Volume 116, Issue 3 p. 705-712
Original Article
Free Access

Intensive chemotherapy improves survival in pediatric high-grade glioma after gross total resection: results of the HIT-GBM-C protocol

Johannes E.A. Wolff MD

Corresponding Author

Johannes E.A. Wolff MD

Department of Pediatrics, The University of Texas M. D. Anderson Cancer Center, Houston, Texas

Department of Pediatric Hematology and Oncology, University of Regensburg

Department of Biostatistics, The University of Texas M. D. Anderson Cancer Center, Houston, Texas

Fax: (713) 792-0608

The University of Texas M. D. Anderson Cancer Center, Department of Pediatrics, 1515 Holcombe Blvd, Unit 87, Houston, TX 77030===Search for more papers by this author
Pablo Hernaiz Driever MD

Pablo Hernaiz Driever MD

Department of Pediatric Oncology and Hematology, Charité-University Medicine Berlin, Berlin, Germany

Pablo Hernaiz Driever, MD and Christof M. Kramm equally contributed to this article.

Search for more papers by this author
Bernhard Erdlenbruch MD

Bernhard Erdlenbruch MD

Department of Pediatrics, Hospital of Minden, Minden, Texas

Search for more papers by this author
Rolf D. Kortmann MD

Rolf D. Kortmann MD

Department of Radiation Oncology, University of Leipzig, Leipzig, Germany

Search for more papers by this author
Stefan Rutkowski MD

Stefan Rutkowski MD

Department of Pediatric Hematology and Oncology, University of Wurzburg, Wurzburg, Germany

Search for more papers by this author
Torsten Pietsch MD

Torsten Pietsch MD

Department of Neuropathology, University of Bonn, Bonn, Germany

Search for more papers by this author
Crystal Parker MSc

Crystal Parker MSc

Department of Biostatistics, The University of Texas M. D. Anderson Cancer Center, Houston, Texas

Search for more papers by this author
Monica Warmuth Metz MD

Monica Warmuth Metz MD

Neuroradiology, University of Wurzburg, Wurzburg, Germany

Search for more papers by this author
Astrid Gnekow MD

Astrid Gnekow MD

Department of Pediatric Oncology, Children's Hospital Augsburg, Augsburg, Germany

Search for more papers by this author
Christof M. Kramm MD

Christof M. Kramm MD

Department of Pediatric Hematology and Oncology, University of Halle, Halle, Germany

Pablo Hernaiz Driever, MD and Christof M. Kramm equally contributed to this article.

Search for more papers by this author
First published: 02 December 2009
Citations: 101

We thank Sabine Wagner, MD, for a great job done during the last year of patient enrollment and during follow-up.

Abstract

BACKGROUND:

The authors hypothesized that intensified chemotherapy in protocol HIT-GBM-C would increase survival of pediatric patients with high-grade glioma (HGG) and diffuse intrinsic pontine glioma (DIPG).

METHODS:

Pediatric patients with newly diagnosed HGG and DIPG were treated with standard fractionated radiation and simultaneous chemotherapy (cisplatin 20 mg/m2 × 5 days, etoposide 100 mg/m2 × 3 days, and vincristine, and 1 cycle of cisplatin + etoposide + ifosfamide 1.5 g/m × 5 days [PEI] during the last week of radiation). Subsequent maintenance chemotherapy included further cycles of PEI in Weeks 10, 14, 18, 22, 26, and 30, followed by oral valproic acid.

RESULTS:

Ninety-seven (pons, 37; nonpons, 60) patients (median age, 10 years; grade IV histology, 35) were treated. Resection was complete in 21 patients, partial in 29, biopsy only in 26, and not performed in 21. Overall survival rates were 91% (standard error of the mean [SE] ± 3%), 56%, and 19% at 6, 12, and 60 months after diagnosis, respectively. When compared with previous protocols, there was no significant benefit for patients with residual tumor, but the 5-year overall survival rate for patients with complete resection treated on HIT-GBM-C was 63% ± 12% SE, compared with 17% ± 10% SE for the historical control group (P = .003, log-rank test).

CONCLUSIONS:

HIT-GBM-C chemotherapy after complete tumor resection was superior to previous protocols. Cancer 2010. © 2009 American Cancer Society.

High-grade glioma (HGG) is a term encompassing various histologically defined brain tumors linked by glial origin and poor prognosis. The most frequent entities are glioblastoma multiforme (World Health Organization [WHO] grade IV1) and anaplastic astrocytoma (WHO grade III). In pediatrics, diffuse intrinsic pontine glioma (DIPG) is sometimes included in this disease group.2-5

For HGG, there is a consensus that chemotherapy plays a limited role.6-10 Temozolomide is the standard treatment for adult patients with HGG.11, 12 The beneficial effect of radiation and complete surgical resection therapy is generally accepted.10, 13, 14 In children, survival rates are higher than in adults, which may be because of more intensive chemotherapy protocols or because pediatric HGG differs biologically from HGG in adults.15

DIPGs are astrocytic tumors that occur almost exclusively in the pediatric population.14, 15 Given the lack of prognostic relevance of histological diagnoses and the impossibility of achieving complete surgical resection, the standard of care is to treat these patients without histological confirmation.15, 16 Radiation therapy is known to extend survival by half a year and to temporarily ameliorate neurological deficits, but the tumors recur.16, 17 A beneficial role of chemotherapy has been suggested,17 but it has yet to be confirmed by prospective phase 3 studies.

The HIT-GBM study is a prospective cohort comparison study.18 After a previous randomized trial had to be closed because of lack of accrual,10 the HIT-GBM concept was to evaluate sequential, consecutive, treatment protocols. The first protocol (HIT-GBM-A) consisted of trophosphamide and etoposide,18-21 the second (HIT-GBM-B) of intensive chemotherapy with simultaneous radiation22 followed by interferon gamma maintenance treatment.23, 24 Only the induction treatment appeared to be superior to the other HIT-GBM protocols.22 Therefore, with the third cohort (HIT-GBM-C), we evaluated the effect of a prolongation of the same type of treatment. Valproic acid was added in the third cohort.25

MATERIALS AND METHODS

Chemotherapy Protocol

The postoperative treatment protocol started after early postoperative magnetic imaging (MRI) scans. Spinal MRI was recommended for staging but not mandatory. Nonsurgical treatment started with simultaneous radiation therapy and chemotherapy induction, followed by maintenance chemotherapy and histone deacetylase inhibition. Radiation was recommended within 14 days after diagnosis and was given to the tumor and a 2-cm margin in 5 1.8-Gy fractions per week, up to a total of 54 Gy for patients 6 years and younger, as well as to brainstem locations and up to 59.4 Gy for older patients with tumors in other locations. Chemotherapy started during the first week of radiation therapy. The first cycle consisted of cisplatin, etoposide, and vincristine (PEV).22 Hydration at 2.4 L/m2/24 hours started at least 8 hours before cisplatin administration and was given until 24 hours after completion of the chemotherapy. Cisplatin was given over 1 hour at a dose of 20 mg/m2/day on Days 1 to 5. Etoposide 100 mg/m2 was given over 2 hours on Days 1 to 3, always after cisplatin. Vincristine was injected at a dose of 1.5 mg/m2 (maximum, 2 mg) on Day 5 only to avoid reducing the DNA binding of cisplatin by means of vincristine-mediated cell arrest. Weekly vincristine followed, and during the last week of radiation therapy, 1 cycle of cisplatin, etoposide, and ifosfamide (PEI) was given. The doses of cisplatin and etoposide were the same as with PEV. Ifosfamide (1.5 g/m2/day) and were given as a 1-hour intravenous infusion on Days 1 to 5. Mesna (500 mg/m2) was given as an injection before and 4 and 8 hours after ifosfamide. The cycle was placed at the end of the radiation treatment because it was expected to cause bone marrow suppression; this would have cause a risk for interruption of radiation if the cycle was given earlier.22 The infusion times of ifosfamide and cisplatin were separated by positioning the etoposide infusion in between, so that mesna was not infused during cisplatin infusion or at times of peak plasma levels.26, 27 After radiation therapy, the chemotherapy continued with further cycles of PEI in Weeks 10, 14, 18, 22, 26, and 30, as well as vincristine in Weeks 13, 17, 21, 25, and 29. Then, valproic acid was started at 10 mg/kg/day orally for 1 week and 20 mg/kg/day in Week 2.25, 28-30 Serum levels were then checked, and the dose was further increased to reach target serum levels of 100-150 μg/mL, unless side effects had occurred at lower levels. Tumor status was re-evaluated after every other chemotherapy cycle.

Eligibility Criteria

The main eligibility criterion was either histologically proven HGG or DIPG. HGG was defined as grade III or grade IV (WHO) astrocytic or oligodendroglial tumors. Central review of histology specimens was provided by the Department of Neuropathology, University of Bonn (Germany). The definition of DIPG included a symptom history of <6 months, tumor centered in the pons, tumor extension over >50% of the pontine cross-sectional area, and no contrast enhancement on MRI. The images were centrally reviewed (Wurzburg, Germany). Biopsy was recommended if pontine tumors appeared to have enhanced contrast, and the patients were excluded if their histology showed anything but astrocytic tumor. However, histological grading was irrelevant to eligibility when the MRI morphology was typical. Also eligible were patients with gliomatosis cerebri, as defined by radiological imaging of diffuse involvement of more than 2 lobes in the presence of astrocytic histology, which had to be WHO grade II or higher. Further eligibility criteria included primary diagnosis of a glioma, age of 3-18 years, institutional review board approval, and informed consent. Secondary gliomas that occurred after malignancies, such as leukemia, were not excluded as this has been the definition of the population in the previous protocols, too. To be eligible in this treatment protocol analysis, one further eligibility criterion was that patients had to receive chemotherapy and radiation therapy simultaneously, defined as starting chemotherapy not earlier than 2 days before and not later than 2 weeks after starting radiation. The completeness of resection was not part of the eligibility criteria, but it was reported following Gnekow criteria.31 Beginning treatment within 14 days after diagnosis was recommended, but delayed treatment was not an exclusionary criteria.

Statistics

Statistical analyses were performed using the Statistical Package for Social Studies software (SPSS 12.0; Chicago, Ill). MRIs and clinical evaluations were completed after every other chemotherapy cycle to evaluate response. On the basis of previous results,10 the survival analysis was performed separately for 1) patients with completely resected tumors, 2) patients with pontine tumors, and 3) others. The previously determined endpoint, ie, overall survival, remained unchanged. By using Kaplan Meier estimates, we compared the historical control group of patients treated on either protocol A20, 21 or protocol B22, 24 with patients treated on protocol HIT-GBM-C in the subgroups a, b, and c. For the main hypothesis that the new protocol resulted in superior overall survival, a P-value below .05 was considered statistically significant. Subgroup analyses (log-rank test) and multivariate analysis (COX regression) were used to explore the relevance of inhomogeneity of risk factors for the conclusions.10 This was exploratory, not hypothesis driven, and, therefore, no P value was set to define significance.

For response description,20, 25 patients with no residual tumors were classified as having continuation of complete remission (CCR) or progressive disease (PD). Patients with measurable tumor after the first surgery were classified as having a complete response (CR), partial response (PR, tumor <50%), stable disease (SD, tumor <125% and ≥50%), and progressive disease (PD), by comparing 1-dimensional measurements to the postoperative tumor size as measured in the postoperative MRI. Criteria remained unchanged all through the prospective cohort comparison study with protocols HIT-GBM-A, -B, and -C.

RESULTS

Patients

The efficacy of the chemotherapeutic approach was evaluated in the 97 patients who were formally enrolled in the prospective HIT-GBM-C trial. Furthermore, 97 patients enrolled in the previous cohorts, HIT-GBM-A and HIT-GBM-B, were used as an historical control group.

Patients in HIT-GBM-C were diagnosed between 1999 and 2003, and 57 were males. Previous malignant diseases were 1 acute lymphoblastic leukemia (ALL), 1 ALL followed by a PNET and then an HGG, and 1 craniopharyngioma. The median age was 9.84 years (range, 3.22-17.93). Tumor locations were as follows: spinal (2), pons (37), mesencephalon (2), cerebellum and fourth ventricle (4), basal ganglia and third ventricle (14), cerebral cortex and white matter (30), overlapping areas restricted to neopallial structures (3), and overlapping areas of neopallium and other structures (5). Surgery yielded gross total resection for 21 patients, subtotal resection for 11, and partial resection for 18; 26 patients had biopsy only and 21 had no surgery. For the outcome analysis, patient data were separated into pontine and nonpontine tumors (Table 1). All patients in the pontine glioma treatment group had reference radiology, and 59 of 60 patients with nonpontine tumors had histology reference evaluation. In the historical control group, the numbers were 37 of 49 and 47 of 48, respectively.

Table 1. Patient Characteristics
Treatment Group Control Group
Pons Nonpons Pons Nonpons
No. of patients 37 60 49 48
Male/Female 21/16 36/24 24/25 28/20
Previous disease 0 3 0 6
Median age, y 7.9 11.4 8.0 10.9
Metastases 0 6 0 3
Surgery
 No surgery 21 0 23 0
 Biopsy 12 14 20 11
 Partial resection 4 14 3 15
 Subtotal resection 0 11 3 6
 GTR 0 21 0 16
Histology (WHO grade)
 No histology 21 0 23 0
 Astrocytomas II 4 1 6 0
 A Astrocytoma III 8 18 10 18
 AGG III 0 1 0 0
 GBM, IV 4 35 10 30
 Gliosarcoma, IV 0 1 0 0
 A Astrobl. III, 0 1 0 0
 A Oligoastr. III 0 1 0 0
 A PXA III 0 2 0 0
WHO grade
 WHO not defined 21 0 23 0
 WHO grade II 4 1 6 0
 WHO grade III 8 22 10 18
 WHO grade IV 4 37 10 30
  • WHO indicates World Health Organization; GTR, gross total resection; A Astrocytoma, anaplastic astrocytoma; AGG, anaplastic ganglioglioma; GBM, glioblastoma multiforme; A Astrobl., anaplastic astroblastoma; A Oligoastr., anaplastic oligoastocytoma; A PXA, anaplastic pleomorphic xanthoastrocytoma.

Toxicity

The toxicity of the chemotherapy used in HIT-GBM-C was as expected. During the first cycle of chemotherapy, nausea was the dominant complaint, and grade IV (NCI-CTC 3.0) neutropenia was observed in only 15% of patients. The second cycle, which contained ifosfamide, resulted predominantly in bone marrow toxicity and fever, from which all patients recovered (Table 2). The feared side effect of severe hearing loss due to adding radiation to the inner ear during treatment with cisplatin did not occur (grade III hearing loss in only 1 of 80, and 74 reported after the first and second cycle, respectively). No toxicity deaths were reported on the protocol. Performance status was described on a qualitative scale and showed an increase in patients confined to the bed, from 4% after the first cycle to 13% after the second.

Table 2. Toxicity of the First Two Cycles of Chemotherapy
Toxicity Cycle 1 (PEV) Cycle 2 (PEI)
No. 0 / I / II / III / IV No. 0 / I / II / III / IV
Hemoglobin 89 20 / 31 / 30 / 07 / 01 85 07 / 11 / 37 / 24 / 04
Leukocytes 89 16 / 08 / 29 / 26 / 10 83 03 / 04 / 06 / 13 / 57
Neutrophils 78 17 / 08 / 14 / 21 / 18 75 04 / 03 / 05 / 12 / 51
Platelets 87 66 / 09 / 01 / 08 / 03 82 23 / 07 / 05 / 35 / 12
Infection 88 70 / 07 / 07 / 04 / 00 85 52 / 12 / 07 / 13 / 01
Fever 87 73 / 12 / 01 / 01 / 00 84 58 / 18 / 04 / 02 / 02
Stomatitis 86 75 / 08 / 03 / 00 / 00 81 60 / 18 / 03 / 00 / 00
Nausea 88 18 / 42 / 23 / 05 / 00 81 10 / 41 / 25 / 08 / 00
Emesis 75 18 / 21 / 35 / 01 / 00 84 13 / 18 / 46 / 06 / 01
Creatinine 83 79 / 04 / 00 / 00 / 00 74 67 / 07 / 00 / 00 / 00
Hematuria 84 79 / 05 / 00 / 00 / 00 80 69 / 11 / 00 / 00 / 00
Creatinine clearance 69 66 / 03 / 00 / 00 / 00 67 59 / 05 / 03 / 00 / 00
Fanconi syndrome 75 75 / 00 / 00 / 00 / 00 74 72 / 01 / 01 / 00 / 00
Hearing loss 80 71 / 07 / 01 / 01 / 00 74 56 / 14 / 03 / 01 / 00
Dermatitis 87 74 / 09 / 04 / 00 / 00 83 71 / 07 / 04 / 01 / 00
Peripheral neuropathy 78 63 / 12 / 02 / 01 / 00 77 62 / 10 / 04 / 01 / 00
Central neuropathy 80 73 / 04 / 01 / 00 / 02 79 70 / 05 / 03 / 01 / 00
Allergy 85 81 / 04 / 00 / 00 / 00 83 81 / 02 / 00 / 00 / 00

Response

MRI studies 1-4 weeks after radiochemotherapy in 75 of 76 patients with residual tumors yielded the following responses: 5 CR, 9 PR, 38 SD, and 23 PD. Early response was more frequently observed in patients with tumors of the pons (8 of 36, 22%) than in others (6 of 39, 15%). Eight patients responded later, bringing the total best response rate to (CR + PR) to 22 of 76 (Table 3). Most patients who started radiochemotherapy after complete resection remained tumor free throughout the protocol (14 of 21); 4 patients had a relapse during the first 2 cycles, and 3 had a later relapse.

Table 3. Tumor Response
Response to Chemotherapy: Pons Tumors Only (n=37)
Cycle 1-2 1-4 1-6 1-8 Best response
CR 1 1 1 1 1
PR 7 6 5 4 8
SD 18 10 7 5 17
PD 10 19 24 27 11
ND 1 1 0 0 0
Response to Chemotherapy: Residual Tumors Outside of the Pons (n=39)
CR 4 3 2 2 5
PR 2 1 1 2 3
SD 20 12 7 5 18
PD 13 23 29 30 13
ND 0 0 0 0 0

Survival

The overall survival rate of all 97 patients treated according to HIT-GBM-C was 91% (SE ± 3%), 56% (±5%), 30%, 23%, 22%, and 19% (±4%) after 6, 12, 24, 36, 48, and 60 months, respectively. The event-free survival (EFS) rate was 52% (±5%), 27%, 16%, 15%, 13%, and 13% (±4%) after the same time periods. Previously described risk factors were confirmed; patients with pontine tumors had worse survival, and patients with completely resected tumors had better survival than others. The original plan of the protocol was to compare overall survival with that of the historical control group. Those survival rates were 53%, 13%, and 7% after 12, 36, and 60 months, which were slightly inferior to rates for the treatment group. The same was true when looking at EFS (24%, 7%, 5%, and 5% after 12, 36, 48, and 60 months, respectively). However, this overall survival comparison was conditional on a subgroup analysis, which showed that patients with pontine tumors needed a separate evaluation,10 and, therefore, the statistical comparison—including the calculation of P values—was inappropriate for the whole group. It was instead performed separately for the following subgroups.

For patients with pontine tumors, overall survival rates did not differ between the HIT-GBM-C and the control groups (Fig. 1). Kaplan-Meier curves crossed with a median overall survival duration of 1.13 years (SE ± 0.10 years; 95% CI, 0.87-1.39) for the treatment group and 0.94 years (SE ± 0.05 years) for the control group. The same was true when we compared EFS (median EFS, 0.40 ± 0.07 years for the treatment group vs 0.55 ± 0.098 for controls). For patients with HGG outside the pons and with residual tumor after surgery, the results were similar. No relevant difference was detected in overall survival duration: the treatment group median survival duration was 0.92 ± 0.06 years (95% CI, 0.71-1.17), compared with 1.17 ± 0.07 years for the control group. When EFS was considered, the treatment group had a tendency toward an inferior outcome (median EFS of the treatment group, 0.44 ± 0.04 years compared with 0.67 ± 0.06 years for controls).

Details are in the caption following the image

The overall survival of 37 patients with pontine tumors treated with HIT-GBM-C chemotherapy did not improve (treatment group, thin black line) compared with 49 patients in the historical control group (broad gray line) treated with previous protocols HIT-GBM-A,21, 20 or HIT-GBM-B.24 “Censored” is defined as alive by the end of the observation time.

Patients with completely resected tumors displayed a different picture. The data from the historical patient group indicated survival almost as poor as in patients with residual tumors, with a median overall survival duration of 1.85 years (SE ± 0.60; 95% CI, 0.78-4.4). In contrast, the HIT-GBM-C treatment group, after complete resection, had an excellent outcome, and the median overall survival was never reached (Fig. 2). After 6, 12, 24, 36, 48, and 60 months, the overall survival rates of the treatment group were 95% (±5), 86% (±8), 76% (±9), 71% (±10), 71% (±10), and 63% (±12), respectively, and rates in the historical control group were 100%, 62% (±12), 44% (±12), 37% (±12), 31% (±12), and 17% (±10), respectively (median 1.85 years ± 0.6 SE; 95% CI, 0.78 -4.42). The analysis of EFS confirmed these findings: after 6, 12, and 60 months, the EFS rates in the treatment group were 86% (±8), 67% (±10), and 48% (±10), respectively (median survival duration, 1.53 ± 0.71 years), versus 69% (±12), 50% (±12), and 13% (±8), respectively, in the historical control group (median, 0.65 ± 0.25 year).

Details are in the caption following the image

HIT-GBM-C protocol was better than the historical control protocols for patients with completely resected HGG. Comparison of overall survival of 21 patients treated with simultaneous radiochemotherapy followed by up to 6 cycles of chemotherapy with cisplatin, etoposide, ifosfamide, and vincristine (treatment group, thin black line) compared with 16 patients in the historical control group (broad gray line) treated with previous protocols HIT-GBM-A,20, 21 or HIT-GBM-B.24 The superior survival rate of patients in the HIT-GBM-C protocol was confirmed by using an exploratory log-rank test, with P = .003.

To ensure that this result was not an artifact, the treatment and control groups of patients with completely resected tumors were compared again with respect to the remaining possible prognostic factors. No difference was found in gender (11:10 male-to-female ratio in the treatment and 9:7 in the control group) or frequency of grade IV histology (16 of 21 vs 14 of 16), and there were only nonsignificant trends in the frequency of primary metastases (3 of 21 in treatment vs 0 of 16 in control group; P = .24 in 2-sided chi-square test) and tumor location (19 of 21 vs 12 of 16). In Cox multiple regression analysis, sex, age, and metastatic status had no influence, but the influence on superior survival of tumor location in the cortex was as strong as that of treatment (HIT-GBM-C superior). A further analysis that restricted the subgroup of cortical locations confirmed the previous finding: after complete resection, patients treated with HIT-GBM-C had better overall survival than those in the historical control group (median survival, 3.0 vs 1.9 years; P = .018 in log-rank test and P = .044 in Cox multiple regressions with metastatic status and tumor grading as additional variables).

Similarly, excluding patients with secondary HGG, those with missing reference histology, or those with gliomatosis did not change the relevant finding that a chemotherapy benefit was demonstrated in patient groups who had complete resection but not in those patients with residual disease.

DISCUSSION

We found that survival was significantly improved for patients with completely resected HGG (Fig. 2) when they were treated with the HIT-GBM-C chemotherapy. The major difference of this protocol compared with the previous ones was the continuation of intensive chemotherapy after radiation.20-22, 24 The current article focuses on the HIT-GBM-C chemotherapy trial; a thorough discussion of the covariates and the heterogeneity of this patient population was recently published.10, 25 As pediatric high-grade glioma are rare, some heterogeneity within the treated patient population has to be accepted to have sufficient numbers of patients. Even with this, the number of patients whose results were used in the final conclusion are small, and the evidence is not as high as a randomized trial with simultaneous protocol arms. However, when such a trial is not possible, the prospective cohort comparison study reported here may be the best possible evidence.

The chemotherapy protocol as a whole was successful. The concept of combining DNA-alkylating agents with topoisomerase inhibitors is based on the paradigm of inhibiting topoisomerases, which are necessary for DNA repair.32 Our data add to the evidence that the concept is valid. However, our results should be interpreted with caution, as they cannot differentiate between the effects of the chemotherapy and long-term treatment with valproic acid, which was used in the latter part of this study.25, 28

The most important finding in this study is that the overall survival of patients with total gross tumor resection was improved by the treatment regimen (Fig. 2), while there was no improvement for treated patients with less than total gross resection. Maximal efforts were undertaken10 and described in the results to assure this conclusion is not biased by inhomogeneous patient populations. Figure 1 shows the survival curve of the 37 patients with pontine glioma. Improvement for patients after complete resection was also shown for other chemotherapy protocols33, 34 but not for all.35 Within the same cooperative group, the effect was historically present in the HIT-91-S protocol33 and not in the concurrently used HIT-91-M protocol, and the effect disappeared in the following HIT-GBM-A and -B protocols and reappeared in the HIT-GBM-C protocol reported here. This timeline makes it unlikely that this finding is related to improvements in surgical or diagnostic technology. Recently supported concepts of various types of tumor stem cells in HGG that have quite distinct biologies provide a possible explanation: the survival of HGG patients after gross total resection depends on the success of the treatment to eradicate any remaining tumor cells. These cells have infiltrated the brain adjacent to the tumor and may be influenced by reactive astrocytes. In contrast, cells in the middle of a measurable tumor lump are exposed to high concentrations of tumor cell products.36, 37 The finding that some—but not all—treatment protocols resulted in increased survival after complete resection indicates that the infiltrating cells are sensitive to certain types of chemotherapy but not to others. The collective experience shows that complete resection is an important step in treatment14, 25 but so is chemotherapy, in particular after complete resection, and that resection alone is not enough.38

The HIT-GBM-C protocol was written in the mid-1990s. At that time, the only regimen with proven efficacy in randomized studies was lomustine, prednisone, and vincristine.6 Since then, the enthusiasm for nitrosourea drugs has decreased as knowledge of their pulmonary side effects has increased.34, 39 Temozolomide3 has since been named the standard of care in adult HGG.11 In pediatrics, temozolomide does not seem to be as successful,40, 41 but other dose schedules are still being explored,42 including high-dose chemotherapy,35, 43 as well as other agents.. Novel approaches such as vaccination with dendritic cells44, 45 and tumor lytic viruses29 appear particularly important for patients with less-than-completely resected tumors.

In conclusion, these data suggest that the HIT-GBM-C protocol is effective for pediatric patients with gross total resected HGG. However, residual cells in measurable tumor after surgery appear to have a different biology and to require other treatment approaches.

CONFLICT OF INTEREST DISCLOSURES

This work was supported by Deutsche Kinderkrebsstiftung.