Papillary meningioma (PM) is a World Health Organization (WHO) grade III tumor defined histologically by a perivascular pseudopapillary growth pattern across most of the tumor (> 50%) [10]. A papillary growth pattern in meningiomas has been associated with brain invasion and aggressive clinical behavior [2, 8, 9, 12]. The standard treatment of PM is surgical resection followed by radiation. However, most patients develop recurrences, and metastatic disease is common, particularly to the lung [9, 19]. Another WHO grade III meningioma is the rhabdoid subtype which often harbors mutations in BAP1 [14, 15]. Interestingly, some meningiomas have cells with rhabdoid cytomorphology arranged in a papillary architecture suggesting a potential molecular and genetic link between the papillary and rhabdoid histologic subtypes of meningioma [8, 14]. The genetic alterations associated with PM remain unclear. Major obstacles to molecular characterization include low incidence of tumor, scarcity of tumor tissues available for genomic analyses, and the presence of artifactual pseudo-papillary features in some meningiomas which thereby confound cohorts [2]. To overcome these challenges and identify potentially recurrent somatic alterations in PM subtype of meningiomas, we mined data collected as part of our clinical comprehensive genomic profiling (CGP) initiative which has to date analyzed 8 PM (> 50% papillary morphology) and 22 meningiomas with focal papillary features (10–50%) amongst over 500 additional meningiomas of other subtypes.

The samples were analyzed in a CAP/CLIA-accredited laboratory (Foundation Medicine, Cambridge, MA). Approval for this study, including a waiver of informed consent and a HIPAA waiver of authorization, was obtained from the Western Institutional Review Board (Protocol no. 20152817). Three board-certified neuropathologists (E.A.W., S.H.R., and S.S.) confirmed the pathologic diagnosis of each case on routine hematoxylin and eosin-stained slides. DNA was extracted from 40-μm-thick paraffin-embedded sections, and CGP was performed on hybridization-captured, adaptor ligation-based libraries to a mean coverage depth of > 650 × for 236 or 315 genes plus the introns from 19 or 28 genes frequently involved in cancer. TMB (tumor mutational burden) was determined on up to 1.14 Mb of sequenced DNA using an estimation algorithm that extrapolates to the genome as a whole [5,6,7].

In our cohort of eight PMs, we identified three cases with inactivation of PBRM1: two cases with a truncating mutation in PBRM1 and one with homozygous deletion of PBRM1 (Table 1). Of the 22 meningiomas with only focal papillary features, 8 cases were PBRM1-mutant. Thus, 11 of 30 cases with at least focal papillary morphology (10–50% of papillary features in the H&E sections) had inactivation of PBRM1 (Supplementary Table 1). In the entire cohort of 562 available meningiomas that represents a general population of all WHO grades, we identified five additional cases with inactivating alterations in PBRM1 that did not display overt papillary morphology in the H&E sections available for analysis. Thus, 11 of 16 PBRM1-mutant cases (69%) occurred in meningioma with papillary histologic features, supporting a significant association between papillary features and PBRM1 mutation (p < 0.0001). Among the 16 PBRM1-mutant cases (2.8% of cohort), the detected PBRM1 alterations included six intragenic deletions, four frame-shifting insertions, four frame-shifting deletions, and two truncating mutations (Fig. 1). All showed biallelic inactivation by SNP array analysis and mutant allele read count data analysis. Median TMB was 2.1 mutations/Mb and all cases were microsatellite stable. Representative histopathology of PBRM1-mutant meningiomas is included in Fig. 2. The majority of PBRM1-mutant meningiomas occurred in female patients (n = 10/16, 62.5%), and median age was 51 years. Most cases were located supratentorially (n = 10). Additional characteristics of PBRM1-mutant meningiomas are shown in Table 1. A notable feature of our cohort was the frequent overlap of PBRM1 mutation with mutations in BAP1 (n = 5). Three of these five cases displayed papillary features while two displayed rhabdoid features. An association between BAP1 mutation and rhabdoid histology has been previously described [14]. This association was confirmed in our cohort of 562 meningiomas, in which 13 of 17 cases that were BAP1-mutant/PBRM1-wt had rhabdoid features. Among the 19 PBRM1-wt meningiomas with papillary histology, two had mutations in BAP1, consistent with prior reports of rare BAP1-mutant cases with papillary morphology [14, 18]. Notably, meningiomas that were BAP1-wt/PBRM1-mutant frequently had papillary morphology (7 of 11). Our findings suggest that BAP1 mutations tend to occur in rhabdoid meningiomas whereas PBRM1 mutations tend to occur in papillary meningiomas, although genetic and histologic overlap is noted.

Table 1 Location, histology, and molecular characteristics of PBRM1-mutant meningioma
Full size table
Fig. 1
figure 1

Schematic diagram of PBRM1 truncating mutations identified in meningioma cases. Six additional cases were identified with two copy-number loss of PBRM1

Full size image
Fig. 2
figure 2

PBRM1-mutant meningioma histopathologic features. a Examination reveals tumor cells arranged in a papillary pattern (H&E 100x). b Higher power image shows fragmentation of tissue architecture with the preservation of perivascular tumor cells with cytoplasm tapering towards a perivascular nuclear-free region (H&E 400x)

Full size image

PBRM1 is a 37-exon gene residing on chromosome 3p21, adjacent to BAP1, separated by ~ 0.135 megabase pairs. PBRM1 encodes the BAF180 protein, the chromatin targeting subunit of the PBAF chromatin remodeling complex [17]. PBRM1 is a tumor suppressor gene, mutated in 40% of clear cell renal cell carcinoma (RCC), as well as a subset of papillary RCC and bladder carcinoma [3, 17]. Mutations in PBRM1 are most often truncations and result in loss of protein expression. Previous studies have illustrated a significant increase in cell proliferation and cell migration after knockdown of PBRM1 [20]. Recent work has also demonstrated that BAF180 is required for centromeric cohesion, and DNA damage in cells lacking PBRM1 results in dynamic chromosome instability [11]. It has been speculated that the latter results in the improved survival of a subset of patients with PBRM1-mutant clear cell RCC cohorts treated with programmed cell death 1 receptor (PD-1) inhibitor [11]. Although our findings suggest that PBRM1 mutation is uncommon in meningioma, the inclusion of patients with this defined genetic subtype of meningioma in similar trials may reveal new therapeutic approaches. Mutations in other genes encoding components of the BAF complex, such as SMARCB1, SMARCE1, and ARID1A, have been previously reported in aggressive meningiomas [1, 4, 13, 16]. Dysregulation of chromatin remodeling is often identified in higher grade meningiomas, regardless of the molecular subgroup. It is postulated that the mutations resulting in disruption of SWI/SNF chromatin remodeling complexes are present in at least 20% of all human cancers [3].

In conclusion, we identify the tumor suppressor gene PBRM1 as a recurrently altered gene in meningiomas with papillary histomorphology. Further investigational studies are needed to assess outcomes of PBRM1-mutant meningioma and to determine whether mutation is an independent negative prognostic biomarker.