Intracranial meningiomas account for 15 to 20% of all primary brain tumors, and most of them are benign [World Health Organization (WHO) grade I]⁴⁶⁾. Aggressive (WHO grade II) or malignant (WHO grade III) meningiomas overall constitute less than 10% of meningiomas, and they are rarely associated with distant metastases^37,50). If metastases are found, the sites are usually the liver, lungs, pleura, and lymph nodes^{1,4,15,34,43)}. This finding suggests that distant metastases usually occur through a blood-borne route. However, some leptomeningeal disseminations (LDs) of intracranial meningioma through cerebrospinal fluid (CSF) have been reported in the literature^{1,2,5,7,8,9,10,13,14,16,17,18,19,20,21,22,23,24,25,26,27,29,34,35,36,39,42,43,44,45,47,48,49)}. In this report, we summarized our experience with 5 intracranial meningiomas with LD through CSF and reviewed the relevant literature to evaluate the characteristics of patients with LD and discuss predisposing factors.

Between 1998 and 2013, 5 patients with intracranial meningiomas demonstrated evidence of LD through CSF at our institutions. We retrospectively reviewed the medical records, imaging studies, and histopathological findings to determine the general features of the patients, presentation of LD, possible predisposing factors, treatment, and outcomes.

We also reviewed PubMed, Medline, Embase, and ISI Web of Knowledge using the keywords 'meningioma', 'cerebrospinal fluid', 'leptomeningeal', 'dissemination' and 'metastasis'. All articles were reviewed, and relevant articles were collected. Forty-five cases of LD through CSF were found from 36 articles since 1950. All cases were reviewed using the same protocol from our own series.

Distant metastases of meningiomas are rare, and most have been reported to occur from malignant (WHO grade 3) tumors¹²⁾. The pathways of metastases are various. The most common destinations for metastasis are the liver, lungs, pleura, and lymph nodes, which indicates that a hematogenous route through the jugular vein may be the most frequent pathway for metastasis¹¹⁾. Less common are metastases through the paravertebral venous plexus, lymphatic vessels and cerebrospinal fluid, which are reported in a small minority of cases. Particularly, metastases through CSF are extremely rare¹⁾. However, considering that meningiomas arise from the arachnoid villi, tela choroidea or leptomeninges, and are naturally exposed to CSF during their growth or at the time of surgical intervention, it is difficult to explain the rarity of LD through CSF.

Since 1950, only 45 cases of postoperative LD of meningioma through CSF have been reported in the literature (Table 3). Despite the extreme rarity of LD of meningioma, CSF is a certain route for meningioma dissemination. Russell and Rubinstein⁴¹⁾ reported that CSF should be expected to provide a good culture medium for tumor cells, and it is apparent that tumor friability may play a great role in determining the potential for CSF dissemination. Chamberlain and Glantz⁵⁾ also reported a large study regarding LD through CSF in 2005. Two hundred meningiomas were followed in their prospective study (median time, 8.5 years). Eight out of 200 patients (4%) had LD and positive CSF cytology.

In light of the physiologic pathways involving CSF, the risk of intraventricular meningiomas to metastasize to leptomeninges should be high. However, only 9 cases with intraventricular meningiomas were described in the literature. Miller and Ramsden³⁰⁾ conjectured the dynamics of the CSF system might prevent fragment formation and the settlement of tumor cells. However, considering the number of intraventricular meningiomas with LD and the rarity of meningioma metastases, 9 cases (25% of 36 cases where initial location was reported) represent a relatively significant sample size. And, as mentioned above, time to LD involving intraventricular meningiomas is the shortest when compared to other intracranial and spinal meningiomas. These findings suggest that the ventricular location is an important site for the LD through CSF of meningiomas.

Histologically, several factors are predictive for the metastasis of meningiomas, according to the literature. They include characteristic features of the malignant tumor, such as high cellularity, cellular heterogeneity, high mitosis rate, nuclear pleomorphism, tumor necrosis, and invasion of adjacent blood vessels^12,15). Although these factors do not explain metastasis of benign meningiomas, the proportion of malignant meningiomas with LD is high when compared to the general proportion of subtypes of meningioma. Twenty-six (58%) out of 45 cases were WHO grade II or III. With regard to histologic transformation, 6 of 8 histologic transformations were aggressive transformations (from WHO grade I to II or III) (Table 3). In the presented cases, 3 cases were grade II or III, and one case with WHO grade I had histological transformation to WHO grade II at the time of LD. These findings suggest that a malignant meningioma tends to metastasize to the leptomeninges more frequently.

The genetic characterization of meningiomas involving histologic transformation has been studied during the past decade. The cumulative acquisition of genetic alterations in meningiomas might be related to the development of LD, especially in cases with malignant transformation. The most common allelic loss in benign meningiomas is alteration in the long arm of chromosome 22 (22q), and this alteration is considered to be the earliest initiating event in meningioma formation³⁾. Whereas allelic loss at other loci, such as 1p, 3q, 6q, 10q, 14q, 17q, and 18q, is associated with tumor progression from the benign state to the atypical and anaplastic meningiomas³¹⁾. Recently, Nakane et al.³³⁾ suggested that meningiomas similarly progress from the benign to the malignant state by the accumulation of genetic alterations such as 1p loss of heterozygosity and p73/RASSF1A promotor methylation. However, the genetic alterations of meningiomas with LD have not yet been reported in the literature owing to the rarity of LD. The study of a larger number of meningiomas with LD could enable the elucidation of the genetic aspect.

Meningiomas are naturally exposed to CSF at the time of surgical intervention. Theoretically, surgical resection might initiate metastatic spread. However, this theory seems improbable since the incidence is so low in the face of so many tumors being operated on, and there have been cases describing LD through CSF for which there was no surgical intervention³⁹⁾. On the other hand, the surgical resection theoretically could influence malignant transformation of meningiomas. Koenig et al.²¹⁾ reported that growth factors at the site of surgical trauma exacerbate a genetic or environmental predilection for malignancy. This theory was tested in an experimental rat model in which neonatal rats were exposed to intravenously administered high-dose nitrogen-based carcinogens and underwent surgical trauma to the brain. Compared to a control group, which was also exposed to carcinogens, twice as many neoplasms, including gliomas and vestibular schwannomas, developed at the site of injury in the surgical trauma group³²⁾.

When meningiomas are observed at multiple sites, it is difficult to distinguish whether the focus is a multiple tumor or a metastatic tumor. Multiple meningiomas are defined as two or more tumors appearing simultaneously or sequentially in the same patient, and the incidence of multiple meningiomas ranges from 1% to 16% of all meningiomas⁴⁰⁾. Multiple meningiomas usually occur in association with neurofibromatosis type 2, but also have been described in families with no evidence of neurofibromatosis²⁸⁾. However, almost all examples in the literature mentioned that meningioma with LD is not associated with neurofibromatosis and familial multiple meningiomas. And, in our cases, LD is certainly secondary to postoperative seeding through CSF, because histopathlogical features of additional biopsies were similar to the first biopsy and there was no familial history. Therefore, neurofibromatosis and familial multiple meningioma may have little role in LD.

Spinal LDs usually occur in malignant primary brain tumors, and those that arise from intracranial meningiomas are extremely rare. Based on the literature, meningiomas accompanied by spinal LDs were reported in 35 of 45 cases (78%). Especially, 6 out of 9 cases (67%) with intraventrciular meningomas were accompanied by spinal LDs. In addition, 4 of those cases were WHO grade III and the other 2 cases were confirmed as malignant transformation from WHO grade II to III. In the intraventricular meningiomas with LD, the median and mean time interval between local recurrence and spinal LD was 6 months and 10.5 months, respectively (range from 0 to 21 months). And, the mean time interval between intracranial LD and spinal LD was 4 months. In the present cases, a patient with an atypical intraventricular meningioma showed local recurrence and intracranial LD at the same time, and spinal LD was developed approximately 4 months after intracranial LD (Case 3). Considering the literature and the present case, screening spinal MRI could be carefully considered between 6 months and 12 months after local recurrence of a non-benign intraventricular meningioma. When intracranial LDs of a malignant intraventricular meningioma are diagnosed, whole spinal MRI should be performed immediately, and follow-up spinal MRI could be carefully considered between 3 months and 6 months after the time of LD.

So far, multimodal treatments, including multiple surgeries, chemotherapy (systemic or regional intraventricular chemotherapy) and radiotherapy (whole brain, spine and involved-field radiotherapy to sites) were performed for patients with LD. However, the prognoses in the reported literature were dismal. This is because there have been few therapeutic options after a tumor grows following administration of radiation-based therapy. In addition, a few patients received chemotherapy, and the received chemotherapy regimens were limited. Recently, as a new effort, Chamberlain et al.^5,6) aggressively treated with concurrent systemic chemotherapy (temozolomide, irinotecan, interferon-α, cyclophosphamide, and bevacizumab) and regional chemotherapy (intraventricular liposomal cytosine arabinoside, busulfan, and thiotepa). Although there was a short median survival^5,6), 3 patients with disease stability were alive at the time of last follow-up and one patient's survival was 39 months. This finding revealed a possibility for more effective chemotherapy for a patient with LD. Also, Pradat et al.³⁸⁾ demonstrated that some patients with meningeal gliomatosis may benefit from a concurrent radiotherapy to symptomatic areas and chemotherapy with agents that cross the blood-brain barrier or are delivered directly into the CSF. In the present cases, the above treatments were not administered because the patients' condition had deteriorated rapidly. However, in the future, an effective combination of radiotherapy and chemotherapy should be considered for meningiomas with LD through CSF.

Intracranial meningioma with LD is extremely rare. And, the pathogenesis for LD of meningioma remains unknown. Although our study does not demonstrate statistically significant conclusions, we generated two hypotheses based on the presented cases and a literature review. Firstly, intraventricular location and histologically aggressive features of meningiomas may be the predisposing factors for LD. Secondly, although the reported prognoses of LD of meningioma were poor, aggressively adjuvant chemotherapy or effective combination of chemo- and radiotherapy should be considered.