J Clin Oncol 22: by American Society of Clinical Oncology INTRODUCTION

Transcription

1 VOLUME 22 NUMBER 10 MAY JOURNAL OF CLINICAL ONCOLOGY O R I G I N A L R E P O R T Thiotepa-Based High-Dose Chemotherapy With Autologous Stem-Cell Rescue in Patients With Recurrent or Progressive CNS Germ Cell Tumors Shakeel Modak, Sharon Gardner, Ira J. Dunkel, Casilda Balmaceda, Marc K. Rosenblum, Douglas C. Miller, Steven Halpern, and Jonathan L. Finlay From the Departments of Pediatrics and Pathology, Memorial Sloan- Kettering Cancer Center; Department of Neurology, Columbia University; and the Departments of Pediatrics and Pathology, New York University Medical Center, New York, NY. Submitted November 10, 2003; accepted February 24, Presented in part at the First International Symposium on CNS Germ Cell Tumors, September 17-19, 2003, Kyoto, Japan. Authors disclosures of potential conflicts of interest are found at the end of this article. Address reprint requests to Shakeel Modak, MD, Department of Pediatrics, Memorial Sloan-Kettering Cancer Center, 1275 York Ave, New York, NY 10021; modaks@mskcc.org by American Society of Clinical Oncology X/04/ /$20.00 DOI: /JCO A B S T R A C T Purpose To evaluate the efficacy and toxicity of high-dose chemotherapy (HDC) followed by autologous stem-cell rescue (ASCR) in patients with relapsed or progressive CNS germ cell tumors (GCTs). Patients and Methods Twenty-one patients with CNS GCTs who experienced relapse or progression despite having received initial chemotherapy and/or radiotherapy were treated with thiotepa-based HDC regimens followed by ASCR. Results Estimated overall survival (OS) and event-free survival (EFS) rates for the entire group 4 years after HDC were 57% 12% and 52% 14%, respectively. Seven of nine (78%) patients with germinoma survived disease-free after HDC with a median survival of 48 months. One patient died as a result of progressive disease (PD) 39 months after HDC, and another died as a result of pulmonary fibrosis unrelated to HDC 78 months after ASCR without assessable disease. However, only four of 12 patients (33%) with nongerminomatous germ cell tumors (NGGCTs) survived without evidence of disease, with a median survival of 35 months. Eight patients with NGGCTs died as a result of PD, with a median survival of 4 months after HDC (range, 2 to 17 months). Patients with germinoma fared better than those with NGGCTs (P.016 and.014 for OS and EFS, respectively). Patients with complete response to HDC also had significantly better outcome (P.001 for OS and EFS) compared with patients with only a partial response or stable disease. There were no toxic deaths because of HDC. Conclusion Dose escalation of chemotherapy followed by ASCR is effective therapy for patients with recurrent CNS germinomas and might be effective in patients with recurrent NGGCTs with a low tumor burden. J Clin Oncol 22: by American Society of Clinical Oncology INTRODUCTION Primary CNS germ cell tumors (GCTs) are rare, constituting less than 5% of all brain tumors. 1 Fifty to 65% of patients with CNS GCTs 2,3 have been reported to have germinomas, although in a Japanese report 4 only 36% were found to have histologically proven pure germinoma. The remainder is a heterogeneous group of tumors termed nongerminomatous germ cell tumors (NG- GCTs). Germinomas are extremely sensitive to both irradiation and platinum-based chemotherapy. 5-7 Craniospinal irradiation has been reported to achieve 5-year event free survival (EFS) of 91% in patients with pure CNS germinomas. 8 Strategies that used chemotherapy alone with the exclusion of radiation therapy to avoid the morbidity of cranial radiotherapy were effective in achieving remission but long-term outcome has been unsatisfactory. 9 An approach using combined chemotherapy and radiation therapy has led to a 3-year EFS of 96%. 10 In contrast, CNS NGGCTs are relatively radioresistant, with poor outcomes reported when treated with radiation therapy alone. 2,11 CNS NGGCTs are chemotherapy 1934

2 High-Dose Chemotherapy in Recurrent CNS GCT Patient Age (years) Sex Initial Diagnosis Reviewed Diagnosis Table 1. Clinical Characteristics of Patients at Diagnosis Extent of Disease Surgery Radiotherapy Chemotherapy Response Time to Relapse or Progression (months) 1 1 M PNET NGGCT Parieto-occipital Par res None None PR M Germinoma Germinoma Pineal Biopsy Focal Carbo CR M NGGCT NGGCT Pineal Biopsy WB focal None PR M NGGCT NGGCT Pituitary None None CPM, VP, CDDP, carbo, BL CR F NGGCT NGGCT Pineal Par res WB focal None PR M NGGCT NGGCT Pituitary None None CPM, VP, BL, CDDP CR M Germinoma Germinoma Disseminated Biopsy None CDDP, VP, BL CR M Germinoma Germinoma Pineal GTR Focal None CR M NGGCT NGGCT Pineal Biopsy None CPM, CDDP, VP PD Not applicable M Germinoma Germinoma Pineal Biopsy CS focal CPM, VP, BL, CDDP CR M Germinoma NGGCT Pineal Par res None CPM, VP, BL, CDDP CR F Optic glioma Germinoma Pituitary None Focal None PR M Epidermoid tumor NGGCT Pineal Par res None None PR M Germinoma Germinoma Pineal GTR WB focal None PR F Germinoma Germinoma Pituitary Biopsy WB local CDDP, VP PR M Germinoma Germinoma Pineal Biopsy None VP, CDDP, BL CR M Adenoma Germinoma Pituitary GTR Focal None CR M NGGCT NGGCT Pineal GTR WB focal Ifosfamide, VP, CDDP CR M NGGCT NGGCT Disseminated None WB focal None CR M Germinoma Germinoma Pineal GTR Focal None CR M Germinoma Germinoma Pineal None WB focal CDDP, VP, BL CR 37 Abbreviations: PNET, primitive neuroectodermal tumor; NGGCT, nongerminomatous germ cell tumor; Par res, partial resection; PR, partial response; CR, complete response; carbo, carboplatin; WB, whole brain; CPM, cyclophosphamide; VP, etoposide; CDDP, cisplatin; BL, bleomycin; GTR, gross total resection; CS, craniospinal disease; PD, progressive disease. sensitive 12,13 and a multimodal approach using combination chemotherapy and craniospinal irradiation seems to be the most promising. 14,15 However, CNS NGGCTs have an inferior prognosis when compared with CNS germinomas. 2,16 Because of the rarity of CNS GCTs and reduced relapse rates compared with other CNS tumors, few studies have addressed the issue of relapsed disease. High salvage rates have been reported using irradiation with or without cyclophosphamide for patients with germinomas who were treated with only chemotherapy initially. 9,17 Conversely, patients initially treated with radiation therapy have responded to chemotherapy at recurrence, 8 although responses usually are not sustained. However, patients who experience relapse after both modalities of therapy, or those who do not respond to initial treatment, invariably die as a result of progressive disease (PD). A strategy involving chemotherapy dose escalation with autologous hematopoietic stem-cell rescue (ASCR) uses the dose-response relationship observed with alkylating drugs and platinum compounds to overcome possible chemotherapy resistance and has proved to be useful in patients with high-risk systemic GCT. 18,19 CNS GCTs, like their systemic counterparts, have been shown to be highly chemotherapy sensitive. Therefore, in an attempt to improve survival in patients with recurrent CNS GCTs, we evaluated the use of high-dose chemotherapy (HDC) followed by ASCR. PATIENTS AND METHODS Twenty-one patients with recurrent or progressive primary intracranial GCT underwent treatment with high-dose myeloablative chemotherapy followed by ASCR from 1986 to All patients were treated (after written informed consent was obtained) using institutional review board approved protocols designed for therapy of high-risk or recurrent brain tumors. Records on patients with CNS GCTs were retrospectively analyzed. Patients Diagnosis. There were three females and 18 males. Age at diagnosis ranged from 1 to 31 years (median, 16 years). Initial diagnoses were 10 patients with germinoma, seven with NGGCTs, one with epidermoid tumor, one with pituitary adenoma, one with supratentorial primitive neuroectodermal tumor, and one patient was considered (by computed tomography scanning only) to have an optic glioma. All slides were reviewed at Memorial Sloan-Kettering Cancer Center or New York University Medical Center by one of two neuropathologists. Patients with initial diagnoses of epidermoid tumor and primitive neuroectodermal tumor were definitively diagnosed at review to have NGGCTs, and the patient with the initial diagnosis of pituitary adenoma was diagnosed at review to have germinoma. The patient with the initial diagnosis of optic glioma was diagnosed with germinoma at biopsy on recurrence (patient 12; Table 1). Twenty patients experi

3 Modak et al Patient Extent of Disease at Relapse Diagnosis Table 2. Clinical Characteristics of Patients at Relapse No. of Relapses Subsequent Surgery Subsequent Radiotherapy Subsequent Chemotherapy Response to Therapy 1 Parietooccipital NGGCT 3 Par res None None PR 2 Frontal lobe NGGCT 2 Par res None CPM, CDDP, VP, carbo PR 3 Pineal and spinal NGGCT 2 Biopsy None CPM, VP, CDDP PR 4 Pituitary NGGCT 1 None None None Not applicable 5 Pineal NGGCT 2 Par res None CPM, BL, V PR 6 Pituitary NGGCT 1 None None None Not applicable 7 Disseminated intracranial Germinoma 2 None WB focal CPM, topotecan CR 8 Cervical spinal cord NGGCT 2 None Focal VIN, paclitaxel, VP CR 9 Pineal elevated markers NGGCT 0 None None None Not applicable 10 Disseminated intracranial Germinoma 1 Biopsy None CPM CR 11 Third ventricle, abdominal NGGCT 2 None CS focal CPM, CDDP, VP, VIN CR 12 Disseminated intracranial Germinoma 4 None Focal Carbo, VP, BL, chlorambucil PR 13 Disseminated intracranial and spinal NGGCT 2 Par res WB focal CDDP, BL, VP, CPM, ifosfamide PR 14 Spinal cord Germinoma 1 None None CPM, VP, CDDP, BL PR 15 Disseminated intracranial and spinal Germinoma 1 Biopsy None CPM, VP, CDDP, BL PR 16 Pineal, abdominal Germinoma 2 None Craniospinal Ifosfamide, VP, carbo CR 17 Pineal, lateral horn Germinoma 2 Biopsy None CPM, VP, CDDP, BL CR 18 Cervical spinal cord NGGCT 2 None Spinal focal CPM, paclitaxel, VIN CR 19 Pineal NGGCT 2 None Gamma knife CPM, carbo SD 20 Leptomeningeal Germinoma 1 None None CPM, VP CR 21 Lateral ventricle, third ventricle Germinoma 2 Biopsy WB CPM, carbo, VP PR NOTE. Therapies at all relapses are summarized together. Abbreviations: NGGCT, nongerminomatous germ cell tumor; Par res, partial resection; PR, partial response; CPM, cyclophosphamide; CDDP, cisplatin; VP, etoposide; B, bleomycin; WB, whole brain; CR, complete response; VIN, vinblastine; carbo, carboplatin; CS, craniospinal; SD, stable disease. enced relapse or disease progression after completing treatment, with median time to first relapse of 12 months (range, 7 to 120 months). One patient developed PD while receiving initial chemotherapy (patient 9). Patients received a wide variety of therapeutic regimens at initial diagnosis. Additional clinical details of patients at diagnosis are described in Table 1. Relapse. Nine patients were diagnosed with germinoma at relapse and 12 patients were diagnosed with NGGCTs. The median number of relapses before HDC was two (range, one to four). Patient characteristics at relapse are described in Table 2. Four patients experienced local relapse and 16 had disseminated CNS disease at relapse, including five patients with spinal cord involvement. Two patients (patients 11 and 16) had abdominal dissemination of disease: one at the time of intracranial relapse and the other as an isolated abdominal dissemination. One patient (patient 9) had PD while receiving initial chemotherapy. Seventeen patients received conventional-dose chemotherapy with or without irradiation in an effort to reduce tumor burden before proceeding to HDC, and one patient (patient 8) received myeloablative doses of cyclophosphamide and melphalan. Sixteen of 17 patients had chemotherapyresponsive disease: a partial response (PR) was achieved by eight patients and a complete response (CR) was achieved by eight patients; one patient had stable disease (SD) after salvage chemotherapy. One patient (patient 1), who was diagnosed initially as having a primitive neuroectodermal tumor, received only surgery both at initial therapy and at relapse; the diagnosis of NGGCTs was made after pathology review only after HDC had been administered. The remaining three patients did not receive any treatment at relapse before HDC: two had localized pituitary relapses and received focal irradiation after HDC and ASCR (patients 4 and 6), and the third patient (patient 9) had PD as manifested by rapidly increasing serum tumor markers while receiving initial chemotherapy. Eight patients had no assessable disease before HDC. Ten of 21 patients had radiographically assessable disease before HDC, and eight of 20 evaluated patients had elevated CSF markers (Table 3). HDC Protocols All patients received thiotepa-based HDC protocols (Table 3). Regimens and dose modifications are listed in Table 4. Dose modifications in planned protocols were carried out if unacceptable toxicity was encountered or anticipated, or in two patients as a result of a then-prevailing national shortage of thiotepa. ASCR was achieved with bone marrow (BM) in three patients and peripheral-blood stem cells (PBSC) in an additional 17 patients. One patient required a PBSC boost after prior BM rescue because of persistent myelosuppression. Ten patients received radiation therapy after ASCR. Post-HDC Therapy Wherever feasible, efforts were made to consolidate post- HDC remission with radiation therapy either focally or to the entire neuraxis (Table 3). Evaluation of Tumor Response Responses were documented by neuroimaging studies and by measuring tumor markers in the CSF, if positive before ASCR. Responses were graded as follows. CR required the disappearance of all radiographically assessable tumor as well as normalization of CSF tumor markers if previously elevated. PR was considered a greater than 50% decrease in the size of radiographically visible tumor and a reduction in previously elevated CSF markers. SD was 1936 JOURNAL OF CLINICAL ONCOLOGY

4 High-Dose Chemotherapy in Recurrent CNS GCT Patient Diagnosis Pre-HDC Radiology Table 3. Clinical Characteristics of Patients Before and After HDC Pre-HDC CSF Markers HDC Regimen Stem Cell Source Short-Term Nonhematologic Toxicity Radiologic Response 1 NGGCT Parietooccipital Not evaluated TE BM None SD 2 NGGCT NAD Normal CTE PBL None CCR 3 NGGCT Pineal and Not evaluated CTE BM Idiopathic pneumonitis PR spinal cord 4 NGGCT Pineal -HCG elevated TT BM PBL None CR 5 NGGCT Pineal -HCG elevated TE BM Grade 4 mucositis requiring intubation SD 6 NGGCT NAD -HCG elevated TT PBL None CCR 7 Germinoma Pineal -HCG elevated TT PBL None CR 8 NGGCT NAD Normal CTTemo PBL None CCR 9 NGGCT NAD AFP elevated TT PBL None CCR 10 Germinoma NAD Normal CTE PBL None CCR 11 NGGCT NAD Normal CTE PBL None CCR 12 Germinoma Disseminated Normal TT-1 PBL None PR 13 NGGCT Disseminated AFP elevated CTE-2 PBL Transient grade 3 renal, grade 3 mucositis, PR pulmonary hemorrhage, grade 3 hyperbilirubinemia 14 Germinoma NAD -HCG elevated CTE-1 PBL Transient renal dysfunction CCR 15 Germinoma Suprasellar Normal CTE PBL None CR 16 Germinoma NAD Normal CTE PBL None CCR 17 Germinoma NAD Normal CTE PBL None CCR 18 NGGCT NAD Normal CTTemo PBL None CCR 19 NGGCT Pineal AFP elevated TE-2 PBL None SD 20 Germinoma NAD Normal CTTemo PBL None CCR 21 Germinoma Leptomeningeal Normal TT PBL None CR Patient CSF Marker Response Post-HDC Adjuvant Therapy Therapy for Post-HDC Relapse Outcome OS (months) EFS (months) 1 Not evaluated None None Died of PD CCR None Not applicable NAD Not evaluated None None Died of PD CR CS irradiation Not applicable NAD Decrease None None Died of PD CR Focal irradiation Not applicable NAD CR CS irradiation Not applicable NAD CCR None Oral VP, temozolomide; LM recurrence; currently NAD WB focal RT 9 Decrease CS focal Oral VP Died of PD CCR None Not applicable Died of bleomycin-related pulmonary fibrosis 11 CCR None Not applicable NAD CCR None Low-dose thiotepa, Died of PD 39 4 procarbazine 13 Decrease None None Died of PD CR Cranial irradiation Not applicable NAD CCR CS irradiation Not applicable NAD CCR None Not applicable NAD CCR None Not applicable NAD CCR CS focal irradiation Topotecan, irinotecan Died of PD Decrease Stereotactic radiosurgery None Died of PD CCR CS focal irradiation Not applicable NAD CCR None Not applicable NAD Abbreviations: HDC, high-dose chemotherapy; OS, overall survival; EFS, event-free survival; NGGCT, nongerminomatous germ cell tumor; BM, bone marrow; SD, stable disease; PD, progressive disease; NAD, no assessable disease; PBL, peripheral blood mononuclear cells; CCR, continued complete response; PR, partial response; CS, craniospinal; -HCG, beta human chorionic gonadotropin; LM, leptomeningeal; VP, etoposide; WB, whole brain; RT, radiation therapy; AFP, alpha-fetoprotein. Chemotherapy regimens are described in Table 4. a less than 25% decrease in tumor size or no more than 10% increase in tumor size. PD was defined as an increase of at least 25% in the sum of the products of the maximum perpendicular diameters of existing tumors, or any increase in levels of tumor markers. Recurrent disease was defined as the appearance of new radiographic tumor or elevation of tumor markers who had previously experienced CR. Toxicity was graded according to WHO criteria

5 Modak et al Abbreviation Table 4. High-Dose Chemotherapy Regimens Chemotherapeutic Drugs Used CTE Carboplatin 500 mg/m 2 /d or dosed according to Calvert formula with AUC 7 (lower dose used) on days 1-3 followed by thiotepa 300 mg/m 2 /d and etoposide 250 mg/m 2 /d on days 4-6 TE Thiotepa 300 mg/m 2 /d and etoposide 250 mg/m 2 /d on days 1-3 TT Thiotepa 200 mg/m 2 /d for 3 days with ASCR; followed 4 weeks later by thiotepa 200 mg/m 2 /d for 3 days with ASCR CTTemo Carboplatin 500 mg/m 2 /d or dosed according to Calvert formula with AUC 7 (lower dose used) on days 1-3 followed by thiotepa 300 mg/m 2 /d on days 4-6 and temozolomide mg/m 2 /d NOTE. Refer to Table 3 for patients who received specific dose modifications described below. Abbreviations: AUC, area under the time-concentration curve; ASCR, autologous stem-cell rescue. CTE-1: Carboplatin 500 mg/m 2 /d for 3 days followed by only one dose each of thiotepa 300 mg/m 2 /d and etoposide 250 mg/m 2 /d because of renal dysfunction. This was followed by ASCR. The patient subsequently received two additional doses of thiotepa 300 mg/m 2 /d followed by ASCR. CTE-2: Carboplatin 500 mg/m 2 /d for 3 days followed by only one dose each of thiotepa 300 mg/m 2 /d and etoposide 250 mg/m 2 /d because of renal dysfunction. This was followed by ASCR. TE-1: Etoposide 500 mg/m 2 /d for 3 days was administered alone because of a then worldwide shortage of thiotepa, followed by ASCR. The patient subsequently received thiotepa 200 mg/m 2 /d for 3 days followed by ASCR. TT-1: Thiotepa 200 mg/m 2 /d for 3 days with ASCR, followed 4 weeks later by thiotepa 200 mg/m 2 /d for 1 day. Dose reduction was carried out to reduce potential toxicity in this heavily pretreated patient. Statistical Analysis EFS was assessed from the date ASCR after HDC to the date of progression or death (whichever occurred earlier), with censoring on the date of most recent contact. Overall survival (OS) was assessed from the date of ASCR to death, with censoring on the date of most recent contact. In patients receiving two sequential myeloablative treatments, EFS and OS were assessed from the date of first ASCR. EFS and OS data were analyzed as of September 1, Estimated EFS and OS rates were calculated according to the Kaplan-Meier method and log-rank test was used for determination of statistical significance. RESULTS Clinical details before and after HDC are described in Table 3. Disease Response Germinoma. Of the nine patients with germinoma, four had radiologically assessable tumors before HDC. Three of these four patients achieved radiologic CR and one achieved PR after HDC. The remaining five patients without radiologically assessable disease continue to remain disease free for a median of 48 months after HDC (range, 6 to 87 months). Two patients with GGCT had elevated CSF markers that normalized after HDC. The patient with radiologic PR after HDC experienced disease progression 4 months after HDC, with additional elevation of CSF beta human chorionic gonadotropin ( -HCG; patient 12). NGGCTs. Of the 12 patients with NGGCTs, six had radiologically assessable disease before HDC. One of six patients achieved radiologic CR, two achieved PR, and three had SD. The patient with radiologic CR (patient 4) has not experienced disease relapse 31 months after HDC; however, all patients with PR and SD experienced disease progression within short duration of HDC (median, 3 months; range, 2 to 4 months). Six of 10 evaluated patients had elevated CSF markers before HDC. Two had complete normalization of markers and the other four had reduction in levels in CSF. One patient with normalization of CSF markers also had radiologic CR as described above. A second patient who achieved normalization of CSF markers (patient 6) survives free of disease 35 months after HDC. All patients who did not achieve marker normalization had PD within 2 to 6 months after HDC. Survival Survival rates are summarized in Table 5. Diagnosis at relapse predicted OS and EFS; patients with germinomas fared better than those with NGGCTs (P.016 for OS and P.014 for EFS; Fig 1). Germinoma. Seven of the nine patients with germinoma survive disease free for a median of 48 months (range, 6 to 87 months) after HDC. One of the patients died (patient 10) 78 months after HDC without disease recurrence as a result of pulmonary fibrosis secondary to bleomycin therapy administered as part of chemotherapy at initial diagnosis, and spinal irradiation delivered after HDC. A second patient with germinoma (patient 12) experienced disease relapse 4 months after HDC with slow disease progression marked by transient partial disease responses to low-dose thiotepa and procarbazine after HDC. This patient died as a result of disease 39 months after HDC. NGGCTs. Seven of 12 patients with NGGCTs died as a result of PD within a median of 4 months after HDC (range, 2 to 17 months). Four patients survive disease free with a median survival of 33 months after HDC (range, 24 to 55 months). An additional patient (patient 8) with NGGCTs survives despite experiencing disease relapse after receiving thiotepa-based HDC. At his first relapse, he had been treated with a myeloablative regimen of cyclophosphamide and melphalan followed by ASCR. He had two additional relapses, the first of which was treated with a thiotepacontaining myeloablative regimen followed by a second ASCR. He subsequently had a second relapse, responded to 1938 JOURNAL OF CLINICAL ONCOLOGY

6 High-Dose Chemotherapy in Recurrent CNS GCT Table 5. Survival Data All Patients Germinoma NGGCT Outcome No. of Patients % No. of Patients % No. of Patients % OS EFS Kaplan-Meier estimate of OS at 4 years Mean (%) Standard deviation (%) Kaplan-Meier estimate of EFS at 4 years Mean (%) Standard deviation (%) Median follow-up, months Median survival, months Abbreviations: NGGCT, nongerminomatous germ cell tumor; OS, overall survival; EFS, event-free survival. whole-brain radiotherapy followed by oral temozolomide and etoposide, and now survives without assessable disease 31 months after his second thiotepa-based HDC regimen. Survival correlates. Response to HDC in patients with assessable disease was predictive of survival (Fig 2). All six patients who achieved CR after HDC survive disease free with a median survival time of 48 months after HDC, whereas none of the seven patients who had PR or SD after HDC survive (P.001 for both OS and EFS). Furthermore, achievement of a state of no assessable disease (NAD) after HDC was critical to improved survival (P.001 for both OS and EFS when compared with patients who could never be rendered free of disease). Eleven of 14 patients (79%) with NAD after HDC (either those rendered disease free by HDC or those administered HDC in the absence of assessable disease) survive disease free with a median survival of 45 months. One of the remaining three patients (patient 10) experienced a toxic death 78 months after HDC without assessable disease. An additional patient (patient 7) responded to salvage therapy with whole-brain irradiation followed by oral temozolomide alternating with oral etoposide and survives without assessable disease 31 months after HDC. Conversely, none of the seven patients who had evidence of disease after HDC survive (Fig 3). There was no statistically significant difference in survival between patients who were administered HDC in the presence of residual disease, defined as presence of either radiographic abnormalities or elevated CSF markers, as compared with those who were disease free (P.45 for OS and.68 for EFS). Fig 1. Kaplan-Meier survival curve comparing event-free survival for patients with germinoma to patients with nongerminomatous germ cell tumors (NGGCTs). HDC, high-dose chemotherapy. Fig 2. Kaplan-Meier survival curve comparing event-free survival for patients who had a complete response (CR) to high-dose chemotherapy (HDC) to patients who did not achieve CR

7 Modak et al spinal irradiation, and died as a result of respiratory failure 78 months after HDC. No additional toxicities attributable to chemotherapy were observed. No HDC-related toxic deaths occurred. DISCUSSION Fig 3. Kaplan-Meier survival curve comparing overall survival for patients who did not have assessable disease after high-dose chemotherapy (HDC) compared with patients who had residual disease. Toxicity As expected, all patients experienced grade 4 hematologic toxicity. There was delayed engraftment in two patients. Patient 10, who received PBSC rescue, was dependent on platelet and red cell transfusions for 18 months after HDC. Patient 5 was dependent on platelet and red cell transfusion 5 months after autologous BM rescue, necessitating a boost of stored PBSC. In another heavily pretreated patient who also received post-ascr therapy with oral etoposide, chlorambucil, and nonmyeloablative doses of thiotepa, a need for occasional transfusion of red cells was noted (patient 12). In all other patients, engraftment of all cell lines was observed. Two patients (patients 8 and 9) developed reversible elevations of blood urea nitrogen and creatinine first noted after receiving two doses of carboplatin, and required modifications of their myeloablative chemotherapy. Grade 4 pulmonary toxicities related to HDC were noted in two patients. One patient (patient 3) developed transient idiopathic pneumonitis in the post-hdc period. Another patient developed a pulmonary hemorrhage in the post-ascr phase, which resolved with platelet transfusions (patient 13). This patient also developed transient spontaneously resolving hyperbilirubinemia in the immediate post-hdc phase. A third patient developed grade 4 mucositis requiring transient intubation and ventilation (patient 4). In all patients pulmonary toxicities had resolved by the time progressive CNS disease was noted. One patient (patient 10), who had prolonged myelosuppression after ASCR, developed severe pulmonary fibrosis related to initial treatment with bleomycin and post-hdc Although patients with CNS GCTs who experience relapse after chemotherapy can, with some success, receive salvage radiation therapy, 9,17 the prognosis in patients with relapsed NGGCT, those with multiply relapsed NG- GCT, and those who experience relapse after initial combined irradiation and chemotherapy remains bleak. A radical approach to these patients is therefore justified in an attempt to improve survival. Our rationale for the use of HDC in this group was based on the following observations: CNS GCTs have similar histologic and biologic characteristics, and share the chemotherapy sensitivity of their systemic counterparts; BM has only rarely been reported as a metastatic site, 24 permitting hematopoietic reconstitution after dose escalation of chemotherapy with minimal risk of tumor dissemination; and myeloablative chemotherapy has proven effective in retrieving patients who experience relapse with systemic GCTs 18,19,25 and for the treatment of systemic GCTs metastatic to the brain. 26 Although the choice of chemotherapeutic agents in our patients was dictated by existing protocols for recurrent CNS tumors in general, GCTs have shown sensitivity to the chemotherapy drugs used in our protocols. Alkylating agents have been shown to be active against both CNS and systemic GCTs 27 and given that their dose-limiting toxicity is primarily hematologic, these are attractive agents for high-dose therapy followed by ASCR. Thiotepa, 28 specifically, has been shown to cross the blood-brain barrier extremely efficiently and has a mild spectrum of nonhematologic side effects. Dose-response effects have been suggested for both cisplatin 29 and carboplatin 30 in patients with systemic GCTs. Carboplatin has been reported to be effective in phase II trials in CNS GCT. 22 Etoposide has the ability to cross the blood-brain barrier, and because its mechanism of action is inhibition of cellular DNA repair, it has a theoretical advantage when used along with alkylators, which inflict DNA damage. A dose-response effect has also been suggested for etoposide in systemic GCTs. 31 Despite using several different HDC regimens, objective responses to HDC were achieved in 12 of 13 patients with assessable disease in an extremely high-risk patient population, most of whom had experienced multiple relapses of disease. The chemotherapy sensitivity demonstrated by tumors in our heavily pretreated patients is in keeping with several studies demonstrating the benefit of an approach using HDC with autologous hematopoietic stemcell support in patients with refractory and recurrent systemic 1940 JOURNAL OF CLINICAL ONCOLOGY

8 High-Dose Chemotherapy in Recurrent CNS GCT (non-cns) GCTs. Various combinations using cisplatin, carboplatin, etoposide, cyclophosphamide, and ifosfamide in high doses followed by stem-cell rescue have shown efficacy even in heavily pretreated patients. 18,19,30,32 A large retrospective study did not favor any particular regimen. 25 Recognizing that toxicity data obtained from a retrospective review may be incomplete, in general, in this heavily pretreated group of patients, HDC was well tolerated. In the three patients who developed significant mucositis and pulmonary complications, toxicities resolved with supportive care. Although one patient died as a result of pulmonary fibrosis secondary to prior bleomycin therapy and after HDC spinal irradiation, no toxic deaths could be attributed to the HDC regimens used and no chemotherapy-related long-term morbidity has been observed to date in patients who have survived their disease. Similar HDC protocols have been used by our group and others for the treatment of other high-risk brain tumors and their toxicities have been well described The estimated OS and EFS rates at 4 years after HDC in our study were 58% 12% and 52% 14%, respectively. We used chemotherapy and/or radiation therapy in 18 of 21 patients in an attempt to reduce tumor burden before proceeding for HDC. Two of the remaining three patients had modest elevations of CSF -HCG and pineal and pituitary abnormalities on magnetic resonance imaging before HDC. Both attained CR, were consolidated with radiotherapy, and survive without assessable disease 31 and 35 months after HDC (patients 4 and 6). The third patient was the only patient who received HDC with PD. Of the eight patients who achieved CR with salvage protocols before consolidation with HDC, five survive without assessable disease 6 to 48 months after HDC (median, 41 months). One patient did not have assessable disease when he died of bleomycin-induced pulmonary fibrosis 78 months after transplantation. Similar results have also been reported by the French Society of Pediatric Oncology in a preliminary report. 37 Six of six patients with recurrent CNS GCT, when treated with HDC when in CR, remained in remission. Conversely, both patients with PD treated with HDC, although responding to HDC, experienced disease progression and died. In our study, the patient with PD responded initially but then experienced disease progression within 4 months and died as a result of disease. Overall, the French Society of Pediatric Oncology study reports on 13 patients (four with germinoma and nine with secretory tumors) treated with a pilot HDC regimen consisting of etoposide 1.5 g/m 2 and thiotepa 900 mg/m 2 during 3 days followed by ASCR. Ten patients survived, with a median follow-up of 16 months after HDC. Mahoney et al 38 from the Pediatric Oncology Group have reported CRs to a myeloablative regimen of cyclophosphamide and melphalan in two patients with recurrent CNS GCTs. Both patients had pure germinomas; one had a progression-free survival of 11 months after ASCR and the other had a progression-free survival of 30 months. Graham et al 39 from the same group reported on two patients with NGGCTs who were treated with HDC with ASCR in first remission; the patients had previously received chemotherapy and radiation therapy without any residual disease before HDC. Although the authors do not state the particular high-dose regimen received by the patients, they indicate that both patients remain disease free 30 and 22 months after therapy. We observed better responses in patients with recurrent and progressive germinomas (three of four achieved CR and one achieved PR) as compared with patients with NGGCTs (two of seven achieved CR, two achieved PR, and three achieved SD). Estimated OFS and EFS were also superior for patients with germinoma (86% 13% and 67% 21%, respectively) versus NGGCTs (42% 14% and 33% 14%, respectively). The four patients with NGGCTs who survive either did not have assessable disease before HDC or had modest -HCG elevation and radiographic abnormalities. Our results are in keeping with the better prognosis at diagnosis for patients with CNS germinomas. In our study, no correlation was observed between survival and prior treatment, site of relapse, or age of patients, nor with specific histologic diagnosis in patients with NGGCTs (data not shown). Although wherever possible, efforts were made to administer post-hdc radiation therapy to consolidate remission, in this heavily pretreated group of patients, irradiation could be administered safely to only seven of 14 patients with no assessable disease after HDC. There was no correlation between survival and post- HDC radiation therapy (data not shown). It can be speculated that the demonstrated chemotherapy sensitivity of tumors in our patients treated with HDC can have implications for the therapeutic approach to highrisk patients at initial diagnosis. If such high-risk patients can be defined, it is conceivable that they might benefit from an approach using HDC with ASCR as a consolidative regimen after conventional induction chemotherapy or radiation therapy, rather than awaiting relapse. Prognostic markers identified in patients with poor-prognosis systemic GCTs include disease refractory to conventional chemotherapy and those with high tumor markers. 25 Investigators in Japan and Germany have postulated that high CSF or serum tumor markers at diagnosis are associated with poor prognosis. 14,16 Using a similar approach, Tada et al 40 treated six patients who had secreting CNS GCTs containing highly malignant components with HDC after CR was achieved using multimodality therapy. All patients remain disease free after discontinuing therapy 1 to 7 years after diagnosis. Patient numbers in our study were small and the population was highly heterogeneous. Treatment regimens varied because patients had differing histologies and different relapse patterns, and the prevailing regimens available

9 Modak et al and/or in use for patients experiencing disease relapse changed during a 17-year period. However, we have shown that a myeloablative chemotherapeutic approach is feasible in this group of patients with poor prognosis and can be safely administered; no toxic deaths were encountered. Objective clinical responses were observed in a majority of patients and encouraging OS and EFS was observed, particularly for patients with relapsed or progressive CNS germinoma. The small population size precludes us from drawing definitive conclusions about the role of HDC and ASCR in patients with CNS GCTs. Nevertheless, on the basis of our study, and on the well-established chemotherapy sensitivity of these tumors at diagnosis, HDC might be beneficial for patients with germinomas that are recurrent after irradiation alone or after combination radiation and chemotherapy. For patients with NGGCTs at relapse, if tumor burden can be minimized, HDC may be considered for consolidating remission. Authors Disclosures of Potential Conflicts of Interest The authors indicated no potential conflicts of interest. REFERENCES 1. Bloom HJ: Primary intracranial germ cell tumors. Clin Oncol 2: , Jennings MT, Gelman R, Hochberg F: Intracranial germ-cell tumors: Natural history and pathogenesis. J Neurosurg 63: , Bjornsson J, Scheithauer BW, Okazaki H, et al: Intracranial germ cell tumors: Pathobiological and immunohistochemical aspects of 70 cases. J Neuropathol Exp Neurol 44:32-46, Matsutani M, Sano K, Takakura K, et al: Primary intracranial germ cell tumors: A clinical analysis of 153 histologically verified cases. J Neurosurg 86: , Shibamoto Y, Abe M, Yamashita J, et al: Treatment results of intracranial germinoma as a function of the irradiated volume. Int J Radiat Oncol Biol Phys 15: , Patel SR, Buckner JC, Smithson WA, et al: Cisplatin-based chemotherapy in primary central nervous system germ cell tumors. J Neurooncol 12:47-52, Kobayashi T, Yoshida J, Ishiyama J, et al: Combination chemotherapy with cisplatin and etoposide for malignant intracranial germ-cell tumors: An experimental and clinical study. J Neurosurg 70: , Bamberg M, Kortmann RD, Calaminus G, et al: Radiation therapy for intracranial germinoma: Results of the German cooperative prospective trials MAKEI 83/86/89. J Clin Oncol 17: , Balmaceda C, Heller G, Rosenblum M, et al: Chemotherapy without irradiation A novel approach for newly diagnosed CNS germ cell tumors: Results of an international cooperative trial The First International Central Nervous System Germ Cell Tumor Study. J Clin Oncol 14: , Bouffet E, Baranzelli MC, Patte C, et al: Combined treatment modality for intracranial germinomas: Results of a multicentre SFOP experience Societe Francaise d Oncologie Pediatrique. Br J Cancer 79: , Dearnaley DP, A Hern RP, Whittaker S, et al: Pineal and CNS germ cell tumors: Royal Marsden Hospital experience Int J Radiat Oncol Biol Phys 18: , Allen JC, Walker R, Luks E, et al: Carboplatin and recurrent childhood brain tumors. J Clin Oncol 5: , Yoshida J, Sugita K, Kobayashi T, et al: Prognosis of intracranial germ cell tumours: Effectiveness of chemotherapy with cisplatin and etoposide (CDDP and VP-16). Acta Neurochir (Wien) 120: , Matsutani M: Combined chemotherapy and radiation therapy for CNS germ cell tumors: The Japanese experience. J Neurooncol 54: , Calaminus G, Bamberg M, Baranzelli MC, et al: Intracranial germ cell tumors: A comprehensive update of the European data. Neuropediatrics 25:26-32, Calaminus G, Andreussi L, Garre ML, et al: Secreting germ cell tumors of the central nervous system (CNS): First results of the cooperative German/Italian pilot study (CNS sgct). Klin Padiatr 209: , Merchant TE, Davis BJ, Sheldon JM, et al: Radiation therapy for relapsed CNS germinoma after primary chemotherapy. J Clin Oncol 16: , Motzer RJ, Mazumdar M, Bosl GJ, et al: High-dose carboplatin, etoposide, and cyclophosphamide for patients with refractory germ cell tumors: Treatment results and prognostic factors for survival and toxicity. J Clin Oncol 14: , Nichols CR, Andersen J, Lazarus HM, et al: High-dose carboplatin and etoposide with autologous bone marrow transplantation in refractory germ cell cancer: An Eastern Cooperative Oncology Group protocol. J Clin Oncol 10: , van Gurp RJ, Oosterhuis JW, Kalscheuer V, et al: Biallelic expression of the H19 and IGF2 genes in human testicular germ cell tumors. J Natl Cancer Inst 86: , Bussey KJ, Lawce HJ, Olson SB, et al: Chromosome abnormalities of eighty-one pediatric germ cell tumors: Sex-, age-, site-, and histopathology-related differences A Children s Cancer Group study. Genes Chromosomes Cancer 25: , Allen JC, DaRosso RC, Donahue B, et al: A phase II trial of preirradiation carboplatin in newly diagnosed germinoma of the central nervous system. Cancer 74: , Peckham MJ, Horwich A, Hendry WF: Advanced seminoma: Treatment with cisplatinum-based combination chemotherapy or carboplatin (JM8). Br J Cancer 52:7-13, Delahunt B: Suprasellar germinoma with probable extracranial metastases. Pathology 14: , Beyer J, Kramar A, Mandanas R, et al: High-dose chemotherapy as salvage treatment in germ cell tumors: A multivariate analysis of prognostic variables. J Clin Oncol 14: , Kollmannsberger C, Nichols C, Bamberg M, et al: First-line high-dose chemotherapy /- radiation therapy in patients with metastatic germ-cell cancer and brain metastases. Ann Oncol 11: , Allen JC, Bosl G, Walker R: Chemotherapy trials in recurrent primary intracranial germ cell tumors. J Neurooncol 3: , Strong JM, Collins JM, Lester C, et al: Pharmacokinetics of intraventricular and intravenous N,N,N -triethylenethiophosphoramide (thiotepa) in rhesus monkeys and humans. Cancer Res 46: , Samson MK, Rivkin SE, Jones SE, et al: Dose-response and dose-survival advantage for high versus low-dose cisplatin combined with vinblastine and bleomycin in disseminated testicular cancer: A Southwest Oncology Group study. Cancer 53: , Nichols CR, Tricot G, Williams SD, et al: Dose-intensive chemotherapy in refractory germ cell cancer: A phase I/II trial of high-dose carboplatin and etoposide with autologous bone marrow transplantation. J Clin Oncol 7: , Wolff SN, Johnson DH, Hainsworth JD, et al: High-dose VP monotherapy for refractory germinal malignancies: A phase II study. J Clin Oncol 2: , Siegert W, Rick O, Beyer J: High-dose chemotherapy with autologous stem cell support in poor-risk germ cell tumors. Ann Hematol 76: , Dunkel IJ, Finlay JL: High-dose chemotherapy with autologous stem cell rescue for brain tumors. Crit Rev Oncol Hematol 41: , Gururangan S, Dunkel IJ, Goldman S, et al: Myeloablative chemotherapy with autologous bone marrow rescue in young children with recurrent malignant brain tumors. J Clin Oncol 16: , Dunkel IJ, Boyett JM, Yates A, et al: High-dose carboplatin, thiotepa, and etoposide with autologous stem-cell rescue for patients 1942 JOURNAL OF CLINICAL ONCOLOGY

10 High-Dose Chemotherapy in Recurrent CNS GCT with recurrent medulloblastoma: Children s Cancer Group. J Clin Oncol 16: , Mason WP, Grovas A, Halpern S, et al: Intensive chemotherapy and bone marrow rescue for young children with newly diagnosed malignant brain tumors. J Clin Oncol 16: , Baranzelli MC, Pichon F, Patte C, et al: High-dose etoposide and thio-tepa for recurrent intracranial malignant germ cell tumours: Experience of the SFOP. Childs Nerv Syst 14:520, Mahoney DH Jr, Strother D, Camitta B, et al: High-dose melphalan and cyclophosphamide with autologous bone marrow rescue for recurrent/progressive malignant brain tumors in children: A pilot pediatric oncology group study. J Clin Oncol 14: , Graham ML, Herndon JE II, Casey JR, et al: High-dose chemotherapy with autologous stem-cell rescue in patients with recurrent and high-risk pediatric brain tumors. J Clin Oncol 15: , Tada T, Takizawa T, Nakazato F, et al: Treatment of intracranial nongerminomatous germ-cell tumor by high-dose chemotherapy and autologous stem-cell rescue. J Neurooncol 44:71-76,