THE EWING S SARCOMA family of tumors, which

Transcription

1 EWS-FLI1 and EWS-ERG Gene Fusions Are Associated With Similar Clinical Phenotypes in Ewing s Sarcoma By Jill P. Ginsberg, Enrique de Alava, Marc Ladanyi, Leonard H. Wexler, Heinrich Kovar, Michael Paulussen, Andreas Zoubek, Barbara Dockhorn-Dworniczak, Herbert Juergens, Jay S. Wunder, Irene L. Andrulis, Rajesh Malik, Poul H.B. Sorensen, Richard B. Womer, and Frederic G. Barr THE EWING S SARCOMA family of tumors, which includes Ewing s sarcoma and peripheral neuroectodermal tumor, has specific chromosomal translocations that result in the fusion of the EWS gene and a gene encoding a member of the ETS family of transcription factors. Chromosomal analysis of Ewing s sarcoma has demonstrated a t(11;22)(q24;q12) translocation that generates an EWS-FLI1 fusion protein in 90% to 95% of cases. In 5% to 10% of cases, a variant EWS-ERG fusion protein resulting from a t(21;22)(q22;q12) translocation is formed. 1,2 In both fusion proteins, the N-terminal EWS domain is joined to the C-terminal ETS-type DNA binding domain of FLI1 or ERG, resulting in the replacement of N-terminal FLI1 or ERG transcription activation domains by EWS sequences. Reversetranscriptase polymerase chain reaction (PCR) methods have been used to detect these two chromosomal translocations in clinical specimens. 3 In addition to the Ewing s sarcoma family, a number of other sarcoma categories have alternative gene fusions and thus underlying molecular heterogeneity. Furthermore, in several entities, these alternative fusions are associated with different clinical outcomes. Within fusion-positive alveolar rhabdomyosarcoma, 80% of cases are positive for the PAX3-FKHR fusion and 20% are positive for the PAX7- FKHR fusion. Patients with PAX7-FKHR tumors are younger at diagnosis, more often have extremity tumors, less often have metastases, and have a better prognosis than PAX3-FKHR positive patients. 4 In patients with synovial sarcoma, the SYT-SSX2 fusion is associated almost exclusively with the monophasic histologic subtype and may also be predictive of longer metastasis-free survival than the SYT-SSX1 fusion. 5 Finally, the precise exon composition of Purpose: There are a variety of solid tumors in which alternative chromosomal translocations generate related fusion products. In alveolar rhabdomyosarcoma and synovial sarcoma, these variant fusions have been found to have major clinical significance. We investigated whether the two alternative gene fusion products, EWS-FLI1 and EWS-ERG, define different clinical subsets within the Ewing s sarcoma family of tumors. Patients and Methods: We selected 30 cases of Ewing s sarcoma with the EWS-ERG gene fusion and 106 cases with the EWS-FLI1 fusion. Clinical data were obtained for each case and compared with the molecular diagnostic findings. Results: There were no significant clinical differences observed between the two groups in age of diagnosis, sex, metastasis at diagnosis, primary site, event-free survival, or overall survival. Conclusion: Differences in the C-terminal partner in the Ewing s sarcoma family gene fusions are not associated with significant phenotypic differences. J Clin Oncol 17: by American Society of Clinical Oncology. the EWS-FLI1 fusion transcript has been found to be a prognostic factor in Ewing s sarcoma. For localized Ewing s sarcoma, patients with tumors that express the type 1 fusion, where exon 7 of EWS is joined to exon 6 of FLI1, have an improved overall survival over those patients with tumors that express non type 1 EWS-FLI1 fusions. 6,7 These reports of clinical differences between alternative gene fusions are the precedent for the current study. Therefore, we investigated whether the EWS-ERG fusion in the Ewing s sarcoma family predicts a clinical phenotype distinct from that of the EWS-FLI1 fusion. Our comparison of retrospective data from 30 EWS-ERG cases and 106 EWS-FLI1 cases revealed similar presentation and outcome. From the The Children s Hospital of Philadelphia, Philadelphia; University of Pennsylvania School of Medicine, Philadelphia, PA; Clinica Universitaria de Navarra, Pamplona, Spain; Memorial Sloan- Kettering Cancer Center, New York; Columbia University College of Physicians and Surgeons, New York, NY; Children s Cancer Research Institute, St. Anna Children s Hospital, Vienna, Austria; Department of Pediatric Hematology and Oncology, and Gerhard Domagk Institute of Pathology, Westfaehlische Wilhelms University, Muenster, Germany; University of Virginia, Charlottesville, VA; Mount Sinai Hospital, Toronto, ON, Canada; and British Columbia s, Canada Children s Hospital, Vancouver, BC, Canada. Submitted July 31, 1998; accepted February 1, Supported in part by National Institutes of Health grants no. CA64202 and CA71838 (to F.G.B.) and grants no. T32-CA09615 and K12-CA76931 (to J.P.G.). Address reprint requests to Frederic G. Barr, MD, PhD, Department of Pathology and Laboratory Medicine, University of Pennsylvania, 36th and Hamilton Walk, Philadelphia, PA ; barrfg@mail.med.upenn.edu by American Society of Clinical Oncology X/99/ Journal of Clinical Oncology, Vol 17, No 6 (June), 1999: pp

2 1810 GINSBERG ET AL This finding thus contrasts with those in alveolar rhabdomyosarcoma and synovial sarcoma in which alternative fusions are associated with differing clinical phenotypes. PATIENTS AND METHODS Clinical Material and Demographics Data on 136 patients were obtained from the following institutions: The Children s Hospital of Philadelphia/Hospital of the University of Pennsylvania, Philadelphia, PA (n 28); Memorial Sloan-Kettering Cancer Center, New York, NY (n 51); British Columbia s Children s Hospital, Vancouver, Canada (n 16); Mount Sinai Hospital, Toronto, Canada (n 8); Clinica Universitaria de Navarra, Pamplona, Spain (n 12); Westfaehlische Wilhelms University, Muenster, Germany, and Childrens Cancer Research Institute, St. Anna Children s Hospital, Vienna, Austria (n 10); The Babies and Children s Hospital of New York, NY (n 3); Vanderbilt University, Nashville, TN (n 2); A.I. Dupont Institute, Wilmington, DE (n 5); and University of Virginia, Charlottesville, VA (n 1). Requirements for inclusion in this study were a diagnosis of Ewing s sarcoma or peripheral neuroectodermal tumor, the presence of either the EWS-FLI1 fusion or EWS-ERG fusion, and availability of clinical data. All institutions contributed patients with the EWS-FLI1 fusion and the EWS-ERG fusion except the Westfaehlische Wilhelms University and the Children s Cancer Research Institute. These two institutions contributed only EWS-ERG patients that met the above criteria. Treatment and Clinical Follow-Up Patients were diagnosed between July 1, 1985 and August 27, 1997, and follow-up information was collected until March Adequate data for both initial therapy and overall survival were available for 133 patients. All patients were treated with curative intent except for one patient with a pelvic primary tumor treated by surgery alone. Thus the remaining 132 cases were used for the overall survival analysis. Adequate data for event-free survival (EFS) were available for 127 patients. The patients received neoadjuvant and/or adjuvant chemotherapy according to the following protocols: T9, T11, P6, 7007, 7881, 7942, 8851, VACIME, HD CAV, EVAIA, or VAIA. The primary agents in these protocols are alkylating agents (cyclophosphamide and/or ifosfamide) and anthracyclines. For local control of primary tumors, most patients were treated with surgical excision, radiation therapy, or both. Three EWS-ERG patients (10%) and nine EWS-FLI1 patients (8.5%) received only chemotherapy. Management of recurrences was either palliative or multimodal. Molecular Genetic Analyses Molecular genetic analyses were performed in six different laboratories. Total RNA was isolated from snap-frozen tumor material by the guanidinium thiocyanate-phenol/chloroform method. 3,8,9 Reverse transcription was performed using random hexamer primers and reverse transcriptase. The cdnas were then amplified by PCR using previously published primers and conditions. Amplified products were separated by agarose gel (2%) electrophoresis and visualized by staining with ethidium bromide. 3,8-10 The identity of the fusion gene type was determined either by Southern blotting PCR products to FLI1- or ERG-specific oligonucleotide probes or by sequencing the PCR products either directly or after cloning. Fusion sizes were determined by measuring electrophoretic mobility of PCR fragments and/or sequencing. Statistical Analyses A 2 test was used to test for differences in patient characteristics and local control measures. A t test was used to compare the dose intensities of alkylating agents and anthracyclines between the EWS-FLI1 and EWS-ERG patients. EFS was defined as the time from diagnosis until the first occurrence of an event (death, relapse, or disease progression). Patients who did not relapse were censored at the time of their last contact. EFS and overall survival rates were estimated by the method of Kaplan and Meier, and comparisons of life-table outcome were performed with the log-rank test. The median follow-up duration was calculated from the time of diagnosis for the patients who are event-free survivors. RESULTS Molecular Analyses We selected all possible patients with an EWS-ERG fusion (n 30) and then 106 patients with an EWS-FLI1 fusion for comparison. Among the two groups, the proportion of different fusion types detected is listed in Table 1. The percentage of EWS-FLI1 type 1 (7-6 fusion) (62%) is representative of what has been described in the literature. Furthermore, there was a similar fraction of type 1 fusions in the EWS-ERG group. However, there is a substantial difference in the distribution of non type 1 fusions (P.002). The majority of non type 1 EWS-FLI1 fusions are larger than the type 1 fusion (34 larger v six smaller) whereas the majority of non type 1 EWS-ERG fusions are smaller than the type 1 fusion (seven smaller v three larger). Clinical Characteristics Clinical characteristics of patients are listed in Table 2. Among the 136 patients, 81 were male and 55 were female, with no significant difference in sex between the EWS-ERG patients and the EWS-FLI1 patients (P.79). The mean age Table 1. EWS-FLI Transcript Types and EWS-ERG Transcript Types No. of Patients Spacer Size in AA EWS exon/fli1 exon * EWS exon/erg exon * NOTE. N 135; one patient without size information. Abbreviation: AA, amino acids. *Type 1 fusion.

3 EWS-FLI1 AND EWS-ERG IN EWING S SARCOMA 1811 Table 2. Comparison of Fusion Types and Clinicopathologic Features Characteristic Total No. EWS-ERG No. EWS-FLI1 No. Age, years Sex Male Female Location of tumor Extremities Central axis Mets at diagnosis No Yes EWS-ERG EWS-FLI1 % SE % SE P OS at 2 years EFS at 2 years Loc ERG Loc FLI1 Total % SE % SE P OS at 2 years EFS at 2 years Loc Type 1 ERG Loc Type 1 FLI1 Total % SE % SE P OS at 2 years EFS at 2 years Abbreviations: OS, overall survival; Mets, metastasis; Loc, localized. at diagnosis for the entire group was 19.2 years (SEM 11.4 months). There was no significant difference in age at diagnosis between the EWS-ERG patients and the EWS-FLI1 patients (P.49). Primary sites were grouped into two main categories: central axis (skull, spine, pelvis, and surrounding soft tissues; n 88) and extremities (bone or soft tissues; n 47). There was no significant difference between P patients with EWS-ERG and EWS-FLI1 in terms of primary site at diagnosis, a comparable portion of patients with each fusion having central axis disease and extremity disease (P.20). There was no significant difference between the number of EWS-ERG patients with metastatic disease at diagnosis compared with the number of EWS-FLI1 patients with metastatic disease at diagnosis (P.33). Treatment Regimens We compared chemotherapy treatment regimens between EWS-FLI1 and EWS-ERG patients. The mean alkylating dose intensity was g/m 2 /wk, and the mean anthracycline dose-intensity was mg/m 2 /wk for EWS-FLI1 patients. In the EWS-ERG group, the mean alkylating dose-intensity was g/m 2 /wk, and the mean anthracycline dose-intensity was mg/m 2 / wk. A t test demonstrated that there were no statistically significant differences in the alkylating dose-intensity (P.40) or anthracycline dose-intensity (P.14) between the EWS-ERG patients and the EWS-FLI1 patients. Assessment of local control measures by 2 analysis yielded no significant differences in the number of EWS-ERG patients and EWS-FLI1 patients receiving radiation therapy (P.80) or surgery (P.77) EFS and Overall Survival of EWS-FLI1 Versus EWS-ERG Similar relapse rates were seen among the two groups. Relapses were observed in 33% (42 of 127) of EWS-FLI1 patients (local, n 25; distant, n 18; one patient relapsed locally and distally) and 50% (15 of 30) of EWS-ERG patients (local, n 8; distant, n 8; one patient relapsed locally and distally). For 127 patients suitable for EFS analysis, EFS for the EWS-ERG group was not significantly different than the EFS of the EWS-FLI1 group (Fig 1A). Fig 1. (A) Kaplan-Meier curve for EFS of patients with EWS-FLI1 and EWS-ERG fusions; (B) Kaplan-Meier curve for overall survival of patients with EWS-FLI1 and EWS-ERG fusions.

4 1812 GINSBERG ET AL In the EWS-FLI1 group, 36 (35%) patients died of disease, and in the EWS-ERG group, 11 (33%) patients died of disease. Overall survival between the two groups did not differ (Fig 1B). The median follow-up duration of event-free survivors is months (SE 6.89) (range, 2.4 to 94.1 months) for the EWS-ERG patients and 23 months (SE 3.37) (range, 1 to 99 months) for the EWS-FLI1 patients. In an effort to increase the number of EWS-ERG patients, 10 EWS-ERG cases available from institutions in Germany and Austria were included in the data set. Because we did not collect EWS-FLI1 cases from these institutions, the data were reanalyzed without these 10 cases to ensure that this selection bias did not skew our outcome results. The EFS analysis (P.65) and the overall survival analysis (P.71) again demonstrated that the EWS-ERG group did not differ significantly from the EWS-FLI1 group, even without these 10 EWS-ERG patients. Clinical Outcome of Patients With Localized Disease and Type 1 Fusions Because metastatic disease at diagnosis is the main negative prognostic factor within Ewing s sarcoma, we reanalyzed our data looking specifically at the patients with localized disease. There were 98 patients with localized disease, 79 with EWS-FLI1 fusions and 19 with EWS-ERG fusions. There were 76 EWS-FLI1 patients and 19 EWS- ERG patients eligible for overall survival analysis and 71 EWS-FLI1 patients and 19 EWS-ERG patients eligible for EFS analysis. There was no significant difference between the two groups in either EFS or overall survival (Figs 2A and 2B). The type 1 (7-6 fusion) EWS-FLI1 fusion has now been shown to have independent prognostic significance regarding improved overall survival. 7 For this reason, we analyzed our data looking at this subset of patients. Fifty-five patients with localized disease had either a type 1 EWS-FLI1 fusion (n 46) or a type 1 EWS-ERG fusion (n 9). There was no significant difference in the EFS or overall survival of these two groups (Figs 3A and 3B). However, the number of patients in this subset of EWS-ERG is small (n 9), and it is possible that the nonsignificant difference may be due to a lack of power. There were too few patients with localized disease and EWS-ERG positivity (n 19) to analyze type 1 EWS-ERG patients versus non type 1 EWS-ERG patients. DISCUSSION As a result of chromosome translocations in Ewing s sarcoma, the 5 portion of the EWS gene that encodes a putative transactivation domain is fused to the 3 regions encoding the DNA-binding domain of several ETS family transcriptional regulators, most notably FLI1 and ERG. 11 The EWS-FLI1 and EWS-ERG fusions result in a number of different transcripts based on the exact fusion site between EWS and the ETS family member. 12 At the molecular level, there are as many as 18 possible in-frame EWS-FLI1 chimeric transcripts that represent different combinations of exons from EWS and FLI1. In this study, we observed eight different in-frame EWS-FLI1 fusion transcripts. Recently, it has been found that the type 1 fusion that joins exon 7 of EWS and exon 6 of FLI1 portends a favorable prognosis in patients with localized Ewing s sarcoma in comparison with patients with non type 1 EWS-FLI1 fusions. 6,7 Fig 2. (A) Kaplan-Meier curve for EFS of patients with localized disease and EWS-FLI1 and EWS-ERG fusions; (B) Kaplan-Meier curve for overall survival of patients with localized disease and EWS-FLI1 and EWS-ERG fusions.

5 EWS-FLI1 AND EWS-ERG IN EWING S SARCOMA 1813 Fig 3. (A) Kaplan-Meier curve for EFS of patients with localized disease and type 1 EWS-FLI1 and type 1 EWS-ERG; (B) Kaplan-Meier curve for overall survival of patients with localized disease and type 1 EWS-FLI1 and type 1 EWS-ERG. Molecular heterogeneity is also found within EWS-ERG. Five different EWS-ERG junctions were observed in this study, all producing in-frame transcripts. Like EWS-FLI1, all the chimeric products contain the DNA binding domain of ERG and the N-terminal domain of EWS. In fact, the fusion points in the ERG and FLI1 genes are comparable, which suggests that ERG and FLI1 gene structures are quite similar. 12 However, the distribution of fusion sizes between the EWS-ERG and EWS-FLI1 groups was different in this series. Although the type 1 fusion was the most common in both groups, the non type 1 EWS-FLI1 fusions tended to be larger, whereas the non type 1 EWS-ERG fusions were smaller. This finding may reflect structural differences between the two genomic loci, such as differing relative sizes of introns in the ERG and FLI1 genes. Alternatively, there may be unknown biologic constraints that tend to select smaller EWS-ERG fusions or larger EWS-FLI1 fusions. The gene fusions in Ewing s sarcoma are useful in the diagnostic differentiation of this tumor from other primitive tumors. Although the fusion transcripts vary in their exon composition, the deduced encoded protein products have constant features. EWS-FLI1 and EWS-ERG are similar in that they share the same N-terminal region, and they have similar DNA binding domains with 98% identity. 2,13 In contrast to alveolar rhabdomyosarcoma, in which the PAX3- FKHR and PAX7-FKHR fusions are expressed from different promoters, the EWS promoter drives expression of both EWS-ERG and EWS-FLI1. In addition, EWS-FLI1 and EWS-ERG have similar transforming abilities and have been found to inhibit apoptosis of NIH3T3 cells. 14,15 We have shown that unlike alveolar rhabdomyosarcoma and synovial sarcoma, where the associated PAX3-FKHR and PAX7- FKHR gene rearrangements and the SYT-SSX1 and SYT- SSX2 gene fusions predict distinct clinical phenotypes, EWS-ERG and EWS-FLI1 chimeric gene products do not result in different clinical presentations, nor do they predict different clinical outcomes. These observations indicate that EWS-FLI1 and EWS-ERG have highly comparable function in the Ewing s sarcoma family of tumors. ACKNOWLEDGMENT We thank Dr A. Kawai, Dr R. Meek, C. Sweeney, J. Mathers, Dr J. Whitlock, and Dr A. Chauvenet for assistance with clinical follow-up data; A. Hamelin, Q.-B. Xiong, and D. Strzelecki for technical assistance; and Dr D. Friedman for statistical advice. 1. Delattre O, Zucman J, Plougastel B, et al: Gene fusion with an ETS DNA-binding domain caused by chromosome translocation in human tumours. Nature 359: , Sorensen PHB, Lessnick SL, Lopez-Terrada D, et al: A second Ewing s sarcoma translocation, t(21;22), fuses the EWS gene to another ETS-family transcription factor, ERG. Nat Genet 6: , Barr FG, Chatten J, D Cruz CM, et al: Molecular assays for chromosomal translocations in the diagnosis of pediatric soft tissue sarcomas. JAMA 273: , 1995 REFERENCES 4. Kelly KM, Womer RB, Sorensen PH, et al: Common and variant gene fusions predict distinct clinical phenotypes in rhabdomyosarcoma. J Clin Oncol 15: , Kawai A, Woodruff J, Healey JH, et al: SYT-SSX gene fusion as a determinant of morphology and prognosis in synovial sarcoma. N Engl J Med 338: , Zoubek A, Dockhorn-Dworniczak B, Delattre O, et al: Does expression of different EWS chimeric transcripts define clinically distinct risk groups of Ewing s tumor patients? J Clin Oncol 14: , 1996

6 1814 GINSBERG ET AL 7. de Alava E, Kawai A, Healey JH, et al: EWS-FLI1 fusion transcript structure is an independent determinant of prognosis in Ewing s sarcoma. J Clin Oncol 4: , Barr FG, Xiong QB, Kelly K: A consensus polymerase chain reaction-oligonucleotide hybridization approach for the detection of chromosomal translocations in pediatric bone and soft tissue sarcomas. Am J Clin Pathol 104: , de Alava E, Ladanyi M, Rosai J, et al: Detection of chimeric transcripts in desmoplastic small round cell tumor and related developmental tumors by reverse transcriptase polymerase chain reaction: A specific diagnostic assay. Am J Pathol 147: , Sorensen PH, Liu XF, Delattre O, et al: Reverse-transcriptase PCR amplification of EWS/FLI-1 fusion transcripts as a diagnostic test for peripheral primitive neuroectodermal tumors of childhood. Diagn Mol Pathol 2: , Delattre O, Zucman J, Melot T, et al: The Ewing family of tumors: A subgroup of small-round-cell tumors defined by specific chimeric transcripts. N Engl J Med 331: , Zucman J, Melot T, Desmaze J, et al: Combinatorial generation of variable fusion proteins in the Ewing family of tumours. EMBO J 12: , May WA, Gishizky ML, Lessnick SL, et al: Ewing sarcoma 11;22 translocation produces chimeric transcription factor that requires the DNA-binding domain encoded by FLI1 for transformation. Proc Natl Acad Sci U S A 90: , Yi H, Fujimura Y, Ouchida M, et al: Inhibition of apoptosis by normal and aberrant FLI1 and ERG proteins involved in human solid tumors and leukemias. Oncogene 14: , Braun BS, Frieden R, Lessnick SL, et al: Identification of target genes for the Ewing s sarcoma EWS/FLI fusion protein by representational difference analysis. Mol Cell Biology 15: , 1995