|
|
| |||||||||||||||||||||||||||||||||||||||
Background Among patients with malignant brain tumors, infants and very young children have the worst prognosis and the most severe treatment-related neurotoxic effects. Therefore, in 1986, the Pediatric Oncology Group began a study in which postoperative chemotherapy was given in order to permit a delay in the delivery of radiation to the developing brain.
Methods Children under 36 months of age with biopsy-proved malignant brain tumors were treated postoperatively with two 28-day cycles of cyclophosphamide plus vincristine, followed by one 28-day cycle of cisplatin plus etoposide. This sequence was repeated until the disease progressed or for two years in 132 children under 24 months of age at diagnosis and for one year in 66 children 24 to 36 months of age at diagnosis. After this, the patients received radiation therapy. The response to the first two cycles of chemotherapy was measured in 102 patients with residual postoperative disease.
Results The first two cycles of cyclophosphamide and vincristine produced complete or partial responses in 39 percent of the 102 patients who could be evaluated. The response rates were highest among patients with medulloblastomas, malignant gliomas, or ependymomas. Patients with brain-stem gliomas or embryonal tumors (primitive neuroectodermal tumors) had little or no response. The progression-free survival rate was 41 percent at one year for children who were 24 to 36 months old at diagnosis and 39 percent at two years for those under 24 months of age at diagnosis. Multivariate analysis identified embryonal tumors as a significant adverse prognostic feature (relative risk, 2.2; 95 percent confidence interval, 1.4 to 3.4) and complete resection as a favorable feature (relative risk, 0.33; 95 percent confidence interval, 0.20 to 0.54). Complete responses to chemotherapy were associated with a progression-free survival rate approaching that achieved with gross total resection. A comparison of cognitive evaluations obtained at base line and after one year of chemotherapy revealed no evidence of deterioration in cognitive function.
Conclusions Chemotherapy appears to be an effective primary postoperative treatment for many malignant brain tumors in young children. Disease control for one or two years in a large minority of patients permitted a delay in the delivery of radiation and, on the basis of preliminary results, a reduction in neurotoxicity. For patients who had undergone total surgical resection or who had a complete response to chemotherapy, the results are sufficiently encouraging to suggest that radiation therapy may not be needed in this subgroup of children after at least one year of chemotherapy.
In response to these problems, members of the Pediatric Oncology Group initiated a pilot study in 1985 in which prolonged postoperative chemotherapy was given in an effort to delay exposing the infant brain to radiation8,9,10,11,12,13,14,15,16,17. The pilot study documented that the chemotherapy regimen was tolerated and seemed to be efficacious. These findings led to a prospective groupwide study, the results of which form the basis of this report.
Methods
Patient Eligibility
Children under 36 months of age at diagnosis who had pathologically proved malignant intracranial tumors were eligible for the study, after surgical resection or biopsy (Table 1). (For children with brain-stem gliomas, a biopsy was not mandatory.) Patients with low-grade astrocytomas were excluded unless there was evidence of neuraxial dissemination (one patient). Written informed consent was obtained from the parents or guardians of each child in accordance with institutional guidelines.
|
Maximal surgical removal was recommended with the caveat that the anatomical location of the tumor and the condition of the child might preclude extensive resection. The degree of surgical resection was assessed by a central review of operative reports and preoperative and postoperative imaging (Table 2). When the results of scans and operative reports differed, the degree of surgical resection was based on the results of computed tomography (CT) or magnetic resonance imaging (MRI).
|
The slides from all patients were reviewed by one of us. Medulloblastoma and well-differentiated glioma were classified according to widely accepted criteria18,19. Densely cellular supratentorial neoplasms were referred to as embryonal tumors (primitive neuroectodermal tumors). Attempts were made to assign these lesions to established categories of embryonal tumors (e.g., cerebral neuroblastoma, medulloepithelioma, ependymoblastoma, and pineoblastoma), but this was not always possible in tumors with mixed composition or those that were entirely undifferentiated. For the purposes of statistical analysis, these and all supratentorial embryonal tumors were considered as one entity -- embryonal tumors (primitive neuroectodermal tumors).
Treatment Regimen
Chemotherapy was started 2 to 4 weeks after surgery and was given in alternating 28-day cycles in the sequence AABAAB. Cycle A comprised vincristine (0.065 mg per kilogram of body weight; maximal dose, 1.5 mg), given by rapid intravenous infusion on days 1 and 8, and cyclophosphamide (65 mg per kilogram), infused over a 30-minute period on day 1. Cycle B consisted of a six-hour infusion of cisplatin (4 mg per kilogram) on day 1 and a one-hour intravenous infusion of etoposide (6.5 mg per kilogram) on days 3 and 4. The planned duration of chemotherapy was 24 months for children under 24 months of age at diagnosis and 12 months for those 24 to 36 months of age. If there was evidence of disease progression or unacceptable toxicity, chemotherapy was discontinued. The patient was then considered for additional surgery, if appropriate, and radiation therapy.
Radiation therapy was started three to four weeks after the last cycle of chemotherapy. The volume of radiation treatment was determined by histologic analysis and the extent of disease as assessed by contrast-enhanced imaging at the time of diagnosis. The radiation dose was based on each patient's disease status at the completion of chemotherapy. Standard dose levels were used for patients with residual or progressive disease or with subarachnoid metastases at initial diagnosis. Children with medulloblastomas, embryonal tumors (primitive neuroectodermal tumors), anaplastic ependymomas, or any tumor with subarachnoid seeding received 35.2 Gy to the craniospinal area with a boost to the primary site (cumulative dose, 54 Gy). Patients with ependymomas, malignant gliomas, or brain-stem gliomas received local radiation therapy to a total dose of 54 Gy. Children who had no residual or recurrent disease after chemotherapy received reduced doses: the dose to the craniospinal area was decreased to 24 Gy, and the dose to the primary site was decreased to 50 Gy. Infants under 24 months of age at the time of radiation therapy received 90 percent of these doses.
Evaluation Procedures
Before chemotherapy and at specific times during and after treatment, the patients underwent neurologic assessment, CT or MRI of the head, myelography or MRI of the spine, bone marrow aspiration, cytologic analysis of cerebrospinal fluid, and bone scanning. The following were determined: height, weight, complete blood count, blood urea nitrogen level, creatinine concentration, electrolyte levels, liver function, and calcium and magnesium concentrations. Audiograms were obtained, and brain-stem auditory evoked responses were assessed. Neurodevelopmental tests were scheduled to be administered postoperatively (before chemotherapy was begun) when the children were neurologically stable, and annually thereafter. Children under 30 months of age were initially tested with the Bayley Scales of Infant Development20. Older children were tested with the McCarthy Scales of Children's Abilities21. If a child crossed the upper age limit for the Bayley Scales of Infant Development during the study, the McCarthy Scales of Children's Abilities were administered. After adjustment for age with test norms, the scores obtained were considered clinically abnormal when they differed from the mean of the test-standardization sample by 15 points (approximately 1 SD) or more.
Assessment of Response
Postoperative CT and MRI scans of the patients with measurable residual tumor were compared with scans obtained after two cycles of cyclophosphamide and vincristine to gather preradiation phase II data on this chemotherapy combination. When it was unclear whether abnormalities on postoperative scans represented residual tumor or postoperative artifacts, the patients were deemed not able to be evaluated for either the degree of surgical resection or the response to chemotherapy. Children who had clinical evidence of disease progression, despite scans indicating apparently stable disease, were considered to have progressive disease if no other cause of clinical deterioration could be identified.
The response of metastatic lesions, as assessed by myelography, MRI of the spine, cytologic analysis, and bone scanning, was determined by the individual investigators during chemotherapy. Patients were considered to have had a complete response if cytologic analysis revealed the absence of malignant cells in two separate samples of cerebrospinal fluid. If patients continued to have positive cytologic results, they were classified as having stable disease, and patients with negative cytologic results that subsequently became positive were identified as having progressive disease.
Statistical Analyses
The method of Kaplan and Meier22 was used to construct life tables, and the log-rank chi-square statistic was used for comparisons23. Prognostic factors were evaluated with the Cox proportional-hazards general linear model24. When the frequencies were sufficiently large, the classic chi-square statistic25 was used to analyze contingency tables. Otherwise, an exact test was used26.
Height and weight measurements obtained within 1 month of the beginning of chemotherapy were compared by the t-test with those obtained after 1 year (±30 days) in children who remained free of tumor progression during chemotherapy. To assess changes in the study group relative to those in normal populations, height and weight z scores were calculated from standard growth curves27.
Analyses of change in the developmental scores were based on t-tests. When numbers were small or there was concern about the lack of statistical normality or the presence of outliers, the analyses were repeated with a nonparametric test (Wilcoxon rank-sum test), which yielded similar results.
Results
Between August 1986 and April 1990, 206 children were enrolled in the study. Eight were excluded from the analysis because of low-grade abnormalities on histologic analysis or, in one case, the presence of multiple primary tumors with various histologic findings. The patients were evenly distributed according to age: two thirds were under two years of age at diagnosis, with half of these under one year (Table 1). The presence of subarachnoid metastasis correlated with histologic findings, and subarachnoid metastasis was detected more often among children with primitive neuroectodermal tumors (33 percent) or medulloblastomas (42 percent) than among children with all other types of tumors (15 percent).
Response to Chemotherapy
Of the 198 patients whose responses to chemotherapy could be evaluated, 57 had undergone complete surgical resection (Table 2) and 39 had incomplete imaging data. Complete or partial responses were achieved in 39 percent of the 102 patients who could be evaluated. The response rate differed significantly according to the type of tumor (P = 0.05), ranging from 0 percent in children with brain-stem gliomas to 60 percent in children with malignant gliomas (Table 3).
|
Progression-free and Overall Survival
As shown in Table 4, the rate of progression-free survival was 41 percent at one year for children who were 24 to 36 months of age at diagnosis and 39 percent at two years for children given a diagnosis before the age of 24 months. Thus, it was possible to delay radiation therapy in a substantial proportion of patients: one year for children 24 to 36 months of age at diagnosis and two years for those less than 24 months of age. Progression-free and overall survival did not differ significantly between the two age groups despite the fact that infants continued to receive chemotherapy for a second year, whereas the older children received radiation therapy after one year of chemotherapy (Figure 1 and Table 4).
|
|
|
|
|
|
There were 13 children with no evidence of disease after at least 12 months of chemotherapy who subsequently declined treatment with radiation. The disease progressed in 1 child, who died 1.5 years after leaving the study; a second had a second cancer (sarcoma); and 11 were disease-free at the most recent follow-up visit 3 months to 3.5 years (median, 1 year) after therapy was stopped.
A multivariate analysis of progression-free survival revealed that complete resection was the only significant favorable factor (P<0.001; relative risk, 0.33; 95 percent confidence interval, 0.20 to 0.54), and a histologic finding of primitive neuroectodermal tumor was the single unfavorable factor (P<0.003; relative risk, 2.2; 95 percent confidence interval, 1.4 to 3.4).
Toxicity
The most frequently reported toxic effect was myelosuppression. Admissions for fever and neutropenia were common. In addition to undiagnosed causes of fever, infections of central venous catheters, shunt infections, pneumonia, sepsis, and bacteremia developed. There was one fatal fungal infection, one death from varicella, one death from pseudomembranous colitis, and one death from pneumonia. Hemorrhagic cystitis developed in approximately half the children but was readily prevented in later cycles with mesna. Seizures associated with serum sodium concentrations below 130 mmol per liter occurred in conjunction with treatment with either cisplatin or cyclophosphamide in nine children. Nausea and vomiting were almost universal. High-frequency hearing loss was reported in 44 patients. Twenty-one patients (15 of whom were infants less than 24 months of age at diagnosis) were excluded from the study because their parents did not grant permission for further therapy.
Of the 198 children in the study, height was measured in 166 and weight in 186 before chemotherapy was begun. Of the 77 children who were still following the protocol at the end of one year (but before radiation therapy was begun), height was assessed in 59 and weight in 69. The average change (-0.08) in height z scores after one year of chemotherapy was not statistically different from that in a normal population (P = 0.7), suggesting that aggregate growth was maintained. In a second analysis limited to 54 children whose heights were assessed both at the start of treatment and one year later, the average change in height z scores (-0.3) was significant (P = 0.008), suggesting that in this subgroup of children, the rate of growth failed to keep pace with that in the normal population. The significant difference in height z scores at one year was limited to children who were 24 to 36 months of age at diagnosis (P = 0.01). Younger children grew at a normal rate after one year of chemotherapy.
There was an average weight gain of 2.8 kg between the start of treatment and the assessment one year later. The weight gain was most dramatic (P = 0.01 for the comparison with weight z scores in a normal population) among children who were less than 24 months of age at diagnosis as compared with children who were older at diagnosis (P = 0.99). A paired analysis of 68 children whose weights were assessed at the start of treatment and one year later showed a significant increase in weight (P<0.001) but no change in the weight z scores (P = 0.37), suggesting that in this subgroup of children, the rate of weight gain kept pace with that in the normal population.
Neurodevelopmental function was evaluated in 112 children after surgery and in 34 children one year later. At the initial assessment, the distribution of scores of global cognitive development (measured by the Mental Development Index or the General Cognitive Index) was abnormal, with 50 patients (45 percent) scoring 15 or more points below the norm and 32 (29 percent) scoring 30 or more points below the norm. After one year of chemotherapy, 21 (62 percent) and 13 (38 percent) of the 34 children were similarly affected. No significant change was noted in the distribution of scores (P = 0.21). Of the 34 children evaluated at one year, 27 had been tested initially. Their mean change of -2.5 points in cognitive function was neither clinically important nor statistically significant (P = 0.65).
A second cancer developed in two children. Acute myelogenous leukemia developed in one child who had no response to chemotherapy or radiation, and a sarcoma developed in the second at the site of a previous choroid-plexus carcinoma after chemotherapy alone.
Discussion
In this large multi-institutional study, chemotherapy was the primary postoperative treatment for children with malignant brain tumors. Objective responses were observed at sites of both primary disease and metastatic disease within the neuraxis. Chemotherapy prevented disease progression and permitted radiation therapy to be delayed for one to two years in a substantial proportion of patients. The best results were seen among patients who had apparently undergone total surgical resection of localized disease. In this subgroup, 74 percent of children who were less than 24 months of age at diagnosis and 91 percent of those who were 24 to 36 months of age had no progression of disease during treatment with chemotherapy alone for one year. Among the younger patients who received chemotherapy for two years, two thirds of those with completely resected or localized tumors had no progression of disease at the end of chemotherapy.
No treatment for infants with brain tumors has been universally accepted, since radiation therapy has been considered to be too toxic. Where possible, we compared our results with those of multi-institutional studies of children with specific types of brain tumors that included a subgroup of either infants or older children with similar risk factors.
A subgroup of children less than four years of age was described in the results of the Children's Cancer Study Group medulloblastoma trial, in which postoperative craniospinal irradiation was given with or without vincristine, lomustine, and prednisone2. Although radiation was delayed in our patients for one to two years, progression-free survival was almost identical to that in the Children's Cancer Study Group. Any comparison of our results with those achieved in older children must be tempered by the recognition that young age is a poor prognostic factor in patients with medulloblastoma. Despite this, the very encouraging results in this study for children with medulloblastomas who had undergone gross total resection of localized disease compare favorably with those in older children with totally resected tumors in cooperative group studies from the Children's Cancer Study Group and the International Society for Pediatric Oncology28,29. Moreover, in patients with no evidence of residual disease after prolonged chemotherapy, a reduced dose of neuraxial radiation (2400 cGy) has achieved results comparable to those in similarly selected older children treated with a full dose of craniospinal radiation (3600 cGy)29.
The three-year survival rate of 61 percent for infants and very young children with ependymomas compares favorably with the Children's Cancer Study Group ependymoma trial, in which children less than seven years of age were treated with immediate postoperative radiation with or without lomustine, vincristine, and prednisone30. In contrast to this other study, our study found that the degree of surgical resection was an important prognostic variable (Figure 5).
Although the numbers of children studied were limited, we had very encouraging results in children with malignant gliomas and brain-stem gliomas. The two-year progression-free and overall survival rates of 54 percent and 65 percent, respectively, in children with malignant gliomas are superior to those achieved in older children treated with postoperative radiation alone (progression-free survival, 20 percent; overall survival, 40 percent)31 and are comparable to those in children treated with postoperative radiation and adjuvant chemotherapy (a combination of vincristine, lomustine, and prednisone, or an eight-drug regimen)32. Similarly, the 42 percent two-year survival rate in patients with brain-stem gliomas is superior to results reported after hyperfractionated radiation therapy directed at the posterior fossa in children less than 6.7 years of age with unselected brain-stem tumors33.
This chemotherapy regimen was least effective in children with embryonal tumors (primitive neuroectodermal tumors), who had significantly poorer responses to chemotherapy and progression-free survival than children with other types of tumors. Because this pathologic classification encompasses a variety of immature malignant tumors, our results cannot be compared with those of previous reports. However, it seems clear that this chemotherapy regimen cannot be recommended for patients with these primitive tumors.
This study has demonstrated that chemotherapy is effective against malignant brain tumors in infancy. Our overall results compare favorably with those obtained in young children treated with standard postoperative radiation. In selected children without postoperative residual disease, chemotherapy alone provided excellent control of malignant brain tumors. The goal of this study was to determine whether a delay in radiation therapy, brought about by the use of postoperative chemotherapy, would limit treatment-induced neurotoxicity without jeopardizing survival. That goal has been achieved. Furthermore, although long-term follow-up will be necessary, there has been no clinically important neurotoxicity to date. Appropriate neurodevelopmental milestones have been reached and, in the aggregate, heights and weights have been maintained relative to those in a normal population. On the basis of the success of this approach, subsequent studies will seek to determine whether intensification of chemotherapy can improve efficacy, thereby allowing the delay or elimination of radiation in a larger proportion of infants and very young children.
We are indebted to Dr. Teresa Vietti, Chair, Pediatric Oncology Group, for her vision and support.
Source Information
From the State University of New York at Buffalo School of Medicine and Biomedical Sciences and the Roswell Park Cancer Institute, Buffalo (P.K.D., M.E.C.); the Pediatric Branch of the National Cancer Institute, Bethesda, Md. (M.E.H.); the Pediatric Oncology Group Statistical Office, University of Florida, Gainesville (J.P.K.); Duke University Medical Center, Durham, N.
(H.S.F., P.C.B.); St. Jude Children's Research Hospital, Memphis, Tenn. (R.A.S., R.K.M., L.E.K.); Children's Mercy Hospital, Kansas City, Mo. (F.G.S.); University of California Medical Center, San Diego (H.E.J.); and Montreal General Hospital, Montreal (C.R.F.).
Address reprint requests to Dr. Duffner at the Pediatric Oncology Group Operations Office (8633, 8634), 4949 W. Pine Blvd., St. Louis, MO 63108.
References
-D-thenylidene-glucoside (PTG) in the treatment of malignant intracranial neoplasms. Cancer 1974;33:460-467. [CrossRef][Medline]
| |||||||||||||||||||||||||||||||||||||||
Related Letters:
Brain Tumors in Infants
Jenkin D., Longcope J. C., Watson S. S., Duffner P. K., Allen J. C., Kun L. E.
Extract |
Full Text
N Engl J Med 1993;
329:1963-1964, Dec 23, 1993.
Correspondence
This article has been cited by other articles:
HOME | SUBSCRIBE | SEARCH | CURRENT ISSUE | PAST ISSUES | COLLECTIONS | PRIVACY | TERMS OF USE | HELP | beta.nejm.org Comments and questions? Please contact us. The New England Journal of Medicine is owned, published, and copyrighted © 2009 Massachusetts Medical Society. All rights reserved. |
