Primary pediatric brain tumors

Statistics of Taipei VGH, Taiwan (1975–2004)


  • Tai-Tong Wong M.D.,

    Corresponding author
    1. Department of Neurosurgery, Neurological Institute, Taipei Veterans General Hospital and National Yang Ming University, School of Medicine, Taiwan, Republic of China
    • Division of Pediatric Neurosurgery, Neurological Institute, Taipei Veterans General Hospital, VACRS, No. 210, Sec 2, Shih-Pai, Taipei, Taiwan, Republic of China, 11217
    Search for more papers by this author
    • Fax: (011) 886-2-28757587

  • Donald M. Ho M.D.,

    1. Department of Pathology and Laboratory Medicine, Taipei Veterans General Hospital and National Yang Ming University, School of Medicine, Taiwan, Republic of China
    Search for more papers by this author
  • Kai-Ping Chang M.D.,

    1. Department of Pediatrics, Taipei Veterans General Hospital and National Yang Ming University, School of Medicine, Taiwan, Republic of China
    Search for more papers by this author
  • Sang-Hue Yen M.D.,

    1. Cancer Center, Taipei Veterans General Hospital and National Yang Ming University, School of Medicine, Taiwan, Republic of China
    Search for more papers by this author
  • Wan-You Guo M.D., Ph.D.,

    1. Department of Radiology, Taipei Veterans General Hospital and National Yang Ming University, School of Medicine, Taiwan, Republic of China
    Search for more papers by this author
  • Feng-Chi Chang M.D.,

    1. Department of Radiology, Taipei Veterans General Hospital and National Yang Ming University, School of Medicine, Taiwan, Republic of China
    Search for more papers by this author
  • Muh-Lii Liang M.D.,

    1. Department of Neurosurgery, Neurological Institute, Taipei Veterans General Hospital and National Yang Ming University, School of Medicine, Taiwan, Republic of China
    Search for more papers by this author
  • Hung-Chi Pan M.D.,

    1. Department of Neurosurgery, Neurological Institute, Taipei Veterans General Hospital and National Yang Ming University, School of Medicine, Taiwan, Republic of China
    Search for more papers by this author
  • Wen-Yuh Chung M.D.

    1. Department of Neurosurgery, Neurological Institute, Taipei Veterans General Hospital and National Yang Ming University, School of Medicine, Taiwan, Republic of China
    Search for more papers by this author



The purpose of the current study was to investigate a hospital series of 986 cases of primary pediatric brain tumors in Taiwan.


The authors reviewed the database of primary pediatric brain tumors in patients < 18 years of age collected in Taipei Veterans General Hospital (Taipei VGH) from 1975 to May 2004. Age and gender distribution, location, and classification of brain tumors were analyzed. Intracranial tumors with diagnostic imaging were included. Nontumoral lesions, cysts, and vascular malformations were excluded.


The mean age of these 986 patients was 7.8 years, and the male to female ratio was 1.4:1. Supratentorial (including pineal–quadrigeminal) located tumors (58.3%) was predominant to infratentorial tumors (41.1%). In these series, 886 patients had either histologic diagnosis (842 patients) or clinical diagnosis (44 patients). The most common 5 categories of tumors were astrocytic tumors (31.1%), germ cell tumors (14.0%), medulloblastomas (13.3%), craniopharyngiomas (8.3%), and ependymal tumors (5.8%). Atypical teratoid/rhabdoid tumors (AT/RTs), a rare but highly malignant tumor, were 2.1%. The high incidence of primary intracranial germ cell tumors correlated with reported series from Japan and Korea. For the remaining 100 patients without diagnostic classifications, the majority were most likely astrocytic tumors in brain stem.


The authors analyzed a large hospital series of primary brain tumors in children. Both histologically verified and unverified tumors were recruited to avoid selective bias. Although it was not a study of a population-based brain tumor registry, it could still be representative of primary pediatric brain tumors in Taiwan. Cancer 2005. © 2005 American Cancer Society.

In Taiwan, the earliest reported series of pediatric brain tumor was 126 cases at the National Defense Medical Center (NDMC) in 1977 and 87 cases at the National Taiwan University Hospital (NTUH) in 1980.1, 2 Taipei VGH later reported a series of 171 CT scan detected childhood brain tumors in 1987.3 Since then, large series of pediatric brain tumors has not been studied. From 1975 to May 2004, we collected 986 cases of primary brain tumors in children younger than 18 years of age in Taipei VGH. Demographic data (age distribution, gender ratio, location, and classification of tumors) of these patients were reviewed and analyzed. We were interested in classifying either histologically or clinically the tumors that were detected by neuroimaging studies.


In this study, we reviewed the database of 986 cases of primary pediatric brain tumors of younger than 18 years of age collected in Taipei VGH from 1975 to May 2004. Primary intracranial tumors that were diagnosed by CT scan or MRI were included. Nontumoral lesions, cysts, and vascular malformations were excluded. Among them, 849 (86.1%) patients received surgeries for tumors, and 842 (85.4%) had histologic diagnosis. Of the 144 patients without histologic diagnosis, 44 had clinical diagnosis. Eighty-four patients were referred after initial tumor surgeries in other hospitals, and 46 of them received further tumor surgeries at Taipei VGH for residue or recurrent tumors. Twenty-six patients died within 1 month after tumor surgeries with a surgical mortality rate of 3.2%. Demographic data (age, gender, location, and classification of tumors) of these patients were analyzed. The distribution of tumors was defined by imaging studies and/or operative findings. Tumor locations were counted as supratentorial, infratentorial, both supratentorial and infratentorial, and also their related neuroanatomic locations. Tumors with histologic diagnosis were classified according to the World Heath Organization 2000 classification as closely as possible.4 For tumors that were operated upon in Taipei VGH, the histologic diagnosis was reviewed and reconfirmed by one senior neuropathologist. For tumors without tissue proof, the diagnoses and classifications were made from related specific clinical features, image findings, and locations of tumors. For germ cell tumors (GCTs) without histologic verification, the diagnosis and classification were achieved through image findings, locations, tumor markers and response to radiotherapy and/or chemotherapy. Brain tumors of phacomatoses without surgical resection were classified according to specific locations, and entities of central nervous system (CNS) tumors occurred in different syndromes. For hypothalamic hamartomas, the classification could be made simply through magnetic resonance imaging (MRI) findings. We were also interested in studying data of several specific entities: GCTs, atypical teratoid/rhabdoid tumors (AT/RTs), phacomatoses, and radiation-induced second tumors.


Age Distribution and Gender Ratio

Of these 986 patients, the mean age was 7.8 years (range from the first day of life to 18 yrs) with peak frequency at 2, 6, 10, and 13 years. The tumors were diagnosed within 1-year-olds in 61 (6.2%) patients. There were 579 (58.7%) boys and 407 (41.3%) girls, with male to female ratio of 1.4:1. Except the age groups of 2 and 8 years, there was male predominance in all other age groups (Fig. 1, Table 1).

Figure 1.

Chart graphically depicts the age distribution of 986 patients.

Table 1. Age and Sex Distribution of Pediatric Brain Tumors (N = 986)
Age yrsMaleFemaleTotal
  1. Average age: 7.8 yrs Male to female ratio: 1.4


Age distribution of different types of tumors

According to the mean age at diagnosis, immature teratomas in ventricles (2.1 yrs), choroids plexus tumors (2.2 yrs), supratentorial primitive neuroectodermal tumors (PNET) (3.5 yrs), AT/RTs (4.2 yrs), pineoblastomas (4.8 yrs), and ependymal tumors (6.0 yrs) occurred in younger age groups. Medulloblastomas (7.0 yrs), astrocytic tumors (7.5 yrs), craniopharyngiomas (7.8 yrs), and oligodendroglioma (7.9 yrs) were common tumors in children 7–8 years old. Neuronal and mixed neuronal glial tumors (8.6 yrs), germ cell tumors (10.6 yrs), tumor of meninges (10.7 yrs), and pituitary tumors (14.3 yrs) were usually diagnosed in older children. The age distribution and gender ratio of different types of tumors are listed in Tables 2–3.

Table 2. Classification and Age and Gender Distributions of Primary Pediatric Brain Tumors in Taipei VGH
Types of tumorsNo. of patientsCases of histology diagnosisCases of clinical diagnosisMean age (range)Gender ratio (M/F)Percentage histologic diagnosisPercentage N = 986
  • a

    PXA: pleomorphic xanthoastrocytoma.

  • b

    Mixed GCTs: G + T(MT/IT) represents mixed GCTs with components of germinoma (G), mature teratoma (MT) or immature teratoma (IT); G + IT + YST represents mixed GCTs with components of germinoma (G), immature teratoma (IT), and yolk sac tumor (YST); G + AFP/HCG represents mixed GCTs with components germinoma with elevated serum titer of AFP with/without elevated serum titer of HCG; T(MT/IT)+YST represents GCTs with component of teratoma (mature teratoma or immature teratoma) and yolk sac tumor; MT + unclassified GCT represents mixed GCT with components of mature teratoma and unclassified GCT.

  • c

    MPNST: malignant Triton tumor arising from trigeminal nerve.

  • d

    Tumors without histologic or clinical diagnosis included 73 tumors in brainstem, 5 tumors in cerebral hemisphere, 5 tumors in thalamus, 1 tumor in hypothalamus, 5 tumors in sellar-suprasellar region, 5 tumors in pineal region, 2 tumors in basal ganglia, and 4 tumors in cerebellar vermis and 4th ventricle. The majority of these tumors may be astrocytic tumors.

Gliomas40340127.0 yrs (3 days–17.8 yrs)1.199.540.9
Astrocytic tumors30730527.5 yrs (3 days–17.8 yrs)1.099.331.1
 Pilocytic astrocytoma13113107.0 yrs (4.4 mos–16.4 yrs)1.010013.3
 Astrocytoma747407.5 yrs (4.4 mos–15.7 yrs)1.01007.5
 Anaplastic astrocytoma454508.6 yrs (3 days–16.6 yrs)1.31004.6
 Glioblastoma393907.0 yrs (2.8 mos–17.8 yrs)0.71004.0
 Giant cell astrocytoma131128.0 yrs (1–11.6 yrs)3.384.61.3
 PXAa4407.8 yrs (7.9 mos–13 yrs)4/01000.4
 Primary leptomeningeal gliomatosis11012.5 yrs0/11000.1
Oligodendroglial tumors121207.9 yrs (1.1 mos–15.1 yrs)1.41001.2
Ependymal tumors565606.0 yrs (3.8–17.4 mos)1.31005.7
 Ependymoma252505.8 yrs (10.1 mos–16.8 yrs)1.3 2.5
 Anaplastic ependymoma313106.0 yrs (3.8 mos–17.4 yrs)1.4 3.1
Choroid plexus tumors191902.2 yrs (0.5 mos–12.4 yrs)1.41001.9
 Choroid plexus papilloma171702.4 yrs (0.5 mos–12.4 yrs)1.1 1.7
 Choroid plexus carcinoma2201.2 yrs (9.0 mos–1.6 yrs)2/0 0.2
Glial tumor of uncertain origin1109.0 yrs0/11000.1
Mixed gliomas7706.1 yrs (5.6 mos–12.7 yrs)2.51000.7
Embryonal tumors16516506.4 yrs (1 day–18 yrs)1.710016.7
 Medulloblastomas13113107.0 yrs (2.5 mos–18 yrs)1.910013.2
 PNET/supratentorial121203.5 yrs (1 day–13 yrs)0.71001.2
 Ependymoblastoma1101.7 yrs1/01000.1
 AT/RTs212104.2 yrs (0.5 mos–10.7 yrs)1.61002.1
Germ cell tumors1381053310.6 yrs (3 days–18 yrs)3.875.514.0
 Germinoma75522311.9 yrs (1.3–18 yrs)
 Teratoma, mature8806.9 yrs (8.5 mos–15.3 yrs)8/01000.8
 Teratoma, immature141406.8 yrs (7 days–14.4 yrs)131001.4
 YST, pure1010010.0 yrs (2.6–14.5 yrs)1.01001.0
 Mixed GCTsb1919010.1 yrs (1.6–15.4 yrs)2.81001.9
  G + T (MT2/IT3)550    
  G + IT + YST440    
  G + AFP/HCG220    
 MT + unclassified GCT110    
 Unclassified GCT2207.2 yrs (5.8–8.7 yrs)2/01000.2
 Diagnosed by markers1001011 yrs (3 days–16.5 yrs2.301.0
Craniopharyngioma828117.8 yrs (7.3 mos–17.7 yrs)1.598.88.3
Neuronal and mixed neuronal-glial tumors212108.6 yrs (1.3–17.8 yrs)0.81002.1
 Ganglioglioma121208.5 yrs (1.4–17.8 yrs)0.71001.2
 Anaplastic ganglioglioma2205.2 yrs (2.6–7.8 yrs)11000.2
 DNT55010.4 yrs (6.6–14.6 yrs)0.71000.5
 Desmoplastic infantile       
  astrocytoma/ganglioglioma1105.2 yrs1/01000.1
 Central neurocytoma1108.6 yrs0/11000.1
Hamartoma9454.5 yrs (8.8 mos–12.8 yrs)
Pineal parenchymal tumors9905.2 yrs (1.6–13 yrs)0.81000.9
 Pineoblastoma8804.8 yrs (1.6–13 yrs)1.01000.8
 Pineocytoma1108.7 yrs0/11000.1
Tumors of meninges1919010.9 yrs (11.9 mos–17.4 yrs)0.71001.9
 Meningiomas1313010.6 yrs (11.9 mos–17.4 yrs)0.61001.3
 Dural sarcoma55010.6 yrs (1.6–14.6 yrs)0.71000.5
 Chondroma11015.3 yrs1/01000.1
Tumors of peripheral nerve1414012.7 yrs (5.4–17.9 yrs)1.0501.4
 Vestibular schwannoma1313012.8 yrs (5.4–17.9 yrs)
 MPNSTc11011.5 yrs0/11000.1
Pituitary adenomas1010014.3 yrs (12.4–17.7 yrs)2.31001.0
Primary melanocytic lesions2207.7 yrs (1.5–13.9 yrs)0/21000.2
 Malignant melanoma       
Tumors of uncertain histogenesis20010.8 yrs (6.7–14.8 yrs)1/11000.2
Chordoma5507.6 yrs (4.4–13 yrs)1.51000.5
Tumor without histology clinical diagnosisd100007.8 yrs (0.9 mos–17.9 yrs)1.5010.1
Table 3. Classification, Locations, Age Distribution, Gender Ratio, and Percentage of Specific Types of Primary Intracranial Germ Cell Tumors (N = 138/986)
Types of tumorsNo. of casesCases of histologic diagnosisMean age (range)Sex ratio (M/F)Percentage histologic diagnosisPercentage N = 986
Total13810510.6 yrs (3 days–18 yrs)3.8/17614.0
Germinoma755211.9 yrs (1.3–18 yrs)469.37.6
 Suprasellar191511.8 yrs (5.6–16.5 yrs)0.978.912.9
 Pineal291911.7 yrs (1.3–17.7 yrs)8.765.52.9
 Suprasellar + pineal7614.5 yrs (8.7–18 yrs)2.585.70.7
 Basal ganglia181011.1 yrs (7.3–14.4 yrs)18/055.61.8
 Cerebral lobes2214.4 yrs (14–14.7 yrs)2/01000.2
Teratoma, mature886.9 yrs (8.5 mos–15.3 yrs)8/01000.8
 Pineal777.8 yrs (1.9–15.3 yrs)7/01000.7
 Deep hemisphere118.5 mos.1/01000.1
Teratoma, immature14146.8 yrs (7 days–14.4 yrs)131001.4
 Suprasellar116.0 yrs0/11000.1
 Pineal4412.0 yrs (7.2–14.3 yrs)41000.4
 Basal ganglia2213.1 yrs (11.7–14.5 yrs)2/01000.2
 Ventricle772.1 yrs (7 days–13.2 yrs)7/01000.7
YST, pure101010.0 yrs (2.6–14.5 yrs)11001.0
 Suprasellar5511.3 yrs (7.7–14.5 yrs)1/41000.5
 Pineal4410.3 yrs (8.3–12.5 yrs)31000.4
 Cerebellum112.6 yrs1/01000.1
Mixed GCT191910.1 yrs (1.6–15.4 yrs)2.81001.9
 Suprasellar668.9 yrs (1.6–13.1 yrs)1/11000.6
 Pineal8810.5 yrs (3–15.4 yrs)3/11000.8
 Suprasellar + pineal1115.0 yrs1/01000.1
 Basal ganglia4410.1 yrs (8.8–11.5 yrs)4/01000.4
Unclassified GCT227.2 yrs (5.8–8.7 yrs)2/01000.2
 Suprasellar118.7 yrs1/01000.1
 Pineal115.8 yrs1/01000.1
Diagnosed by markers10011 yrs (3 days–16.5 yrs)2.301.0
 Suprasellar4011.7 yrs (7.1–16.5 yrs)100.4
 Pineal509.3 yrs (3 days–13.5 yrs)400.5
 Suprasellar + pineal1016.4 yrs1/000.1

Gender ratio (male to female ratio) of different types of tumors

Of the five most common types of tumors in children, the gender ratios of astrocytic tumors, ependymal tumors, craniopharyngiomas, medulloblastomas, and germ cell tumors were 1.0, 1.4, 1.5, 1.9, and 3.8, respectively. Except astrocytic tumors, all other four types of tumors were predominantly in males. This tendency of male predominance (above the average ratio of 1.4) was clear in medulloblastoma and germ cell tumors. For germ cell tumors, only yolk sac tumors or tumors in the suprasellar region were equal in gender ratio or female predominant (Tables 2–3).

Classification of Tumors

With reference to the World Heath Organization 2000 classification,4 we classified and counted relative incidence of different types of brain tumors with either histologic diagnosis or clinical diagnosis in the whole series. In this series of 986 patients, 842 (85.4%) patients had histologic diagnoses, and the other 44 patients had clinical diagnoses. The common entities of tumors were astrocytic tumors (31.1%), germ cell tumors (14.0%), medulloblastomas (13.2%), craniopharyngiomas (8.3%), and ependymal tumors (5.7%). Relative rare entities of tumors included neuronal and mixed neuronal glial tumors (2.1%), AT/RTs (2.1%), tumors of meninges (1.9%), tumors of choroids plexus (1.9%), tumors of cranial nerves (1.4%), oligodendroglioma (1.2%), supratentorial PNET (1.2%), pituitary adenomas (1.0%), pineal parenchymal tumors (0.9%), hamartomas (0.9%), chordomas (0.5%), hemangioblastomas (0.3%), primary melanocytic tumors (0.2%), ependymoblastomas (0.1%), and astroblastoma (0.1%). For remaining 100 cases with tumors having neither histologic nor clinical classification, there were 73 brainstem tumors. The majority of these unclassified tumors in brainstem, cerebral hemisphere, optic apparatus, and thalamus were most likely astrocytic tumors (Table 2). Classified gliomas constituted 40.8% of tumors in this series. If most of the tumors without histologic diagnosis or clinical diagnosis were astrocytic tumors or other gliomas, the proportion of gliomas would be up to 50% of primary brain tumors in children (Tables 2–3).

Specific entities of tumors

Germ cell tumors (GCTs).

In this series, a total of 138 GCTs were diagnosed at an average age of 10.7 years with a male to female ratio of 3.8:1. One hundred-five patients had histologic verification, and the remaining 33 were diagnosed clinically. Of these 33 histologically unverified GCTs, 23 tumors were defined as germinomas, and 10 tumors were defined as nongerminomatous malignant germ cell tumors (NG-MGCTs). The diagnosis was made according to clinical features, neuroimaging, serum tumor markers (alpha-fetaprotein [AFP], beta human chorionic gonadotropin [beta-hCG]), and response to radiation therapy and/or chemotherapy. The diagnosed entities of these tumors included germinomas (54.4%), mature and immature teratomas (15.9%), pure yolk sac tumors (7.3%), mixed GCTs (13.8%), unclassified GCTs (1.5%), and tumors diagnosed by tumor markers (7.3%). Among 19 cases of mixed GCTs, 16 cases had component of teratoma, 11 cases had component of germinoma, 11 cases had component of yolk sac tumor, and 1 case had component of unclassified GCT. In two cases with component of germinoma, the diagnosis of mixed GCTs relied on elevated serum titer of AFP. Five of the 19 mixed GCTs were initially teratomas (3 immature teratomas and 2 mature teratomas) in their first tumor resection. The pathologies changed to other types of GCTs on recurrence.

Common locations of GCTs in these series were pineal (58, 42%), suprasellar–pineal (9, 6.5%), sellar–anterior third ventricle–hypothalamic region (36, 26.1%), cerebral hemisphere including 25 basal ganglia tumors and 2 cerebral lobe tumors (27, 19.6%), and ventricle (7, 5.1%). The location, age distribution, gender ratio, and percentage to the whole series of different types of GCTs are shown in Tables 3–8. In these 138 GCTs, 111 tumors had serum tumor markers (AFP, beta-hCG) studies at diagnosis. Eighty-two patients had both histologic diagnosis and serum tumor marker studies. Correlation with different histologic types of GCTs and serum titers of tumor markers is shown in Table 4.

Table 4. Correlation of Serum Tumor Markers (AFP, HCG) and Different Types of Intracranial GCTs with Histologic Diagnosis
Histologic typesNo. of patientsSerum AFP ng/mLSerum HCG mIU/mL
  1. AFP: alpha-fetoprotein; HCG: human chorionic gonadotropin; ND: not done; N: normal (within reference range); E: elevated (higher than reference range)

Teratoma, mature7250  070  
Teratoma, immature900921.511626127.6 
YST, pure90091293265015328.8804
     13555427   187 
Mixed GCT17131321.514637710.5175
     24.9365   30.82802
     38366   61.43312
     49.41874   156 
Table 5. Phacomatoses and Brain Tumors
PhacomatosesNo. of casesIntracranial tumors
  • a

    Three sellar region tumors in NFI patients without histologic diagnosis were classified as optic gliomas.

  • b

    Bilateral tumors in internal meatus of 7 NFII patients without histologic diagnosis were classified as vestibular schwannomas.

  • c

    Among 13 patients with SEGA, 12 patients presented clinical features of tuberous sclerosis, and 2 of these SEGAs had no histologic diagnosis.

NFI10Optic gliomas 5
   pilocytic astrocytoma 2
   no histologic diagnosis 3a
  Pilocytic astrocytoma 3
   thalamus, cerebellar vermis, pons
  Astrocytoma 1
  Germinoma 1
   Basal ganglia
NFII13Bilateral vestibular schwannomas
   histologic diagnosis 6
   no histologic diagnosis 7b
  Meningiomas 6
   histologic diagnosis 6
Tuberous sclerosis12Subependymal giant cell astrocytomas 10
   histologic diagnosis 10
   no histologic diagnosis 2c
Basal cell nevus syndrome2Medulloblastomas 2
   Cerebellar vermis 1
   Cerebellar hemisphere 1
Neurocutaneous melanosis2Malignant melanomas 2
Table 6. Distribution of Pediatric Brain Tumors in Taiwan
 No. supratentorial tumors (%)No. infratentorial tumors (%)No. supra- & infratentorial tumors (%)
  • a

    If 7 arteriovenous malformations and 4 arachnoid cysts were excluded, the ratio of supratentorial and infratentorial tumors would be 49.6–50.4%.

  • b

    If 22 histologically unverified brain stem gliomas were included, the ratio of supratentorial and infratentorial tumors would be 48.6–51.4%.

  • c

    Data from study of 171 cases of pediatric brain tumors of < 15 yrs of age after the CT brain scan was used as the diagnostic tool in Taipei VGH.

  • d

    Data from study of 501 cases of pediatric brain tumors of < 15 yrs of age with diagnostic images.

  • e

    Data from study of 935 cases of pediatric brain tumors of < 15 yrs of age with diagnostic images.

  • f

    Data from study of 986 cases of pediatric brain tumors of < 18 yrs of age with diagnostic images.

Shih et al. NDMC, Taipei, 1977aN = 12667 (53.2)59 (46.8) 
Ho et al. NTUH, Taipei, 1969–1978bN = 8753 (60.9)34 (39.1) 
Wong et al. Taipei VGH, 1978–1985 Age ≤ 15 yrscN = 17199 (57.9)72 (42.1) 
Wong et al. Taipei VGH, 1978–1994 Age ≤ 15 yrsdN = 501291 (58.1)207 (41.3)3 (0.6)
Wong et al. Taipei VGH, 1975–2004 Age ≤ 15 yrseN = 935540 (57.8)390 (41.7)5 (0.5)
Wong et al. Taipei VGH, 1975–2004 Age ≤ 18 yrsfN = 986575 (58.3)405 (41.1)6 (0.6)
Table 7. Neuroanatomic Locations of Pediatric Brain Tumors
  • a

    Not otherwise specified in location.

Cerebral hemisphere19.7%
 Deep hemisphere4.5% 
Sellar-anterior III ventricle-hypothalamus 19.2%
Thalamus 4.7%
Pineal 7.5%
Suprasellar-pineal 1.0%
Cerebellum 24.2%
Brainstem 12.2%
Ventricle 8.3%
Extraparenchymatous 1.9%
 Bil, vestibular schwannoma1.3% 
Dura 1.5%
Table 8. Types of Tumors in Specific Neuroanatomic Locations
LocationsNo. of casesCases of tissue diagnosisCases of clinical diagnosisTypes and percentage (%) of tumors
  1. Ast: astrocytic tumor; AT/RT: atypical teratoid/rhabdoid tumor; Chon: chondroma; CPT: choroid plexus tumor; Cran: craniopharyngioma; D.I. Gang: desmoplastic infantile gangliglioma; EB: ependymoblastoma; Epen: ependymal tumor; Gang: ganglioglioma; GCT: Germ cell tumor; Ham: hamartoma; Hem: hemangioblastoma; MDB: medulloblastoma; Men: meningioma; Olig: oligodendroglial tumor; PB: pineoblastoma; PC: pineocytoma; Pit: pituitary adenoma; PNET: primitive neuroepithelial tumor; Sar: sarcoma; SEGA: subependymal giant cell astrocytoma.

  2. aNOS: not otherwise specified in location.

Cerebral hemisphere1941797AstGCTEpenGangOlig
Sellar-anterior IIIV-Hypothalamus18916617CranAstGCTPitHam
Pineal & Suprasellar-pineal745415GCTPBAstPC 
    MDBAstAT/RTGangD.I. Gang
 Brachium44025 50 25
 Pons9427024.52.1 1.0 
Meninges15150MenSarChon Ast
    53.333.36.7 6.7
Atypical rhabdoid/teratoid tumors (AT/RTs).

Among 165 embryonal tumors, there were 131 medulloblastomas and 21 AT/RTs. AT/RTs constituting 2.1% of primary brain tumors in this series. The mean age of diagnosis was 4.2 years (range, 0.5 mos to 10.7 yrs) with a male to female ratio of 1.6:1. These AT/RTs of the brain were located predominately in the cerebellum (71.4%). Other locations included cerebral hemisphere, pineal–quadrigeminal region, and lateral ventricle. Cerebellar AR/RT mimics medulloblastoma. The ratio of cerebellar medulloblastoma to cerebellar AT/RTs was 131/15 (8.7:1) (Tables 2, 8).


In this series of primary pediatric brain tumors, 39 (4.0%) children had phacomatoses. The average age of diagnosis was 9.6 years (range, 1–18 yrs). Male to female ratio was 1.5:1. Ten children had NFI. Five of them had optic gliomas, with histologic diagnosis of pilocytic astrocytomas in 2 patients. Four patients had astrocytic tumors including 1 in pons (pilocytic astrocytoma), 1 in cerebellar vermis (pilocytic astrocytoma), 1 in thalamus (pilocytic astrocytoma), and 1 in hypothalamus (astrocytoma). The remaining 1 patient had a germinoma at the basal ganglia.5 Thirteen children had NFII. All of them had bilateral vestibular schwannomas. Cranial meningiomas were also recorded in 6 patients. Twelve patients had tuberous sclerosis with subependymal giant cell astrcytomas (SEGAs). Ten of them had histologic diagnosis. Two patients had neurocutaneous melanosis and histologic proof of intracranial malignant melanomas. Basal cell nevus syndrome (Gorlin syndrome) was observed in 2 children with medulloblastomas (Table 5).

Radiation-induced tumors.

We observed 7 patients who developed second tumors after radiation therapy for primary brain tumors. Their primary tumors were cerebellar medulloblastomas in 4 patients, sellar craniopharyngioma in 1 patient, pineal germinoma in 1 patient, and posterior temporal residue pilocytic astrocytoma in 1 patient. The durations of time after resection and radiation therapy to diagnosis of second tumors were from 2.8–21.1 years. These radiation-induced tumors included basal cell epithelioma of scalp and meningioma of cerebral convexity developed in a girl at 16.3 years after radiation for her medulloblastoma at 2 years old. Multiple supratentorial meningiomas were found at 21.1 years after craniospinal irradiation (CSI) for medulloblastoma at 5.5 years of age. An anaplastic oligodendroglioma with tumoral hemorrhage was found 20.5 years after CSI for medulloblastoma in a girl aged 9 years. A left frontal pole meningioma was diagnosed 20 years in a girl after CSI for medulloblastoma at 5.1 years of age. Atypical meningioma of the tentorium was found 13.1 years after local radiation of a residue pilocytic astrocytoma of the posterior temporal region. A glioblastoma multiforme of pons was diagnosed 8 years after radiation therapy for a craniopharyngioma in a girl 4 years old. Angiosarcoma in soft tissue of occipital region occurred at 2.8 years after radiotherapy for a pineal germinoma. These tumors were excluded from the study series of primary brain tumors.

Locations of tumors

Including 91 cases of pineal–quadrigeminal tumors, there were 575 (58.3%) supratentorial tumors and 405 (41.1%) infratentorial tumors. Another 6 (0.6%) patients had both supra- and infratentorial tumors and all of them had traits of neurofibromatosis (Table 6). Infratentorial tumors predominated only in the groups aged 5, 6, and 8 years. There was almost equal distribution of supratentorial and infratentorial tumors in the age groups of 4 and 18 years. Supratentorial tumors predominated in all of the other year groups (Fig. 2). The neuroanatomic distributions of tumors were cerebellum 24.2% (239), cerebral hemisphere 19.7% (194), sellar-suprasellar–anterior third ventricular–hypothalamus region 19.2% (189), brain stem 12.2% (120), ventricles 8.3% (82), single pineal region tumors 7.5% (74), tumors of both suprasellar and pineal regions 1.0% (10), thalamus 4.7% (46), dura 1.5% (15), and bilateral vestibular schwannoma 1.3% (13) (Table 7).

Figure 2.

Chart shows the distribution of supratentorial (including pineal–quadrigeminal) versus infratentorial tumors.

Types of tumors in specific neruoanatomic locations.

Common tumors in cerebral hemisphere were astrocytic tumors (49.0%), GCTs (13.9%), ependymal tumors (10.3%), gangliogliomas (7.2%), and oligodendroglioma (6.2%). In cerebral hemisphere, the majority of GCTs (23/25 patients) originated in the deep hemisphere—the basal ganglia region. Common tumors of sellar–anterior III ventricular–hypothalamic region were craniopharyngiomas (43.4%), astrocytic tumors (21.2%), GCTs (19.1%), pituitary adenomas (5.3%), and hamartomas (4.2%). For gliomas in this location, most were optic pathway gliomas, and 3 of 37 patients had no histologic confirmation. In suprasellar region, 28 of 36 GCTs had histologic diagnosis. Types of GCTs in this area were germinomas (52.8%), immature teratoma (2.8%), mixed GCTs (16.7%), pure yolk sac tumors (13.9%), unclassified GCT (2.8%), and tumors diagnosed by tumor markers (11.1%). Two types of tumors occurred in the hypothalamus, the astrocytic tumors and the hamartomas. For pineal tumors, there were 74 patients with tumors in the pineal region alone. Another 10 patients had both suprasellar and pineal tumors. The average age at diagnosis of these tumors was 10.3 years with a male to female ratio of 5.5:1. Tumors in the pineal or suprasellar–pineal regions were GCTs (79.8%), pineoblastomas (9.5%), astrocytic tumors (2.4%), and pineocytoma (1.2%). Common tumors in thalamus region were astrocytic tumors (71.7%) and PNETs (15.2%). Cerebellar tumors were mainly medulloblastomas (54.8%), astrocytic tumors (33.5%), and AR/RTs (7.1%). In cerebellar vermis, 65.3% tumors were medulloblastoma. In cerebellar hemisphere, 63.5% were astrocytic tumors. For brain stem tumors, among 120 patients of primary brain stem tumors, only 46 (38.3%) had histologic diagnosis, and the majority was astrocytic tumors. These tumors were distributed at the midbrain (7), pons (94), and the medulla and cervicomedullary junction (18). The distribution of 46 histologically proved tumors in brainstem were 41 astrocytic tumors (midbrain 3, pons 23, medullar/cervicomedullary 15), 3 gangliogliomas (pons 2, medulla 1), 1 ependymoblastomas (pons 1), and 1 hemangioblastoms (pons 1). These histologic findings demonstrated that the majority of brainstem tumors (89.1% in this series) were actually brainstem astrocytic tumors. Ventricular tumors were mainly posterior fossa ependymomas (43.9%), choroids plexus tumors (23.2%), SEGAs (14.6%), GCTs (8.5%), and meningiomas (4.9%). The two major types of tumors of meninges were meningiomas (56.3%) and meningeal sarcomas (31.3%) (Table 8).

Location distributions of specific primary brain tumors.

Astrocytic tumors and gangliogliomas distribute both supratentorially and infratentorially in brain. Other types of primary brain tumors in children tend to arise in specific neuroanatomic locations. In this series, the most common distributions of astrocytic tumors were cerebral hemisphere (30.9%), cerebellum (26.1%), brain stem (13.4%), optic appraratus-sellar–hypothalamus region (13.0%), thalamus (10.7%), and SEGAs in the lateral ventricle (4.2%). Ependymal tumors distributed in cerebral hemisphere (35.7%) and in fourth ventricle of the posterior fossa (64.3%). All oligodendrogliomas and mixed gliomas originated in cerebral hemisphere. The 3 most common locations for GCTs were pineal and suprasellar–pineal region (48.6%), sellar–anterior third ventricle–hypothalamic region (26.1%), and basal ganglia and cerebral lobes (19.6%). All medulloblastomas occurred within cerebellum, and the majority of them in cerebellar vermis (86.3%). AT/RTs, which may mimic medulloblastomas, in cerebellum (71.4%), but they were also distributed in cerebral hemisphere, ventricle, and pineal–quadrigeminal region. Supratentorial PNETs were found mainly in thalamus (58.3%), cerebral hemisphere (16.7%), and sellar region (16.7%). All craniopharyngioma in this series were sellar or suprasellar tumors with various extensions and directions of extension. Neuronal and mixed neuronal glial tumors were mainly in cerebral hemisphere (66.7%), cerebellum (14.3%), brain stem (14.3%), and hypothalamus (4.8%). Meninigiomas were tumors arising from dural matter (64.3%), but there were also tumors in ventricle or cerebral hemisphere. Meningeal sarcomas all originated from dura of anterior temporal region (Table 9).

Table 9. Location Distribution of Specific Primary Brain Tumors in Children
Types of tumorsNo. of casesCases with histologic diagnosisNeuroanatomic location and percentage of tumors
Cerebral HemSellar-anterior III V-hypothalamusPineal and suprasellar-pinealVentricleCerebellumBrain stemThalamusMeninges
  • Ast: astrocytic tumor, AT/RT: atypical teratoid/rhabdoid tumor; Cran: crnaiopharyngioma; Epen: ependymal tumor; GCT: germ cell tumor; MDB: medulloblastoma; M. glioma: mixed glioma; NMNGT: neuronal and mixed neuronal glial tumor, Men: meningioma; M. Sa: meningeal sarcoma; Olig: oligodendroglial tumor; PNET: primitive neuroepithelial tumor.

  • a

    4th ventricular ependymoma.

  • b

    cerebellar vermis 86.3%, cerebellar hemisphere 12.2%, cerebellar location undetermined 1.5%.

  • c

    cerebellar vermis 11 (52.4%), cerebellar hemisphere 4 (19.1%), brachium cerebellum 2 (9.5%).

Epen565635.7  64.3a    
M. glioma77100       
MDB131131    100b   
AT/RT21219.5 4.84.881.0c   
PNET121216.716.7 8.358.3   
Cran8281 100      
NMNGT212166.7   9.514.34.8 
Men14147.1  28.6   64.3
M. Sa55       100


In this study, specific tumor data that highlight the character of this series were 1) average age of diagnosis = 7.8 years; 2) male to female ratio of 1.4/1; 3) high incidence of germ cell tumors (14%); 4) AT/RTs representing 2.1% of tumors in this series; 5) Thirty-nine (3.9%) patients with traits of phacomatoses; 6) seven patients who developed second tumors after radiotherapy for primary tumors; 7) supratentorial tumors (58.3%) predominated over infratentorial tumors (41.8%); 8) neuroanatomical locations of tumors common in cerebellum, cerebral hemisphere, sellar–anterior third ventricular–hypothalamic region and brain stem; 9) tumors in pineal and basal ganglia regions mostly germ cell tumors.

Increased incidence of brain tumors in children has been observed and reported in the West. Also reported is an increased incidence of astrocytoma and medulloblastoma and/or primitive neuroectodermal tumor (M/PNET).6–11 In Japan, a study of incidence of childhood brain tumors in Hokkaido Prefecture showed increased incidence.12 According to the register of the childhood brain tumor (age < 18 yrs) from 17 representative hospitals by the Childhood Cancer Foundation, R.O.C. Taiwan from 1995 to 2002, the annual registered cases varied from 116 to 144 with annual incidence rate of 1.84, 2.02, 2.03, 2.4, 2.09, 2.25, 2.03, 2.23/100,000, respectively. There was no evidence of trend increase. This collection represented the majority of childhood brain tumors diagnosed in Taiwan but was still not a national registration.

We compared the reported data of Childhood Brain Tumor Consortium (CBTC), hospital series of University Hospital of Munster (UHM), the collection of Brain Tumor Registry of Japan (BTRJ), the hospital series of Seoul National University Hospital (SNUH) of Korea, and our series.13–16 The CBTC collected data on 3291 children with CNS brain tumors. These patients received first tumor surgery at one of the CBTC centers before January 1, 1979 before age 21 years. The series of UHM consisted of 319 cases of brain tumors of up to 17 years of age (age 0–17 yrs) with biopsy samples. Patient data were collected over a 17-year period (1984–2000) by the Institute of Neuropathology Munster from 7 regional neurosurgical departments. In BTRJ (1969–1996), there were 4280 cases of childhood brain tumors under the age of 15 years (age 0–14 yrs). Among them, 210 patients had no histologic diagnosis. In the SNUH series, a series of 677 patients with age younger than 16 years old, only surgically treated patients were included. The Taipei VGH series, a series of 986 patients, age younger than 18 years, and all of the patients with computed tomography (CT) or MRI diagnosis were recruited. Among them, 953 patients were younger than 16 years of age. The three series in Asia had similar mean age and gender ratio of patients. The mean age of diagnosis, gender ratio, and relative frequency of common tumor entities of the five comparative series are listed in Table 10. Except the series of CBTC, the other 4 series were predominant in supratentorial tumors that were 53.3%, 52.8%, 60.0% and 58.3%, respectively. In the recently reported series of primary pediatric brain tumors in children, supratentorial tumors predominated over infratentorial tumors.14, 16, 17 By using chi-square test, statistical comparison of relative incidence of tumors in these series was performed (data not shown). Except choroid plexus tumor and pineal parenchymal tumor, significant statistical difference (P < 0.05) of relative incidence existed in astrocytic tumor, medulloblastoma, germ cell tumor, craniopharyngioma, ependymoma, and oligodendroglioma. Comparing our series with the series of CBTC and the series of UHM individually, significant statistical difference (P < 0.0125) of relative incidence existed in both astrocytic tumors and germ cell tumors. There was no such statistical difference between our series and the series of BTRJ and the series of SNUH. The results showed geographic differences in incidences existed between countries and geographic regions for astrocytic tumor and other tumors. We have no explanation for the relatively higher incidence of astrocytic tumors in the two series from Western countries. For the extreme significant difference of incidence in germ cell tumor between the two series of Western countries and the 3 Asian series (P < 0.0001), genomic difference should be considered.

Table 10. Comparison of Case Collections for Average Age, Gender Ratio, and Relative Frequency of the Most Common Childhood Brain Tumors
SeriesCBTC before 1979UHM (1984–2000)BTRJ (1969–1996)SNUH (1959–2000)Taipei VGH (1975–2004)
  1. Case collection of Childhood Brain Tumor Consortium (CTBC), the hospital series of University Hospital of Munster (UHM), Brain Tumor Registry of Japan (BTRJ), Seoul National University Hospital of Korea (SNUH) and the Taipei VGH series of Taiwan (Taipei-VGH).

No. of patients32913194080699986
Gender ratio M/F1.
Age range in yrs0–200–170–14< 16< 18
Average age   7.87.8
Supratentorial tumors40.9%53.3%52.8%60.0%58.3%
 %%%%Mean ageM/F%Mean ageM/F
Astrocytic tumor40.347.328.
Germ cell tumor2.32.515.311.210.01.714.010.63.8
Ependymal tumor10.
Oligodendroglial tumor1.
Choroid plexus tumor2.
Pineal parenchymal tumor1.

In this series, we classified part of those histologic unverified tumors through their own specific clinical features. These tumors were GCTs, specific entities of CNS tumors of phacomatoses, and hypothalamic hamartomas. For tumors of neither histologic diagnosis nor clinical classification, the majority of these tumors were brainstem tumors, and most of them were actually astrocytic tumors. For the present clinical practice, histologic diagnosis is not absolutely necessary for diagnosis and management of GCTs, optic pathway gliomas, vestibular schwannomas, SEGAs, hypothalamic hamartomas, and pontine gliomas.

Primary intracranial GCTs consist of a variety of tumor types with different degrees of malignancy. The reported incidences of GCTs in children were high in Japan and Korea as compared with incidence in Western countries.14–16, 18 We had a similar high incidence. For GCTs, over 40% of these tumors were germinomas.15, 19 It was 54.3% in our series in children. Intracranial GCTs mainly appear in pineal, sellar region, basal ganglia and ventricules. In the pineal region, usually > 70% tumors were GCTs.20, 21 In our series, it was 79.8%. The region of basal ganglia is a significant location for germ cell tumors. In a reported series of 107 patients with primary germ cell tumors, 16 were located in the basal ganglia or the thalamic region.22 In our series, among 138 children with GCTs, 25 were located in the basal ganglia, which constituted of 67.6% of tumors in the basal ganglia.

Serum titers of AFP and beta-hCG in 82 GCTs in this series correlated with their histologic entities and an earlier report from this institution.23 Serum AFP levels were usually higher in pure yolk sac tumors than in immature teratomas and mixed GCTs. The serum AFP titers in pure yolk sac tumors were more than 1000 ng/mL in 9 of 9 cases versus 1 of 9 in immature teratoma and 3 of 13 in mixed GCTs. Normal serum levels of AFP were observed in 5 of 5 mature teratomas. In 2 of 40 germinomas, a low level of elevation of serum AFP titers was found, but staining for AFP was negative in these 3 tumors. The serum hCG levels were high in tumors with syncytiotrophoblastic giant cells.23 In this series, there was no histologic diagnosis of choriocarcinomas. Elevated serum titers of beta-hCG occurred in 7 of 37 germinomas (range, 44.6–6555 mIU/mL), 3 of 8 pure yolk sac tumors (range, 28.8–804 mIU/mL), and 7 of 14 mixed GCTs (range, 10.5–3312 mIU/mL). Serum beta-hCG titers were normal in 7 of 7 mature teratomas. Only 1of 7 immature teratoma had a low elevation of serum beta-hCG. (Table 4).

CNS AT/RT was a rare entity of embryonal tumors. The incidence in children in 2 reported series were 1.3% and 1.6%.14, 24 In this series, it was 2.1% of all primary intracranial tumors in children. These tumors are most common in infants of < 2 years of age. The average ages of diagnosis were 17 months and 29 months in 2 reported series as compare with 4.2 years in ours.25, 26 The diagnosis relied on the specific interest of experienced neuropathologists to separate this entity of tumor from PNET/MB. The diagnosis criteria included morphologic features (rhabdoid cells, small PNET/MB cells, epithelial components, spindle cells), and immunohistologic profiles.26, 27

Phacomatoses represents a diverse group of syndromes that is characterized by retinal lesion, cutaneous signatures, and CNS manifestation, especially tumors. In the current series, the incidence of phacomatoses was 3.9% including NF1, NF2, tuberous sclerosis, BCNS, and neurocutaneous melanosis. Clinical diagnosis for brain tumors could be made in some of these syndromes, such as optic gliomas in NF1, meningiomas and bilateral vestibular schwannomas in NF2, and SEGAs in tuberous sclerosis. From a series in Denmark, among 911 children under 15 years of age, there were 21 patients who had neurofibromatosis. Among them, 16 had an astrocytoma (12 in the optic pathway), 2 had a supartentorial meningioma, and 1 had a vestibular schwannoma.17 von Hippel–Lindau disease with CNS hemangioblastomas usually presented in adult patients. This entity did not occur in the current series.

Radiation-induced brain tumor is a potential complication of radiation therapy for brain tumors in adult and children. In the current series, radiation-induced meningiomas, glioblastoma, anaplastic oligodendroglioma, and soft tissue angiosarcoma were observed in 7 patients after radiotherapy for malignant or benign brain tumors. All of these second tumors have been reported in the literature. Among them, radiation-induced gliomas and meningiomas were most common.28–31


Epidemiologic studies of pediatric brain tumors are usually based on population-based surveys of childhood cancer registries, childhood brain tumor registries, or hospital series. Annual incidence rates or age-specific incidence rates in children can be obtained in a country or in a geographic region. The pitfalls for population-based survey of brain tumors in children through childhood cancer registries or childhood brain tumor registries were incompleteness of patient registration that may lead to underestimation. Bias also exists for not including benign tumors or tumors without histologic verification for registry.32, 33 For hospital series, incompleteness occurred in series involving only tumors with surgery or with histologic diagnosis. In the current series from Taipei VGH, we collected data on 986 patients with primary pediatric brain tumors in the years of 1975–2004. Among them, 842 (85.4%) patients had histologic diagnosis. This large hospital series included all primary brain tumors diagnosed by CT or MRI. Although it is not a study of population-based brain tumor registry, it may still be data representative of primary pediatric brain tumors in Taiwan.