Only a limited number of reports on cerebellar glioblastoma exist, owing to its rarity. Recently, multiple reports on prognostic data have emerged [1, 4–6, 12, 14, 17, 19, 28, 31]. However, several unknown aspects of its clinical and genetic characteristics still remain. In this retrospective study, we analyzed cases of cerebellar glioblastoma in accordance with the WHO 2021 criteria [18]. We found several important aspects of cerebellar glioblastoma.
First, recurrent cerebellar glioblastoma is characterized by a high rate of distant metastasis and meningeal dissemination, which differs from that of supratentorial glioblastoma. Generally, most recurrent patterns have been reported as local recurrences (79.3–80%), with distant recurrences being limited (10.3–20%) in glioblastomas [7, 20]. In our series of cerebellar glioblastomas, seven of the eight cases (87.5%) that recurred were distant or disseminated recurrences, and four of these cases involved the spinal region. Few reports exist on the recurrence patterns of cerebellar glioblastomas. Akimoto et al. reported disseminated recurrence in two cerebellar glioblastoma cases [2]. Additionally, Picart et al. reported that more than half of recurrence cases in cerebellar glioblastoma were distant or meningeal recurrences [19]. We believe that this difference in relapse patterns is of great clinical importance. Glioblastoma has an extremely high recurrence rate; therefore, it is essential to consider factors beyond local recurrence during follow-up. The high incidence of distant and disseminated recurrences may be associated with the anatomical location. Medulloblastoma, which shares the same CNS WHO grade 4 as glioblastoma and occurs in the posterior fossa, demonstrates a high rate (58–78%) of distant recurrence [10, 11]. Additionally, leptomeningeal relapse has been reported to often occur (33–50%) in metastatic brain tumors in the posterior fossa [3, 25, 29]. As malignant tumors in the posterior fossa often recur distantly or spread, cerebellar glioblastomas likely demonstrate similar tendencies.
Second, differences exist in genetic features. In glioblastomas, TERT promoter mutations have been reported to occur in approximately 80% of cases. [8, 15] In the present study, the TERT promoter was wildtype in all cases. Only a few reports exist on the genetics of cerebellar glioblastoma. One prior study reported that TERT promoter mutations occur in only one of four cases of cerebellar glioblastoma, IDH-wildtype. [12] Another reported this alteration in two of 19 cases of cerebellar glioblastoma [6]. Reinhardt et al. reported that TERT promoter mutations were observed in 31% (9/29) of cerebellar glioblastomas, in contrast to 77% (98/127) of supratentorial glioblastomas [21]. They used DNA methylation profiles to clarify the differences between cerebellar and supratentorial glioblastoma. IDH wildtype glioblastoma can be classified into seven DNA methylation subgroups. Among these, “glioblastoma IDH wildtype midline (GBM MID)” exhibits a low TERT promoter mutation rate (8%), whereas “glioblastoma IDH wildtype subclass mesenchymal (GBM MES)”, “glioblastoma IDH wildtype subclass RTK Ⅰ (GBM RTK Ⅰ)” and “glioblastoma IDH wildtype subclass RTK Ⅱ (GBM RTK Ⅱ)” show higher TERT promoter mutation rates (78%, 77%, and 83%, respectively) [26]. Cerebellar glioblastomas are characterized by a higher prevalence of GBM MID and a lower frequency of GBM MES and GBM RTK II [21]. This distribution likely accounts for the lower incidence of TERT promoter mutations in cerebellar glioblastoma. These findings suggest that glioblastomas arising in the cerebellum and supratentorial regions exhibit distinct genetic profiles.
Third, the prognosis of cerebellar glioblastomas is similar to that of supratentorial glioblastomas. Several previous reports exist on cerebellar glioblastoma [1, 4–6, 12, 14, 17, 19, 28, 31]. Among these, cerebellar glioblastoma prognosis has been reported to be poor, good, or similar to that of supratentorial glioblastoma [1, 4–6, 14, 17, 19, 28]. However, most reports include a mixture of different subgroups, such as those diagnosed solely based on pathological results without IDH status or those containing H3K27M mutations. IDH and H3K27M mutations are factors associated with survival prognosis [24, 32]. This report is based exclusively on the WHO 2021 diagnostic criteria for cerebellar glioblastoma. Compared to supratentorial glioblastoma at the same time, OS and PFS were similar. Contrastingly, the median OS following Bev therapy was 6.6 months, although distant metastasis and meningeal dissemination are common recurrence forms. No significant difference was observed compared to supratentorial glioblastoma. The only report on the treatment outcomes of Bev for recurrent cerebellar glioblastoma is a case report. Linsenmann et al. reported achieving 12 months of control by administering Bev and radiation therapy for spinal metastases of cerebellar glioblastoma [16]. Based on these findings, Bev’s effectiveness for distant and disseminated recurrence of cerebellar glioblastoma may be similar to that of supratentorial glioblastoma.
At our facility, we treat cerebellar glioblastoma in the same manner as supratentorial glioblastoma. We conducted surgical resection followed by adjuvant treatment consisting of TMZ radiochemotherapy as per the Stupp regimen [27]. Based on the treatment findings, treatment according to supratentorial glioblastoma can be considered reasonable at present. Nevertheless, given the various recurrence patterns, we propose that a follow-up MRI should encompass an examination of the spine and spinal cord in addition to the brain.
Limitations
This retrospective analysis has some limitations, including a small sample size and a single-center design. Additionally, a search for EGFR amplification and alterations in chromosome 7 and 10 copy numbers has not been possible, and DNA methylation analysis was not performed. Larger cohorts are required to elucidate the clinical and genetic features of cerebellar glioblastomas. Furthermore, a possibility of selection bias exists in defining the cohort by tumor type, which can be confirmed by pathological and genetic information obtained via surgical resection. Thus, the study likely did not include patients with poor prognosis who were not eligible for surgery. Thus, prospective trials are necessary to demonstrate the implications of these findings.