We reviewed the medical records of all patients younger than 18 years, who were diagnosed with HGGs at Samsung Medical Center between February 2006 and October 2015. Inclusion criteria were WHO grade 3 or 4 astrocytic tumors (glioblastoma, anaplastic astrocytoma, and gliomatosis cerebri) and other rare HGGs. Pontine glioma was not included in the analysis. During the study period, most patients were recommended to undergo HDCT/auto-SCT if they remained progression free during induction treatment. The subjects were retrospectively identified through survey of our institutional database. A detailed review of the clinical data was performed to ascertain the presenting features, degree of surgical resection, pathology, chemotherapy regimen, RT, and the response to treatment before and after HDCT/auto-SCT.

We evaluated disease response using brain magnetic resonance imaging (MRI) with or without spine MRI. We estimated tumor size by MRI as the product of the greatest diameter and the longest perpendicular diameter. We categorized disease response as follows: 1) progressive disease (PD): greater than 25% increase in tumor size or the appearance of a new tumor; 2) stable disease (SD): less than 50% reduction in tumor size or less than 25% increase in tumor size; 3) partial response (PR): greater than 50% decrease in tumor size; and 4) complete response (CR): complete disappearance of all previously measurable tumors. Toxicities during tandem HDCT/auto-SCT were graded using the National Cancer Institute's Common Terminology Criteria (version 4.0).

Event-free survival (EFS) was calculated from the date of diagnosis until the date of relapse, progression, or death, whichever occurred first. Overall survival (OS) was calculated from the date of diagnosis until death from any cause. Survival rates and standard errors were estimated using the Kaplan-Meier method. Differences in survival rates between groups were compared using the log-rank test. We performed multivariate analysis using Cox-regression analysis to find independent prognostic factors for survival. We analyzed differences in the frequency of toxicity between the first and second HDCT/auto-SCT using the χ² test or Fisher's exact test. Differences in continuous variables between the first and second HDCT/auto-SCT were analyzed using the Mann-Whitney U test. P values less than 0.05 were considered significant.

The study protocol was approved by the Institutional Review Board at Samsung Medical Center, Seoul, Korea (IRB No. 2016-05-009). The need for informed consent was waived by the board.

A total of 30 patients (21 boys and 9 girls) were diagnosed with HGGs during the study period. Patient characteristics are summarized in Table 1. The median age at diagnosis was 12.5 years (range 0.3–18.0), and 3 patients were younger than 3 years at diagnosis. Glioblastoma (n = 16) was the most frequent pathology, followed by anaplastic astrocytoma (n = 7), high-grade astroblastoma (n = 3), gliomatosis cerebri (n = 2) and other HGGs (n = 2). Primary tumor originated from cerebral hemisphere in 19 patients, midline structures including basal ganglia in 9, and cerebellum in 2. Gross total resection or near total resection (> 90% resection) was possible in 11 patients and subtotal resection (50%–90% resection) or biopsy (< 50% resection) was performed in the remaining 19 patients. Four patients had leptomeningeal seeding at diagnosis.

Chemotherapy was the front-line treatment after surgery in 16 patients including 3 patients younger than 3 years. Among them, concomitant (n = 8) or subsequent (n = 4) RT was given later except 4 patients who refused further treatment after progression (n = 1) or who were younger than 3 years (n = 3). RT was the front-line treatment after surgery in the remaining 14 patients and subsequent chemotherapy was given except 2 patients who was transferred to other hospital or refused further treatment after RT. Cisplatin + etoposide + cyclophosphamide + vincristine (CECV) and vincristine + ifosfamide + carboplatin + etoposide (VICE) regimens were used as front-line chemotherapy in 24 patients (Table 2). Both regimens were used in alternation (1718). Other front-line regimens were prednisolone + CCNU + vincristine (PCV) in 2, carboplatin + etoposide (CE) in one, and TMZ in one. Local RT dose to the primary site was in the range of 54.0–60.0 Gy and craniospinal RT (23.4–44.0 Gy) was given in 4 patients with leptomeningeal seeding.

Fig. 1 summarizes the flow of patients. A total of 12 patients remained progression free during induction treatment. Among them, 8 patients (4 CRs and 4 PRs) proceeded to the first HDCT/auto-SCT and one is waiting HDCT/auto-SCT. However, the remaining 3 patients could not proceed to the HDCT/auto-SCT due to death from intracranial hemorrhage during induction treatment in one, transfer to other hospital in one, and refusal of further treatment in one. The remaining 18 patients experienced progression during induction treatment. Among them, 13 received conventional salvage treatment including second-look surgery, RT, or chemotherapy, and 3 of them achieved second PR (PR2) and proceeded to the first HDCT/auto-SCT. Another 2 patients proceeded to the first HDCT/auto-SCT as salvage treatment without preceding conventional salvage treatment. The other 3 patients gave up further treatment after initial progression.

Tumor progression during induction treatment was more frequent in subtotal or less resection group than in gross or near total resection group (78.9% vs. 27.3%, P = 0.009). Glioblastoma group as compared to other HGG group and chemotherapy as compared to RT as front-line treatment after surgery were not associated with a higher progression rate during induction treatment. Peripheral blood stem cells (PBSCs) were collected during the recovery phase of chemotherapy cycle and the aim was to collect a minimum of 2 × 10⁶ CD34⁺ cells/kg, with an optimal collection of greater than 5 × 10⁶/kg to be used for bone marrow rescue during tandem HDCT/auto-SCT. The median number of CD34⁺ cells collected was 40.0 × 10⁶ cells/kg (range 2.4–127.1).

Fig. 2 summarizes the flow of patients during and after tandem HDCT/auto-SCT. A total of 13 patients (CR in 4, PR in 4, PR2 in 3, and PD in 2) underwent the first HDCT/auto-SCT and 11 of them proceeded to the second HDCT/auto-SCT. The remaining 2 patients could not proceed to the second HDCT/auto-SCT due to insufficient stem cells in one and tumor progression after the first HDCT/auto-SCT in 1. Tumor status at the second HDCT/auto-SCT was CR in 5, PR in 2, PR2 in 2, and PD in 2. Carboplatin + thiotepa + etoposide (CTE) was used for the first HDCT/auto-SCT and cyclophosphamide + melphalan (CyM) regimen was used for the second HDCT/auto-SCT (Table 2). The median interval from the first PBSC infusion to initiation of the second HDCT/auto-SCT was 84 days (range 79–102). Table 3 compares the frequency of grade 3 and 4 toxicities that developed during tandem HDCT/auto-SCT. A median of 20.8 × 10⁶ CD34⁺ cells/kg (range 1.1–56.7), and 15.2 × 10⁶ CD34⁺ cells/kg (range 2.2–44.6) were infused for the first and second HDCT/auto-SCT, respectively. Neutrophil and platelet counts recovered rapidly during the first and second HDCT/auto-SCT. The number of days with fever was higher in the first HDCT/auto-SCT than in the second (P < 0.001); however, we found no difference in the number of positive blood cultures. Grade 3 and 4 stomatitis and elevation of liver enzymes were more frequent in the first HDCT/auto-SCT than in the second. In the second HDCT/auto-SCT, the frequency and severity of mucositis-related toxicity was lower; however, the frequency of hepatic veno-occlusive disease (VOD) was higher with borderline significance (P = 0.082) and one patient died from hepatic VOD during the second HDCT/auto-SCT.

Overall, 9 patients remain progression free, 19 patients experienced progression at least once, and 2 patients died from toxicities (intracranial hemorrhage and hepatic VOD). Two of 19 patients who experienced progression remain alive after salvage treatment (one in CR and one in PD). Therefore, a total of 11 patients remain alive for a median of 38 months (range 3–82) after diagnosis. The probabilities of 3-year OS and EFS after diagnosis were 31.5% ± 9.2% and 23.7% ± 8.1%, respectively. The probability of 3-year OS is higher than that in 14 patients during 5 years (2001–2005) prior to the present study period, most of whom were treated with conventional modalities alone (Fig. 3A), although the EFS rate was not different (Fig. 3B). Subtotal or less resection of the primary tumor was associated with inferior EFS (Fig. 3C). However, glioblastoma (Fig. 3D), age under 3 years, chemotherapy compared to RT as front-line treatment after surgery were not associated with inferior OS or EFS. In the multivariate analysis for EFS, subtotal or less resection was the only significant unfavorable factor (hazard ratio [HR] 3.95, 95% confidence interval [CI] 1.19–13.18, P = 0.025, Table 4). In analysis confined only to 13 patients who underwent the first HDCT/auto-SCT, 8 patients remain alive for a median of 27 months (range 12–67) after the first HDCT/auto-SCT. Seven of them remain progression free for a median of 33 months (range 12–67) after the first HDCT/auto-SCT, the other one patient remains alive with disease after progression. The remaining 5 patients died due to progression in 4 and toxicity in 1. The probabilities of 3-year OS and EFS after the first HDCT/auto-SCT were 55.4% ± 16.1% and 43.1% ± 14.8%, respectively (Fig. 4A). In analysis confined only to 11 patients who were in CR, PR, or PR2 at the first HDCT/auto-SCT, the probabilities of 3-year OS and EFS after the first HDCT/auto-SCT were 58.2% ± 16.9% and 49.9% ± 16.4%, respectively. Tumor status less than PR at the first HDCT/auto-SCT was associated with inferior EFS (Fig. 4B); however, there was no difference in survival between glioblastoma and other HGGs (Fig. 4C).