Hemangioblastoma, a rare tumor of the central nervous system (CNS), was first reported by von Hippel–Lindau in 1895 (1). It is a benign, vascular tumor that accounts for 1%–3% of intracranial space-occupying lesions (1, 2), and recent studies have demonstrated that hemangioblastoma arises from arrested mesoderm-derived hemangioblasts during embryonic development (3–5). Hemangioblastoma occurs frequently in the cerebellum, brainstem, and spinal cord, but rarely does supratentorial hemangioblastoma occur either sporadically or in conjunction with von Hippel–Lindau (VHL) disease, which accounts for approximately one-third of cases (6–8).

VHL disease is an autosomal-dominant inherited disorder, mainly caused by germline mutations of the VHL gene located on the short arm of chromosome 3. The prevalence of VHL is 1/(91,000–36,000) (9) and penetrance can reach 87%–97% (10, 11). A VHL patient often presents with a multiorgan tumor syndrome, including CNS hemangioblastoma, retinal hemangioblastoma, renal carcinoma or renal cyst, pancreatic neuroendocrine tumor or cysts, and adrenal pheochromocytoma (12, 13). The development and progression of CNS-associated hemangioblastoma is the most common cause of morbidity and mortality in VHL disease, among which supratentorial hemangioblastoma accounts for only 1%–6% of VHL-related hemangioblastoma (7, 14). However, studies focusing on supratentorial hemangioblastoma have been rarely reported, which makes it difficult to provide valuable information on clinical diagnosis and treatment.

This study was performed mainly to summarize the clinical characteristics of sporadic and VHL-related supratentorial hemangioblastoma and explore the differences in clinical characteristics between cystic and solid tumors in supratentorial hemangioblastoma in order to enrich the current evidence of biological and clinical features and further optimize the management of intracranial hemangioblastoma, especially supratentorial hemangioblastoma.

CNS hemangioblastomas are the most common clinical manifestations of VHL disease. Previous studies have indicated that the growth rate of intracranial hemangioblastomas is substantially greater than that of other benign intracranial tumors (17, 18). However, CNS-associated hemangioblastomas are highly variable in supratentorial and infratentorial anatomical distribution. Lonser et al. (19) analyzed 225 VHL patients with CNS hemangioblastomas, and supratentorial patients accounted for 10.2% (23 cases in total). In terms of the number of tumors, there were 1921 tumors in 225 patients, of which 1% (21 supratentorial hemangioblastomas) were supratentorial locations. By analyzing 152 CNS-associated hemangioblastomas, Peyre et al. (7) showed that VHL accounted for 54% and supratentorial hemangioblastoma occurred in 3.2%, which is in agreement with reports on VHL patients, with supratentorial hemangioblastoma occurring in 1%–6% (14, 15). In addition, they found that only 13 (3.2%) of 409 VHL patients had supratentorial tumors, of which 4 were supratentorial multiple hemangioblastomas. A further analysis of the imaging data of 14 tumors revealed that 13 tumors had progressed (92.9%); in contrast, of 373 infratentorial cases reported by Wanebo et al. (20), only 44% reported tumor growth. Thus, based on the progression of supratentorial hemangioblastoma reported by Peyre et al. the possibility of supratentorial hemangioblastoma developing further seems to be higher (92.9%) than infratentorial and spinal hemangioblastoma (44%).

It is of great practical significance to clarify the clinical characteristics between cystic and solid components in hemangioblastoma. Our data showed that among supratentorial tumors, cystic tumors accounted for 27%, compared with the rate of 33% reported by Mills et al. (21). We further analyzed the location distribution of supratentorial cystic tumors and found that cystic tumors were mainly located in the cerebrum. Similar to cerebellar hemangioblastoma, which is mainly located in the posterior (74%) and superficial (68%) regions of the cerebellar cortex rather than in the anterior or deep region (6), supratentorial cystic hemangioblastomas are more frequently located in the superficial (58.2%) than in the deep cerebral cortex. This may be due to differences in the hydraulic conductivity of the cerebrum compared with other supratentorial locations. These intrinsic anatomic differences in fluid transport may make the cerebrum more prone to the generation of peritumoral cysts (22). In addition, peritumoral edema precedes and underlies the propagation of peritumoral and intratumoral cysts. Our data showed that there is a significant correlation between the presence of edema and the presence of cystic components in supratentorial tumors (r = 0.307, p < 0.01). Previous research has provided insights into the potential mechanisms leading to the formation and progression of tumor-related peritumoral cysts, which has an association with high intratumoral pressure, increased vascular permeability, and the levels of vascular endothelial growth factor (VEGF) (23, 24).

The formation of peritumoral edema or cyst is an important reason for the occurrence of symptoms of patients, but this edema will subside with tumor resection. For example, in patient 7 from our institute, the preoperative massive peritumoral edema disappeared after total tumor resection. Moreover, Peyre et al. (7) found no significant correlation between the formation of cysts and tumor size in a small series of cases. Similarly, consistent with their results, our data showed no correlation between supratentorial cyst formation and tumor size (p = 0.053). In view of its highly vascular characteristics, hemangioblastoma often shows obvious contrast enhancement on T1-MRI (25, 26). The signal of flow void within the tumor on the T2-weighted images is occasionally observed and regarded as a characteristic feature of the tumor (27, 28). These regions are considered to be intratumoral vascular structures. In our data, the hemangioblastoma with the flow void effect accounted for 13.5%. The typical vascular flow void signal was shown in patient 2, which is a low signal on T2 MRI and postcontrast enhancement. In addition, by analyzing the surgical resection degree of supratentorial tumors, we found that the total resection of cystic tumors located in the cerebral lobe could be easily performed, while for sellar tumors, whether cystic or solid type, the total resection of GTR was more difficult to perform compared with that of the other parts (29). Therefore, it is suggested that both cystic and solid tumors in the sellar region should be inspected before surgery to evaluate whether they can be completely resected.

Previous studies have confirmed that, during the assessment of the progression from the asymptomatic to the symptomatic state, the tumor size gradually increases, and the volume of the cyst is significantly larger than that of the tumor (20, 22). We found that in patients with supratentorial solitary tumor, the size of the cystic tumor was significantly larger than that of solid tumors; however, this result had no statistical difference. A further analysis of the difference in tumor size between symptomatic and asymptomatic hemangioblastoma showed that the mean size of symptomatic hemangioblastoma was larger than that of asymptomatic hemangioblastoma.

In our analysis, we combined STR and RT as an independent treatment modality to compare with GTR in extending PFS. The results showed that STR and RT were significantly inferior to GTR. In previous studies, high-dose fractioned external beam radiotherapy (EBRT) has been demonstrated to improve overall survival (OS) and disease-free survival (DFS) for CNS-related hemangioblastoma (30, 31). Recently, numerous studies of stereotactic radiosurgery (SRS) for CNS hemangioblastoma showed that local tumor control rates at 5 and 10 years ranged from 83% to 94% and 61% to 80%, respectively (32–34), with a 0%–7% risk for adverse radiation effects. Of these, Gamma Knife Radiosurgery (GKRS) is considered an acceptable treatment modality for CNS hemangioblastoma, the 5-year tumor control rate was 74%–85% (35, 36), and compared with patients with a solitary hemangioblastoma, those with multiple lesions in CNS were more likely to show progress after GKRS treatment (35). Liebenow et al. (37) held the idea that GKRS altered the clinical course from a saltatory growth pattern to a reduction of tumor size. They reported that the formation rates of new hemangioblastomas at 1, 3, and 5 years were 97%, 80%, and 46%, respectively. In addition, GKRS is not an ideal treatment for cystic hemangioblastoma (32, 34, 37), especially for small mural nodules with large cysts and for those who need urgent relief from symptoms (38). As shown in patient 11 from our institution (Supplementary Figure S3), changes occurring in tumors were highly uncertain during gamma knife treatment, and it was shown during dynamic observation that the tumor could shrink, stabilize, or increase in size, and even cystic changes occurred after this treatment. Gamma knife treatment did not prevent the development of new lesions. If the tumor appears to undergo a cystic change, surgery should be the only curative treatment modality.

However, it is worth noting that GKRS may be an effective treatment for cystic lesions, when they are located in an unfavorable surgical location, or for multiple intracranial lesions in an individual patient (39–41). Despite few studies having reported the efficacy of RT for supratentorial hemangioblastoma, studies on CNS hemangioblastoma have shown that the 5-year OS of the VHL subgroup was better than that of the non-VHL when patients were stratified by VHL status (32, 34). Evidence suggests that surgical resection can be the first line of treatment of supratentorial hemangioblastoma, and GTR should be the first priority. However, in those in whom lesions occur in the eloquent area or in whom only STR can be achieved, EBRT, SRS, or GKRS could be recommended as an alternative treatment modality for small, solid, or noncystic and VHL-associated lesions.

We explored the relationship between different age subgroups and cystic or solid tumors. The results showed that younger symptomatic patients tended to have cystic hemangioblastoma, while older symptomatic patients tended to have solid hemangioblastoma. Furthermore, survival analysis showed that the PFS of cystic patients was better than that of solid patients (p < 0.05). Combined with the Cox regression analysis, it was found that cystic tumor and GTR were two factors of PFS. Based on these results, we recommend that, for young patients with supratentorial cystic hemangioblastoma accompanied by edema, if the tumor is not located in the sellar region, for which the total resection rate was 65% in our study, which was lesser than that for cerebrum (89%), ventricle (69%), and optic nerve (80%), GTR could be the first priority for surgery for achieving a better prognosis.

Craniospinal hemangioblastoma occurs in over 80% of patients with VHL-associated hemangioblastoma, and multiple lesions occur in over 90% of patients (25). The results of our data analysis showed that sporadic patient cases (91.82%) were found for supratentorial single lesions. While sporadic and VHL-related hemangioblastomas share histological characteristics in the CNS, the clinical course of these tumors may differ significantly (42). Compared with sporadic patient cases, VHL patients, whose lesions were detected at a younger age, are more likely to develop multiple tumors. Lonser et al. (19) also found that in patients with VHL, new tumorigenesis seemed to correlate with younger age. Patients younger than 20 years were more likely to develop new tumors than those older than 40 years. In our study, we found that the average age of VHL patients developing supratentorial hemangioblastoma was younger than that of their sporadic counterparts, indicating that, during the lifetime of a VHL patient, the risk of new tumors decreases with age. This age-related tumorigenesis may be due to VHL gene missense mutated pVHL (VHL protein) (8), which still potently participates in the course of hypoxia inducible factor (HIF) degradation. However, due to the instability of the mutant protein, it is often degraded by chaperone proteins (e.g. Hsp70 and Hsp90) shortly after transcription (43). Previous studies have confirmed a progressive decline in the function of these chaperones with aging, which may contribute to the prolongation of missense VHL protein activity and attenuation of the new tumor growth (44). This age-related decline in proteasome function may partially account for the difference in the onset age of supratentorial hemangioblastoma between the VHL and the sporadic patient cases.

In conclusion, in this study, we mainly performed a detailed and in-depth analysis of the clinically relevant features of supratentorial hemangioblastoma by summarizing previous case reports from the literature and case series from our institute. We found that, in supratentorial hemangioblastoma, GTR significantly prolonged patients’ PFS compared with STR and RT. Meanwhile, we also found that cystic hemangioblastoma existed as a separate entity and had significant differences with solid tumors in terms of preoperative KPS, treatment modality, and PFS. However, due to the limited number of cases and insufficient clinical data in this study, further multicentric prospective studies and in-depth exploration are needed to refine the management of supratentorial hemangioblastoma.