extraction kit (Invitrogen, Carlsbad, CA) according to the manufacturer's instructions. Before extraction, the proportion of tumor cells was verified to exceed 90%. RNA concentrations were measured using GeneQuant pro spectrophotometer (Amersham Pharmacia, Cambridge, United Kingdom). RNA quality was assessed using Agilent's 2100 Bioanalyzer (Agilent, Palo Alto, CA) and gel electrophoresis. RT-PCR amplification of the house-keeping gene β-actin was also used to confirm the quality of cDNAs. EWS tissue samples and CD34+ cell line, as putative control sample, were hybridized to Affymetrix Human Genome U133 Plus 2.0 oligonucleotide microarrays (Affymetrix, Santa Clara, CA) according to the manufacturer's GeneChip One-Cycle Target Labeling protocol (www.affymetrix.com). The arrays were then scanned using a confocal laser GeneChip Scanner 3000, and image analysis was performed using GeneChip Operating Software (Affymetrix, Sacramento, CA). Microarray data are available at Gene Expression Omnibus21 public database with the accession number GSE12102.
Global array processing and following analysis were performed using R Bioconductor 23 and MATLAB (MathWorks Inc, Natick, MA) software packages. Data were re-scaled within each array and between arrays through an affine transformation.25
A one-way analysis of variance was performed to check single probe significance in no evidence of disease (NED)/relapse (REL) cases discrimination. A list of 2,092 significant probes (P ≤ .05, referred to as list 1) was used for pathway significance analysis based on Kyoto Encyclopedia of Genes and Genomes database.26 A more refined list of 260 probes (list 2) was obtained by requiring a ratio ≥ 2 and a minimum mean expression value of 10 for the sample probes, from which genes were selected for RT-PCR and survival analysis. A further restricted signature of 35 probes was obtained as an optimal classifier for good/bad prognosis, considering an innovative method combining quadratic discriminant analysis with leave-one-out cross-validation performed over every couple of probes and graph theory.
Unsupervised biclustering (using Pearson correlation metrics) and principal component analysis were considered to discriminate between NED and REL cases (Fig 2) and metastases, NED, and REL cases (Fig 3).
Total RNA from frozen clinical samples and cell lines was extracted using the TRIzol extraction kit (Invitrogen Ltd, Paisley, United Kingdom) and the quality of the RNA samples was determined by electrophoresis through agarose gel. A total of 500 ng of total RNA for each sample were reverse-transcribed to cDNA in a 50-μL reaction mixture using the High Capacity cDNA Reverse Transcription Kit (Applied Biosystems, Foster City, CA).
Predesigned TaqMan probes and primers sets for genes were chosen, and three replicates per gene were considered. Samples were analyzed using an ABI Prism 7900 Sequence Detection System (Applied Biosystems) according to manufacturer's instructions. Expression levels of target genes were normalized to that of glyceraldehyde 3-phosphate dehydrogenase, and the relative quantification analysis was performed on the basis of the ΔΔCT method. cDNA from human bone marrow CD34+ cells was used as calibrator for the comparative analysis.
The EWS cell lines TC-71, 6647, SK-N-MC, SK-ES-1, and RD-ES were obtained from the American Type Culture Collection (Rockville, MD). LAP35, IOR/BRZ, IOR/CLB, IOR/CAR, IOR/NGR, and IOR/RCH were obtained and characterized at the Laboratory of Oncologic Research, Rizzoli Institute.
Doxorubicin was purchased from Sigma-Aldrich (Milano, Italy). 6-(7-nitro-2,1,3-Benzoxadiazol-4-ylthio)hexanol (NBDHEX) was synthesized as reported by Ricci et al.19 Stock solution of doxorubicin (2 mg/mL) was stored at 4°C. NBDHEX was dissolved in dimethyl sulfoxide at 50-mmol/L concentration, and stock solution aliquots were stored in darkness at room temperature. Immediately before use, the integrity of NBDHEX was verified by spectrophotometry. After dilution of NBDHEX stock solutions to the appropriate concentration required for in vitro experiments, the final dimethyl sulfoxide concentration never exceeded 0.01% and had no effect on cell growth inhibition. Working concentrations were prepared by diluting stock solutions in culture medium immediately before use.
To study in vitro cytotoxicity of NBDHEX, a panel of EWS cell lines were seeded in Iscove modified Dulbecco medium plus 10% fetal bovine serum in 96-well plates. After 24 hours, medium was replaced by Iscove modified Dulbecco medium plus 10% fetal bovine serum with or without (control) various concentrations of the compound (100 nmol/L to 24 μmol/L) up to 72 hours. Effects on cell growth were determined with MTT assay (Roche Diagnostics GmbH, Mannheim, Germany) and expressed as drug concentration resulting in 50% inhibition of cell growth after 96 hours of in vitro treatment. To evaluate the in vitro interactions between NBDHEX and DXR, cells were seeded as previously described. After 24 hours, cells were treated with combinations in which the two drugs were used at equitoxic concentrations, corresponding to dosages resulting in 10%, 20%, 30%, 50%, 75%, 90% growth inhibition. After 72 hours of treatment, effects of combination were evaluated with 3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide assay (Invitrogen) according to manufacturer's instruction.
To define the type of interaction between the two drugs (in terms of synergism, additivity, or antagonism), the combination index (CI) was calculated with the isobologram equation27 by using the CalcuSyn software (Biosoft, Stapleford, United Kingdom). Drug-drug interactions were classified as synergistic when CI was less than 0.90, as additive when 0.90 ≤ CI ≤ 1.10, or as antagonistic when CI was ≥ 1.10. The individual effective doses of NBDHEX and doxorubicin to achieve 90% (straight line) growth inhibition (ED90), 75% (hyphenated line) growth inhibition (ED75), and 50% (dotted line) growth inhibition (ED50) were plotted on the x- and y axes. CI values calculated using Calcusyn software are represented by points above, on, or below the lines that indicate antagonism, additivity, or synergy, respectively (Fig 4D).
Fig A1. Scatterplot of principal component analysis performed on 37 samples: 30 EWS primary tumor samples and seven metastases. NED, no evidence of disease; REL, relapse; MET, metastasis.
