• solid tumor;
  • brain metastases;
  • child;
  • relapse


  1. Top of page
  2. Abstract


Brain metastases are uncommon among children with solid tumors. However, improvements in survival have increased the period of time during which children are at risk for developing these metastases. The authors reviewed brain metastases in children with solid tumors treated at the Centre Léon Bérard during the 9 years between 1987 and 1995.


Among 486 patients with solid tumors, 162 eventually developed distant metastases in their disease process, including 12 brain metastases detected by imaging. The tumor type, clinical setting, imaging characteristics, treatment modalities, and outcome were assessed for each patient.


The most common tumors causing brain metastases were Ewing's sarcoma (in three patients), neuroblastoma (in three patients), and osteogenic sarcoma (in three patients). At the time of initial diagnosis, 9 of the 12 patients had metastatic disease. All but one patient initially received intensive multiagent chemotherapy, including high dose chemotherapy with bone marrow rescue in six patients. The median time from initial diagnosis to the detection of brain metastases was 15 months. These metastases were clinically detectable in 10 patients and subclinical in 2 patients. Brain metastases were present at the time of first relapse in five patients. In two patients, the brain was the only site of relapse. All other patients had extensive systemic disease. Seven patients had multiple brain metastases. Two children underwent surgical resection of solitary metastases, and eight were irradiated. One child achieved complete remission following chemotherapy and irradiation. All other children died, mostly of their systemic disease, within a median period of 3 months.


The introduction of effective systemic chemotherapy has changed the patterns of brain metastases in children. The increasing incidence of these metastases in patients with sarcoma and neuroblastoma suggests that the brain is a pharmacologically protected site in patients initially diagnosed with metastatic disease. Cancer 1997; 79:403-10. © 1997 American Cancer Society.

Brain metastases occur in 20-30% of adult patients with systemic cancer.1 The point in the disease process at which these metastases appear can vary, but they sometimes reveal the underlying disease. The most common primary tumors associated with brain metastases are melanoma and lung, breast, kidney, and colon carcinoma.2 Because brain metastases represent a challenge in adults with cancer, many studies have focused on these areas in the past few decades. The occurrence of brain metastasis in childhood is more unusual, and only a few reports have described their clinical and radiologic features. The findings for adults, however, cannot be extrapolated to the pediatric population because the spectrum of primary tumors in children differs markedly. Reports of the dissemination of pediatric solid tumor in the brain were anecdotal until recently. In the early 1970s, reports were concerned with the propensity of Ewing's sarcoma to develop intracranial metastases, and prophylactic whole-brain irradiation and intrathecal chemotherapy has been advocated in the management of newly diagnosed Ewing's sarcoma patients.3 Few overviews of brain metastases in children have been previously reported,4-6 and little is known about the changes that have come with the advent of aggressive multimodal therapy, such as intensive chemotherapy and high dose chemotherapy with bone marrow rescue. As the survival of children with malignancies increases, the natural history of these diseases is changing and might increase the risk for the development of brain metastases. In addition, improved neuroimaging techniques, such as magnetic resonance imaging (MRI), can detect brain metastases earlier with greater certainty and thus appear to increase the reported incidence of these metastases. This report summarizes 9 years of experience at the Centre Léon Bérard in dealing with metastases to the brain in children with solid tumors.


  1. Top of page
  2. Abstract

The records of children with metastases of solid malignant tumors to the brain who were treated at the Centre Léon Bérard between 1987 and 1995 were reviewed. Patients with primary brain tumors and lymphoma were excluded. Patients with lesions secondary to direct extension from the skull or dura and those with isolated cerebrospinal fluid (CSF) seeding were also excluded. The diagnosis of brain metastasis was ascertained by reviews of computed tomography (CT) and MRI scans. Autopsy reports were not examined for this study. Tumor type, location, and initial staging, patient's age at diagnosis, interval from diagnosis to brain metastases, the extent of systemic disease at time of brain metastases, clinical symptoms, treatment, and outcome were recorded. The treatment modalities were determined jointly by the oncologist, the neurosurgeon, and the radiotherapist. Treatment varied according to the potential for resection, chemotherapy or irradiation, the extent of the underlying disease, and the wishes of the family. Because patients often die of their systemic disease, we examined the success of treatment modalities according to deaths from neurologic causes separately from deaths due to systemic disease.


  1. Top of page
  2. Abstract

Of the 486 children with solid malignant tumors included in the study, 205 (41%) developed tumor recurrences, including 162 with distant metastases. Among those with distant metastases, 12 patients (7.4%) presented with brain metastases (Table 1). The overall frequency of brain metastases at our institution was 2.4% (12 of 486 patients). There were 9 boys and 3 girls, ranging in age from 3 to 17 years (median, 9 years). The incidence of brain metastases differed according to the type of primary tumor. The most common tumors causing brain metastases were Ewing's sarcoma, neuroblastoma, and osteosarcoma (Table 2). The other occurrences included one child with Wilms' tumor, one with synovial sarcoma, and one with retinoblastoma. At time of initial diagnosis, nine patients had metastatic disease. Three children had Stage IV neuroblastoma with bone metastases and bone marrow involvement. Two children with Ewing's sarcoma, two with osteosarcoma, one with Wilms' tumor, and one with synovial sarcoma had isolated metastases to lung at diagnosis.

Table 1. Incidence of Brain Metastases in Children with Solid Tumors
DiagnosisNo. of patientsNo. of patients with met to the brain (%)No. of relapsing patients with distant met% with distant brain met
  • Met: metastases.

  • a

    Represents mean percentage.

Ewing's sarcoma543 (5.5)319.7
Neuroblastoma1603 (1.9)624.8
Wilms' tumor781 (1.3)137.7
Soft tissue sarcoma531 (1.9)166.2
Osteosarcoma663 (4.5)2313
Germ cell tumor270 (0)30
Other cancers481 (2.1)147.1
Total48612 (2.4a)1627.4a
Table 2. Patients' Characteristics, Treatment, and Outcome
Patient no.Gender/ageDiagnosisMet at diagnosisInterval (mos)Location of brain metNo. of metTreatmentOutcome
  1. M: male; F: female; BM: bone marrow; XRT: radiation therapy; CR: complete remission; DSD: dead of systemic disease; DBM: dead of brain metastases; Met: metastases; Supra: supratentorial; Infra: infratentorial.

1M/8RetinoblastomaBilateral84Supra1Chemo + XRTCR 76+ mos
2M/17OsteosarcomaLocalized15Supra2NoneDSD 2 mos
3M/9Ewing's sarcomaLung18Supra1Surg + XRTDSD 15 mos
4M/3NeuroblastomaBone, BM11Infra and supra4NoneDBM 1 day
5M/3NeuroblastomaBone, BM17Infra and supra + spine5XRTDSD 4 mos
6M/9OsteosarcomaLung14Supra1XRTDBM 2 mos
7F/17Ewing's sarcomaLocalized51Infra3XRTDSD 3 mos
8M/16Synovial sarcomaLung24Supra1XRTDSD 38 mos
9F/4NeuroblastomaBone, BM9Supra3NoneDSD 2 mos
10M/17OsteosarcomaLung12Supra2XRTDSD 4 mos
11M/17Ewing's sarcomaLung11Infra and supra + spine2XRTDBM 2 mos
12F/6NephroblastomaLung10Supra1SurgDBM 4 mos

Clinical Features

The median interval from the diagnosis of the primary tumor to the detection of brain metastases was 15 months (range, 9-84 months). In five patients, brain metastases were the unique sites of first relapse (two patients) or discovered at time of first relapse (three patients). In four patients these metastases were detected in the context of refractory disease or poor response to initial treatment. In the last three patients, brain metastases occurred respectively at time of the second, third,and fifth relapses. Two patients had isolated brain metastases without any other sign of systemic disease. All other patients had systemic metastases involving lungs (in six patients), bone and bone marrow (in three), and cervical lymph nodes (in one). Two patients had brain metastases and intraspinal metastases simultaneously; one of them also had positive CSF.

Imaging Characteristics

Brain metastases were clinically detectable in 10 patients. The delay from first symptoms to the radiologic diagnosis of brain metastases was short, ranging from 1-30 days in symptomatic patients (median, 5 days). In two patients with spinal metastases, they were disclosed by MRI of the brain that was performed routinely. Neurologic symptoms related to brain metastases were lethargy (in four patients), symptoms of increased intracranial pressure (in two patients), diplopia (in 2 patients), speech disorders (in 2 patients), hemiparesis (in 2 patients), facial palsy (in 1 patient), and seizures (in 1 patient). One child with neuroblastoma had an acute onset of lethargy followed by profound coma as a first symptom of isolated central nervous system relapse. He died 24 hours after the first symptoms appeared, despite high doses of steroids and mannitol.

Brain metastases were detected by CT scan in 10 patients and MRI in 2. The patient with the longest delay to metastases (30 days) had a CT scan that failed to show any lesion and a later MRI that revealed 3 infratentorial metastases. Five patients had solitary metastases. The number of metastases detected in the seven remaining patients ranged from two to five. All three of the patients with neuroblastoma had multiple metastases, as did two patients with Ewing's sarcoma and two with osteosarcoma. In eight patients, the supratentorial area was the sole site of brain metastases. One patient had isolated infratentorial metastases. Three patients had both supratentorial and infratentorial metastases. Infratentorial metastases were located in the brain stem or the cerebellum (Fig. 1). Their sizes ranged widely, from 0.2 to 20 cm2. Most of them were small; only 5 measured more than 5 cm2. Metastases were spontaneously dense and homogeneous before contrast (Fig. 2). No calcification and no cyst was noted. One patient exhibited a pattern of hemorrhagic metastases that was confirmed by histologic examination. Mild to moderate edema was present in seven patients, and diffuse brain edema was present in one patient, along with radiologic signs of subarachnoidal seeding. Edema was absent in four patients, including two with subclinical metastases. Following contrast, lesions were uniformly nodular in 11 patients and heterogeneous in 1.

thumbnail image

Figure 1. Patient 7, a girl age 17 years, had Ewing's sarcoma of the left arm with perivertebral soft tissue recurrence, as well as a 3-week history of facial palsy and diplopia with normal computed tomography scan. This gadolinium-enhanced magnetic resonance imaging scan shows several nodular and pericentimetric foci of enhancements in pontocerebellar angles, the brainstem, and the left cerebellar peduncles (black arrows). These enhancements disappeared after radiotherapy.

Download figure to PowerPoint

thumbnail imagethumbnail image

Figure 2. Patient 1, a boy age 8 years presented with bilateral retinoblastoma seven years ago. He had a history of anterior headache, vomiting, and seizures. Cervical adenomegalies were positive for retinoblastoma on cytologic smears. (a) This cranial computed tomography (CT) scan shows spontaneous hyperdense solid mass in the frontal lobes (30 mm in diameter) without edema, brightly enhanced after contrast. (b) This postchemotherapy CT scan shows complete disappearance of the tumor mass.

Treatment and Outcome

Three patients received palliative treatment with steroids. Two underwent surgical resection of a single lesion. Irradiation was administered to eight patients as sole treatment (for six patients), after surgery (for one patient), or after chemotherapy (for one patient). Five of six patients treated with irradiation alone improved clinically. One patient treated with stereotactic radiosurgery developed further brain metastases and received whole-brain irradiation. One child treated with surgery alone had an early recurrence at the same site and then underwent radiotherapy without any evidence of improvement. The other child who underwent surgery developed metastases to bone 3 months afterwards but never showed evidence of recurrent brain tumor. Finally, the child with retinoblastoma achieved complete remission after two courses of intensive chemotherapy with etoposide and carboplatin (Fig. 2). He was then treated with high dose chemotherapy followed by bone marrow rescue and focal irradiation on the frontal area. He is the only long term survivor, with a 6-year continuous complete remission. The other children died within a median of 3 months (range, 1 day to 38 months). Seven deaths were related to the evolution of the systemic disease and four to neurologic causes.


  1. Top of page
  2. Abstract

Brain metastases have been reported in most pediatric solid tumors: osteosarcoma,7-9 Ewing's sarcoma and peripheral neuroectodermal tumor,3 soft tissue sarcoma,10-13 Wilms' tumor,14 hepatoblastoma,15 neuroblastoma,16-18 germ cell tumor,19 retinoblastoma,20 and melanoma.21 However, only 3 previous series have focused on patterns of brain metastases in children with solid tumors in the past 20 years.4-6 Because brain metastases in children were not registered before 1987 at our department, we were not able to make comparisons with data from our institutions from before that year. Initial reports in the literature were based on postmortem examinations and revealed an incidence ranging from 6-10% in children younger than 15 years who had died of their malignancies.5 However, the past few decades encompassed an era when the introduction of effective chemotherapy dramatically improved survival. Several new agents were introduced for the management of cancer patients, such as doxorubicin, cisplatinum, carboplatin, and etoposide. More intensive protocols have been assessed for patients with advanced disease, and high dose chemotherapy followed by bone marrow rescue is increasingly used in an attempt to achieve a better survival rate for patients with a poor prognosis.22 Neuroimaging techniques have also changed. CT and MRI scans have replaced the obsolete brain scintigraphy and pneumoencephalography as the optimal means of brain imaging. Consequently, the patterns of brain metastases may have changed through this period. To illustrate these changes, Vannucci and Baten found Wilms' tumor to be the most common cause of brain metastases, representing 13% of their autopsy study.4 The survival rate for patients with Wilms' tumor approaches 90% today, and reports of metastases to the brain are uncommon.14

The current series is based on clinical and radiologic data, whereas previous reports were based mainly on postmortem examinations (Table 3). Studies of the autopsies of adult patients have been thought to lead to overestimations of the clinical incidence of brain metastases. This is easily explained by the finding that oncologists do not systematically assess neurologic problems in palliative care patients; this makes comparison with historical data difficult. However, when the previous report of Vannucci and Baten4 was compared with the current series, the incidence of brain metastases in children did not differ substantially. The reasons for this are unclear. There are three possible explanations: (1) imaging techniques are as capable of identifying brain lesions as autopsy studies; (2) progress in chemotherapy, surgery, and radiotherapy has changed the progress of pediatric solid tumors, and current high risk patients are defined differently from those of 20 years ago; and (3) the natural history of some diseases is changing, and the brain is a sanctuary where relapse may occur in patients with prolonged survival, as we are about to discuss.

Table 3. Incidence of Brain Metastases in Children with Solid Tumors in Three Different Series
 No. of patients with brain met
DiagnosisVannucci and Baten (1974)4 aGraus et al. (1983)5 bCurrent series (1996)
  1. Met: metastases.

  2. a

    Examinations in this study were postmortem.

  3. b

    CT scans or postmortem examinations.

Ewing's sarcoma0/152/123/54
Wilms' tumor4/31 1/78
Soft tissue sarcoma3/453/211/53
Germ cell tumor 4/80/27
Other cancers3/394/211/48

The patterns of brain metastases in children are changing. Cerebral metastases are seen as a part of the natural history of sarcoma.10 However, they are unusual in patients with neuroblastoma. Since chemotherapy has prolonged the survival of patients with metastatic neuroblastoma, increasing numbers of reports have described the brain as a site of tumor recurrence.16-18, 23 Brain metastases are often multiple in these patients and are sometimes associated with spinal metastases, as in our Patient 5.17, 24 An unusual feature of brain metastasis in neuroblastoma patients is a diffuse meningeal involvement causing increased intracranial pressure and coma. Previous reports have focused on this occurrence.17, 25 Our Patient 4 had a Stage IV neuroblastoma with N-myc amplification that has been suggested to be a risk factor for central nervous system (CNS) relapse.26

The manifestations of brain metastases in children are strikingly different from those in adults. They are the initial manifestation of the systemic neoplasm in 5-10% of adults,27 whereas, to our knowledge, this has never been reported in children. Moreover, the origin of brain metastases remains unknown in 10-25% of adult cases. In children, previous reports have emphasized that they occur in the setting of widely metastatic disease. However, there is an increasing incidence of reports of brain metastases as the first site of relapse.7, 8, 12, 14, 17, 28 In our series, they occurred at the time of first recurrence in five patients. This suggests that though chemotherapy has improved the control of systemic disease, the CNS remains a pharmacologically protected site. Similar situations were encountered in childhood acute lymphocytic leukemia in the 1970s29; thus, the question of CNS prophylaxis has been raised for patients with solid tumors, especially Ewing's sarcoma. Trigg et al. reported administering whole-brain irradiation and intrathecal chemotherapy to 62 patients with Ewing's sarcoma.30 They concluded that CNS relapses were not prevented by this prophylaxis. Unfortunately, more intensive chemotherapy protocols do not seem to prevent their occurrence. In our series, six patients developed brain metastases after high dose chemotherapy, including one after total-body irradiation. For patients at risk for brain metastasis (e.g., metastatic sarcoma and neuroblastoma patients), further chemotherapy studies might identify new agents that are likely to cross the blood-brain barrier.

The rapidity of the onset of signs and the frequency of intracranial hypertension or lethargy in children differs from neurologic manifestation observed in adult patients. The rapid kinetics of the tumor growth of childhood malignancies is likely to explain such differences. Graus et al. reported an incidence of 36% of seizures in their cohort; the seizures occurred most commonly in children younger than 15 years and in patients with germ cell tumors.5 Only one patient in the current series developed seizures, and this could be explained by the absence of patients with germ cell tumors. An acute onset with coma and subsequent death seems to be specific to childhood. Two children had subclinical brain metastases. Autopsy reports by Graus et al. and Vannucci and Baten also described such findings.4, 5 However, in our patients they were found 2 and 5 months before death. Because many children are given aggressive treatments at the time of tumor recurrence, a systematic assessment of subclinical brain metastases should be appropriate for high risk neuroblastoma and sarcoma patients.

New imaging technology has improved the diagnostic yield for both primary central neoplasms and brain metastases. The dramatic increase in the observed incidence of primary intracranial neoplasm in adult patients during the past few decades has been attributed in part to improved detection with new imaging procedures.31 Moreover, MRI scan appears to be more accurate than CT scan in detecting intraparenchymal metastases.32 For children, more extensive restaging that includes the CNS at the time of tumor recurrence might also influence the incidence of secondary brain lesions. Two children in our study had subclinical brain deposits associated with spinal metastases. This indicates that in children the whole CNS is likely to be the site of recurrent disease. Imaging characteristics of brain metastases in children have been previously analyzed. Pedersen et al. reported multiple metastases in 25% of the children in their study, whereas 7 out of 12 children in our series had multiple metastases.33 However, the study of Pedersen et al. included patients with skull lesions with contiguous parenchymal extension, whereas these patients were excluded from our analysis. Based on our experience and on the studies of Vannucci and Baten and Graus et al., we conclude that the rate of occurrence of multiple metastases in children is not different from the rate in adult patients (at least 50%). The main differences with adults concern the small size of the lesions and the frequent absence of peritumoral edema.

Treatment of metastases to the brain has yielded disappointing results. The studies published up to now have reported only a short relapse-free period of survival. Few studies have reported prolonged survival.8, 9, 12, 18 This is mostly related to the common preclusion of any aggressive management of extensive metastatic disease. Moreover, most patients were heavily pretreated with multiagent chemotherapy, which limits the potential for additional chemotherapy. For the few patients with solitary metastases and no systemic disease, surgery may be indicated. For other patients, any conclusion regarding the benefit of surgery, chemotherapy, or radiotherapy is impossible. Whole-brain irradiation may benefit those patients who have multiple metastases or whose condition precludes neurosurgery. Five of six patients in the current series improved neurologically with radiotherapy but subsequently died of their systemic disease. Surgery alone was followed by an early local recurrence in one child; therefore, we conclude that radiotherapy has to be considered following resection. Finally, chemotherapy induced a complete remission in a child with metastatic retinoblastoma. This prolonged survival indicates that opportunities for salvage therapy have to be examined for every child with brain metastases.

As the natural history of pediatric malignancies changes as a result of the introduction of effective systemic chemotherapy, brain metastases may become a new concern for pediatric oncologists. The increasing number of affected neuroblastoma patients illustrates these changes in pediatric malignancies. Further studies should assess the high risk patients and treatment opportunities to prevent CNS recurrences.


  1. Top of page
  2. Abstract