Patients were identified for study through a review of the clinical records of the Department of Neurosurgery, Huashan Hospital of Fudan University, Shanghai, China. A total of 172 HGMs (Supplementary Table S1) were included following study approval by the Human Subjects Institutional Review Board at Huashan Hospital, Fudan University (KY-2017-09). Clinical characteristics including age, gender, tumor location, extent of surgical resection, and outcome data were collected. In general, adjuvant radiation was recommended to both atypical and anaplastic meningioma patients, regardless of gross total resection (GTR; Simpson grades I–III) or subtotal resection (STR; Simpson grades IV–V). And the final decision was made based on the negotiation with the relatives of patients. The details of postoperative RT were described in our previous work (5). Tumor pathological subtypes were reconfirmed by at least two experienced neuropathologists. Follow-up was conducted routinely according to the guidelines of Huashan Neurosurgical Center. Written informed consent was obtained from all patients involved in our study.

A total of 172 patients with a confirmed diagnosis of HGM who met inclusion criteria (141 WHO grade 2 meningiomas and 31 WHO grade 3 meningiomas; Supplementary Table S1). Out of 172 cases, 87 cases received RT after surgery. Among the 87 patients, 42 (48.3%) were de novo meningiomas, while the remaining 45 patients presented with recurrent meningioma following prior surgical resection. And finally, 37 de novo meningioma samples with adequate quality of DNA concentration were included for further next-generation sequencing.

Tumor genotyping was performed on formalin-fixed paraffin-embedded (FFPE) tumor tissue by next-generation sequencing (NGS) covering 184 genes, including common pathological relevant genes of meningiomas (Supplementary Table S2) (13–21, 23). Five DNA samples were excluded for sequencing due to inadequate quality of concentration. High-throughput sequencing was performed on Illumina miniseq platform by KuoRan Biomedical Technology as previously described (Supplementary Material) (24).

The TERT promoter mutations were evaluated using Sanger sequencing. Genomic DNA was obtained from FFPE using the HiPure FFPE DNA Kit (Magen, D3126-03) following polymerase chain reaction-based amplification of the target region (forward primer: GGATTCGCGGGC ACAGAC; reverse primer: CAGCGCTGCCTGAAA CTC; details on PCR conditions are available upon request).

The specific details of statistical tests are included in the figure legends. A two-tailed Fisher’s exact test was used to calculate statistical significance between different groups using a χ² 2 × 2 table. Categorical variables were compared with the Fisher’s exact tests, and continuous variables with the independent-samples Student’s t-test (data with normal distribution) or Mann–Whitney U-test (data with skewed distribution). Continuous data were expressed as the mean ± standard deviation (SD). Overall survival (OS) and progression-free survival (PFS) curves were plotted using the Kaplan–Meier method. Statistical analysis was performed using Statistical Package for Social Sciences (SPSS, Version 20.0, Chicago, IL, USA). Data were considered to be significant when p < 0.05.

We identified 172 patients with a confirmed diagnosis of HGM who met inclusion criteria (141 WHO grade 2 meningiomas and 31 WHO grade 3 meningiomas; Supplementary Table S1). Out of 172 cases, 87 cases received RT after surgery, including external beam radiation therapy (EBRT), stereotactic radiosurgery (SRS), and cyber knife (CK). Among the 87 patients, 42 (48.3%) were de novo meningiomas while the remaining 45 patients presented with recurrent meningioma following prior surgical resection. Five of the 42 (11.9%) meningiomas were excluded for sequencing due to inadequate quality of DNA concentration. Thus, tumor genotyping covering 184 genes was performed on the 37 de novo meningioma cases (Supplementary Tables S2, S3). In total of the 37 cases, 23 males (62%) and 14 females (38%) with a median age of 45 years (range: 34–73) harboring 31 atypical (83.8%), five anaplastic (13.5%), and one atypical/chordoid coexisting (2.7%) meningiomas were included. Tumors were located at the convexity (n = 17, 45.9%), the falx/parasagittal (n = 13, 35.1%), the skull base (n = 5, 13.5%), or in other locations (n = 2, 5.5%). Among these patients, 28 (75.7%) were treated with EBRT, six (16.2%) were treated with SRS, and three cases were treated with CK (8.1%) (Figure 1).

In order to identify tumor mutations associated with efficacy of postoperative RT, we determined the association between recurrent mutations and tumor progression in the cohort of 37 patients (Figure 1). Of the 37 de novo HGMs following RT, 19 cases (51.4%) had tumor progression. Progression individuals was defined as tumor regrowth within the radiation field based on the criterion of Response Assessment in Neuro-Oncology Working Group (RANO) radiologic criteria for meningiomas (25). Part of the genomic mutational landscape of 37 patients with de novo meningioma is displayed in Figure 2. Nine mutations met our predetermined recurrence frequency threshold of >20% (Figure 2 and Supplementary Table S2). Consistent with previous studies, high mutational rates of NF2 (n = 22; 59%) were discovered in this cohort. Additional common pathological relevant genes of meningiomas, including AKT1 (n = 3; 8%), CDKN2A (n = 2; 5%), SMO (n = 0; 0%), SUFU (n = 0; 0%), POLR2A (n = 6; 16%), TRAF7 (n = 1; 3%), and SMARCB1 (n = 2; 5%), were observed as well (Supplementary Table S2). Besides, the most frequently altered genes including ATRX (n = 13; 35%), ARID1A (n = 11; 30%), ATM (n = 11; 30%), NF1 (n = 11; 30%), ROS1 (n = 10; 27%), KDM6A (n = 9; 24%), FAT1 (n = 8; 22%), and TERT (n = 8; 22%) were observed (Figure 2).

Strikingly, only TERT mutations were significantly associated with tumor progression (n = 8, adjusted p = 0.031), and all these mutations were present in tumors that progressed after RT (Figure 3A). Of the TERT mutant cases, 87.5% (7/8) presented with TERT promoter mutations (62.5% C228T variant and 25% C250T variant; Supplementary Figure S1), and 12.5% (1/8) harbored copy number amplification (CN amp.). As many tumor driver genes are co-occurring or show strong exclusiveness in their mutation pattern, we next explored the potential different somatic interactions in the cohort. None of the gene mutations presents co-occurring or mutually exclusive in HGM cases (Supplementary Figure S2). Thus, TERT mutations appear to be the dominant cause of tumor progression among mutations in this cohort. Furthermore, we found that TERT mutation status (p = 0.003), as well as extent of resection (EOR) (p < 0.001), was significantly associated with tumor progression by Fisher’s exact tests (Table 1). Neither tumor location (p = 1.00) nor WHO grade (p = 1.00) predicted tumor progression after postoperative RT. After adjustment for multiple comparisons, these associations were not significant. Importantly, TERT mutation cases were more likely to have unfavorable time to recurrence (TTR) over the entire cohort, with a mean observation time of 47.6 months (Figure 3B). However, no significant differences in the average time to recurrence between TERT mutant and wild-type cases were observed (Wilcoxon rank sum test with continuity correction W = 491, p = 0.0864).

Meningioma with TERT alterations, regardless of WHO grades or pathological subtypes, had a highly significant risk of recurrence (26). To exclude the disruption to RT efficacy might be caused by TERT-related malignant biological behavior, we thus performed analysis on newly diagnosed HGMs depending on TERT alterations only in progression group. Of the 19 cases in the progression group, mitotic index (ki-67%) depending on TERT alterations did not show any significantly difference (unpaired t-test, p = 0.051; Supplementary Figure S3A). Additionally, TERT alterations of de novo HGMs had no predictive effect on tumor recurrence in progression group following postoperative RT (p = 0.074 with log-rank test; Supplementary Figure S3B).

With these findings, we next analyzed the effect of TERT mutation status on PFS in patients with de novo HGMs after RT. The primary endpoint of PFS, defined as time from surgery to date of progression, was assessed on the basis of progression of meningioma after initial surgery on imaging follow-up. The median PFS of 75 months (range: 7–109 months) was observed for the entire cohort. Subgroup analysis revealed a median PFS of 25 months (range: 10–79 months) in TERT mutant group and 77 months (range: 7–109 months) in TERT wild-type group. TERT mutants were significantly associated with decreased PFS in de novo HGM cases that underwent postoperative RT (p = 0.0114 with log-rank test; Figure 4A).

The relationship between TERT mutations and OS was analyzed using the log-rank test (Figure 4B). The median OS of 84 months (range: 13–123 months) was observed for the entire cohort. Subgroup analysis revealed a median OS of 66 months (range: 13–123 months) in TERT mutant group and 85 months (range: 24–109 months) in TERT wild-type group. Patients with de novo HGMs after RT harboring TERT mutations had worse OS (p = 0.0562; Figure 4B).