Risk Factors for Tumor Recurrence in Meningiomas: Multicenter Multivariate Analysis

Abstract

Meningiomas are among the most frequent primary brain tumors, so knowing the association of different factors with the imaging recurrence of meningiomas is important for post-surgical decision-making. We conducted a retrospective analysis in six high specialty hospitals in Mexico with 190 patients diagnosed with meningiomas and who underwent resection surgery from 2009 to 2019. A univariate and multivariate analysis was performed to identify different risk factors, a simple linear regression analysis for quantitative variables, and a comparison of means to identify the impact of the different factors with the months of survival free of tumor recurrence. A total of 190 patients were analyzed, with an overall recurrence rate of 38.9% (74 patients). Regarding risk factors, the onset of intracranial hypertension syndrome was associated with a higher risk of tumor recurrence at follow-up. The simple linear regression analysis found that the greater the intraoperative bleeding, the lower the survival rate of tumor recurrence. In the comparison analysis of means, the intracranial hypertension syndrome and the petroclival location significantly reduced the months of free survival of the tumor recurrence. We report that a clinical presentation such as intracranial hypertension syndrome increases the risk of recurrence in meningiomas up to eight times more and that the greater the intraoperative bleeding, the lower the survival free of recurrence.

Introduction

Meningiomas are one of the most common primary brain tumors, with a prevalence of 6.59 cases per 100,000 population [1]. Most meningiomas are histopathologically benign (up to 92.8%), only 2.2% are considered grade 2, and 5% are malignant (grade 3) [2]. Only a tiny subset of meningiomas follow an aggressive clinical course characterized by local recurrence and poor survival [3]. A series of tumor characteristics have been evaluated as possible predictors of recurrence, with special attention to histopathological findings; however, there are areas of uncertainty about the factors involved [4]. Many factors, such as the degree of surgical resection, tumor size, and proliferation rate, have been shown to affect tumor recurrence potentially [5]. The challenge in the study of meningiomas is that they are relatively rare diseases, so multicenter studies are required for their evaluation due to the long latency of these tumors [6, 7]. We performed a univariate and multivariate multicenter retrospective analysis of possible risk factors for recurrence in meningiomas and a regression analysis to demonstrate the entities negatively involved and their impact on the months of recurrence-free survival in the Mexican population.

Material and Methods

Study Design

According to the STROBE recommendations, a retrospective analysis was performed in six hospitals in Mexico between 2009 and 2019 [8]. Convenience sampling was used on patients diagnosed with meningiomas and tumor recurrence. The provisions of the Declaration of Helsinki followed the research standards. This study was approved by the institutional review board with the waiver of informed consent or exemption at our institution.

Criteria Selection

Inclusion Criteria

Eligible patients aged between 18 and 90 years old who had a diagnosis of single meningioma defined by the history of histopathologically proven meningioma resection surgery in the last 10 years and a follow-up with imaging studies (Magnetic Resonance) at three, six, and 12 months.

Exclusion Criteria

Patients who did not have complete imaging studies (magnetic resonance imaging controls) within their clinical files were excluded. Patients with clinic history uncompleted or histopathological samples with incomplete reports were also excluded.

Data Collection

Demographic data (age, sex, histopathological grade, location, degree of resection achieved, and months of survival) were retrospectively reviewed from the records. For this study, tumor recurrence was defined as any lesion that was imaging compatible with a meningioma; hypo- or isointense lesions in T1, iso or hyperintense in T2 and that had an enhancement to the contrast administration, and that, in addition, they had a history of histopathologically proven meningioma resection surgery in the last 10 years. Data collection was done manually and daily by verification of the electronic clinical file.

Statistical Analysis

The statistical analysis was carried out with the support of the SPSS computer program in version 21.0 (IBM Corp., Armonk, NY, USA). The graphs were generated using the Microsoft Office 2020 ™ package spreadsheet. (Excel 2020 program). First, a frequency analysis was performed using measures of the sociodemographic data’s central tendency (mean and median), such as age, sex, histopathological grade, location, degree of resection achieved, and months of survival. Subsequently, a univariate analysis was performed, and the analysis was completed with the multivariate analysis. For the survival analysis, we excluded patients in whom there was no recurrence since this analysis aimed to determine whether the identified factors negatively impact recurrence-free survival. Therefore, this analysis was completed with only 74 patients with proven imaging recurrence. For this analysis, the Pearson correlation coefficient was used for quantitative variables: bleeding, time of surgery, age, and resected tumor size as a function of the months of recurrence-free survival. For the comparison of means of independent samples, the Student’s t-test was used to compare differences between months of survival in the different sub-groups.

Results

An initial screening of hospital databases detected 253 patients with single meningioma surgical resection. Of these, 35 patients did not have complete magnetic resonance studies. Twenty-eight more were excluded because of a lack of data, did not complete the necessary time to follow-up, or did not complete pathological studies. One hundred ninety histopathological samples were obtained, 116 corresponding to women and 74 to men. Of the 190 tumors analyzed, 165 (86.8%) were grade 1 of the World Health Organization 2021 classification, 22 (11.6%) grade 2, and 3 (1.6%) grade 3. The tumor size in cubic centimeters obtained was from 0.05 to 880 cc with a mean of 63.8 cc (SD ± 100 cc). The mean size for men was 71.5 (SD ± 93.8 cc), and for women, it was smaller, with a mean of 58.9 (SD 104.8 cc) (p < 0.05). The mean age of the patients analyzed was 56 years (SD ± 15.9 years). The results of tumor size, location, histological pattern, and their respective significance value are shown in Table 1.

Table 1.

Tumor size by location and its significance value and tumor size by histological pattern and its significance value. *Statistically significant value. SD standard deviation, CPA cerebellopontine angle

Localization (total)	Percentage (%)	Size in cubic centimeters (SD)	P value
Convexity (56)	29.5%	95 cc (15 cc)	0.012*
Sickle (17)	8.9%	81 cc (17 cc)	0.323
Parasagittal (27)	14.2%	88 cc (32 cc)	0.401
Tentorial (13)	6.8%	37.3 cc (13 cc)	0.071
Clinoid (11)	5.8%	29.7 cc (10 cc)	0.010*
Sphenoid (18)	9.5%	57.22 cc (16.9 cc)	0.671
Petroclival (10)	5.3%	27.2 cc (19.4 cc)	0.084
CPA (9)	4.7%	18.1 cc (10.4 cc)	0.001*
Sellar tuberculum (7)	3.7%	14.2 cc (6.2 cc)	0.000*
Olfactory groove (6)	3.2%	67.8 cc (30.8 cc)	0.918
Cervical (5)	2.6%	0.68 cc (0.19 cc)	0.000*
Thoracic (8)	4.2%	2.26 cc (0.85 cc)	0.000*
Lumbar (1)	0.5%	No data	No data
Ventricular (2)	1.1%	119.5 cc (80.5 cc)	0.615
Histology	Percentage (%)	Size in cubic centimeters (SD)	P value
Meningothelial (47)	24.7%	47,9 cc (12 cc)	0.160
Fibroblastic (22)	11.6%	51 cc (10.53 cc)	0.271
Transitional (75)	39.5%	76.4 cc (15.6 cc)	0.250
Psammomatous (12)	6.3%	34.4 cc (13.7 cc)	0.05*
Angiomatous (6)	3.2%	52.6 cc (22 cc)	0.628
Microcystic (3)	1.6%	58.5 cc (10.3 cc)	0.659
Metaplastic (2)	1.1%	98.4 cc (98.3 cc)	0.786
Chordoid (2)	1.1%	19.66 cc (19.3 cc)	0.220
Clear Cells (4)	2.1%	126 cc (19.5 cc)	0.039*
Atypical (14)	7.4%	87.5 cc (17.1 cc)	0.203
Rhabdoid (1)	0.5%	35 cc	No data
Anaplastic (2)	1.1%	57.8 cc (45.8 cc)	0.907

We excluded 16 patients with spinal meningioma for the univariate analysis to avoid bias in the results. The risk analysis was performed with 174 cranial meningioma patients. The following were identified as risk factors for tumor recurrence: Simpson 4 (OR 57; p-value = 0.00), clinically onset as a cranial nerve disorder (OR 2.1; p-value = 0.02), or as intracranial hypertension syndrome (OR 5; p-value 0.00). The following were identified as protective factors: immediate post-surgical improvement (OR 0.14, p-value 0.00) or a Simpson resection degree 1 (OR 0.15, p-value 0.00); the rest of the values are studied and presented in Table 2.

Table 2.

Univariate analysis (UA) of clinical, surgical and demographic factors for tumor recurrence, with their respective confidence Interval (CI) and “P” value. Multivariate analysis (MA) of clinical, surgical and demographic factors of tumor recurrence, with their respective confidence interval (CI) and “P” value. *Statistically significant value. OR, odds ratio; CC, cubic centimeters; CPA, cerebellopontine angle

Factor for UA	OR (CI 95%)	P value
Age 60 years	1.74 (0.81–3.75)	0.15
Simpson 1	0.15 (0.07–0.34)	0.000*
Simpson 2	0.50 (0.27–0.93)	0.026*
Simpson 3	1.26 (0.32–4.88)	0.729
Simpson 4	57 (13.01–248)	0.000*
Simpson 5	1.04 (0.98–1.04)	0.209
Pain/headache	0.65 (0.36–1.18)	0.164
Motor deficit	1.00 (0.53–1.88)	0.995
Sensitive deficit	0.41 (0.13–1.31)	0.126
Seizure	0.76 (0.37–1.54)	0.45
Cranial nerve deficit	2.1 (1.10–4.27)	0.023*
Intracranial hypertension syndrome	5 (2.06–12.08)	0.000*
Cerebellar syndrome	1.05 (1–1.11)	0.011*
Post-surgical improvement	0.14 (0.06–0.35)	0.000*
Tumor size resected 60 cc	1.13 (0.61–2.08)	0.686
Surgical time > 280 min	3.12 (1.70–5.72)	0.00*
Bleeding > 850 ml	1.34 (0.74–2.41)	0.32
Meningothelial	1.08 (0.55–2.12)	0.811
Fibroblastic	0.70 (0.27–1.81)	0.466
Transitional	0.98 (0.54–1.78)	0.949
Psammomatous	2.31 (0.70–7.06)	0.155
Angiomatous	0.30 (0.03–2.65)	0.255
Microcystic	0.78 (0.07–8.76)	084
Metaplastic	1.57 (0.97–25.57)	0.74
Chordoid	1.57 (0.97–25.57)	0.74
Clear cells	0.51 (0.53–5.05)	0.56
Atypical	1.19 (0.39–3.58)	0.75
Rhabdoid	0.99 (0.97–1)	0.42
Anaplastic	1.57 (0.09–25.57)	0.74
Convexity	0.23 (0.11–0.51)	0.00*
Sickle	1.10 (0.40–3.05)	0.84
Parassagital	0.50 (0.20–1.25)	0.13
Tentorial	1.37 (0.44–4.26)	0.58
Clinoid	1.95 (0.57–6.66)	0.27
Sphenoid	2.10 (0.79–5.61)	0.12
Petroclival	15.92 (1.97–128.58)	0.00*
CPA	5.95 (1.20–29.50)	0.01*
Sellar tuberculum	4.13 (0.78–21.87)	0.07
Olfactory groove	0.77 (0.13–4.35)	0.77
Factor for MA	OR (IC 95%)	P value
Pain/headache	1.34 (0.40–4.51)	0.63
Motor deficit	1.22 (0.32–4.60)	0.76
Sensitive deficit	0.65 (0.08–5.13)	0.68
Seizure	1.94 (0.47–8.25)	0.36
Cranial nerve deficit	1.06 (0.14–7.29)	0.95
Intracranial hypertension syndrome	8.58 (1.64–44.68)	0.01*
Post-surgical improvement	0.15 (0.04–0.60)	0.007*
Tumor size resected 60 cc	2.95 (0.80–10.85)	0.10
Surgical time > 280 min	2.19 (0.66–7.29)	0.19*
Bleeding > 850 ml	1.01 (0.31–3.32)	0.97
Meningothelial	2.36 (0.03–167)	0.69
Fibroblastic	1.42 (0.01–144)	0.88
Transitional	1.97 (0.02–135)	0.75
Psammomatous	11.1 (0.10–1246)	0.31
Microcystic	5.92 (0.04–865)	0.48
Metaplastic	0.69 (0.00–1440)	0.92
Chordoid	0.09 (0.00–56,267)	0.72
Clear Cells	0.19 (0.00–67)	0.58
Atypical	6.59 (0.07–575)	0.40
Convexity	0.307 (0.27–3.46)	0.34
Sickle	0.63 (0.04–8.56)	0.73
Parassagital	0.37 (0.35–4.04)	0.42
Tentorial	2.38 (0.17–33.06)	0.51
Clinoid	0.58 (0.02–13.92)	0.74
Sphenoid	0.20 (0.01–3.33)	0.26
Petroclival	16.91 (0.31–906)	0.16
CPA	5.01 (0.15–165.80)	0.36
Sellar tuberculum	1.19 (0.02–59.15)	0.92
Olfactory groove	0.01 (0.00–1.36)	0.06

For the multivariate analysis, it was found that the clinical debut as intracranial hypertension syndrome is considered a risk factor for recurrence (OR 8.58, p-value 0.01) and that the post-surgical improvement is a protective factor so that there is no tumor recurrence (OR 0.15, p-value 0.00).

For the subgroup analysis of recurrence, patients who did not present it were excluded, so this analysis was carried out with 74 patients, excluding 100 patients who had not presented recurrence. The mean of recurrence was 42 months (SD ± 39). For the simple line regression analysis, bleeding was identified as a factor that negatively impacts survival; the more significant the intraoperative bleeding, the shorter the survival time free of tumor recurrence. The rest of the analysis is presented in Table 3 and Fig. 1. Finally, when comparing the mean survival for the different subgroups, it was obtained that the mean survival for petroclival meningiomas was 21 months; for those presenting clinically as intracranial hypertension syndrome, it was 14 months, and for those presenting as a cerebellar syndrome of 29 months, all three significantly (p < 0.05). The rest of the mean comparison analysis is in Table 4 and Fig. 2.

Table 3.

Simple linear regression. *Statistically significant value

Variable	Correlation coefficient	P value	R2
Age	− 0,01	0.46	0.00
Resected tumor size	0.00	0.17	0.012
Surgery time	− 0.12	0.14	0.01
Bleeding	− 0.23	0.02*	0.05

Fig. 1.

Simple linear regression. Correlation coefficient and R² value

Table 4.

Comparison of means. *Statistically significant value. IHS intracranial hypertension syndrome

Variable	Mean survival in control group	Mean group survival in exposed group	F value	P value
Clinoid	41.7 (± 38)	54 (± 57)	4.01	0.04*
Petroclival	45.6 (± 40.9)	21.8 (± 17)	7.68	0.00*
IHS	53.3 (± 37.3)	14.3 (± 30.9)	3.87	0.05*
Cerebellar syndrome	43.5 (± 40.3)	29 (± 16)	3.99	0.05*
Fibroblastic	40.7 (± 36.9)	62.5 (± 59.1)	7.69	0.00*

Fig. 2.

Comparison of means of significant tumor characteristics with the mean global recurrence-free survival. x = mean;—= median

Follow-up

All patients with meningioma grade 1 were selected for clinical observation and with magnetic resonance after surgical resection; independently, it was total or subtotal resection. In the three cases with meningioma grade 3 (two anaplastic and one rhabdoid), conformal radiotherapy was given as adjuvant therapy as soon as possible. In meningioma grade 2, the 22 patients were only observed through time, and no radiotherapy was considered in the residual tumor or recurrence.

Discussion

In meningiomas, some histopathological patterns appear predisposed to complex anatomical locations, surgically speaking, although histologically, they are considered benign, leading to an increased risk of recurrence [6]. In the case of meningiomas of the convexity, for example, recurrence will largely depend on the degree of resection achieved; however, higher rates of malignancy have been reported in these tumors, higher mitotic indices, and invasion of brain parenchyma, which would make complete resections challenging to achieve and would favor recurrence [9]. One of our two cases of anaplastic meningiomas was found in the convexity location, achieving a Simpson 3 resection. The patient was managed with postoperative radiotherapy 4 weeks after surgery, so no recurrence was reported during follow-up. A significant subset of patients is not successfully treated with a single surgery or in whom complete resection is not possible due to the tumor’s relationship to complex anatomy [10]. In addition, during surgery, the surgeon must maximize the extent of tumor resection but also be aware of the factors involved in possible recurrence and thus reduce the chances of recurrence [11].

Several factors have been associated with a high risk of meningioma recurrence, including imaging peritumoral brain edema, cellular pleomorphism, neovascularization, hyperostosis, the presence of macronuclei, and brain invasion [9, 12]. The relevance of the degree of resection achieved in meningiomas to prevent their recurrence is more than proven [5, 13–15]. However, other risk factors for tumor recurrence have also been described, such as age, location, or tumor size [16–19]. Few authors have reported that large tumors were associated with a higher incidence of tumor infiltration in brain tissue and adherence to arachnoid membrane [20, 21]. Han et al. have reported that size is an important factor associated with recurrence. Tumors larger than 5 cm suggest aggressive behavior, high degrees of malignancy, and, therefore, higher recurrence rates. On the other hand, meningiomas of the base of the skull also present higher recurrence rates. However, these tumors are not more significant in size; on the contrary, they are smaller. Given the anatomical conditions of the base of the skull, a complete Simpson 1 resection is more complicated than in other locations [22, 23]. In our population, we found that neither sex nor tumor size is related to higher recurrence rates. However, one of the two high-grade tumors was found in a meningioma of the convexity, with a reported size of 6.2 cm. Similarly, numerous studies have reported higher recurrence rates for males than females. Nevertheless, other studies, including ours, have found no significant difference based on sex.

In the case of location, lesions at specific sites were associated with a higher recurrence or progression rate; the location of the meningioma is suggested to be a determining factor in surgical resectability and prognosis. Extreme examples in literature are meningiomas of the sphenoid ridge and petroclival meningiomas, which present an increased risk of tumor recurrence [14, 24, 25]. Also, those of larger size and brain invasion have higher recurrence rates [11]. In our results, the convexity and the petroclival localizations were risk factors for higher recurrence rates.

Several studies have focused on the close association between histological grade and risk of recurrence, and few have focused on symptoms or clinical presentation. Haddad et al. exposed the most common clinical features at the initial diagnosis and showed that only 22.2% of the patients were diagnosed as an incidental finding; it also showed that the headache associated or not with the focal neurological deficit was present in at least a third of cases [26]. Moreover, this is how it has been shown that presentation as an altered state of consciousness and having a palpable cranial mass was associated with an increase in tumor recurrence [25, 27]. On the other hand, within the clinical aspects, it has been found that, in meningiomas, the appearance of headache has been associated with a higher recurrence rate [28]. Unlike our study, where we found that the single presentation with headache is a protective factor, the presentation of intracranial hypertension syndrome carries a risk of up to eight times more recurrence. Albakr et al. devised a computer tool for predicting recurrence in meningiomas. They considered factors such as malignancy grade, gender, ethnicity, age, degree of resection, association with genetic syndrome, and cell proliferation. However, their analysis did not consider clinical presentation or comorbidities, which have also been described as important factors associated with recurrence [29–31]. Very few studies speak of the relevance of the clinical presentation, which, to a great extent, is an important part of a correct surgical decision [26]. Therefore, we focus our attention on an aspect as essential as the clinic, which, from the patient’s reception, can guide us in knowing the possibility of recurrence and the long-term prognosis.

There are studies so far that base risk factors for recurrence on the degree of resection achieved (gross total resection vs. subtotal resection), even specifically in atypical meningiomas, achieving a degree of total resection, have been considered the only factor significantly associated with recurrence [32–34]. The neoplastic field hypothesis, based on the grade 0 resection theory, suggests that recurrence arises from clusters of meningothelial cells located within the dura near meningiomas [35]. Therefore, some authors consider the degree of resection the best predictor of recurrence of the Simpson grading system, which determines complete or partial resection [19]. In our study, this grading scale for resection is an adequate predictor for recurrence since statistical significance was obtained in most of them. However, we must comment that there is the possibility of a subjective error between surgeons that could represent a bias within said determination since the Simpson scale only determines the degree of resection for convexity meningiomas and the resected margins according to the technique—surgery for each surgeon [3].

The Simpson scale has been evaluated in multiple prospective and retrospective studies. In some cases, no prognostic differences regarding recurrence in Simpson 1, 2, or 3 have been found. Ultimately, the Simpson scale is a subjective scale for intraoperative evaluation, which depends entirely on the surgeon’s criteria, so exact measurement and validity could be difficult. For this reason, in all possible cases, the imaging evaluation should always be complemented with an MRI study based on the RANO criteria, and, in this way, determine the presence or absence of residual, response, or recurrence in a more objective way [36, 37]. On the other hand, it is well known that hypoxia and related endothelial growth factors, such as VEGF, are related to different prognostic aspects in meningiomas, such as recurrence, the degree of edema, higher histological degrees, and even non-total resections [21, 38, 39]. For example, if a resection surgery lasts more than 4 h, or there is more significant intraoperative bleeding, which favors the use of more bipolar energy to avoid bleeding and which in turn perpetuates changes hypoxic in the surgical anatomical site, it was correlated with fewer months of tumor-free survival.

Limitations of the study

The study’s main limitation is that it is retrospective, so we cannot generalize the data reported here. In addition, intraoperative bleeding, which was reported as a risk factor for recurrence, has multiple factors, such as the surgeon’s skill, the type of surgical approach, the tumor location, and arterial nutrition, among others, so we consider it should be taken into consideration but with caution, since it cannot be extrapolated to the rest of the population. Regarding the intracranial hypertension syndrome as a clinical presentation and risk factor for recurrence, it is evident that the larger the size, the greater the probability of intracranial hypertension and the greater the surgical complexity. However, we consider in this data that intracranial hypertension is very likely also related to a more significant amount of tumor blood nutrition, greater bleeding, and a greater risk of tumor residual or recurrence. The data presented here should be considered delicately and carefully like any retrospective study since they lack the validity of a prospective or observational study.

Conclusions

Many factors that affect recurrence-free survival in meningiomas have been described, including molecular, surgical, and sociodemographic factors. The form of clinical presentation is essential for recurrence in meningiomas, and as we evidenced in our multivariate analysis, the onset as an intracranial hypertension syndrome hurts the recurrence-free months after tumor resection surgery. That is why the complete analysis of patients with tumor pathology, especially meningiomas, should be done based on as many characteristics as possible: histological grades, location, clinic, and the degree of resection achieved. The clinical presentation should also be considered an important factor of recurrence since it reflects the extension of the tumor and its behavior. There is a complete agreement in the literature that radical surgery is one of the positive factors influencing the prognosis of benign and malignant meningiomas.

Abbreviations

SD

Standard deviation

OD

Odds ratio

VEGF

Vascular endothelial growth factor expression

cc

Cubic centimeters

Declarations

Conflict of interest

The authors declare no competing interests.