Eligible patients were age 3 to 30 years at the time of diagnosis with average-risk, histologically proven medulloblastoma as determined by each participating institution. Average-risk was defined as less than 1.5 cm³ residual tumor following resection by neuroimaging, no evidence of metastases by brain and spine MRI, and no tumors cells present on cerebrospinal fluid (CSF) cytology. CSF was obtained by lumbar puncture between 10 days and 4 weeks after surgery. Postoperative MRI of the primary site with and without contrast was required within 72 hours of surgery, preferably within 48 hours; volumetrics were performed by neuroradiologists at their respective institutions.

This single-arm study was developed by the authors and conducted at three institutions in the United States. The primary objective was to reduce the late toxicity of CSI without compromising the therapeutic efficacy compared with standard therapy [86% ± 4% standard error of the mean (SEM) three-year relapse-free survival per CCG 9892 (12)]. The secondary objectives were to assess relapse-free survival and overall survival 5 years from time of study enrollment.

Surgical resection was carried out according to institutional standards with staging as stated above. Following institutional diagnosis of medulloblastoma, histologic slides were sent to the Children’s Hospital of Philadelphia for correlative studies.

Patients began CSI within four weeks of surgical resection; concurrent weekly intravenous (IV) vincristine 1.5mg/m² (maximum single dose 2 mg) was administered for 6 doses. Radiation therapy consisted of 18 Gy photons to the whole brain and spine in 10 fractions with a boost to the tumor bed in 21 fractions for a total dose of 55.8 Gy to the tumor bed. Four weeks after completion of radiotherapy, maintenance chemotherapy was initiated, consisting of 9 cycles of regimens A and B in an AABAABAAB pattern. Regimen A included oral lomustine 75 mg/m² and IV cisplatin 70 mg/m² on day 0, and IV vincristine 1.5 mg/m² (maximum single dose 2 mg) on days 0, 7, and 14. For regimen B, patients received IV cyclophosphamide 1000 mg/m² on days 0 and 1 and IV etoposide 150 mg/m² on days 0 and 1, followed by oral etoposide 50 mg/m² on days 14-34.

All patients were monitored for adverse events according to the National Cancer Institute Common Terminology Criteria for Adverse Events, version 3.0. MRI of the brain and spine with and without gadolinium was obtained prior to surgery, post-operatively, in 3-month intervals during maintenance chemotherapy and in the first year after the end of treatment. Participants then underwent imaging every 6 months for years 2 and 3, and yearly thereafter or if clinically indicated for suspected relapse. Relapse was defined as a greater than 25% increase in the product of the longest two perpendicular measurements of residual tumor on MRI, evidence of new tumor by MRI, or de novo tumor cells in the CSF. Patients were removed from study for relapse, unacceptable toxicity or other extraordinary medical circumstances, inability or refusal to comply with study treatment or follow up with required observations, or death.

Audiograms were obtained prior to the start of each chemotherapy cycle, every 6 months from one to three years after the completion of chemotherapy and yearly thereafter. Laboratory studies to evaluate liver, kidney and endocrinologic function as well as MRI were obtained serially according to study protocol. Neurocognitive testing utilizing age-appropriate Wechsler Scales was obtained within 6 months of diagnosis and at one, two and three years after the completion of chemotherapy.

The study was approved by institutional review boards at the participating institutions and conducted in accordance with the Declaration of Helsinki. All patients or their legal guardians provided written informed consent and age-appropriate assent was obtained. ClinicalTrials.gov identifier: NCT00031590.

The target accrual of the study was 50 patients over 3 years, assuming that 15% represents an acceptable difference in relapse-free survival between the study and the standard survival rate of 86 ± 4% at 3 years in CCG 9892 (12). With a sample size of 50, 11 events would represent relapse-free survival of 78% with a 90% confidence interval (CI) of 70.8% - 85.8%. Stopping rules were designed to close accrual if local relapses, exoprimary relapses or toxicity of the protocol exceeded those reported in CCG 9892 (12). Relapse-free survival (RFS) was defined as the time to first disease progression, disease recurrence, death from any cause, or occurrence of a subsequent malignant neoplasm. Overall survival (OS) was defined as time to death from any cause. Time to first relapse was defined as the time from diagnosis to the time of tumor relapse as previously defined. RFS and OS and were calculated using the methods of Kaplan and Meier.

Thirty patients were assessed for eligibility; 2 were excluded as they did not meet inclusion criteria: one patient due to presence of tumor cells in CSF by cytology at diagnosis, and the other due to a delay in starting radiation therapy (Figure 1). A total of 28 patients were enrolled on study between April 2001 and December 2010 across three institutions (Children’s Hospital of Philadelphia, Children’s Hospital of Atlanta, and Lucile Packard Children’s Hospital at Stanford University). Median age at enrollment was 9 ± 0.8 years (range: 2.8 to 17.7 years). Collected patient characteristics are listed in Table 1. The study was terminated early due to slow accrual in light of evolving knowledge of molecular subgrouping and risk stratification in medulloblastoma.

Two patients (7%) withdrew from study; one due to toxicity from chemotherapy and the other due to withdrawal of consent. Both patients were included in the intent-to-treat analysis. Eight patients (29%) received all 6 cycles of cisplatin-containing chemotherapy whereas the remainder (20, 71%) had dose reductions, omissions or were changed to carboplatin for renal toxicity, ototoxicity, or parental preference, with approval of the study chair. No dose reductions in regimen B (IV cyclophosphamide, IV and oral etoposide) were required. Of the 25 patients with endocrinologic monitoring data, 15 (60%) required thyroid hormone replacement and 7 (28%) required growth hormone replacement at some point during or after their treatment course. Collected toxicity data are summarized in Table 2. There were no grade 5 toxicities.

The median follow-up time was 140 months (11.7 years; interquartile range 111 to 158 months) by reverse Kaplan-Meier estimator (15). The 3-year RFS and OS were 79% (95% CI 58% to 90%) and 93% (95% CI 74% to 98%), respectively. The 5-year RFS and OS were 71% (95% CI 50% to 85%) and 86% (95% CI 66% to 94%), respectively (Figures 2A, B). Nineteen patients (68%) were alive at the time of last contact with no evidence of disease. Seven patients (25%) died of disease, whereas one patient died of secondary glioblastoma and one (who was later diagnosed with a germline TP53 mutation) of secondary skull base sarcoma.

Eighteen patients (64%) were greater than 7 years of age at enrollment. For this subgroup, the 3- and 5-year OS were 83% (95% CI 68% to 100%) and 78% (95% CI 61% to 100%). Of these 18 patients, 11 (61%) were alive at the time of last contact with no evidence of disease at a median follow up time of 11.7 ± 1.0 years. There was no difference in 3- and 5-year OS when compared to patients less than or equal to 7 years of age or in all patients when stratified by sex.

Seven patients (25%) developed relapsed disease: two patients with diffuse or leptomeningeal relapse including the primary site; three with isolated intraspinal tumor; one with exoprimary brain relapse solely in the pituitary; and one patient who initially developed primary site relapse and subsequently disseminated disease.

Partial neurocognitive data was available on 15 patients at multiple different time points, limiting complete analysis. Collected data demonstrated mixed changes over time (Supplementary Figure 1). One year from the initiation of therapy, one patient showed an increase of greater than 10 IQ points, one patient demonstrated a decrease of greater than 10 IQ points. The remaining six patients from whom data was collected remained within 10 IQ points of their baseline. At 3 years, two patients experienced a decline in greater than 10 IQ points, one remained stable and another had an increase of greater than 10 IQ points from baseline.