Medulloblastoma is a malignant neuroectodermal tumor uncommon in adults. Median age at diagnosis is 5 years. Medulloblastoma accounts for 30% of brain tumors of childhood but fewer than 3% of all primary neoplasms of the central nervous system (CNS) in adults. Most children treated for medulloblastoma are included in prospective studies. Clinical prognostic factors have been identified in children and guide therapeutic strategies. In contrast, in adults, the majority of reported cases of medulloblastoma are single-institution, retrospective series including only a few patients.

Adult and pediatric medulloblastomas are radiologically distinct. Adult tumors usually involve the lateral cerebellar hemispheres, and pediatric tumors occur in the vermis. Histology also differs between adult and pediatric tumors.

Discussion

We report the largest series of adult medulloblastomas, with a median follow-up of 7 years. Five-year and 10-year survival rates were 72% and 55%. This is comparable to results reported in two other adult series (72% and 83%). Large series of children’s medulloblastomas have been reported, with 5-year survival rates of 65% and 70.7%. The prognostic factors identified in this multicenter series were all similar to those observed in children.

Histology differs somewhat between adults and children. More adult medulloblastomas are desmoplastic (25–40%), and 30% in the present series were desmoplastic. In a series of children’s medulloblastomas, 15% were desmoplastic. The prognostic value for desmoplastic subtype is not established. Some investigators consider desmoplastic histology a favorable prognostic factor, but others disagree. In the present series, desmoplastic histology did not influence prognosis. Lower growth rate parameters have been described in adult medulloblastoma. For example, Giordana  described a low rate of MDM2 amplification and p53 gene mutation in adult medulloblastomas. In univariate analysis, MDM2 and Tp 53 had poor outcome in adults. This may explain the late recurrence of adult medulloblastomas. In the present study, most relapses occurred 2–7 years after treatment, as described in the other series.

Adult medulloblastomas occurred more often in the medial part of the PCF. In the present series, 44% of patients experienced a laterally located tumor, but tumor location was not a prognostic factor, agreeing with results published by Park  and Bloom.

In this study, the presence of surgical residue had borderline significance. In contrast, Chan’s  reported a series of 32 adult medulloblastomas in which gross total resection was the most important prognostic factor. Hartsell et al. reported similar results. Similarly, in children, the presence of residue >1.5 cm2 has prognostic value. Our retrospective study included patients treated between 1975 and 2002; the evaluation of surgical residue was probably better performed with MRI after 1990. The issue of locoregional invasion may be crucial: brainstem and V4 floor involvement were negatively correlated with survival in multivariate analysis in this study, with a high correlation between V4 floor involvement and surgical residue. These prognostic factors are similar to those identified in pediatric medulloblastoma series. The true prognostic significance of the extent of surgical resection is still unclear. Similarly, postoperative performance status was found to be a significant prognostic factor only in univariate analysis, but was strongly correlated with the extension of disease and so not found to be an independent prognostic factor. One may wonder whether locoregional extension and the presence of residue may depend on intrinsic biologic factors that would render the tumor more invasive.

Chemotherapy did not increase survival in either the whole population or in the patients designated as having standard or high risk in this series. One probable explanation is the heterogeneity of regimens in this multicenter study that included patients treated over a long period. Efficiency of chemotherapy has been well established in the pediatric population with metastases, but it is still controversial in the subgroup designated as being at standard risk for recurrence. For example, Taylor’s group  reported a series of primitive neuroectodermal tumors that compared three cycles of vincristine, etoposide, cyclophosphamide, and carboplatin followed by radiotherapy with radiotherapy alone. The 5-year PFS rate was 74.2% with combined therapy, compared with 59.8% in the group receiving radiotherapy alone; but no difference was observed regarding overall survival. Similarly, the eight drugs in 1 day used in most patients of the current series may not be the best protocol for treatment of medulloblastoma: Zelter’s group  randomized 203 patients to receive vincristine, lomustine, and prednisone or eight drugs in 1 day after standard-dose radiotherapy. The 5-year PFS rate was 63% with the three-drug regimen, compared with 45% in the eight-in-1-day regimen (p = 0006).

In the present study, 124 patients were classified with standard-risk medulloblastoma. Their survival was significantly different from that of patients with high-risk tumors, with respective 10-year survival rates of 62% and 49% (p = 0.03). In this series, 6 patients did not receive radiotherapy, 4 because of chemotherapy toxicities. The toxicity of chemotherapy delaying or suppressing radiation in standard-risk tumors was described as negatively impacting prognosis by Kortman. They reported a 3-year PFS rate of 65% in patients receiving postoperative high doses of chemotherapy (methotrexate, ifosfamide, etoposide, cisplatin, and cytarabine) delivered before radiotherapy, compared with 78% in patients receiving postirradiation chemotherapy (cisplatin, vincristine, and lomustine). Results of the last French pediatric protocol purposed to suppress chemotherapy and to treat patients with standard risk with hyperfractionated radiation showed a PFS rate of 81% at 3 years. This rate compared favorably to those published by Thomas et al. for the Children’s Cancer Group with standard dose and by Kortman et al. for the HIT protocol. Moreover, preliminary results from the M-SFOP 98 evaluation of intelligence quotient (IQ) seemed to demonstrate no decrease of IQ in the 22 evaluated patients after 2 years, whereas during this same period in children treated with standard irradiation, cognitive impairment is usually already evident. The impact of the timing of chemotherapy in the treatment or the type of regimens is not clear; chemotherapy must not delay the start of radiation. Hyperfractionated treatment for adult standard-risk medulloblastoma should be evaluated with prospective investigations.

In univariate analysis, radiotherapy dose delivered to the spinal axis and on the PCF are significant prognostic factors. In multivariate analysis only, radiation doses <50 Gy to the posterior fossa are significant. These results suggest that dose to the PCF was superior at 50 Gy. Moreover, most relapses occurred in the PCF, as reported in major studies for adult medulloblastomas. The importance of dose delivered to the PCF is already established in children.

Only one prospective clinical trial has been reported for adult medulloblastoma. Between 1988 and 2002, 36 patients received 36 Gy to the craniospinal area and 54.8 Gy to the posterior fossa. Patients at high risk received neoadjuvant and maintenance chemotherapy. Five-year PFS was significantly better for patients at low risk (75%) than for those at high risk (61%). Patients with nonmetastatic medulloblastoma had better outcome (5-year PFS 75% vs. 45%; p = 0.01).

A usual dose of 36 Gy is established for the brain and spinal axis. Pediatric protocols try to evaluate the possibility of reducing craniospinal dose in patients at standard risk. Exclusive low-dose irradiation is not optimal according to the results of the Pediatric Oncology Group/Children’s Cancer Group. This study randomized 126 children to receive standard- or low-dose (23.4 Gy) craniospinal irradiation. The PFS rate was 67% in patients receiving the standard dose, compared with 52% in the low-dose group (p = 0.08). In a recent study by the French Society of Pediatric Oncology, 136 patients between the age of 3 and 18 years with standard-risk medulloblastoma received two courses of eight drugs in 1 day followed by two courses of etoposide plus carboplatin. Radiotherapy was administered with reduced dose to the craniospinal region. Overall and 5-year survival was 73.8%. The investigators concluded that reduced-dose craniospinal radiation therapy can be proposed in standard-risk medulloblastoma.

For patients with standard-risk medulloblastoma in the present series, no significant difference in spinal or brain relapse was observed between the patients treated with 36 Gy to the craniospinal area and those treated with lower doses of radiotherapy and concomitant chemotherapy. In this favorable group, only 3 patients experienced spinal relapse with 36 Gy, and 2 with 30 Gy. Several studies showed a decline in IQ in children of different ages who received craniospinal radiation. A study purposed to evaluate the impact of therapy on neurocognitive outcome in children treated for CNS malignancies showed a significant difference in IQ decline between children treated by craniospinal radiation and children treated without radiation (p = 0.005). In this study, decline of IQ in patients receiving cranial dose of ≥33 Gy was greater than in those receiving ≤24 Gy. Although IQ evaluation was impossible in the current retrospective study because of lack of data, these results suggest the neurocognitive impact in decreasing the craniospinal radiation dose in adult patients. Similar to the pediatric series, it seems possible to reduce the craniospinal dose in the favorable group of adult patients receiving chemotherapy by maintaining at least 50 Gy to the posterior fossa. However, the several limitations to this include the lack of quality control of dosimetry. This series is a retrospective study, and no quality control of treatment planning was performed, although several trials have suggested that the quality of radiation impacts outcome. In a series published by Carrie, the number of major deviations in radiotherapy treatment correlated strongly with the risk of tumor relapse (67% 3-year relapse rate in the patient group with two major deviations, 78% in those with three major deviations). Treatment planning should be reviewed by a panel of experts to decrease the target deviations. Quality control would improve survival to an extent equivalent to that attributed to adjuvant chemotherapy.

Conclusion

Medulloblastoma is a malignant neuroectodermal tumor uncommon in adults. We present the largest multicenter series of adult medulloblastomas, from 253 patients aged >18 years. The median overall survival at 5 and 10 years was comparable to survival reported in children (72% and 55%). Prognostic factors identified in this multicenter series were all similar to prognostic factors observed in childhood. As in pediatric series, the posterior fossa remains the major site of relapse. In patients with tumors at standard risk for recurrence, results suggested that treatment should consist of either lower-dose irradiation to the craniospinal area with at least 50 Gy to the posterior fossa plus chemotherapy, or hyperfractionated radiation without chemotherapy. Only a large, randomized study could ensure the role of chemotherapy, but because of the rarity of this disease, the results would remain uncertain. Moreover, the use of different techniques of irradiation, such as a hyperfractionated regimen, and the development of quality control should improve survival to an extent equivalent to that attributed to adjuvant chemotherapy. The irradiation techniques need strict control and should only be performed by centers that treat pediatric patients. As in pediatric trials, radiation treatment planning should be reviewed by a panel of consultant radiation oncologists to decrease the target deviations before the beginning of treatment.