The Cytogenetic Relationship between Primary and Recurrent Meningiomas Points to the Need for New Treatment Strategies in Cases at High Risk of Relapse

Meningiomas are usually considered as benign tumors. Nevertheless, recurrences occur in between 10% and 25% of cases undergoing a complete resection of the tumor (1–4) and they represent the major factor influencing patient's outcome (3, 5–7). In the last decade, important advances have been made as regards the identification of prognostic factors for predicting recurrence of meningiomas. Thus, the combination of malignant histology, younger age, and monosomy 14 is currently considered as the most relevant triad of adverse prognostic factors for recurrence of meningiomas (3, 4, 7–11). Despite this, recurrences are still observed in histologically benign tumors occurring in older patients who do not carry monosomy 14 (12–14). Because of this and the fact that recurrent meningiomas usually show identical histopathologic characteristics to those observed at diagnosis, the question remains about whether tumor recurrence, even many years after surgery, could be due to regrowth of the tumor, emergence of another meningioma, or both. In fact, among other additional prognostic factors, the extent of the surgical resection has been considered as an obvious prognostic factor because tumors that undergo a wider excision will be less likely to recur (6, 7, 12, 15, 16). In turn, the effectiveness of tumor resection could also be related to its localization (3, 6–8, 15, 17) although this still remains controversial (2, 18). Therefore, with the exception of those cases in which complete surgical resection of the tumor cannot be achieved, no definitive systematic demonstration of the exact relationship between diagnostic and recurrent tumors occurring in the same individual has thus far been obtained.

At present, it is well established that meningiomas are genetically heterogeneous tumors which usually show variable patterns of intratumoral clonal evolution due to chromosomal instability (1, 3, 4, 10, 18–25). Interestingly, previous reports suggest that different tumor cell clones could display a distinct distribution throughout the tumor (21, 26). Because of this, detailed analysis of the patterns of intratumoral clonal evolution in diagnostic and subsequent recurrent meningioma samples provides a unique tool for establishing a link between primary and recurrent tumor lesions in individual patients. Previous reports which have explored the chromosome abnormalities in meningioma patients studied at diagnosis and later on at recurrence (19, 27–29) have either included few cases (17, 28, 30, 31) or used conventional cytogenetic techniques which hamper the identification of all tumor cell clones present in a sample due to the relatively low percentage of neoplastic metaphases analyzed from all tumor cells (29). In turn, multicolor interphase fluorescence in situ (iFISH) analyses have not been systematically employed.

In the present study, based on a series of 157 patients, we compared the cytogenetic patterns obtained through iFISH analysis of primary versus recurrent meningiomas sequentially developing in the same individual. Our results indicate that the development of recurrent meningiomas after complete tumor resection is usually due to regrowth of the primary tumor and rarely due to the emergence of an unrelated meningioma. In these patients, tumor size together with the prognostic score proposed by Maillo et al. (8), based on age of patient, tumor histopathology, and the presence of monosomy 14, represented the best combination of independent prognostic factors for predicting recurrence-free survival.

Patients. A total of 157 consecutive meningioma patients who underwent a complete surgical resection according to the Simpson criteria (32) at the Neurosurgery Service of the University Hospital of Salamanca (Spain) were included in this study (Table 1). Follow-up controls, done with magnetic resonance imaging techniques, showed absence of any residual tumor lesion after surgery in all cases, including patients showing recurrence of the tumor. At the closing of this study, 25 patients had relapsed with a median follow-up of 91 ± 46 months (range, 25-181 months) versus 83 ± 49 months (range, 13-170 months) for the nonrecurrent tumors (P > 0.05). The overall number of recurrences observed for the 25 relapsed patients was 51, ranging between 1 and 6 recurrences per patient. From them, 12 patients (48%) were males and 13 (52%) were females with a mean age of 46 ± 19 years (range, 13-76 years); there was no family history of meningioma for any of the patients. Tumor localization at diagnosis for recurrent versus nonrecurrent meningioma patients was distributed as follows: cranial base including posterior fossa, 36% versus 33% of the cases; cerebral convexity, 28% versus 24%; parasagittal and tentorial, 20% versus 33%; ventricular, 12% versus 1%; and spinal, 4% versus 10% of the patients. Histologic diagnosis was established according to the WHO criteria (33). At diagnosis, 17 of 25 relapsed patients (68%) and 125 of the 132 nonrelapsed cases (94%) were classified as grade 1 or benign subtypes with a median relapse-free survival of 34 versus 84 months; 7 (28%) versus 6 (5%) were grade 2 or atypical with a median relapse-free survival of 21 versus 94 months; and 1 (4%) versus 1 (1%) were diagnosed as grade 3 or anaplastic tumors with a relapse-free survival of 31 versus 55 months.

From the prognostic point of view, all cases were classified according to the Maillo et al. (8) scoring system based on age of patient (<45, score = 1; ≥45, score = 0), tumor grade (grade 1, score = 0; grade 2 or 3, score = 1), and the presence (score 1) or absence (score 0) of monosomy 14.

Tumor specimens were obtained by conventional surgical procedures. Part of the tumor showing both macroscopic and microscopic infiltration was divided into two parts. One was fixed in formalin and embedded in paraffin using conventional procedures and the other was employed to prepare single-cell suspensions as previously described in detail (34).

iFISH studies. iFISH analyses were done in diagnostic tumor samples from all 150 patients. Sequential iFISH studies were done in a total of 53 samples (16 diagnostic and 37 relapse tumor samples) corresponding to 19 of the 25 patients studied. A minimum of two different samples was analyzed per patient. In the remaining six recurrent meningiomas, iFISH studies were done exclusively at diagnosis. In addition, to establish the potential existence of preferential tissue localization of the different clones present in a meningioma tumor sample, formalin-fixed, paraffin-embedded tissue sections from 47 samples corresponding to 37 patients with at least two distinguishable clones by iFISH were analyzed. In these latter studies, nine different tumor areas were screened per sample.

iFISH assays for the detection of numerical abnormalities of chromosomes 1, 7, 9, 10, 11, 14, 15, 17, 18, 22, X, and Y were done according to previously reported techniques (34) both on freshly obtained single-cell suspensions, after fixation in methanol/acetic (3:1, v/v), and on unstained 5-μm formalin-fixed, paraffin-embedded tissue sections placed on poly-l-lysine–coated slides (BioGenex, San Ramon, CA). The following commercially available probes, all obtained from Vysis, Inc. (Downers Grove, IL) except the 1p36 Midi-Satellite fluorescein-labeled probe which was purchased from Q-BIOgene (Carlsbad, CA), were systematically used in double stainings: for chromosomes 9 and 22, LSI BCR/ABL dual-color probe; for chromosomes 15 and 17, LSI PML/RAR-α dual-color probe; for chromosomes 11 and 14, LSI IgH/CCND1 dual-color probe; for chromosomes 14 and 18, LSI IgH/BCL2 dual-color probe; for chromosomes X and Y, CEP X DNA probe, conjugated with Spectrum Orange, and CEP Y DNA probe, conjugated with Spectrum Green; for chromosomes 7 and 10, CEP 7 DNA probe, conjugated with Spectrum Orange, and CEP 10 DNA probe, conjugated with Spectrum Green; for chromosome 1, both the 1p36 midi-satellite fluorescein-labeled probe and the CEP 1q12 DNA probe. Once hybridized, the cell nuclei were counterstained with 4,6-diamidino-2-phenylindole (0.1 μg/μL) and Vectashield (Vector Laboratories, Inc., Burlingame, CA) was used as antifading agent.

A BX60 fluorescence microscope (Olympus, Hamburg, Germany) equipped with a 100× oil objective was used for counting the number of hybridization spots per nuclei; for each slide, a minimum of 200 nuclei were counted. Only those spots with a similar size, intensity, and shape in nonoverlapping nuclei with a distinct nuclear border were evaluated; doublet signals were considered as single spots. The criteria used for the definition of the presence of numerical abnormalities for each individual chromosome, as well as for the definition of a tumor cell clone, were based on the analysis of normal interphase nuclei as previously described in detail (34).

Flow cytometric analysis of tumor cell DNA contents. In all patients studied, flow cytometric measurement of tumor cell DNA contents was done using aliquots of the same freshly frozen diagnostic tumor samples used for iFISH studies as described elsewhere (34, 35) for a minimum of 10⁴ events per sample. The presence of DNA aneuploidy was defined based on the identification of two G₀-G₁ populations of cells with different DNA contents. In DNA aneuploid cases, a second sample aliquot containing normal diploid control cells added to the tumor sample was used to confirm which of the two G₀-G₁ populations corresponded to the DNA aneuploid tumor cells. DNA index, as well as the cell cycle distribution of tumor cells, was calculated as previously described in detail after excluding cell doublets and debris (35).

Statistical methods. For all variables included in the present study, their mean and median values, SD, and range were calculated using the SPSS 11.0 software package (SPSS, Inc., Chicago, IL). The χ² and either the Mann-Whitney U or the Kruskal-Wallis test were used to assess the statistical significance of the differences observed between groups for qualitative and quantitative variables, respectively. Survival curves were plotted according to the method of Kaplan and Meier and the log-rank test was used for their comparison. Cox regression was used for the multivariate analysis of prognostic factors for relapse-free survival. In the multivariate analysis, only those variables showing a significant effect on relapse-free survival in the univariate study were included. Statistical significance was considered to be present when P < 0.05.

Chromosomal abnormalities in recurrent meningioma patients as detected by iFISH. iFISH analysis of tumor samples from those patients who relapsed showed variable cytogenetic patterns in the first tumor sample analyzed. Interestingly, only 8 cases (32%) showed either a normal diploid chromosome pattern (n = 5) or monosomy 22 associated with small tetraploid clones (n = 3). All other cases (n = 17; 68%) displayed complex iFISH patterns consisting of multiple chromosome aberrations (Tables 1 and 2). From these 17 cases, 3 (18%) showed one clone with simultaneous loss of between two and six different chromosomes, 7 (41%) displayed two clones, 3 (18%) had three different tumor cell clones, and in the remaining 4 cases (23%), four different tumor cell clones were identified. The exact abnormalities detected in each of these tumor cell clones are shown in Table 2.

Monitoring of chromosomal abnormalities in recurrent follow-up tumor samples. Interestingly, 14 of the 19 (74%) meningioma patients in whom sequential analysis of recurrent tumor samples was done showed tumor cell clones displaying identical cytogenetic aberrations by iFISH as those observed in the initial study. In three of the remaining five patients, at least one of the tumor cell clones identified in the first sample studied showed acquisition of one or more additional aberrations. The chromosomal aberrations acquired included (a) monosomy 14 in association or not with trisomy 9 and 22 and disappearance of the two initially predominant clones in one patient (case 20); (b) variations in two tetraploid clones with acquisition of monosomy X in a second individual (case 21); and (c) acquisition of additional chromosome 1 abnormalities in a tumor cell clone carrying −14, in association with the disappearance of two minor clones initially detected, in the third case (case 19). One additional patient showed emergence of two new clones: a tetraploid tumor cell clone which was not detected initially, together with acquisition of monosomy 10 and 14 in the most represented clone in the first study (case 7). The fifth case showed the disappearance of the most abundant tumor cell clones observed in the first sample studied (case 24; Table 2).

Tissue localization of different tumor cell clones. Only 1 of 47 samples analyzed showed the clear presence of different tumor cell clones at variable percentages in different areas of the tumor tissue (Fig. 1A and B). In all other tumor samples, simultaneous coexistence of all tumor cell clones was observed in all areas analyzed. However, it should be noted that nuclei from small tetraploid tumor cell clones tended to have larger nuclei and to predominate in areas showing lower cell densities.

Clinicobiological characteristics of recurrent versus nonrecurrent meningioma patients.Table 3 shows the clinical and biological characteristics of the 25 recurrent meningioma patients studied in comparison with a group of 132 patients diagnosed during the same period and who were disease-free by the end of the study. As illustrated, recurrence was associated with a higher frequency of males (48% versus 30%; P = 0.08) and younger patients (mean age, 46 ± 19 versus 61 ± 14 years; P < 0.001). Moreover, three of four ventricular tumors had recurrence whereas this was observed in only 1 of the 14 spinal meningiomas (P = 0.01). In comparison with the nonrecurrent tumor group, recurrent patients also showed larger tumors (average tumor size, 55 ± 19 versus 42 ± 16 mm; P = 0.003), a higher incidence of multifocal (20% versus 6%; P = 0.04), and histologically aggressive tumors (32% versus 4%; P < 0.001), as well as a greater percentage of cases with an adverse prognostic score (score 2 and 3) according to the classification of Maillo et al. (ref. 8; 36% versus 3%; P < 0.001). Regarding the frequency of numerical abnormalities for the individual chromosomes analyzed, recurrent meningiomas showed a greater proportion of cases with del(1p36) (P = 0.006), del(10q)/−10 (P = 0.05), del(14q)/−14 (P < 0.001), del(18q)/−18 (P = 0.01), gains of chromosomes 1q (P = 0.004), 7 (P = 0.02), 9 (P < 0.001), and 22 (P = 0.005) and of women with monosomy X (P = 0.005) with respect to nonrecurrent cases (Table 4). In addition, recurrent meningiomas showed a higher proportion of cases carrying three or more tumor cell clones by both iFISH and flow cytometry DNA analysis (P = 0.05 and P = 0.01, respectively). By contrast, recurrent meningiomas had diploid or −22 ancestral tumor cell clones less frequently than the nonrecurrent tumor group (32% versus 56%; P = 0.03; Table 3). As expected, a higher incidence of deaths (40% versus 3%; P < 0.001) and a lower overall survival were observed among recurrent cases (median overall survival, 255 months versus not reached; P = 0.0001). Multivariate analysis of prognostic factors for relapse-free survival showed that the prognostic score proposed by Maillo et al. (8), which takes into account tumor grade, age of patient, and chromosome 14 abnormalities (P = 0.003), together with tumor size (P = 0.003) was the best combination of independent variables for predicting patient outcome.

Recurrence is the major clinical complication of meningiomas occurring in up to one quarter of all patients undergoing curative surgery (1–4) as also found in our series. Despite the identification of predictive factors for tumor recurrence, its exact nature remains largely unknown. Accordingly, tumor recurrence after complete surgical resection could be due to the local persistence of tumor cells with clonogenic capacity because of tumor seeding during surgery, microinvasion of lymphatic vessels and other tissues, or an incomplete microscopic resection of the tumor. Alternatively, a new tumor, independent from the first meningioma, could develop in the same individual because of a favorable genetic and/or environmental background.

The aim of the present study was to provide evidence about the exact nature of recurrent meningiomas. To the best of our knowledge, this is the first report in which genetic evidence for a direct link between primary and recurrent meningiomas developing in the same individual is provided through the comparison of the tumor cell clones present in each tumor sample, as identified by multicolor iFISH. In line with our previous observations (8, 34), the systematic use of 13 probes specific for an identical number of chromosome regions commonly deleted or gained in meningiomas allowed the identification of two or more tumor cell clones within a tumor sample in the majority of the meningiomas, particularly in recurrent cases. Investigation of the distribution within a tumor sample of the different clones identified showed the admixture at different proportions of all tumor cell clones in most (46 of 47) samples analyzed. Apparently, these results are in contrast with those reported by Pfisterer et al. (26) who found different chromosomal abnormalities in different areas of the tumor. However, it should be noted that in those tumors showing a small fraction of tetraploid cells, which usually had a higher nuclear size, these cells displayed a tendency to predominate in areas with lower cell densities. Based on these results, the investigation of the persistence of the same tumor cell clones or the appearance of new related clones in recurrent meningiomas developed within the same individual after complete tumor resection could be a potentially useful tool to establish the relationship between primary and recurrent meningiomas. Interestingly, in most individuals analyzed, exactly the same tumor cell clones identified in the initial lesion were also detected in the recurrent meningiomas. Moreover, in the remaining cases, we observed tumor cell clones related to those observed in the primary tumor but which had acquired one or more additional chromosome aberrations related to either the appearance of new clones or the disappearance of the most representative clones from the primary lesions. Altogether, these results provide strong evidence for a direct relationship between primary tumors and recurrent lesions in virtually all meningiomas. Of note, in those cases showing the emergence of new chromosomal abnormalities in recurrent meningiomas, the presence of more complex cytogenetic iFISH patterns was observed, in keeping with the notion that tumor progression is associated with the sequential accumulation of additional genetic abnormalities, most often involving deletions of chromosomes 1p and monosomy 14 (1, 3, 8, 10, 21, 23, 24, 27, 29).

Interestingly, comparison between the initial tumors from individuals showing recurrence versus those from nonrecurrent patients showed a significantly higher frequency of cases displaying complex iFISH patterns and a greater number of tumor cell clones among the former group. This finding supports previous observations showing that the presence of specific chromosome abnormalities, such as monosomy 14, del(1p36), gains of chromosome 22, and complex karyotypes, but not monosomy 22, is associated with a worse clinical outcome, as reflected by both higher recurrence and shorter relapse-free survival rates (1, 8, 10, 19, 21, 36–38). However, none of the chromosomal abnormalities detected could on its own explain all recurrences. In fact, one quarter of all recurrent meningiomas studied showed a diploid or −22 ancestral tumor cell clones. In such cases, additional genetic abnormalities together with other disease features would probably explain the occurrence of tumor recurrence. In line with this, mutations of the NF2 gene at chromosome 22q12.2 have been reported in a variable percentage of cytogenetically diploid tumors, this molecular abnormality being not explored in our cases. In addition, some of these latter recurrent meningiomas corresponded to younger patients, another well-recognized adverse prognostic factor (8, 11).

Other clinical characteristics of meningioma patients which have been associated with recurrent tumors include tumor histology (3, 4, 7–10), patient gender (4, 6, 7, 20, 39, 40), tumor localization (3, 6–8, 15), and proliferation rate (6, 13, 41, 42). In the present study, recurrent meningiomas also occurred more frequently in histologically malignant tumors and among males whereas they showed similar proliferation rates to nonrecurrent meningiomas. In turn, most ventricular meningiomas relapsed whereas all, except one, spinal tumors did not. Although it could be speculated that some specific tumor localizations (e.g., ventricular) could be associated with more conservative surgical resection procedures than others (e.g., spinal), most ventricular tumors were histologically malignant, showed adverse cytogenetic features, and displayed an adverse prognostic score according to Maillo et al. (8) whereas all, except one, spinal meningiomas were histologically benign and displayed a diploid or −22 homogeneous tumor cell population. These findings could help to explain the different recurrence rates observed for tumors occurring at different localizations in the present study. In fact, multivariate analysis of prognostic factors showed that the prognostic score proposed by Maillo et al. (8), in which age of patient, tumor grade, and monosomy 14 are considered, together with tumor size was the best combination of independent variables for predicting tumor recurrence.

Interestingly, no differences were observed as regards the prognostic value of the Maillo et al. score once each score group was separately considered or when the two adverse prognostic factors (scores 2 and 3) were considered together (data not shown). Altogether, these results indicate that younger patients who have tumors with a large size, adverse histologic features, and/or monosomy 14 are more prone to develop recurrence which, according to our data, could be due to the persistence of tumor cells even after a macroscopically complete tumor resection. As expected (3, 5–7, 43), tumor recurrence was also associated with a significantly higher frequency of deaths and shorter overall survival rates, confirming that recurrence represents a major prognostic factor for overall survival in meningioma patients.

In summary, our results provide strong evidence about the existence of a direct relationship between primary and subsequent recurrent meningiomas in patients undergoing curative surgery, with younger patients who have larger tumors with adverse histologic features and/or monosomy 14 being more prone to develop tumor recurrence. These observations, together with the confirmed adverse prognostic effect of tumor recurrence on the overall survival of meningioma patients, underline the need for alternative therapeutic strategies in patients at high risk of relapse.