Impaired survival and long-term neurological problems in benign meningioma

Abstract

Purpose: To assess long-term functional outcome and survival among patients with meningioma World Health Organization (WHO) grade I. Methods: Retrospective analysis of 205 patients after resection of WHO grade I intracranial meningioma from 1985 through 2003. Expected age- and sex-specific survival was calculated by applying Dutch life-table statistics to each patient for the individual duration of follow-up. Long-term functional outcome was assessed using a mailed questionnaire to the general practitioner. Results: The mean duration of follow-up was 11.5 years. Survival at 5, 10, 15, and 20 years was 92%, 81%, 63%, and 53%, respectively, which is significantly lower than the expected survival (94%, 86%, 78%, and 66%, respectively). Survival was worse with higher age (P < .001). Survival among patients younger than 45 years and older than 65 years was comparable to the expected survival but significantly worse among patients aged 45–65 years. Analysis of the cause of death suggests an excess mortality associated with both brain tumor death and stroke (P = .07). Recurrence rates at 5, 10, and 15 years were 18%, 26%, and 32%, respectively. Higher Simpson grade (P < .001) and lower age (P = .02) were associated with a higher recurrence rate. In 29 patients (14%) receiving radiotherapy, the 5-year recurrence rate was 18% and the 5-year survival was only 58%. Long-term functioning (≥5 years after last treatment) could be assessed in 89 long-term survivors: 29 patients (33%) showed no deficits, and 60 (67%) showed at least 1 neurological symptom, of whom 24 (27%) were unable to perform normal daily activities. Conclusion: Long-term survival in WHO grade I meningioma is challenged in patients more than45 years of age. Excess mortality seems to be associated with both tumor recurrence and stroke. The majority of patients have long-term neurological problems.

Meningiomas are the second most common primary brain tumors, with incidence rates of 3–4 cases per 100 000 males per year and 9–13 cases per 100 000 females per year.^1,2 The incidence increases with age and reaches its peak in the sixth and seventh decades. Histologically, most meningiomas are World Health Organization (WHO) grade I and, therefore, are regarded as benign tumors.³ Neurosurgical resection is the mainstay of primary treatment. The role of radiotherapy is still debated and is mostly reserved for the adjuvant treatment of nonradically removed or recurrent meningiomas.^4–6 The reported 5-year recurrence/progression rates are approximately 10% after complete resection and 45% after nonradical resection (Table 1).^4,6

Table 1.

Literature overview of 5-years recurrence risk in patients with meningioma (adapted from Marosi et al., 2008 and Gondi et al., 2010.^4,6)

First author	Year	No. of patients	Follow-up (months)	Radical surgery	Nonradical surgery	Surgery + radiotherapy
Adegbite27	1983	114	10–276	10%	55%	18%
Forbes28	1984	27	45	–	–	28%
Mirimanoff29	1985	225	65%>60	7%	37%
Barbaro30	1987	136	78	4%*	41%	23%
Taylor31	1988	132	60%>60	5%	57%	14%
Glaholm32	1990	186	80	–	–	22%
Miralbell33	1992	115	57	–	52%	12% @ 8 years
Goldsmith34	1994	140	40	–	–	11%
Mahmood35	1994	254	61	2%	46%	2/6 progression
Peele36	1996	86	46	–	48%	100%
Condra37	1997	262	98	7%	47%	10%
Stafford38	1998	581	55	12%	39%
Maguire39	1999	28	41	–	–	8% @ 4 years
Nutting40	1999	82	108	–	–	8%
Vendrely41	1999	156	40	–	–	11%
Kondziolka42	1999	99	>60	–	–	7% (radiosurgery)
Wenkel43	2000	46	53	–	–	0% (proton)
Dufour44	2001	17	73	–	–	7% @ 10 years
Debus45	2001	180	35	–	–	3% (radiosurgery)
Pourel46	2001	26	30	–	–	5%
Pollock47	2003	198	64	4%	42%
Selch48	2004	45	36	–	–	2% @ 3 years
Soyuer16	2004	92	92	23%	48%	9% (radiosurgery)
Henzel49	2006	36	36			3%
Milker-Zabel50	2007	94	53	–	–	4%
Sughrue26	2010	373	44	5–15%	19%	–
Honig51	2010	73	30	–	–	14%
Present study	2011	205	138	10%	34%	18%
Average				≈10%	≈ 45%	≈10%

Although high survival rates are reported for WHO grade I meningiomas, complications and long-term disability occur frequently, which decrease quality of life.^7–9 Studies on functional outcome of meningioma in patients usually include few patients and/or a short period of follow-up.

This study reports long-term results in terms of survival, tumor recurrence, and functional outcome measured in a large cohort of patients.

Material and Methods

Patients

A retrospective analysis was performed of all 233 patients who underwent neurosurgical resection of a WHO grade I intracranial meningioma in the AMC–University of Amsterdam from 1985 through 2003. Excluded were 28 patients with an optic nerve meningioma, with neurofibromatosis, or with in sufficient follow-up information (these were mostly patients from abroad). The minimal duration of follow-up for surviving patients was 6 years.

Patients' characteristics and preoperative and follow-up data were extracted from the written and digital patient files (summarized in Table 2). The date of first surgery was defined as the date of diagnosis. All patients had a histologically confirmed WHO grade I meningioma.^3,10 The tumor location was determined using radiological and surgical description and divided into 4 main groups: convexity, falx and sagittal sinus, tentorium, and skull base. The extent of surgery was determined from surgical records and was graded according to the Simpson criteria.¹¹ The preoperative performance status was graded according to the Karnofsky Performance Scale (KPS).¹² Neurological functioning and deficits at presentation were classified into 1 of 8 groups, according to the most common functions or symptoms: speech, motor function, vision, cognition, epilepsy, pain, cerebellar, and other symptoms. Symptoms and adverse effects were graded according to the modified Common Toxicity Criteria (CTC): 0 =no symptoms, 1 = subjective but nonobjective symptoms, 2 = mild objective symptoms not interfering with daily activities, 3 = symptoms interfering with daily activities, and 4 = severely disabling, not able to perform daily activities.¹³

Table 2.

Characteristics of the 205 meningioma patients at time of diagnosis

Variable
Follow-up	Mean: 11 years
	Range: 0–24 years
Gender
Male	60 (29%)
Female	145 (71%)
	Male-Female ratio: 1:2.4
Age	Mean: 53.7 years
	Range: 4.5–84 years
≤45 years	57 (28%)
46–55 years	62 (30%)
56–65 years	45 (22%)
>65 years	41 (20%)
Karnofsky Performance Status	Mean: 74
	Range: 20–100
20–50	24 (12%)
60–70	60 (29%)
80–100	121 (59%)
Tumor localization
Convexity	66 (32%)
Falx & sagittal sinus	35 (17%)
Tentorium	24 (12%)
Skullbase	80 (39%)
Treatment
Surgery	205
+Radiotherapy	29 (14%)
Adjuvant	6 (3%)
Salvage after re-surgery	10 (5%)
Salvage without re-surgery	13 (6%)
Surgical radicality (Simpson)
Grade I	41 (20%)
Grade II	96 (47%)
Grade III	22 (11%)
Grade IV/V	43 (21%)
Unknown	3 (1%)

Symptomatic response 12 months after treatment was compared with signs and symptoms before treatment. Symptomatic response was rated as completely resolved, partially resolved, or progression of symptoms after treatment. A recurrence was defined as progressive disease of residual tumor and a recurrence after macroscopically radical resection on neuroradiological imaging.

Long-Term Follow-Up

The vital status of each patient was verified in the national population registry.

Death records with the causes of death of all deceased patients were retrieved from and analyzed by the Dutch national vital statistics registry (Statistics Netherlands). Causes of death were categorized in 5 main groups: brain tumor, stroke, other cardiovascular death, other benign or malignant tumor, and all other causes (see Supplementary Table 2). We also combined brain tumor and stroke into a single brain disease group and compared this with all other causes.

To more accurately evaluate long-term neurological outcome, a questionnaire was mailed to the general practitioners (GPs) of all patients surviving at least 5 years after last treatment. GPs were asked to grade the current neurological symptoms with the same CTC-based score used to evaluate preoperative functioning. In addition, an estimation of general performance was requested by choosing 1 of 3 performance groups: KPS 80–100 (able to care for him/herself), 60–70 (requires some assistance in daily activities), and ≤50 (disabled, requires special care or hospitalization). Use of anti-epileptic drugs was also recorded.

Statistical Analysis

Data were entered into a database (Microsoft Access 2002) and analyzed using SPSS, version 18.0, for Windows (SPSS).

Differences in proportions and contingency tables were compared using Fisher's exact probability test, and differences in means were compared using t tests (normally distributed) or analysis of variance (not normally distributed).

Survival and tumor recurrence times were calculated from surgical diagnosis with use of the Kaplan Meier method.

The prognostic influence of different variables (age, sex, preoperative functioning, Simpson grade, localization, and number of tumors) on survival and recurrence were determined by univariable analysis and compared by the log-ranktest, followed by a multivariable analysis using the Cox' proportional hazards model applying forward analysis. Age and KPS were tested both as continuous and as grouped variables; surgical radicality (Simpson grade) and tumor localization were treated as categorical variables. Because age, radicality, and localization were not mutually independent, interaction terms were created; these did not change the significance of the final model as presented.

Because this study describes a long-term follow-up of up to 23 years from initial diagnosis and the mean age of our patients was 53 years, a considerable number of patients had died from causes unrelated to the meningioma. To estimate the age- and sex-specific survival, we calculated the age- and sex-specific annual death rate based on Statistics Netherlands life-tables for 2000 for each individual patient for each year from diagnosis until death or end of follow-up (http://statline.cbs.nl) (for more details, see Supplementary Methods).

Results

Patient Characteristics

All written and digital files for the 205 evaluable patients were reviewed; Table 2 presents their characteristics. A total of 183 patients (89%) had a follow-up period of more than5 years, with a mean follow-up period of 11 years (range 0–24 years). Of these, 13 patients (6%) died; 9 (4%) moved out of the Netherlands within 5 years and were included as lost to follow-up. The mean age of men and women at radiological diagnosis was 53.9 and 53.6 years, respectively (NS). The age was normally distributed from 22 through 84 years, with one outlier at 4.5 years. The file and pathology of this exceptionally young patient was reviewed. After cerebellar symptoms, the patient underwent surgery for a midline meningioma in the posterior fossa; the tumor was radically removed, and the patient is alive and well at the age of 21 years.

Clinical Presentation

Patients presented with a variety of neurological symptoms (Table 3). The most common complaint at presentation was pain, which was mostly attributable to episodic headache, although pain due to nerve compression was also frequently reported. Of the 24 patients with a KPS less than 60, 4 (2%) were unconscious because of increased intracranial pressure. In 4 other patients, a meningioma was diagnosed incidentally, because radiological imaging was performed for nontumoral reasons.

Table 3.

Neurological symptoms at presentation

Symptoms	None	Mild	Severe	Total
Pain	116 (57%)	71 (35%)	18 (9%)	89 (43%)
Visual disturbances	141 (69%)	25 (12%)	39 (19%)	64 (31%)
Cognitive deficits	142 (69%)	40 (20%)	23 (11%)	63 (31%)
Epilepsy	144 (70%)	14 (7%)	47 (23%)	61 (30%)
Motor deficits	148 (72%)	35 (17%)	22 (11%)	57 (28%)
Cerebellar problems	157 (77%)	39 (19%)	9 (4%)	48 (23%)
Speech problems	186 (91%)	11 (5%)	8 (4%)	19 (9%)
Other	149 (73%)	43 (21%)	13 (6%)	56 (27%)

Tumor Characteristics

A total of 195 patients presented with a single lesion. Six patients presented with 2 tumors, without further evidence of neurofibromatosis; 4 patients had 3 or more tumors. Tumors location showed no significant difference between sexes.

Survival

Of the 205 patients, 58 (28%) died during follow-up. Fig. 1 shows the curve for the observed survival, together with the expected age- and sex-specific survival. Perioperative mortality up to 3 months after neurosurgery was observed in 4 patients (2%). The overall survival at 5, 10, 15, and 20 years was 91.5% (95% confidence interval [CI], 87.4%–95.5%), 81.4% (95% CI, 75.6%–87.2%), 62.5% (95% CI, 52.5%–71.5%), and 53.4% (95% CI, 41.4%–65.4%), respectively. This is increasingly lower than the expected age- and sex-specific survival for this specific group, calculated at 94.7%, 84.3%, 74.9%, and 67.3%, respectively. Fig. 1 shows that, after 14 years, the observed survival curve becomes significantly worse than the expected survival curve.

Fig. 1.

Observed and expected age- and sex-specific survival. The bars indicate the 95% confidence intervals (CIs) at 5, 10, 15, and 20 years. Despite increasing CIs, after 13 years, the observed survival lies significantly below the expected survival curve.

Tables 4 and 5 give a summary of the univariable and multivariable analysis for survival and progression. Fig. 2 shows univariable survival curves and cumulative progression rates for age, localization, and surgical radicality by Simpson grade. Stratified by age group (Fig. 2), the prognostic value of age becomes clear. To correct for normal aging, Fig. 2 also gives the survival curves expected for age and sex, for patients aged 45 years and younger and 65 years or older; the observed survival is as expected. However, the observed survival is particularly worse than the expected survival among patients aged 46–65 years. In the patient group aged 46–55 years, survival was significantly lower among patients who developed a recurrence, compared with those who did not (P = .04). However, older or younger patients without a recurrence did not have a better survival than did those who did develop a recurrence. In the Cox multivariable analysis, survival was significantly better with a lower age at diagnosis (P < .001) and with a trend for a higher KPS (P = .06) (Table 5).

Table 4.

Summary of univariable analysis of recurrence and survival

Variable	Category	Recurrence rate					Overall survival
		5-year (%)	95% CI (%)	10-year (%)	95% CI (%)	Univariable P-value	5-year (%)	95% CI (%)	10-year (%)	95% CI (%)	Univariable P-value
Age	≤45 years	22	11–33	37	22–51	P = 0.22a	100		95	86–100	P < 0.001b
	46–55 years	21	10–22	25	13–37		93	87–99	83	72–94
	56–65 years	12	2–22	26	11–41		84	73–95	77	64–90
	>65 years	11	0–11	15	2–28		85	73–97	60	44–76
Gender	Female	22	15–29	29	21–37	P = 0.10	94	85–100	83	72–94	P = 0.18
	Male	7.3	1–14	19	7–31		86	82–90	79	72–86
Karnofsky	0–50	14	0–29	37	12–62	P = 0.43	75	57–93	55	33–77	P < 0.001
Performance	60–70	24	12–36	31	18–54		89	81–97	77	65–89
Score	80–100	15	8–22	22	14–30		96	92–100	89	83–95
Localization	Convexity	14	5–23	18	8–28	P = 0.22	92	85–99	70	58–82	P = 0.085
	Falx/Sagittal Sinus	12	1–23	27	11–43		91	81–100	82	69–95
	Tentorium	22	5–39	43	21–65		100		95	85–100
	Skull base	22	2–22	27	16–38		88	81–95	87	79–95
Radicality (Simpson)	Grade I	5	0–12	8	0–17	P < 0.001	97	92–100	82	69–95	P = 0.40
	Grade II	12	5–19	23	13–33		93	88–98	86	79–93
	Grade III	20	2–38	26	5–47		86	71–100	80	62–98
	Grade IV-V	41	26–56	50	34–66		87	77–97	71	57–85

^aThe recurrence risk is lower at higher age (see Supplementary Fig. 2B).

^bCompared to the age and gender-specific survival in the Dutch population, survival is significantly worse in meningioma patient groups aged 46–65 years (see Fig. 2).

95% CI = 95% confidence interval.

Table 5.

Summary of multivariable analysis of recurrence and survival

		Recurrence				Survival
Variable	Group	Univariable		Multivariable		Univariable		Multivariable
		P-value	HR (95% CI)	P-value	HR (95% CI)	P-value	HR (95% CI)	P-value	HR (95% CI)
Age	Continuous	P = 0.11	0.98 (0.97–1.00)	P = 0.03a	0.97 (0.96–1.00)	P < 0.001	1.10 (1.07–1.13)	P < 0.001	1.09 (1.06–1.12)
Gender	Male		Reference group		Reference group		Reference group		Reference group
	Female	P = 0.11	1.76 (0.88–3.53)	P = 0.43	1.34 (0.65–2.78)	P = 0.18	0.69 (0.40–1.19)	P = 0.18	0.67 (0.37–1.20)
Karnofsky Performance Score	Continuous	P = 0.81	1.00 (0.98–1.02)	P = 0.57	1.00 (0.98–1.01)	P < 0.001	0.97 (0.96–0.99)	P = 0.06	0.98 (0.97–1.00)
Localization	Convexity	P = 0.24	Reference group	P = 0.87	Reference group	P = 0.10	Reference group	P = 0.40	Reference group
	Falx/Sagittal Sinus	P = 0.32	1.59 (0.65–3.91)	P = 0.95	0.97 (0.37–2.52)	P = 0.43	0.75 (0.37–1.53)	P = 0.54	0.77 (0.34–1.78)
	Tentorium	P = 0.05	2.48 (1.00–6.09)	P = 0.56	1.35 (0.50–3.63)	P = 0.04	0.27 (0.08–0.91)	P = 0.10	0.35 (0.10–1.23)
	Skull base	P = 0.11	1.83 (0.87–3.86)	P = 0.92	0.96 (0.42–2.20)	P = 0.08	0.59 (0.32–1.06)	P = 0.84	0.93 (0.45–1.91)
Radicality (Simpson grade)	Grade I	P < 0.001	Reference group	P < 0.001	Reference group	P = 0.41	Reference group	P = 0.19	Reference group
	Grade II	P = 0.21	2.00 (0.67–5.89)	P = 0.29	1.87 (0.59–5.93)	P = 0.83	1.09 (0.51–2.33)	P = 0.72	1.16 (0.52–2.59)
	Grade III	P = 0.66	3.27 (0.92–11.6)	P = 0.09	3.18 (0.85–11.9)	P = 0.53	1.39 (0.50–3.91)	P = 0.34	1.75 (0.56–5.48)
	Grade IV-V	P < 0.001	7.26 (2.50–21.1)	P = 0.001	7.34 (2.26–23.9)	P = 0.18	1.74 (0.78–3.93)	P = 0.07	2.24 (0.93–5.40)

^aThe recurrence risk is lower at higher age (see Fig. 2B).

HR, hazard ratio; CI, confidence interval.

Fig. 2.

Univariable curves for survival by age group. Dotted lines give the age- and sex-specific survival.

To explain the excess mortality among patients aged 46–65 years, the causes of death were compared with those for patients 46 years and younger and those older than 65 years of age. Patients aged 46–65 years showed a somewhat higher excess mortality associated with brain tumors (11 [35%] of 31) and stroke (3 [10%] of  31; χ² = 3.36; P = 0.07) (see Table 6).

Table 6.

Cause of death of the 58 deceased (from 205) meningioma patients. The data suggest a relative excess mortality from brain diseases in patients aged 46–65 years (χ² = 3.36; P = 0.07), i.e., from both a brain tumor (11/31 = 35%) and stroke (3/31 = 10%)

Cause of death	≤45 years and >65 years (n = 98)a	46–65 years (n = 107)a	All ages (n = 205)a
Brain disease	6 (22%)	14 (45%)	20 (34%)
Other causes	21 (78%)	17 (55%)	38 (66%)
All causes of death	27 (100%)	31 (100%)	58 (100%)

^aOriginal group sizes = alive + death.

Recurrence or Progression

Fifty patients (24%) experienced recurrent or progressive disease after their first surgery. The mean time to recurrence was 59 months (range, 6 months–13 years). Recurrence rates at 5, 10, and 15 years were 17.5% (95% CI, 11.9%–23.1%), 26.4% (95% CI, 19.6%–33.2%), and 31.9% (95% CI, 33.5%–40.3%), respectively. In the multivariable analysis, the recurrence rate was lower after more radical surgery (P < .001) and, surprisingly, for a higher age at surgery (P < .03). Neither sex, KPS, tumor localization, nor the number of tumors was an independent prognostic factor.

To facilitate comparison with the literature, we also regrouped patients into radically (Simpson I-II, 137 patients) and nonradically resected tumors (Simpson III-V; 65 patients). The 5-year recurrence rates after radical and nonradical surgery were 9.7% (95% CI, 4.3%–15.1%) and 33.7% (95% CI, 21.7%–45.7%), respectively (Table 1).

Radiotherapy

Radiotherapy (RT) was given to 29 patients, particularly older patients with a worse KPS, of whom 5 (17%) developed a subsequent recurrence. Only 6 patients had adjuvant RT within 6 months after a first nonradical resection; 23 patients had RT for recurrent meningioma, either after resurgery (10 patients) or as salvage treatment for an inoperable recurrence (13 patients). The prescribed dose was 50.4–54.0Gy in daily fractions of 1.8Gy, CT-based (and from the mid-1990s also with MRI) conformal RT. The 5-year recurrence rate after RT was 17.7% (95% CI, 3.3%–32.1%), and the 5- and 10-year survival rates were 80.1% (95% CI, 63.9%–96.3%) and 42.2% (95% CI, 29.0%–45.4%), respectively.

Short-Term Symptomatic Changes

Within 1 year after surgery, 126 patients (61%) reported that their signs or symptoms had partially or completely resolved, 27 (13%) had no change of symptoms, and 45 (22%) had progression of symptoms. The response of 7 patients (4%) was unknown. After RT (n = 29), symptoms improved in 9 patients (31%), remained stable in 13 (45%), and were progressive in 6 (21%).

Patients with progression of symptoms after surgery did not have a significantly higher risk of recurrence or a worse mortality.

Long-Term Functioning

Of the questionnaires sent to the GPs regarding the 139 long-term surviving patients, 89 (64%) were returned. For all 89 cases, it had been at least 5 years after the last treatment. Table 7 summarizes the remaining symptoms. In total, 29 patients (33%) were reported as not showing any long-term deficits; 67% of the patients experienced at least 1 neurological symptom at long-term follow-up, and 24 (27%) experienced severe symptoms and were not able to perform normal daily activities.

Table 7.

Current neurological symptoms reported by general practitioners (n = 89)

Symptoms	None	Mild	Severe	Unknown	Total (Mild + Severe)
Pain	61 (69%)	15 (17%)	3 (3%)	10 (11%)	20%
Visual disturbances	62 (70%)	10 (11%)	9 (10%)	8 (9%)	21%
Cognitive deficits	61 (69%)	19 (21%)	3 (3%)	6 (7%)	25%
Epilepsy	66 (74%)	12 (13%)	5 (6%)	6 (7%)	19%
Motor deficits	60 (67%)	12 (13%)	12 (13%)	5 (2%)	27%
Cerebellar problems	79 (89%)	4 (4%)	1 (1%)	5 (6%)	5%
Speech problems	71 (80%)	12 (13%)	2 (2%)	4 (4%)	4%
Other	79 (89%)	4 (4%)	2 (2%)	4 (2%)	7%

To exclude potential responder bias, we compared some important patient and treatment characteristics between the patients for whom a questionnaire was returned and patients for whom a questionnaire was not returned (nonresponders). There were no apparent differences in age (P = .80), sex (P = .70), KPS before surgery (P = .40), tumor localization (P = .37), and Simpson score (P = 0.97). However, the duration of follow-up after surgery was longer for nonresponders (13.6 years) than for responders (11.3 years).

Discussion

This study presents long-term treatment results in a large series of patients with a histologically confirmed WHO grade I intracranial meningioma. All patients primarily received a diagnosis on the basis of CT or MR scan and were followed for 6 to 23 years or more after their first surgery. Although 5-year survival and recurrence rates are comparable to those of other series (Table 1), when compared with the age- and sex-specific mortality of each individual patient, survival among patients with meningioma was increasingly worse. Subgroup analysis showed that survival is comparable to the expected survival among patients aged 45 years and younger and 65 years and older but is significantly worse among patients aged 46–65 years.

That survival is impaired in patients with benign meningioma was earlier suggested by data from the Swedish Cancer Registry¹⁴ but is not further described in a modern series.¹⁵ This disturbing (and largely unknown) finding may have been obscured because of the following reasons.

First, few series have reported long-term survival results. In a study by Soyuer et al. (2004), survival among 70 patients with meningioma was compared with age- and sex-specific survival, but no significant differences were found.¹⁶ However, on careful re-examination of the printed curves in that study, there is a similar divergence of the curves 8 years after diagnosis, as found in the present study. In the study by Soyuer et al., the differences may not have appeared to be significant because of the small number of patients still at risk and the shorter duration of follow-up, compared with the present series.

Second, comparison with age- and sex-specific mortality is a procedure not generally included in standard statistical software. Therefore, we designed a separate spreadsheet method to calculate the annual age- and sex-specific mortality risk for each individual patient for the duration of follow-up until death or until censoring of the patient. This method takes into consideration that a survival curve is confounded by 2 naturally competing death risks (i.e., the increasing death rate by age and the decreasing death rate by a relatively higher loss of older persons in the earlier years of follow-up).

Third, from a pathologist's viewpoint, the WHO grade I meningioma is a benign tumor and is therefore not recorded as cancer according to the WHO ICD-O morphology code (code M 9530-0).¹⁷ Therefore, benign meningioma has long been excluded from cancer registration and statistics and, in some countries, still is excluded¹⁸. In addition, from a neuro-oncologist's perspective, survival among patients with meningioma is far better than that among the majority of patients with glioma. From that perspective, our own 5-year survival and recurrence rates compare favorably with normal survival and are in agreement with most reports on WHO grade I meningioma. Although this seems reassuring, placed in a wider perspective, the tumor-associated survival rate is comparable to or worse than that among patients with early stages of breast or prostate cancer, seminoma testis, or melanoma.^14,19,20 It has only recently been recognized that meningiomas constitute 33%–50% of newly diagnosed intracranial tumors and that patients with meningioma constitute the vast majority of long-term brain tumor survivors.^19–22

Lower survival rates are associated with higher age and lower performance status. The negative impact of high age on survival, surgical mortality, and morbidity is well known.^23–25 Extensive resection is therefore discouraged in patients aged 70 years or older. We also found that a higher age is associated with worse overall survival. However, the excess mortality that is only found after comparison with age- and sex-specific survival was most pronounced in patients aged 46–65 years and, to a much lesser extent, in the younger or older patients.

That survival among patients aged 46–55 years was significantly higher in those who developed a recurrence, compared with those who did not, supports recurrent tumor as a cause of late excess mortality; however, such a causative relation should be considered with caution. It is methodologically incorrect to use an event later in time (i.e., recurrence) as an upfront prognostic factor. Particularly in older age groups, recurrence may then not be recognized as a risk factor simply because patients may have died from other causes before a recurrence could have been diagnosed. Therefore, we also compared the cause of death among patients aged 46–65 and older and younger patients and found a (nonsignificant) excess mortality associated with both recurrent brain tumor and stroke.

That survival was not worse than expected among patients aged >65 years might be explained by careful selection of patients fit for surgery; in addition, in the present study, older patients had relatively less radical surgery and more often received adjuvant RT. The latter may also explain why we found that a higher age is associated with a lower recurrence rate.

Although the prognostic significance of the Simpson grade was debated by Sughrue et al. (2010),²⁶ most studies acknowledge that the extent of surgery is a risk factor for recurrence. Some of the controversy can be explained by patient selection and confounding of competing risk factors, as described above for our older patients; surgical radicality; and adjuvant RT. In the multivariable analysis, the Simpson grade for radicality remains a highly significant factor for recurrence but not for survival.

Only a small minority of our patients received additional RT and mostly as salvage treatment in poor-risk patients after 2 or more resections. The remarkably low 5-year recurrence risk is comparable to those in series using adjuvant conventional RT, but survival is poor.

We reported earlier that, compared with matched healthy control subjects, more than50% of patients with meningioma experience impaired cognition and worse quality of life.^7–9 The present study also shows that about one-third had stable or worse neurological symptoms shortly after surgery and that two-thirds of the patients had long-term neurological sequelae; 1 in 4 patients is chronically disabled. The long-term neurological outcome was evaluated using a mailed questionnaire to the GP of all patients surviving at least 5 years after last treatment. The GPs' response rate was high (64%). To exclude potential response bias, we compared some important patient and treatment characteristics between the patients for whom a questionnaire was returned and patients for whom a questionnaire was not returned. Except for a longer duration of follow-up in nonresponders, we found no differences, compared with responders. Two possible explanations for the longer follow-up from nonresponding GPs is that more patients had moved to another medical practice or to a home for the elderly and/or that GPs had discarded these patient files.

Conclusion

In the long term, survival is progressively challenged among patients with WHO grade I meningioma aged 45–65 years, and a majority of patients have long-term neurological deficits. The excess mortality seems to be associated with both tumor recurrence and stroke. Because patients with meningioma are the largest group of long-term brain tumor survivors, more research is urgently needed to improve both tumor control and to reduce long-term sequelae.

Supplementary Material

Supplementary material is available online at Neuro-Oncology (http://neuro-oncology.oxfordjournals.org/).

Funding

All authors are students (HvA, MdL) or employees of the Academic Medical Center - University of Amsterdam or of Statistics Netherlands (JK). No external funding, no conflict of interest.