Child's Nervous System

, Volume 23, Issue 4, pp 399–405

Outcome of children with posterior fossa medulloblastoma: a single institution experience over the decade 1994–2003


  • D. Kombogiorgas
    • Department of NeurosurgeryBirmingham Children’s Hospital
    • Department of NeurosurgeryBirmingham Children’s Hospital
  • A. R. Walsh
    • Department of NeurosurgeryBirmingham Children’s Hospital
  • A. D. Hockley
    • Department of NeurosurgeryBirmingham Children’s Hospital
  • M. Stevens
    • Department of OncologyBristol Children’s Hospital
  • R. Grundy
    • Department of OncologyQueen’s Medical Centre
  • A. Peet
    • Department of OncologyBirmingham Children’s Hospital
  • M. English
    • Department of OncologyBirmingham Children’s Hospital
  • D. Spooner
    • Department of RadiotherapyBirmingham Children’s Hospital
Original Paper

DOI: 10.1007/s00381-006-0258-5

Cite this article as:
Kombogiorgas, D., Sgouros, S., Walsh, A.R. et al. Childs Nerv Syst (2007) 23: 399. doi:10.1007/s00381-006-0258-5



While the impact of radiotherapy in the management of medulloblastoma was recognised, the introduction of chemotherapy was investigated in clinical trials and shown to confer an additional advantage. We reviewed the outcome of a series of consecutive patients to assess the impact in a population-based clinical establishment.

Materials and methods

A series of 38 children treated for medulloblastoma at Birmingham Children’s Hospital between 1994 and 2003 was analysed. The effect of surgery, radiotherapy, chemotherapy and metastasis on survival was analysed.


The overall 5-year survival rate was 61.4% for the 36 patients who had resective surgery, while 2 patients had biopsy only and died within a few months. There was no operative mortality. The incidence of hydrocephalus needing permanent shunting was higher in the first 3 years of life (p = 0.007, chi-square). The 5-year survival rate of patients with total and sub-total excision of medulloblastoma was 61.1% and 61.8%, respectively. The 5-year survival rate of patients older than 3 years was 73.4% and for patients under 3 years was 36.3% (p = 0.007, log rank). Metastases at presentation did not influence survival. All deaths occurred in the first 32 months.


The contribution of chemotherapy in the improvement of the overall survival appears more evident in children younger than 3 years or presenting with metastases. The absence of significant difference in survival between patients with total or sub-total excision of medulloblastoma supports the view that total excision of medulloblastoma can be avoided when the risk for potential intra-operative damage and consequent neurological deficits is high.




Medulloblastoma is a malignant brain tumour that affects mainly patients in childhood and constitutes approximately 20% of childhood cerebral tumours [29]. During the past 40 years, the survival improved as a result of radical surgical removal of the tumour followed by radiotherapy to the craniospinal axis with or without chemotherapy [3, 10, 18, 2224, 30, 31, 35, 37, 39, 40]. However, there is considerable variation on the 5-year survival rates reported from various centres ranging from 50% to 80% [2, 23].

This study focuses on the survival outcome of consecutive patients treated during the decade 1994–2003, when MRI had been employed routinely pre-operatively and post-operatively, re-operations for residual tumour have increased and chemotherapy was adopted as standard adjuvant treatment, in line with established United Kingdom Children’s Cancer Study Group (UKCCSG) protocols. Therefore, we wanted to establish the impact on survival of such intense modern treatment.

Materials and methods

Patient population

As our patients are derived exclusively from the population of the West Midlands region of Britain (approximately 5.2 million people/1 million children) and were followed-up exclusively in our hospital since the date of diagnosis, this is a population-based study. Each child with a brain tumour is registered routinely with the Cancer Registry of our hospital and the UKCCSG (United Kingdom Children’s Cancer Study Group) Registry upon admission and diagnosis. Patients for this study were obtained from the computer database of the Cancer Registry at Birmingham Children’s Hospital and actively followed-up in oncology and neurosurgery clinics. The above patients were operated on during 1994–2003 by three neurosurgeons (ADH, ARW, SS) and were subsequently treated by three paediatric oncologists (RG, ME, MS) and one radiotherapist (DS).

The study includes 38 patients with histologically proven posterior fossa medulloblastoma who completed their treatment at our institution. All patients had pre-operatively CT scan of the brain in the early years and since 1997 MRI scan. Most of them before 1997 and all after 1997, had post-operative MRI imaging at 48 h and then every 3 months for the first 2 years and every 6 months thereafter until 5 years from operation. Most of these patients had been recruited to the UKCCSG clinical trials that were active at the time of presentation and hence most information had been prospectively collected.


Variables considered as possible predictors of survival were: age at diagnosis, gender, duration of symptoms, presence of metastasis or hydrocephalus at time of diagnosis, extent of resection, post-operative adjuvant therapy as radiotherapy or chemotherapy. The follow-up period was defined as the period from the date of diagnosis up to the date of last medical review of the patient until May 2005 or the date of death. Post-treatment morbidity was studied such as CSF leak, post-operative haemorrhage, neurological deficits, ambulatory and educational difficulties.

Statistical analysis was performed with the statistical program SPSS [version 12.0, SPSS, IL, USA]. The chi-square was used to identify factors of significance. Linear regression analysis was used to investigate the effect on survival of all variables that proved significant on the chi-square. The analysis of variance (ANOVA) was used to assess the influence of different factors on survival and the Kaplan–Meier survival analysis was performed for sub-groups. Significance was set at p = 0.05.


There were 22 (57.9%) male and 16 (42.1%) female (ratio 1.4:1) patients with the mean age of 70.6 months at diagnosis (range: 1–169 months, SD = 44.9); 25 (66%) patients were over 3 years old and 13 (34%) patients were under 3 years of age and 5 patients were under 2 years of age at the time of diagnosis. The mean follow-up time was 43.7 months (range: 2–131 months, SD = 36.8). Of the 38 patients, 15 (39.5%) died. The mean survival time of the deceased patients was 12.8 months (range: 2–32 months, SD = 8.117).

The most common symptoms at the time of presentation included: vomiting (73.7%), headache (63.2%), ataxia/ataxic gait (55.3%), sleepiness (23.7%), reduced appetite (18.4%), weight loss (15.8%), diplopia (13.2%), irritability (7.9%), blurred vision (5.3%), increased head size (5.3%), change of personality (2.6%) and others (21.1%).

Symptoms lasted from 1 to 52 weeks (mean = 8.8 weeks, SD = 11.8). The most common neurological signs at the time of presentation included: ataxia/ataxic gait (52.6%), nystagmus (21.1%), cranial nerve palsy (21.1%), papilloedema (42.1%), increased head circumference (10.5%), torticollis (7.9%), limitation of upward gaze (7.9%), reduced visual acquity (5.3%) and others (10.5%). Two children less than 2 years old had tense anterior frontanelle at presentation.


Radiologically proven hydrocephalus at the time of presentation was found in 33 patients (87% of the whole group); 21 patients over 3 years of age (84% of the patients over 3 years) and 12 patients under 3 years of age (92% of the patients under 3 years). In total, 12 patients (31.6% of the whole group) required permanent treatment of hydrocephalus before or after tumour excision; 4 patients over 3 years of age (16% of the patients over 3 years) and 8 patients under 3 years of age (61.5% of the patients under 3 years). This difference in the incidence of hydrocephalus needing permanent treatment between the two age groups (under and over 3 years of age at presentation) was statistically significant (p = 0.007, chi-square). One patient had endoscopic third ventriculostomy pre-operatively, which was unsuccessful in controlling the hydrocephalus and required a subsequent shunt, 11 patients had ventriculoperitoneal shunt and 1 patient had lumbar peritoneal shunt. The time of shunt operation ranged from 1 to 112 days (mean = 21.9) after the posterior fossa tumour operation in 10 patients within 30 days. Of the 12 shunted patients, 7 died within 19 months from diagnosis (mean = 9 months). The need for shunting did not correlate with the presence of metastases at presentation or later in the course of the disease.


All 38 patients underwent surgery in the form of posterior fossa craniotomy or craniectomy for histological diagnosis and removal of as much tumour as was possible; 36 of the patients had an excision of the tumour and 2 patients only had biopsy. The degree of tumour excision as assessed by the operating surgeon was total in 16 patients (42.1%) and subtotal (greater than 70% of the tumour but less than total) in 20 (52.6%) patients. The extent of surgical re-section did not correlate with the age group at presentation.

Post-operative complications occurred in 15 (39.5%) patients. They included: haematoma (2.6%), infection/meningitis (2.6%), hydrocephalus (10.5%), pseudomeningocele (10.5%), transient CSF leak (13.2%), mutism (7.9%), cranial nerve palsy (2.6%), hemiparesis (7.9%) and others (5.2%). There was no operative mortality. The relatively high rate of CSF leak may be related to the underlying hydrocephalus. Most patients did not have insertion of external ventricular drain at the time of tumour excision even if the ventricles were enlarged and most of the patients who required ventricular shunting were shunted after tumour excision.

The histology report showed classic medulloblastoma in 35 (92.1%) patients and desmoplastic medulloblatoma in 3 (7.9%) patients.


A total of 10 patients (26%) developed metastases either before or after tumour surgery. Radiologically proven tumour metastasis was found in 5 (13%) of 38 patients at the time of presentation. Four patients had spinal metastases and three of them died within 18 months of diagnosis, while one was alive at 18 months. One patient had cerebrospinal metastases and was alive at 45 months. The presence of metastases at presentation did not correlate with the duration of symptoms. Of the remaining 33 patients, another 5 patients (14.7%) developed metastases during the follow-up period. The type of metastases was: cerebral in one patient, cerebrospinal in one patient, extra-axial in two patients and systemic in one patient. Three of the patients died within 10 months of appearance of metastases (mean survival time of 3.7 months), 1 with cerebrospinal metastases was alive at 12 months and 1 with cerebral metastasis was alive at 74 months. It is interesting to note that the development of metastases after surgery adversely correlated with the extent of surgical resection (p = 0.048, chi-square), as none of the patients who had subtotal excision developed metastases after surgery.

Only 5 (13.15%) of the 38 patients had lumbar puncture for CSF cytology because it was not done routinely in the early 1990s.


Radiotherapy was administered to 29 patients (76.3%), 23 of these patients were over the age of 3 years. Of the 29 patients, 28 completed their course. Craniospinal radiotherapy and booster dose in the posterior cranial fossa was given to 26 patients, 24 of them were treated according the PNET-3 protocol [35, 36]. Two patients only had posterior fossa irradiation of 45 Gy.


Chemotherapy was administered to 24 patients (63.2%), 11 were under the age of 3 years. Of the 24 patients, 15 completed their course. Chemotherapy was discontinued in nine patients due to complications or poor tumour response. The type of chemotherapy regimes included: PNET-3 protocol (six cases) [35], infant brain tumour protocol in three cases (UKCCSG study CNS 9204: vincristine 1.5 mg/m2, carboplatin 550 mg/m2, methotraxate 8 g/m2, folic acid 15 mg, vincristine 1.5 mg/m2, cyclophosphamide 1,500 mg/m2, cisplatin 40 mg/m2), SFOP infant brain tumours protocol in seven cases (carboplatin 15 mg/kg, etoposide 5 mg/kg/day, vincristine 0.05 mg/kg, procarbazine 4 mg/kg/day, cisplatin 1 mg/kg/day, cyclophosphamide 50 mg/kg), recurrent PNET protocol (UKCCSG study CNS 2000–01) in two cases and Packer’s protocol in three cases [23, 24].

The most common complications after chemotherapy and radiotherapy treatment included: GH deficiency (39.5%), hypothyroidism (34.2%), ovarian failure (5.3%), behavioural problems (13.2%), educational problems (7.9%), hearing problems (7.9%), cataract (7.9%), developmental delay (2.6%), depression (2.6%) and renal failure (2.6%).


The overall 5-year survival rate was 58.1%. Excluding the 2 patients who had biopsy only, the 5-year survival of patients who had resective surgery was 61.4%. All deaths (15 cases) occurred within 32 months since the time of diagnosis. Survival rates were 46.4% for males and 73.86% for females, but this difference was not statistically significant (p = 0.158, log rank). For patients older than 3 years at the time of diagnosis (25 cases), the 5-year survival rate was 73.4%. Patients under 3 years old at the time of diagnosis (13 cases) who had resective surgery had a 5-year survival rate of 36.3%. The difference in survival between the two age groups was statistically significant (p = 0.007, log rank). From the 15 deceased patients, 6 (40%) were over 3 years of age at the time of diagnosis and 9 (60%) were under 3 years. There was no correlation between the duration of symptoms (more or less than 6 weeks) and survival.

While the 5-year survival rate was better for patients without metastases at presentation (61% vs 40% for patients with metastases at presentation), this difference was not statistically significant (p = 0.283, log rank). This applied to both age groups (over and under 3 years of age at presentation).

Hydrocephalus at presentation was not a statistically significant factor affecting the 5 years survival (p = 0.533). While the need for permanent treatment of hydrocephalus was a significant risk factor affecting survival on its own, this was not an independent variable but dependent on age at presentation, as we have already seen.

For the entire group, the 5-year survival rate of patients with total and subtotal excision of medulloblastoma was 61.1% and 61.8%, respectively. The difference was not significant. The two patients who only had biopsy died within 6 months from diagnosis.

For children presenting over the age of 3 years, both radiotherapy+chemotherapy (n = 12) and radiotherapy only (n = 11) groups had a 5-year survival rate of 81.8%. The difference was not statistically significant (p = 0.744, log rank).

For children presenting under the age of 3 years, radiotherapy+chemotherapy (n = 6) and chemotherapy only (n = 5) groups had 5-year survival rates of 50% and 20%, respectively. The difference was statistically significant (p = 0.015, log rank).

Children who presented with radiological metastases and were treated with radiotherapy and chemotherapy (n = 4) had a 5-year survival rate of 50%, in contrast to 1 child who was treated with chemotherapy only and died at 13 months.

Children who developed radiological metastases at follow-up and were treated with radiotherapy and chemotherapy (n = 2) had 100% actuarial 5-year survival rate, in contrast to 1 child who was treated with chemotherapy only and died at 15 months and 2 children who were treated with radiotherapy only and died at 17 and 32 months.


There is a strong indication from published series that children with medulloblastoma presenting over 3 years of age benefit greatly from radiotherapy, but the treatment of younger children remains a challenge. Recent trials demonstrated a beneficial effect of chemotherapy although this was not universally observed [4, 13, 16, 27, 33, 36]. Prognosis of patients with medulloblastoma depends on several factors such as the presence or absence of metastases at diagnosis and the radiotherapy modality [15], use of chemotherapy [23], extent of residual tumour [1], gender [38], duration of the symptoms [17], insertion of ventricular-peritoneal shunt within 30 days since the surgical re-section of tumour [5].

The M stage system has proven to be one of the most important variables in predicting for recurrence-free survival [8, 10, 21, 23, 39]. Currently, children more than 3 years of age are divided into standard risk group and high risk group. The standard risk group includes patients with non-metastatic disease and less than 1.5 cm2 residual tumour based on immediate post-operative gadolinium-enhanced MRI scan of the head and the spine and a negative cytology obtained from a lumbar puncture approximately 10 days after surgical resection of the tumour. The high risk group includes patients with metastatic disease or those with more than 1.5 cm2 residual tumour on immediate post-operative CT or MRI scan [12, 40], although recent studies (e.g. PNET IV) do not use residual disease as a high risk feature because some studies showed this not to be of prognostic value [24]. Children under 3 years of age at presentation are considered high risk group. Brain stem invasion, which was previously thought to be a high risk feature is no longer considered to be so. Gajjar et al. reported that there was no difference in progression-free survival (PFS) for patients who underwent gross total resection compared to those with any detectable residual tumour [12].

A short duration of symptoms was reported to be associated with the diagnosis of more advanced medulloblastoma [17]. Such a correlation was not shown in our study.

Leptomeningeal metastases are usual in children with medulloblastoma and were reported as high as 46% but usually ranges from 10% to 35% [6]. Radiologically proven leptomeningeal metastases were diagnosed in 13% of our patients at the time of presentation. Another 13% of our patients developed metastases at the course of their illness after surgery. In the absence of complete lumbar puncture data at presentation, it is possible that the true incidence of metastatic disease may have been higher.

Currently, the mortality rate from surgical treatment ranges from 1% to 5% [32]. There was no mortality in our series. The extent of surgical resection based on post-operative imaging was considered as an important prognostic factor. Khafaga et al. reported actuarial survival for the whole study group of 149 patients, 53% at 5 years and 38% at 10 years. Survival of patients who had clinical and radiological complete resection of tumour at surgery was significantly better than patients with incomplete tumour removal [20]. Albright et al. reported that the extent of tumour resection as estimated by the neurosurgeon did not correlate with outcome, but that the extent of residual tumour on post-operative imaging did correlate with prognosis in children over 3 years of age without disseminated disease [1].

The prognosis markedly changed since the introduction of irradiation of the whole neuroaxis post-operatively, which improved survival up to 54% [26]. The combination of surgery with radiotherapy and/or chemotherapy has improved the prognosis of children with medulloblastoma [11]. In the more recent years, the overall 5 years survival is approximately 60% ranging from 32% up to 83% in different studies [1, 5, 23, 29, 30, 34, 37].

Although radiation is a highly effective therapy for medulloblastoma, it causes severe damage to the developing brain. Radionecrosis, mineralizing microangiopathy and calcification of brain, hypopituitarism (growth hormone, FSH, LH deficiency) hypothyroidism, cataract and ototoxicity are well-documented long-term complications of radiotherapy for pediatric brain tumours [7]. Perceptual motor task performance was below average in more than 50% of the participants in the series by Johnson et al., but motor dexterity was more severely affected than perception. Also, the majority of the participants have learning problems and delay in both physical growth and development [19]. Palmer et al. support the hypothesis that medulloblastoma patients demonstrate a decline in intelligence quotient (IQ) values because of the inability to acquire new skills and information at a rate comparable to their healthy same-age peers, as opposed to a loss of previously acquired information and skills [25].

At present, the treatment of average risk patients with medulloblastoma includes maximum surgical resection, followed by chemotherapy and whole neuroaxis radiotherapy [14]. In the USA, the standard neuroaxis radiation dose was reduced from 36.0 to 23.4 Gy aiming to reduce the neurological toxic effects, which are still present especially in children younger than 8 years of age [28].

Chemotherapeutic regimens were investigated for the treatment of standard risk medulloblastoma patients such as SIOP-3, CCG9892, SJMB 96, POG 8631/CCG923 [24, 31, 35, 37]. These studies showed that post-operative chemotherapy contributes to improved survival in standard risk patients but which of the adjuvant chemotherapeutic regimen, if any, has the greatest survival advantage is still unclear. Rutkowski et al. reported that chemotherapy (cyclophosphamide, vincristine, methotrexate, carboplatine and etoposide) and intra-ventricular methotraxate alone is a promising treatment for medulloblastoma in young children without metastases [30].

Historically, 5-year progression-free survival in children with high risk medulloblastoma was less than 50% [9]. Several attempts were done to improve the survival of high risk children using pre-irradiation and post-irradiation chemotherapy regimens such as CCG 921, SJMB 96 and Packer protocol with promising results [23, 31, 40]. In our study, the overall 5-year survival was 61% for children who had resective surgery as first treatment, 73.4% for children presenting over the age of 3 years and 36.3% for children presenting under the age of 3 years. Overall treatment morbidity would be described as good. Radiotherapy significantly affected the 5-year survival but chemotherapy appears to improve the survival of children under 3 years of age and also children presenting with metastases because the presence of metastases at presentation was not a significant predictor factor affecting survival of patients either younger or older than 3 years of age. Age at presentation remains the sole most important predictive factor of outcome with children presenting under the age of 3 years having significantly worse outcome. The extent of surgical resection does not appear to influence survival. The reason for this is twofold: as imaging and surgical techniques have improved, in children who had subtotal resection, the residuum is considerably smaller than in the past, typically in the region of 10–15% of the original tumour. This tumour bulk appears to be dealt with successfully by the additional treatment of radiotherapy and/or chemotherapy. In addition, with modern surgical techniques surgical morbidity was reduced and this allows the majority of the patients to start early and complete adjuvant treatment, an important factor in improving survival.


The survival rate results obtained in this study are compatible with current published data. The role of radiotherapy in the treatment of medulloblastoma remains paramount. The contribution of chemotherapy in the improvement of the overall survival seems to concentrate in the younger age group and in children presenting with metastatic disease. In the latter group, it appears that the use of chemotherapy has removed the effect of metastasis as a negative predicting factor for survival. The absence of significant difference in survival rate between patients with total or subtotal excision of medulloblastoma supports the view that the total excision of medulloblastoma can be avoided when the risk for potential neurological deficits is high.


We would like to thank Helen Curry, data manager at the Cancer Registry of the Birmingham Children’s Hospital for the significant help in supplying us with information.

Copyright information

© Springer-Verlag 2006