Role of Fusion Subtype in Ewing Sarcoma

Abstract

This issueof JournalofClinicalOncologypresents twocontemporary manuscripts describing the absence of prognostic importance of gene fusion subtype in Ewing sarcoma (ES). Van Doorninck et al, reporting for the Children’s Oncology Group (COG), claim that this loss is a result of modern, more effective therapy that has eliminated the prognostic advantageof type1fusions inES.Similarly,LeDeleyetal foundthatgene fusion subtype has no prognostic significance in the Euro-E.W.I.N.G. 99 trial. However, they suggest that prior reports may have been premature and based on retrospective analyses—underscoring the need for large prospectivestudies.Regardlessofreasons, thesetwoarticlesare important additions to our understanding of the importance of gene fusion subtype in ES, stress that therapy is always an important prognostic factor, and remind us that the ability of clinicians to better tailor therapy must be based on robust prognostic factors. Although both studies claim to prospectively evaluate gene fusion subtype in ES, the actual sample set in both studies appears to be a convenience sample of the available tumor tissue from these clinical trials, withfusiondataonlyavailable forasubsetof theentirepatientpopulation enrolled. This is both a testament to the patients and families to enroll in these critically important biology studies and to local treating institutions for faithfully submitting tumor samples, but also a caveat that in neither study were the majority of enrollments available for fusion analysis. The Euro-E.W.I.N.G. 99 study had the larger data set (703 v 132). The COG study only used fresh-frozen tissue, while the Euro-E.W.I.N.G. 99 trial used both fresh and formalin-fixed paraffin-embedded tumors, although thevastmajorityoftumorsstudiedwerefresh-frozentumors(82%).Both studiesusedsimilarreverse-transcriptionpolymerasechainreaction(RTPCR) methodology and sequencing to detect fusion subtypes in these samples. Patient demographics were quite similar. From an enrollment perspective, the Euro-E.W.I.N.G. trial evaluated patients with both localized and metastatic disease, while the COG study only enrolled patients with localized disease. Both reports address the statistical power of their research. Assuming a progression-free survival difference of 20% in different fusion types, the Euro-E.W.I.N.G. analysis required a minimal sample size (localized tumors only) of 111 type 1 EWS-FLI1 fusions, 38 type 2 EWS-FLI1 fusions, 19 other EWS-FLI1 fusions, and 18 EWS-ERG fusionstotesttheirhypothesis.TheCOGreportstatesthattheyhada95% power to detect a relative failure rate of 4.5, with an 80% power to detect a relative failure rate of 3.0, using a 5% type 1 error. The treatment eras encompassing the two ES studies were quite similar. Although therapy in the two studies shared similarities, treatments were also clearly quite distinct. Patients in both studies received doxorubicin, etoposide, vincristine, and classic alkylators, although the COG study used both cyclophosphamide and ifosfamide, and some Euro-E.W.I.N.G. patients only received ifosfamide. There are substantial differences in dose, schedule, and timing of therapy; this issue may be important in that the recently completed AEWS0031 study clearly showed that dose-compressed therapy resulted in better outcomes. Some patients enrolled on Euro-E.W.I.N.G. were randomly assigned to either cyclophosphamide or ifosfamide, and many patients also received other agents including dactinomycin or high-dose therapy with stem cell reconstitution; some patients received whole-lung irradiation. Although the comparison of demographics and therapy approaches identified substantial differences, these two studies report remarkably similar results. There was a quite similar distribution of fusion subtypes, with the Euro-E.W.I.N.G. study having a slightly higher proportion of type 2 fusions (Table 1). More importantly, both studies report quite similar outcomes for each of the fusion subtypes, although details of analysis and presentation of data make a head-to-head comparison difficult (Table 2). Both studies therefore come to the same conclusion that fusion subtype is of no prognostic significance with the current therapy. Thesefindingsraiseadditionalconsiderationsconcerningutilization of molecular pathology testing for the fusion products in ES. The major technologies to detect these molecular changes are in situ hybridization (ISH)for thegenefusions,RT-PCRfor the fusiontranscripts, andimmunohistochemistry (IHC) for the fusion proteins. Though several comparisons can be made concerning the relative utility of these assays, an important issuegermanetothisdiscussionis the inabilityofbothISHand IHC assays to provide fusion subtype information. In contrast, as shown in the two articles in this issue, RT-PCR assays provide fusion subtype information by size fractionation after gel electrophoresis and direct sequencing of RT-PCR products. Previous studies established that ES diagnosis does not vary with fusion subtype. Furthermore, the two studies in this issue now conclude that fusion subtype does not impact on ES prognosis. Therefore, the need for collecting fusion subtype information