Introduction

Meningiomas are the most common primary intracranial neoplasms and are usually treated by microsurgical resection and/or radiation therapy. Although the vast majority of meningiomas are characterized by benign biological behavior and therefore correspond to WHO grade I, about 20% of the tumors display frequent recurrences (50–90%) and increased mortality and are therefore classified as grade II and III lesions [11, 27]. Aside from the WHO grade of the tumor, the extent of tumor resection according to the Simpson classification system and, more significantly, the volume of the tumor left behind after microsurgery have been shown to correlate strongly with the risk of tumor recurrence [31, 32]. In fact, proximity to critical neurovascular structures and invasion of the adjacent bone and soft tissue can distinctly alter the extent of tumor resection and therefore directly impact the risk of postoperative tumor relapse. Hence, treatment of tumors refractory to microsurgery and irradiation is a key challenge during neuro-oncological care for meningioma patients, and further, e.g., chemotherapeutical options are urgently needed.

Over the last decades, a number of substances including cyclophosphamide-doxorubicin-vincristine chemotherapy, antiangiogenic drugs (e.g., bevacizumab), and tyrosine kinase inhibitors such as vatalanib, sunitinib, or trabectedin have been investigated in the treatment of meningiomas in mostly small and retrospective series, and only displayed limited effects on tumor control [11]. The efficacy of checkpoint inhibitors such as pembrolizumab (NCT03279692) or nivolumab (NCT02648997, NCT03173950) remains to be determined in currently ongoing clinical trials.

Decitabine (DCT, 5-aza-2’–deoxycytidine) is a demethylating agent commonly used in the treatment of hematopoietic malignancies (e.g., acute myeloid leukemia [16]) and also induces demethylation in soft tissue tumors such as sarcoma, which display both genetic and morphological similarities to meningiomas [3, 15, 19]. Recent in vitro analyses revealed a dose-dependent efficiency of DCT also in high-grade meningiomas, with specific DNA demethylation of several onco- or tumor suppressor genes (TRIM58, FAM84B, ELOVL2, MAL2, LMO3, DIO3), which have been hardly investigated in meningiomas yet [33]. Regarding the promising results from these in vitro analyses, we therefore (i) investigated the expression of the onco- and tumor suppressor genes differentially methylated by DCT in meningiomas in vivo and (ii) analyzed correlations with clinical and histological variables and prognosis.

Materials and methods

Patient selection

Clinical, radiological, and histological data were obtained from the institutional meningioma database, containing information of 1302 surgeries performed in our department between 1991 and 2018. For this study, 111 patients who underwent surgery for primary diagnosed intracranial grade I (N = 54, 49%) and II/III (N = 57, 51%) meningioma with complete information on age, sex, tumor location, extent of resection, and with a postoperative follow-up period of at least 60 months were selected. The collective was intendedly enriched for high-grade lesions to enable expression analyses in a sufficient number of patients.

Clinical, radiological, and histological data recovery

Clinical, radiological, and histological data recovery has been described previously in detail. Briefly, archives of the local Institute of Neuropathology were reviewed for all histopathologically confirmed meningiomas resected in our department between 1991 and 2018. Meningioma subtype and histopathological grading were diagnosed according to the current 2016 WHO criteria. Hence, brain invasion was diagnosed in case of “irregular, tongue-like protrusions of tumor cells infiltrating underlying parenchyma, without an intervening layer of leptomeninges” on hematoxylin and eosin and Elastica van Gieson-stained slides. Maximum safely achievable tumor resection was performed in all patients and intraoperatively quantified according to the Simpson classification, as it is standard in our institution. Adjuvant irradiation was administered for primary diagnosed grade III and recurrent or subtotally resected grade II tumors as well as in benign lesions following simple debulking. Preoperative physical state was quantified according to the Karnofsky Performance Score (KPS). Tumor location was obtained from preoperative imaging and classified as described below.

Patients were followed-up by magnetic resonance imaging (MRI) and physical examination 3 months after surgery. Contrast-enhanced CT scans were performed in case of any contraindications for MRI, and imaging and examinations were repeated in 12- and 6-month intervals for benign and high-grade meningiomas, respectively. After 5 years of an event-free course, follow-up was repeated in bi-annual and annual intervals in grade I and II/III lesions, respectively. Imaging was evaluated by a team of two independent observers (at least one neurosurgeon and one (neuro-)radiologist) and progression or recurrence was diagnosed in case of any detected tumor growth, independent of the indication for subsequent therapy. Progression-free interval (PFI) was calculated from the date of surgery to the date of radiologically confirmed tumor progression or, in case of an event-free follow-up, to the date of the last follow-up.

Expression analyses—immunohistochemistry

Expression of TRIM58, FAM84B, ELOVL2, MAL2, and LMO3 was analyzed by immunohistochemical staining. Hence, formalin-fixed, paraffin-embedded Sects. (3–4 µm) were deparaffinized and rehydrated through a graded alcohol series according to standard protocols. Antigens were retrieved in sodium citrate buffer (pH = 6.0 Target Retrieval Solution S2369 DAKO; Agilent Technologies, Inc., Santa Clara, CA, USA, 1:10 diluted in distilled water) by the method of heat activation (40-min steam cooker, 20 min cooled by room temperature). Staining was performed using an Agilent Autostainer Link 48 with the DCS DetectionLine (CEA1706) Kit. The sections were stained with DAB (3,3-diaminobenzidine)-Chromogene (DC135C006 DCS) and then counterstained with hematoxylin, then dehydrated in an ascending series of alcohols (70%,96%,99%, Xylol), and finally sealed with Eukit and a coverslip for microscopic evaluation (Olympus BX-51). Table 1 specifies the applied antibodies and dilutions. Expression was analyzed by a team of two independent observers (JC and CT) and, for statistical reasons, quantified according to scores by two independent observers as follows. As technical negative controls, staining was performed without the primary antibody, while vascular endothelium served as biological negative controls in all cases. For quantification of oncogene expression, previously published scoring systems were chosen whenever available.

Table 1 Summarization of the used antibodies. The table reveals the analyzed antigen, the manufacturer of the antibody with the corresponding order number, the applied dilution for staining, and the host species
Full size table

For TRIM58, staining intensity was classified as 0 points (negative), 4 points (weak intensity), 8 points (moderate intensity), or 12 points (strong intensity). Staining density was classified based on the percentage of cells stained as 0 points (0%), 4 points (1–25%), 8 points (26–50%), 10 points (51–75%), or 12 points (> 75%). The final score was calculated as the sum of the intensity score and the density score as reported previously [23]. MAL2 expression was dichotomously registered as absent (< 5% of immunopositive tumor cells) or present in five separate 400 × high-power microscopic fields. Expression intensity of FAM84B was quantified as 0 (negative), 1 (weakly positive), 2 (moderately positive), and 3 (strongly positive), and the percentage of positive cells was classified as 0 (negative–10%), 1 (11–25%), 2 (25–50%), and 3 (> 50%). The final score was then calculated by multiplication of the staining intensity with the percentage of positive staining cells according to previous descriptions [37]. For LMO3, staining intensity was quantified as absent (1), weak (2), moderate (3), or strong (4), and the percent of membranous and cytoplasmic staining in tumor cells was classified as 1 (0–25%), 2 (26–50%), 3 (51–75%), and 4 (76–100%). The final score was calculated by multiplication of the staining intensity with density [24]. For ELOVL2, staining intensity was evaluated in absent (0), weak (1), moderate (2), and strong (3), as the percentage of positively stained cells was subdivided in 1 (1–25%), 2 (26–50%), 3 (50–75%), and 4 (76–100%). The final score was calculated by multiplication of staining intensity and percentage of positively stained cells.

Expression analyses—qRT-PCR

As a reliable antibody against DIO3 was not available, we performed quantitative real-time PCR. Fifteen frozen samples (10 × grade I, 5 × grade II/II) were used for RNA extraction (Maxwell 16 simplyRNA Tissue Kit, Promega). QRT-PCR was performed using commercial TaqMan Assays (DIO3: Hs00956431_s1 and GAPDH: Hs02786624_g1 as housekeeping gene) in a StepOne Plus (Applied Biosystems) as described by the manufacturer. Results were normalized to the DIO3 expression of the meningioma cell line Ben-Men-1, that was set to 1.

Statistical analyses

All calculations were performed using standard commercial statistic software (IBM SPSS Statistics, Version 28, IBM, Germany). Data are described by standard statistics with median and range for continuous and absolute and relative frequencies for categorical variables. For statistical reasons, the tumor location was classified as “convexity/parasagittal” vs “skull base”. Similarly, the extent of resection was dichotomously registered as gross (GTR, Simpson grades I–III) and subtotal resection (STR, Simpson grades IV and V). In univariate analyses, correlations between categorical and continuous variables were investigated by Fisher’s exact and Mann–Whitney-U tests, respectively. Distribution of PFS was visualized by Kaplan–Meier plots and compared by Log-rank tests. Multivariate analyses were performed using the Mantel-Cox test and backward Wald logistic regression and characterized by hazard (HR), 95%-confidence intervals (CI), and Wald-test p-values. The following variables were tested in multivariate regression models (ref = reference): age, sex (male (ref) vs. female), WHO-grade (classified into grade I (ref) vs. high-grade, II/III), tumor location (classified as described above, “convexity/parasagittal” = ref), and degree of resection (classified into GTR (ref) vs. STR). A p-value of < 0.05 was considered statistically significant throughout the whole analyses. All reported p-values are two-sided. Data collection and scientific use were approved by the local ethics committee and approved by the patients in each single case (Münster 2007–420-f-S and Münster 2018–061-f-S).

Results

Clinical and histological characteristics

Table 2 summarizes baseline clinical, radiological, and histopathological data. As expected, the extent of resection strongly correlated with tumor location, and GTR was more commonly achieved in convexity/parasagittal than in skull base tumors (N = 48 of 59, 81% vs N = 30 of 52, 57%; p = 0.007). Within a median follow-up of 79 months (mean: 109 months, range: 60–284 months), tumor recurrence was observed in 44 cases (40%) and occurred in 32 of 57 high-grade but in 12 of 54 benign meningiomas (56% vs 22%, p < 0.001). No correlations between the extent of resection (p = 0.210) or tumor location (p = 0.083) and recurrence were found. Multivariate analyses adjusted for patients’ age, sex, tumor location, and extent of resection confirmed high-grade histology as the only independent predictor of tumor recurrence (HR: 2.30, 95%CI 1.17–4.52; p = 0.016).

Table 2 Baseline clinical and histopathological data. Information was available in almost all included patients (KPS, Karnofsky Performance Score)
Full size table

Oncogene expression and correlation with clinical and histological variables

Immunohistochemical staining revealed a distinct expression of all analyzed oncogenes in the majority of tumor samples (Fig. 1). On visual inspection, expression was cytoplasmic in TRIM58 and FAM84B and detected both in the nucleus and the cytoplasm in MAL2 and ELOVL2 and LMO3. Correlations between expression scores and clinical variables were mostly lacking and are summarized in Table 3. However, distinct relations between histology and expression were found.

Fig. 1
figure 1

Representative images from immunohistochemical staining. Expression of the analyzed oncogenes TRIM58 (a), FAM84B (b), ELOVL2 (c), MAL2 (d), and LMO3 (e) was detected in most tumors with variable density and intensity. For illustration, samples with strong expression of all onco-/tumor suppressor genes were selected (magnification 200-fold, corresponding antibodies summarized in Table 1)

Full size image
Table 3 Correlation of expression of TRIM58, FAM84B, ELOVL, or MAL2 with clinical variables (p-values). Although reaching level of statistical significance, median FAM84B expression was only slightly higher in males than in females (6, range 1–9 vs 4, range 0–9; p = .021). No further correlations were found
Full size table

In 101 samples successfully stained for TRIM58 (91%), the median expression score was 20 and ranged from 0 (N = 9) to 24 (N = 19). Although median expression scores were 20 in both high-grade and benign meningiomas, the Kruskal–Wallis test revealed a statistically significant correlation between histology and TRIM58 expression (p = 0.034, Fig. 2A). In fact, mean expression score was 20 (SD ± 4) in benign and 16 (± 8) in high-grade meningiomas (p = 0.002). Correspondingly, all 9 samples lacking TRIM58 expression displayed high-grade histology. Immunohistochemistry for FAM84B was successful in 97 cases (87%) and displayed a median expression score of 6 (range: 0–9). Expression was slightly higher in males than in females (mean score 6, range 1–9, vs 4, range: 0–9; p = 0.009). Moreover, median FAM84B expression scores were increased in high-grade (6, range 0–9) as compared to WHO grade I meningiomas (4, range 0–9; p ≤ 0.001, Fig. 2B). Similarly, immunohistochemistry for ELOVL2 showed expression in all analyzed samples (N = 103) with a median score of 8 (range: 2–12), and expression was higher in grade II/III (9, range: 2–12) than in grade I tumors (6, range 2–12; p < 0.001, Fig. 2C).

Fig. 2
figure 2

Box and whisker plots illustrating correlations of the TRIM58, FAM84B, and ELOVL2 expression with histology. Although median expression score was 20 in both groups, the Kruskal–Wallis test and the corresponding plot were suggestive for higher TRIM58 expression levels in grade I as compared to grade II/III meningiomas (p = .034, A), and mean expression score higher in grade I than in high-grade lesions (20 vs 16, p = .002, indicated with x). In contrast, both median FAM84B (6, range 0–9 vs 4, range 0–9; p ≤ .001, B) and ELOVL2 (9, range: 2–12 vs 6, range: 2–12; p < .001, C). Expression scores were higher in grade II/III than in benign meningiomas. The boxes indicate upper and lower 25% quartile, the whiskers the minimum and maximum value, the dots the outliers, the asterisks the extreme values, and the heavy horizontal line indicates the median

Full size image

As previous reports already demonstrated expression in meningiomas and correlation with histology by microarray analyses and gene arrays, immunohistochemical slides for MAL2 [10] and LMO3 [30] were subjected to interim analyses. For MAL2, all 52 analyzed cases including 32 benign and 20 high-grade meningiomas displayed immunopositivity with strong expression (median 6, range 1–12) in most (N = 45) cases. In samples from six grade I and five grade II/III meningiomas subjected to LMO3 immunohistochemistry, expression was strong in all samples (median score 12, range 4–16) and no further staining was performed.

qRT-PCR showed a median relative expression of DIO3 of 140.15 (range: 3.37–10,286.51), which was distinctly higher as compared to the decitabine-resistant reference cell line Ben-Men 1, in all samples (suppl Fig. 1). Statistical analyses revealed a brought range but similar median expression values in (N = 9) grade I as compared to (N = 6) high-grade meningiomas (140.15, range 3.38–3572.39 vs 263.56, range: 9.65–10,286.51; p = 0.556).

Correlation of oncogenes with recurrence

Correlations between recurrence and TRIM58, FAM84B, and ELOVL2 expression were analyzed in univariate analyses as well as in multivariate tests adjusted for age, sex, tumor location, extent of resection, and, most notably, histology (Table 4). For statistical reasons, expression scores were dichotomized into < vs ≥ median score of each oncogene. Here, an increased ELOVL2 expression (score ≥ 8) was identified as a strong risk factor for tumor relapse in both uni- (HR: 2.42, 95%CI 1.18–4.94; p = 0.015) and multivariate (HR: 2.09, 95%CI 1.01–4.44; p = 0.046) analyses (Fig. 3). TRIM58 expression tended to correlate with recurrence in multi- (HR: 1.86, 95%CI 1.00–3.52; p = 0.056) but not in univariate analyses (HR: 1.74, 95%CI 0.92–3.29; p = 0.086), but without reaching the level of statistical significance. No further correlations between prognosis and the analyzed oncogenes were found.

Table 4 Correlations between the analyzed oncogenes and recurrence. ELOVL2 expression above median was found a strong risk factor for tumor relapse in both uni- and multivariate analyses
Full size table
Fig. 3
figure 3

Correlation between progression-free survival and expression of TRIM58 and ELOVL2. Increased expression of ELOVL2 was correlated with shorter progression-free survival (p = .012, Log-rank test)

Full size image

Discussion

The treatment of meningiomas refractory to surgery and/or irradiation remains challenging during neuro-oncological care. Recently updated treatment guidelines again reported limited efficacy of chemotherapy, hence underlining the urgent need for further laboratory and clinical research [12].

DCT is a DNA methyl transferase-inhibitor commonly applied in the treatment of AML. In vitro analyses have also shown efficiency in further malignancies and, noteworthy, a tumor-specific DNA demethylation [14]. In meningiomas, we could recently demonstrate a distinct and dose-dependent reduction of viability and proliferation in malignant, but not in benign meningioma cell lines following exposition to DCT. Remarkably, genome-wide DNA methylation analyses following drug exposition showed a specific DNA demethylation in DCT-sensitive, but not in DCT-refractory meningioma cells, hence suggesting molecular alterations underlying DCT sensitivity. In fact, demethylated regions included promoter regions of six tumor suppressor/ oncogenes (TRIM58, FAM84B, ELOVL2, MAL2, LMO3, DIO3) [33], whose expression in vivo is further characterized in the present study.

Expression of TRIM58 was found in the majority of analyzed meningioma samples. TRIM58 belongs to the tripartite motif protein (TRIM) family of E3 ubiquitin ligases and is considered a candidate tumor suppressor. Aberrant gene methylation of TRIM58 has been shown in several malignancies including liver [28], lung, and colorectal cancer, induces silencing [18], and is associated with poor prognosis [23]. Correspondingly, expression in our study was higher in grade I than in grade II/III tumors and the only tumors lacking TRIM58 immunopositivity were grade II/III lesions. Correlations with recurrence were rather ambiguous and lacking statistical significance.

FAM84B (Family With Sequence Similarity 84 Member B) is considered an oncogene which has been described in pancreatic [37], gastric [38], prostate [35], and esophageal [6] cancer, and has also been shown to correlate with tumor progression [6, 35]. In our study, FAM84B expression was found in all meningioma samples and was increased in high-grade as compared to benign tumors. However, no correlation with prognosis was found.

ELOVL2 (Elongation Of Very Long Chain Fatty Acids Protein 2) is widely considered a biomarker for aging and silencing by DNA methylation [5, 9] and has been occasionally described in the context of oncogenesis, e.g., in breast and renal cell cancer or neuroblastoma [8, 17, 34]. Remarkably, ELOVL2 has been proposed as both a tumor suppressor [8, 17] and, vice versa, a proto-oncogene [34]. As the first study so far, we revealed a brisk ELOVL2 expression in meningiomas, which was additionally increased in high-grade as compared to benign lesions. ELOVL2 expression was also associated with a > twofold risk of tumor relapse independent of the WHO grade of the tumor.

DIO3 (Iodothyronine Deiodinase 3) plays a major role during embryogenesis and has also been described to promote cancer development by inhibiting tumor-suppressive actions of thyroid hormone T3 in several malignancies such as ovarian [26], lung [25], or prostate [13] cancer. Given the lack of a sufficient antibody for immunohistochemistry, we analyzed the expression in our series by RT-PCR. Here, normalized for DCT-resistant Ben-Men 1 cells, DIO3 expression was distinctly increased in all samples but independent of the WHO grade.

LMO3 (LIM domain only protein 3) is a transcription co-factor interacting with p53 [20] and considered an oncogene, e.g., in neuroblastoma [1]. MAL2 (Mal, T Cell Differentiation Protein 2), a transmembrane protein of the MAL proteolipid family, is upregulated in a number of malignancies, such as breast, colorectal, pancreatic, or ovarian cancer, has been shown to correlate with invasion and worse prognosis [2, 4, 22, 36]. In meningiomas, three previous studies analyzed MAL2 and LMO3 expression. In contrast to studies reporting the promotion of invasion and proliferation by LMO3 in gastric and hepatocellular carcinoma [7, 29], Serna et al. showed a decreased expression in biologically aggressive meningiomas [30]. Our series basically confirmed the brisk expression of LMO3 in meningiomas, while immunopositivity hardly varied in exploratory analyses and correlations with histology or prognosis were not investigated. For MAL2, noteworthy, previous studies reported promotor hypermethylation and downregulation in high-grade and recurrent meningiomas [10, 21]. Our study confirmed a brisk expression, while correlations with prognosis or histology were not found.

The authors are aware of some limitations of the study. The small sample size limits transferability and may lead to selection bias. Although clinically and histopathologically well-characterized, molecular information such as TERT promotor mutation status or DNA methylation classes of the patient collective were not available. Due to methodology, immunohistochemical staining only enables semi-quantitative analyses. Hence, while providing important exploratory analyses, determination of the role of the reported oncogenes for tumorigenesis in meningiomas remains to be further determined, e.g., by RNA sequencing or knock-out models.

In conclusion, this is the first study reporting the expression of the oncogenes TRIM58, FAM84B, ELOVL2, and DIO3 in meningiomas and correlations with histology and prognosis. Moreover, we confirmed a brisk expression of MAL2 and LMO3, underlining the importance during tumorigenesis. Previous molecular analyses already revealed the regulation of some of these oncogenes by DNA hypermethylation. Hence, these results further explain the efficiency of DCT in high-grade meningiomas in vitro and strongly encourage future in vitro and, potentially, in situ investigations.