Marrow-ablative chemotherapy and re-irradiation are associated with improved overall survival in patients with relapsed central nervous system nongerminomatous germ cell tumors.

PURPOSE: Relapsed/refractory central nervous system nongerminomatous germ cell tumors (NGGCTs) have a poor prognosis. We explored prognostic factors that may guide the management of relapsed/refractory NGGCTs. METHODS: We performed a multinational retrospective study of relapsed/refractory NGGCTs. Data were summarized using standard descriptive statistics. Overall survival (OS) was estimated using the Kaplan-Meier method. Prognostic factors were evaluated with univariable and multivariable Cox proportional hazards models, reported as hazard ratios (HR) with 95% confidence intervals (CI). RESULTS: We identified 30 patients from seven centers in four countries. The median age at initial diagnosis was 12 years (IQR: 10,15). Twenty patients experienced relapse, and 10 had treatment-refractory disease. The median follow-up time since relapse/progression was eight months (IQR: 2,29). The median time to relapse was 12 months (IQR: 7,22). Eleven (37%) received myeloablative chemotherapy (HDCx) with autologous hematopoietic stem cell rescue (AuHCR), and 16 (57%) received re-irradiation. Ten patients survived at the last follow-up, all of whom had achieved a complete response at the end of therapy. Among the 10 survivors, three received HDCx, one received re-irradiation, and six received both. HDCx (HR = 0.08, 95%CI: 0.018–0.36; p = 0.001) and re-irradiation at relapse (HR = 0.403, 95%CI: 0.16–1.004; p = 0.051) suggest survival benefit. In multivariable analysis, HDCx was independently associated with improved OS regardless of re-irradiation and relapse/progression status (HR = 0.111, 95%CI: 0.023–0.532; p = 0.006). The 3-year OS of patients who received HDCx or re-irradiation was 76% (95%CI: 51–100%) and 33% (95%CI: 14%-77%) compared with patients who did not receive HDCx 5.3% (95%CI: 0.8%-35%), or re-irradiation 17% (95%CI: 4.7%-59%), respectively (p < 0.05). CONCLUSION: HDCx and/or re-irradiation are associated with improved OS. The retrospective design, small sample size, histologic heterogeneity, and survival bias limit interpretation. To guide future therapies, collaborative studies are needed to elucidate biological differences between long-term survivors and non-survivors.