- Email: [email protected]
Brain metastases in children
Brain metastases in children We reviewed the records of 31 children under the age o f 21 years with parenchymal brain metastases diagnosed by CT scan (13 patients) or necropsy (18). Brain metastases were found in 18 of 139 (13%) children with solid tumors in whom complete postmortem examinations were done. Osteogenic sarcoma and rhabdomyosarcoma were the most frequent primary tumors causing brain metastases in patients younger than 15 years, and testicular germ cell tumor, from age 15 to 21 years. Evidence of intratumoral hemorrhage was found in 50% of autopsy cases. Pulmonary metastases were present in 28 o f 3l (90%), The median interval from recognition of pulmonary metastases to the development of brain metastases was 10 months. No patient had evidence of brain metastases at diagnosis of the systemic cancer. In only one patient was the brain the only site of relapse. Following detection of brain metastases, the median survival was seven months in six patients who underwent surgery and whole-brain radiation therapy and four months in 15 given radiotherapy (3000 rads) alone, patients with relatively radioresistant brain metastases may benefit from surgical excision or higher doses of radiation, or both. (J PEDIATR 103:558, 1983)
Franeesc Graus, M.D., Russell W. Walker, M.D., and Jeffrey C. Allen, M.D. N e w Y o r k , N . Y .
THE MOST COMMON childhood solid tumors other than primary brain tumors are neuroblastoma and sarcomas ~including rhabdomyosarcoma, osteogenic sarcoma, Ewing garcoma, and Wilms tumor. Collectively, these solid tumors constitute 30% of childhood cancers. Intraparenchymal brain metastases constitute more than 50% of all intracranial metastatic disease ~ and are thought to result from hematogenous dissemination. Intracranial tumors m a y also originate as direct extensions of metastatic lesions of the skull or dura mater. The incidence of brain metastases in adults with solid tumors ranges from 10% to 20% t at autopsy, but appears to be lower in children.2 In a large cooperative clinical study of 445 patients with Ewing sarcoma~ brain metastases were detected in 1.8% of patients2 The rarity of this clinical problem in a pediatric oncology population has made a unified approach to its management difficult.
From the Departments of Neurology and Pediatrics, Memorial Sloan-Kettering Cancer Center. Dr. Graus is a recipient of a grant from the Fundacio del Congres de Cultura Catalana Barcelona and the l:ondo de lnvestigaciones Sanitarias, Madrid, Spain. Reprint requests: JeJfrey C. Allen, M.D., Memorial Hospital, 1275 York Ave., New York, N Y I0021.
~$8
The Journal of P E D I A T R I C S
The increasing survival of children and adults with solid tumors in recent years has increased the time at risk for developing brain metastases. In addition, with the use of computed tomography, brain metastases can be recognized with greater certainty and treated earlier. Several reports have suggested that the introduction of effective systemic chemotherapy for the treatment of sarcomas in adults has not only increased survival but has resulted indirectly in an increased incidence of brain metastases? We review brain metastases from solid tumors in children examined at Memorial Sloan-l~ettering Cancer Center during the 10 years from January 1973 to December 1982, with special attention to the incidence, clinical setting, and response to therapy.
METHODS Patients were ascertained by either review of autopsy records or computed tomography scan files. We reviewed the autopsy records of all patients age 21 years and younger in whom the primary diagnosis was a solid tumor other than lymphoma or a primary brain tumor and in whom a complete autopsy was performed. Approximately 30% of autopsies were excluded because the central nervous system was not examined. Patients who had had brain metastases secondary to direct extension from skull or dura
Volume 103 Number 4
Brain m e t a s t a s e s in children
559
Table I. Incidence of brain metastasis in solid malignant tumors examined postmortem Age ~ 1 5 years Brain metastases
Patients
Neuroblastoma Osteogenic sarcoma Rhabdomyosarcoma Ewing sarcoma Germ cell tumor Other Total
Age <~21 years Brain metastases
Patients
n
n
%
n
n
37 28 15 6 2 14 102
0 4 2 0 1 3 10
0.0 14.3 13.3 0.0 50.0 21.6 9.8
39 38 21 12 8 21 139
0 5 3 2 4 4* 18
13.5 14.3 16.6 50.0 19 - 12.9
*Malignant schwannoma (two patients), melanoma (one), angiosarcoma (one). Table It. Interval from diagnosis of primary tumor to detection of lung and symptomatic brain metastases in 25 children Lung metastases*
Osteogenic sarcoma Germ cell tumor Rhabdomyosarcoma Othert
Patients with brain metastases
Median interval
4 8 6 7
8 1 1 1
(mo)
Brain metastases
Range (mo)
5 0 0 0
to to to to
22 12 144 26
Median interval (too)
22 13 8.5 20
Range (too)
10 0 3 10
to to to to
56 39 144 60
*Lung metastases never developed in three patients (two with primary pelvic rhabdomyosarcomas, one with Wilms tumor). "tEwing sarcoma (three), melanoma, Wilms tumor (one), angiosarcoma (one), and fibrosarcoma (one). mater were also excluded. Of 139 patients identified, 18 had had brain metastases. Thirteen additional patients were ascertained via review of C T files over the past 41/2 years, These patients had had lesions on C T that were compatible with intracerebral metastases. In three, histologic confirmation of brain metastases was made at surgery. The clinical records of all 31 patients were subsequently reviewed in detail. Brain metastases were considered responsible for death when they caused cerebral herniation or when the patient with a severe neurologic deficit died of an intercurrent illness? RESULTS The incidence of brain metastases was highest in children with germ cell tumor (Table I). Patients with osteogenie sarcoma, rhabdomyosarcoma, and Ewing sarcoma had a lower incidence (10 of 17), approximately 14%. None of the 39 patients with neuroblastoma had brain metastases. The median interval from the diagnosis of the primary tumor to the detection of lung and symptomatic brain metastases in 25 patients is summarized i n Table II. Pulmonary metastases were present in 22 of 25 patients (88%) at the time intracranial disease was detected. All
patients except one with rhabdomyosarcoma had either active or silent systemic disease at the time diagnosis of brain metastases was made. With the exception of patients with osteogenic sarcoma, the median interval between the diagnosis of the primary neoplasm and the diagnosis of pulmonary metastatic disease was only one month, reflecting the widespread dissemination of the disease at diagnosis. No patient had clinical symptoms of brain metastases as the initiaI manifestation of the primary malignancy. In patients with osteogenic sarcoma, pulmonary metastases tended to be diagnosed later. The median interval between the diagnosis of lung metastases and brain metastases in the whole group was 10 months (range 0 to 60 months). Neurologie symptoms referable to brain metastases were found in 25 of 31 patients (81%). The other 19% harbored subclinical metastatic lesions, which were discovered postmortem. The most common signs and symptoms of brain metastases were headache or signs of increased intracranial pressure (52%), seizures (36%), hemiparesis (36%), and altered mental status (16%). Seizures were more frequent in younger children and in patients with germ cell tumors. Seizures were the initial symptom in 50% of patienis younger than 15 years, but in only 18.5% of patients older than 15 years. Five of eight
560
Graus, Walker, and Allen
(62%) patients with germ cell tumors had seizures. Five patients (two with germ cell tumor, two with Ewing sarcoma, and one with rhabdomyosarcoma) had acute and catastrophic clinical signs and symptoms, including headache, lethargy, and hemiparesis. Three died Within a few days. In all five, autopsy Or CT scan revealed a hemorrhage into a metastasis. Solitary metastases were present in seven patients (six with sarcomas and one with germ cell tumor) in our autopsy series, three of whom had subclinical lesions. Hemorrhagic metastases were found in nine of the 18 brains studied (four patients with germ cell tumor, two with osteogenic sarcoma, two with Ewing sarcoma, and one with rhabdomyosarcoma). In all but two patients With hemorrhagic metastases, the p!ateiet count was >50,000/ mm 3. In two of 20 patients who had CT scans, the initial scan showed a hematoma without obvious radiologic evidence Of an underlying metastasis. CerebeHar metastases were present in seven of 31 (23%) patients. Of the 25 patients in whom the diagnosis of brain metastases was made during life, 21 underwent combinations of surgery or radiation therapy or both. The median survival of six patients with solitary brain metastases who had surgical excisions followed by radiotherapy was seven months. The usual radiation dose was 3000 rads administered to the whole brain over 10 elapsed days. One patient is still alive at 41 months, although he acquired a relapse at the same site. In three of these patients who had postmortem examinations, no microscopic evidence of tumor was found at the surgical site. Only one of the patients died of a recurrent brain tumor. Fifteen patients were treated with radiation but without surgery; the majority also received chemotherapy. The median survival in this group was four months, but death was caused by neurologic disease in only four (27%) patients. A complete response to radiotherapy was detected on CT by three months in four of five patients with germ cell tumors. However, only one of six patients with sarcoma responded in a similar fashion. Of the five patients who experienced a complete response to radiotherapy, none died of recurrent brain disease. Of the six nonresponders, three died of brain metastases. DISCUSSION A survey of brain metastases in autopsy cases of children with solid tumors from 1973 to 1982 revealed an incidence of 18 of 139 (12.9%) in children younger than 21 years and 10 of 102 ( 9 . ' ~ ) in children younger than i5 years. This represents an increase in the incidence of brain metastases at this institution over the past decade. Vannucci and Baten 2 found an incidence of 6% in an autopsy series of 217 children younger than 15
The Journal of Pediatrics October 1983
years of age at our institution between 1951 and 1972. In their Study, Wilms tumor was the most common cause of brain metastases, but contributed to none of the cases in Our study. This tumor represented 13%of all diagnoses in their autopsy study, but only 5% of diagnoses in our study of children younger than 15 years. The reduced frequency of Wilms tumor is most likely related to the introduction of effecti~,e chemotherapy that has drastically reduced the mortality and institutional referrals of this malignancy. In our study, the most common solid tumors causing brain metastases were osteogenic sarcoma, rhabdomyosarcoma, Ewing sarcoma, and germ cell tumors. Brain metastases most frequently arose in the context of extensive metastatic disease, especially involving the lungs. The increased incidence of brain metastases in our study most probably reflects our ability to prolong survival of patients with disseminated disease with more aggressive treatment. The absence of parenchymal brain metastases in patients with neuroblastoma most likely reflects the rarity of pulmonary metastases in this disease. The median interval from the onset of pulmonary metastases to the diagnosis of the brain metastases did not differ significantly among the various tumors (Table II). However, with the exception of osteogenic sarcoma, the median interval from diagnosis of the primary tumor to the detection of pulmonary metastases was one month, indicating that the cancer in these patients had an extremely aggressive course from the start. Pulmonary metastases usually preceded the recognition of brain metastases by 10 months. All but one patient had identifiable systemic disease at the time of diagnosis of brain metastases, and all but three patients had pulmonary metastases. The neurologic manifestations that led to the diagnosis of metastatic brain disease in our study differ from those found among adults in two ways; the higher incidence Of seizures and the abrupt onset of the symptoms.6 Seizures were seen most~bften in children younger than 15 years and in patients with germ cell tumors. There was no correlation between the presence of seizures and the demonstration of hemorrhagic metastases at autopsy or on CT scan. Schold et al. 7 also found a high incidence of seizures as the initial symptoms of brain metastases in a series of adult patients with germ cell tumors. Acute or catastrophic focal neurologic symptoms arose in.20% of patients in our series. A lower incidence (10%) of similar presentations has been reported from our institution in adults with primary brain metastases, predominantly from carcinomas? This clinical presentation usually signaled hemorrhage into tumor. Hemorrhagic metastases were observed at autopsy in 50% of our patients. This high incidence of hemorrhagic lesions arising in children with
Volume 103 Number 4
primarily metastatic sarcomas has not been previously reported, although apoplectic neurologic symptoms with or without hemorrhagic brain metastases have been observed occasionally in patients with brain metastases from sarcomas.3.8 The results of treatment in our study are similar to those reported in other series of children and of adult patients with brain metastases? ,~ The longer survival in patients who underwent surgery probably reflects the better clinical condition and prognosis of patients selected for surgery. The responses observed with radiation therapy are consistent with the known radiosensitivity of germ cell tumors and the relative radioresistance of sarcomas. Surgery would appear to be indicated, therefore, in patients with sarcomas who have single brain metastasis, provided they are good operative risks and have relatively long life expectancy. Higher doses of radiation therapy or the concomitant use of chemotherapy may be of benefit to those patients with radioresistant tumors who are not candidates for surgery. The importance of achieving a complete response to radiotherapy is highlighted by the fact that none of our five patients whose brain metastases responded to radiation died from recurrent brain disease. Deutsch et al., 9 in a review of brain metastases, recommend higher radiation dosages, comparable to those used to control the primary tumor.
Brain metastases in children
561
We thank Dr. Jerome Posner for critical review of the manuscript.
REFERENCES
l. Posner JB: Brain metastases: A clinician's view. In Weiss L, Gilbert HA, Posner JB, editors: Brain metastases. Boston, 1980, GK Hall, pp 2-30. 2. Vannucci RC, Baten M: Cerebral metastatic disease in childhood. Neurology 24:981, 1974. 3. Trigg ME, Glaubiger D, Nesbit ME: The frequency of isolated CNS involvement in Ewing's sarcoma. Cancer 49:2404, 1982. 4. Espana P, Chang P, Wiernik PH: Increased incidence of brain metastases in sarcoma patients. Cancer 45:377, 1980. 5. Cairncross JG, Kim JH, Posner JB: Radiation therapy for brain metastases. Ann Neurol 7:529, 1980. 6. Posner JB: Clinical manifestations of brain metastases. In Weiss L, Gilbert HA, Posner JB, editors: Brain metastases. Boston, 1980, GK Hall pp 189-208. 7. Schold CF, Vurgrin D, Golbey RB, et ah Central nervous system metastases from germ cell carcinoma of testis. Semin Oncol 6:102, 1979. 8. Gercovieh FG, Luna MA, Gottlieb JA: Increased incidence of cerebral metastases in sarcoma patients with prolonged survival from chemotherapy. Cancer 36:1843, 1975. 9. Deutsch M, AIbo V, Wollman MR: Radiotherapy for cerebral metastases in children. Int J Radiat Oncol Biol Phys 8:1141, 1982.
Franeesc Graus, M.D., Russell W. Walker, M.D., and Jeffrey C. Allen, M.D. N e w Y o r k , N . Y .
THE MOST COMMON childhood solid tumors other than primary brain tumors are neuroblastoma and sarcomas ~including rhabdomyosarcoma, osteogenic sarcoma, Ewing garcoma, and Wilms tumor. Collectively, these solid tumors constitute 30% of childhood cancers. Intraparenchymal brain metastases constitute more than 50% of all intracranial metastatic disease ~ and are thought to result from hematogenous dissemination. Intracranial tumors m a y also originate as direct extensions of metastatic lesions of the skull or dura mater. The incidence of brain metastases in adults with solid tumors ranges from 10% to 20% t at autopsy, but appears to be lower in children.2 In a large cooperative clinical study of 445 patients with Ewing sarcoma~ brain metastases were detected in 1.8% of patients2 The rarity of this clinical problem in a pediatric oncology population has made a unified approach to its management difficult.
From the Departments of Neurology and Pediatrics, Memorial Sloan-Kettering Cancer Center. Dr. Graus is a recipient of a grant from the Fundacio del Congres de Cultura Catalana Barcelona and the l:ondo de lnvestigaciones Sanitarias, Madrid, Spain. Reprint requests: JeJfrey C. Allen, M.D., Memorial Hospital, 1275 York Ave., New York, N Y I0021.
~$8
The Journal of P E D I A T R I C S
The increasing survival of children and adults with solid tumors in recent years has increased the time at risk for developing brain metastases. In addition, with the use of computed tomography, brain metastases can be recognized with greater certainty and treated earlier. Several reports have suggested that the introduction of effective systemic chemotherapy for the treatment of sarcomas in adults has not only increased survival but has resulted indirectly in an increased incidence of brain metastases? We review brain metastases from solid tumors in children examined at Memorial Sloan-l~ettering Cancer Center during the 10 years from January 1973 to December 1982, with special attention to the incidence, clinical setting, and response to therapy.
METHODS Patients were ascertained by either review of autopsy records or computed tomography scan files. We reviewed the autopsy records of all patients age 21 years and younger in whom the primary diagnosis was a solid tumor other than lymphoma or a primary brain tumor and in whom a complete autopsy was performed. Approximately 30% of autopsies were excluded because the central nervous system was not examined. Patients who had had brain metastases secondary to direct extension from skull or dura
Volume 103 Number 4
Brain m e t a s t a s e s in children
559
Table I. Incidence of brain metastasis in solid malignant tumors examined postmortem Age ~ 1 5 years Brain metastases
Patients
Neuroblastoma Osteogenic sarcoma Rhabdomyosarcoma Ewing sarcoma Germ cell tumor Other Total
Age <~21 years Brain metastases
Patients
n
n
%
n
n
37 28 15 6 2 14 102
0 4 2 0 1 3 10
0.0 14.3 13.3 0.0 50.0 21.6 9.8
39 38 21 12 8 21 139
0 5 3 2 4 4* 18
13.5 14.3 16.6 50.0 19 - 12.9
*Malignant schwannoma (two patients), melanoma (one), angiosarcoma (one). Table It. Interval from diagnosis of primary tumor to detection of lung and symptomatic brain metastases in 25 children Lung metastases*
Osteogenic sarcoma Germ cell tumor Rhabdomyosarcoma Othert
Patients with brain metastases
Median interval
4 8 6 7
8 1 1 1
(mo)
Brain metastases
Range (mo)
5 0 0 0
to to to to
22 12 144 26
Median interval (too)
22 13 8.5 20
Range (too)
10 0 3 10
to to to to
56 39 144 60
*Lung metastases never developed in three patients (two with primary pelvic rhabdomyosarcomas, one with Wilms tumor). "tEwing sarcoma (three), melanoma, Wilms tumor (one), angiosarcoma (one), and fibrosarcoma (one). mater were also excluded. Of 139 patients identified, 18 had had brain metastases. Thirteen additional patients were ascertained via review of C T files over the past 41/2 years, These patients had had lesions on C T that were compatible with intracerebral metastases. In three, histologic confirmation of brain metastases was made at surgery. The clinical records of all 31 patients were subsequently reviewed in detail. Brain metastases were considered responsible for death when they caused cerebral herniation or when the patient with a severe neurologic deficit died of an intercurrent illness? RESULTS The incidence of brain metastases was highest in children with germ cell tumor (Table I). Patients with osteogenie sarcoma, rhabdomyosarcoma, and Ewing sarcoma had a lower incidence (10 of 17), approximately 14%. None of the 39 patients with neuroblastoma had brain metastases. The median interval from the diagnosis of the primary tumor to the detection of lung and symptomatic brain metastases in 25 patients is summarized i n Table II. Pulmonary metastases were present in 22 of 25 patients (88%) at the time intracranial disease was detected. All
patients except one with rhabdomyosarcoma had either active or silent systemic disease at the time diagnosis of brain metastases was made. With the exception of patients with osteogenic sarcoma, the median interval between the diagnosis of the primary neoplasm and the diagnosis of pulmonary metastatic disease was only one month, reflecting the widespread dissemination of the disease at diagnosis. No patient had clinical symptoms of brain metastases as the initiaI manifestation of the primary malignancy. In patients with osteogenic sarcoma, pulmonary metastases tended to be diagnosed later. The median interval between the diagnosis of lung metastases and brain metastases in the whole group was 10 months (range 0 to 60 months). Neurologie symptoms referable to brain metastases were found in 25 of 31 patients (81%). The other 19% harbored subclinical metastatic lesions, which were discovered postmortem. The most common signs and symptoms of brain metastases were headache or signs of increased intracranial pressure (52%), seizures (36%), hemiparesis (36%), and altered mental status (16%). Seizures were more frequent in younger children and in patients with germ cell tumors. Seizures were the initial symptom in 50% of patienis younger than 15 years, but in only 18.5% of patients older than 15 years. Five of eight
560
Graus, Walker, and Allen
(62%) patients with germ cell tumors had seizures. Five patients (two with germ cell tumor, two with Ewing sarcoma, and one with rhabdomyosarcoma) had acute and catastrophic clinical signs and symptoms, including headache, lethargy, and hemiparesis. Three died Within a few days. In all five, autopsy Or CT scan revealed a hemorrhage into a metastasis. Solitary metastases were present in seven patients (six with sarcomas and one with germ cell tumor) in our autopsy series, three of whom had subclinical lesions. Hemorrhagic metastases were found in nine of the 18 brains studied (four patients with germ cell tumor, two with osteogenic sarcoma, two with Ewing sarcoma, and one with rhabdomyosarcoma). In all but two patients With hemorrhagic metastases, the p!ateiet count was >50,000/ mm 3. In two of 20 patients who had CT scans, the initial scan showed a hematoma without obvious radiologic evidence Of an underlying metastasis. CerebeHar metastases were present in seven of 31 (23%) patients. Of the 25 patients in whom the diagnosis of brain metastases was made during life, 21 underwent combinations of surgery or radiation therapy or both. The median survival of six patients with solitary brain metastases who had surgical excisions followed by radiotherapy was seven months. The usual radiation dose was 3000 rads administered to the whole brain over 10 elapsed days. One patient is still alive at 41 months, although he acquired a relapse at the same site. In three of these patients who had postmortem examinations, no microscopic evidence of tumor was found at the surgical site. Only one of the patients died of a recurrent brain tumor. Fifteen patients were treated with radiation but without surgery; the majority also received chemotherapy. The median survival in this group was four months, but death was caused by neurologic disease in only four (27%) patients. A complete response to radiotherapy was detected on CT by three months in four of five patients with germ cell tumors. However, only one of six patients with sarcoma responded in a similar fashion. Of the five patients who experienced a complete response to radiotherapy, none died of recurrent brain disease. Of the six nonresponders, three died of brain metastases. DISCUSSION A survey of brain metastases in autopsy cases of children with solid tumors from 1973 to 1982 revealed an incidence of 18 of 139 (12.9%) in children younger than 21 years and 10 of 102 ( 9 . ' ~ ) in children younger than i5 years. This represents an increase in the incidence of brain metastases at this institution over the past decade. Vannucci and Baten 2 found an incidence of 6% in an autopsy series of 217 children younger than 15
The Journal of Pediatrics October 1983
years of age at our institution between 1951 and 1972. In their Study, Wilms tumor was the most common cause of brain metastases, but contributed to none of the cases in Our study. This tumor represented 13%of all diagnoses in their autopsy study, but only 5% of diagnoses in our study of children younger than 15 years. The reduced frequency of Wilms tumor is most likely related to the introduction of effecti~,e chemotherapy that has drastically reduced the mortality and institutional referrals of this malignancy. In our study, the most common solid tumors causing brain metastases were osteogenic sarcoma, rhabdomyosarcoma, Ewing sarcoma, and germ cell tumors. Brain metastases most frequently arose in the context of extensive metastatic disease, especially involving the lungs. The increased incidence of brain metastases in our study most probably reflects our ability to prolong survival of patients with disseminated disease with more aggressive treatment. The absence of parenchymal brain metastases in patients with neuroblastoma most likely reflects the rarity of pulmonary metastases in this disease. The median interval from the onset of pulmonary metastases to the diagnosis of the brain metastases did not differ significantly among the various tumors (Table II). However, with the exception of osteogenic sarcoma, the median interval from diagnosis of the primary tumor to the detection of pulmonary metastases was one month, indicating that the cancer in these patients had an extremely aggressive course from the start. Pulmonary metastases usually preceded the recognition of brain metastases by 10 months. All but one patient had identifiable systemic disease at the time of diagnosis of brain metastases, and all but three patients had pulmonary metastases. The neurologic manifestations that led to the diagnosis of metastatic brain disease in our study differ from those found among adults in two ways; the higher incidence Of seizures and the abrupt onset of the symptoms.6 Seizures were seen most~bften in children younger than 15 years and in patients with germ cell tumors. There was no correlation between the presence of seizures and the demonstration of hemorrhagic metastases at autopsy or on CT scan. Schold et al. 7 also found a high incidence of seizures as the initial symptoms of brain metastases in a series of adult patients with germ cell tumors. Acute or catastrophic focal neurologic symptoms arose in.20% of patients in our series. A lower incidence (10%) of similar presentations has been reported from our institution in adults with primary brain metastases, predominantly from carcinomas? This clinical presentation usually signaled hemorrhage into tumor. Hemorrhagic metastases were observed at autopsy in 50% of our patients. This high incidence of hemorrhagic lesions arising in children with
Volume 103 Number 4
primarily metastatic sarcomas has not been previously reported, although apoplectic neurologic symptoms with or without hemorrhagic brain metastases have been observed occasionally in patients with brain metastases from sarcomas.3.8 The results of treatment in our study are similar to those reported in other series of children and of adult patients with brain metastases? ,~ The longer survival in patients who underwent surgery probably reflects the better clinical condition and prognosis of patients selected for surgery. The responses observed with radiation therapy are consistent with the known radiosensitivity of germ cell tumors and the relative radioresistance of sarcomas. Surgery would appear to be indicated, therefore, in patients with sarcomas who have single brain metastasis, provided they are good operative risks and have relatively long life expectancy. Higher doses of radiation therapy or the concomitant use of chemotherapy may be of benefit to those patients with radioresistant tumors who are not candidates for surgery. The importance of achieving a complete response to radiotherapy is highlighted by the fact that none of our five patients whose brain metastases responded to radiation died from recurrent brain disease. Deutsch et al., 9 in a review of brain metastases, recommend higher radiation dosages, comparable to those used to control the primary tumor.
Brain metastases in children
561
We thank Dr. Jerome Posner for critical review of the manuscript.
REFERENCES
l. Posner JB: Brain metastases: A clinician's view. In Weiss L, Gilbert HA, Posner JB, editors: Brain metastases. Boston, 1980, GK Hall, pp 2-30. 2. Vannucci RC, Baten M: Cerebral metastatic disease in childhood. Neurology 24:981, 1974. 3. Trigg ME, Glaubiger D, Nesbit ME: The frequency of isolated CNS involvement in Ewing's sarcoma. Cancer 49:2404, 1982. 4. Espana P, Chang P, Wiernik PH: Increased incidence of brain metastases in sarcoma patients. Cancer 45:377, 1980. 5. Cairncross JG, Kim JH, Posner JB: Radiation therapy for brain metastases. Ann Neurol 7:529, 1980. 6. Posner JB: Clinical manifestations of brain metastases. In Weiss L, Gilbert HA, Posner JB, editors: Brain metastases. Boston, 1980, GK Hall pp 189-208. 7. Schold CF, Vurgrin D, Golbey RB, et ah Central nervous system metastases from germ cell carcinoma of testis. Semin Oncol 6:102, 1979. 8. Gercovieh FG, Luna MA, Gottlieb JA: Increased incidence of cerebral metastases in sarcoma patients with prolonged survival from chemotherapy. Cancer 36:1843, 1975. 9. Deutsch M, AIbo V, Wollman MR: Radiotherapy for cerebral metastases in children. Int J Radiat Oncol Biol Phys 8:1141, 1982.