ORIGINAL ARTICLE

Histopathologic indicators of recurrence in meningiomas:

correlation with clinical and genetic parameters
Yoo-Jin Kim & Ralf Ketter & Wolfram Henn &
Klaus D. Zang & Wolf-Ingo Steudel & Wolfgang Feiden
Received: 11 July 2006 / Accepted: 26 July 2006 / Published online: 3 October 2006
# Springer-Verlag 2006
Abstract Meningiomas in general are circumscribed
slow-growing tumors. However, despite gross total resec-
tion, tumor relapse and patients outcome are still an issue.
Risk stratification based on histomorphology alone
remains problematic. This study explored the independent
prognostic value of potential risk factors among 206
patients who underwent meningioma resection and fol-
lowed-up until death or a median of 44 months. The
statistical analysis considered clinical data, histomorpho-
logic parameters, cytogenetic findings, Ki-67 immunore-
activity, and activity of tissue non-specific alkaline
phosphatase (ALPL). Recurrence-free survival estimates
were computed and prognostic factors were identified
using Cox proportional hazards model. Independent
predictors of recurrence included (1) anaplasia; (2) mitotic
index 20/10 high-power fields; (3) subtotal tumor resec-
tion; (4) loss of short arm of chromosome 1 (1p); and (5)
Ki-67 labeling index (LI) >12%. Among totally resected
WHO grade I meningiomas, neither histopathologic nor
clinical parameters were predictive, whereas 1p was the
only independent prognostic factor. ALPL did not reach
significance in the multivariate modeling, however, the fast
and low-cost histochemical detection of ALPL expression
could be proved as a highly sensitive screening method for
1p. In particular, biologically aggressive meningiomas of
histologically benign or borderline phenotype could be
therefore identified by ALPL detection followed by 1p in
situ hybridization.
Keywords Meningioma recurrence
.
Prognosis
.
Ki-67
.
Alkaline phosphatase
.
Cytogenetics
Introduction
Meningiomas are tumors that arise from the leptomenin-
geal covering of the brain and the spinal cord, accounting
for 15% to 20% of all central nervous system tumors [26].
According to the current World Health Organization
(WHO) histological grading system, three grades of
meningiomas with increasing risk of recurrence are
distinguished [14]. Benign or common-type tumors with
a low rate of recurrence (7% to 20%) are assigned to
Grade I. Grade II meningiomas include atypical meningi-
omas as well as the rare chordoid and clear cell
(intracranial) variants and exhibit a higher risk of
recurrence (29% to 40%). Grade III meningiomas are
anaplastic meningiomas with high mitotic activity (20/10
high-power fields) and/or obviously malignant cytology as
well as the rare variants, papillary and rhabdoid meningi-
omas. Recurrence rate of anaplastic meningiomas is
suggested to be 50% to 78%. While the application of
histomorphologic criteria can be considered as a means of
risk stratification, the prognostic impact of the proposed
criteriaparticularly the mitotic thresholds distinguishing
Virchows Arch (2006) 449:529538
DOI 10.1007/s00428-006-0285-3
Y.-J. Kim (*)
:
W. Feiden
School of Medicine, Institute of Neuropathology,
Saarland University,
Bldg. 90.3,
66421 Homburg-Saar, Germany
e-mail: yoo.jin.kim@uniklinikum-saarland.de
R. Ketter
:
W.-I. Steudel
Department of Neurosurgery, School of Medicine,
Saarland University,
Homburg/Saar, Germany
W. Henn
:
K. D. Zang
School of Medicine, Institute of Human Genetics,
Saarland University,
Homburg/Saar, Germany
between the three different gradesare not irrevocable in
individual cases [25]. Regarding the recurrence rate of up
to 20% in histologically benign meningiomas [8] on the
one hand, and the variable recurrence rate among atypical
meningiomas on the other, accurate risk estimation in
individual cases based on histomorphologic criteria alone
remains problematic. Apart from histomorphologic param-
eters, incompleteness of tumor resection is known to be
the major risk factor of recurrence. However, 720% of
gross totally resected meningiomas tend to recureven
phenotypically benign tumors. Therefore, an adequate
prognosis of biological behavior and individual assess-
ment of recurrence risk requires additional prognostic
factors or markers.
It is well-known, that high expression of immunohisto-
chemically detectable markers of proliferative activity, e.g.
Ki-67 antigen, indicates a more aggressive behavior and
thus, a higher risk of recurrence. However, the determina-
tion of the independent predictive value of Ki-67 as well as
the appropriate quantitation method still needs to be
cleared.
Cytogenetically, meningiomas are characterized by
monosomy 22 as a typical primary aberration [33] and
progression-associated secondary aberrations, of which
monosomy of the short arm of chromosome 1 (1p) is
the most important. Monosomy 1 or loss of the distal part
of the short arm of chromosome 1 was found to be
correlated with higher histologic grade and with a higher
recurrence rate in meningiomas [6, 10, 12, 18, 32]. The
tissue non-specific isoform of alkaline phosphatase
(ALPL), located on 1p36.1p34, is strongly active in
normal arachnoid cells and in cells of common-type
meningiomas, but in contrast, enzyme activity lacks in
meningiomas with deletion of 1p. Since karyotyping by
means of classic cytogenetics is laborious and not a routine
part of tumor diagnostics, the histochemical detection of
the loss of ALPL was introduced by Niedermayer et al.
[20] as a fast and low-cost technique for the detection of
1p. The gene, encoding ALPL, is suggested as a putative
and not yet identified tumor suppressor gene on chromo-
some 1p. Beyond its role as enzyme marker for 1p, loss
of ALPL activity could be applied as a predictor of
recurrence, as it has been recently demonstratedat least
in univariate mannerby Bouvier et al. [4]. However,
neither the independent predictive value of 1p and loss of
ALPL expression nor the correlation of ALPL to 1p has
yet been clarified in a multivariate setting and among a
large cohort.
The presented study aims to identify independent
prognostic factors and to reveal complex interrelationships
among clinical data, histomorphologic parameters, prolifer-
ative marker, histochemical expression of ALPL, and
cytogenetic parameters in a multivariate approach.
Materials and methods
Patients and clinical data
The retrospective cohort study is based on the data of 265
patients with initial diagnosis of meningioma operated on
19972002 at the Department of Neurosurgery, Saarland
University. This subset consists of all meningioma patients
within the given period where vital tumor tissue was available
for histochemical detection of alkaline phosphatase. Mean age
(SD; range) was 59 (13 years; 488 years). The cohort
consisted of 186 females and 79 males (sex distribution 7:3).
13 patients (median age at diagnosis 61 years; SD 9.6; range
5178) died because of disease within a mean period of
11 months after surgery. Patients were followed-up for a mean
period of 44 months (range 696 months). After 6 months, 59
patients were lost to follow-up and were not considered in the
statistical analyses. Clinical data recorded included dates of
birth and death, gender, tumor location, date of surgery for
primary and recurrent tumor resections, date of recurrence/
regrowth as defined radiologically (CT or MRI), extent of
resection, and grossly apparent invasion of adjacent struc-
tures. The extent of resection was documented along the
guidelines suggested by Simpson [28]. Complete surgical
extirpation of the tumor (gross total resection: GTR) was
defined as Simpson grade I or II, corresponding to macro-
scopically determined complete tumor resection with bipolar
coagulation of the dural insertion.
Histomorphology
All available slides from primary meningiomas as well as
additional hematoxylin and eosin or specially stained slides
in all cases were reviewed by two independent pathologists
(Y.K. and W.F.) who were unaware of the patients
outcomes. When there were discrepancies for parameters
between the two investigators, the case was examined
simultaneously by both observers and the results were
recorded after an agreement was reached. According to the
predominant feature or growth pattern, each tumor was
assigned to a histologic prototypical guise (meningo-
theliomatous, fibrous, transitional, microcystic, secretory,
lymphoplasmocyte-rich, xanthomatous or otherwise meta-
plasia, atypical, clear cell, chordoid, papillary, rhabdoid,
and anaplastic) and graded according to the current WHO
criteria [14]. Further histologic parameters, including
hypercellularity, nuclear polymorphism, macronucleoli,
small cell formation with high nuclear to cytoplasmic ratio,
patternless architecture, inflammatory infiltrate, macro-
phages/foamy cells, calcification, necrosis, brain invasion,
invasion of structures other than the brain, and mitotic
index were evaluated in each case. Brain invasion was
defined as irregular projections of tumor or tumor cells into
530 Virchows Arch (2006) 449:529538
adjacent central nervous system parenchyma without an
intervening layer of leptomeninges. Except for mitoses, all
parameters were recorded as either absent or present. The
mitotic index (MI) was defined as the highest count in 10
consecutive high-power fields (HPF; 1 HPF=0.16 mm
2
).
According to the mitotic thresholds as suggested by Perry
et al. [22] and adopted by the WHO, a MI <4 was defined
as 0, MI greater or equal to 4 and less than 20 as 1, and
MI 20 as 2.
Ki-67 immunohistochemical analysis
For each case, formalin-fixed and paraffin-embedded speci-
mens were sectioned at 3 m, deparaffinized, and treated
with microwave antigen retrieval (210 min at 700 W and
350 W, respectively). Slides were incubated for 1 h in a
humid chamber at 37C with a monoclonal antibody
directed against Ki-67 (clone KiS5, 1:50; DakoCytomation,
Glostrup, Denmark). Biotinylated secondary antibodies
were detected using the streptavidin/horseradish peroxidase
method and diaminobenzidine as chromogen. Percentage of
immuno-positive tumor cells were determined in digital
images of 5 HPF of the highest labeling in each case using
an automated image analysis device as described by Kim et
al. [11]. The computer-based method is described and
provided for free at the morphometry section of http://www.
uniklinikum-saarland.de/neuropathologie [11].
Histochemical reaction of ALPL
For each case, histochemical detection of alkaline phosphatase
was performed in cryostat sections and/or smear preparations
Fig. 1 Histochemical expression patterns of ALPL (tissue non-
specific alkaline phosphatase) in meningiomas. (a) Positive enzyme
activity in meningothelial tumor cells. Cryostat section of a common-
type meningothelial meningioma, objective 10. (b) Same tumor
specimen, close up view of ALPL detection obtained in smear
preparation, objective 40. Strong enzyme activity, in particular, in
whorl formations. (c) Atypical meningioma with partial loss of ALPL
activity, displaying an alternate pattern of ALPL expression, objective
10. Well-differentiated whorl formations exhibiting positive enzyme
activity alternate with less differentiated tumor areas with loss of
ALPL activity. (d) Meningioma with complete lack of ALPL activity.
Only the endothelium of blood vessels serving as intrinsic positive
control exhibit enzyme activity (objective 10)
Virchows Arch (2006) 449:529538 531
as described previously by Niedermayer et al. [20]. Briefly,
histochemical detection of ALPL was done at room
temperature with hexazotized triamino-tritolyl-methane-chlo-
ride (Neufuchsin; Merck, Darmstadt, Germany) and naph-
thol-AS-MX phosphate (Sigma-Aldrich, Buchs, Switzerland)
at a pH of 8.8. Staining reaction was stopped after 30 min.
Positive reactions in the endothelium of the blood vessels
served as intrinsic positive control. According to the staining
pattern, meningiomas were assigned to (1) ALPL positive
with uniform reaction in all tumor cells (Fig. 1a,b); (2) partial
loss of ALPL expression with loss of enzyme activity in
more than 50% of tumor cells in clearly identifiable tumor
areas (Fig. 1c), as suggested by Bouvier et al. [4]; and (3)
complete lack of ALPL expression in tumor cells (Fig. 1d).
Karyotyping
Tissue specimens from tumors were obtained freshly during
surgery and processed according to previously described
methods [13]. Briefly, the cell cultures were grown in
Dulbeccos modified Eagles minimal essential medium
supplemented with 10% fetal calf serum and antibiotics.
Chromosome preparation from primary cultures, GTG-
banding, and karyotyping followed standard procedures. In
197 meningiomas of our study group, cytogenetic findings
were obtained. These cases, which represent a part of the
cytogenetic meningioma database, founded and character-
ized by Zang [31], and continuously investigated by Ketter
et al. [10], were subdivided into five cytogenetic subgroups:
(1) meningiomas with normal chromosome set; (2) with
monosomy 22 as the sole cytogenetic aberration; (3)
markedly hypodiploid meningiomas with loss of further
autosomes in addition to monosomy 22, but with intact
chromosome 1; (4) tumors with deletions of the short arm of
a chromosome 1 (1p); and (5) the rare cyogenetic subtype
of meningiomas with hyperdiploid chromosome set, with or
without detectable loss of chromosome 22 or 1p.
Statistical analysis
Survival times and recurrence-free survival times were
computed and distributions were estimated using the
KaplanMeier method. Univariate correlation of each
Table 1 Association of clinicopathological parameters, Ki-67 LI, ALPL expression, and cytogenetic findings with meningioma recurrence in 206
patients; univariate analysis, chi-square test
Relative recurrence risk (%) p
Clinical parameters
Age >65 years 19 0.0011
Sex (female vs male) 11 vs 19 n.s.
Convexity vs other locations 12 vs 9 (WHO I:8 vs 3) 0.02 (0.0152)
GTR vs STR 12 vs 47 <0.001
Local infiltration 28 <0.001
STR and GTR GTR only
Histopathologic parameters
Hypercellularity 26 22 0.004 (<0.001)
Small cell formation 25 23 0.003 (<.001)
Macronucleoli 32 25 <0.001
Necrosis 33 29 <0.001
Brain invasion vs non-invasion 28 vs 10 0.03
WHO grade
II vs I 18 vs 9 18 vs 5 0.09 (0.004)
III vs I and II 75 vs 11 75 vs 8 <0.001
Mitotic index
4/10 HPF 38 39 <0.001
20/10 HPF 83 83 <0.001
Ki-67 LI
>8% 29 27 0.006 (<0.001)
>12% 42 42 0.0052 (<0.001)
ALPL
At least partial loss 21 18 0.004 (<0.001)
Complete loss 27 25 0.003 (<0.001)
Karyotype
1p vs others 45 45 <0.001
Ki-67 LI=Ki-67 labeling index; ALPL=tissue non-specific alkaline phosphatase; p=p value of chi-square test; n.s.=not significant; GTR=gross
total resection; STR=subtotal resection; 1p=partial or complete deletion of the short arm of chromosome 1; p values calculated for the
GTR subset only are put in parenthesis
532 Virchows Arch (2006) 449:529538
parameter regarding recurrence was analyzed by using the
chi-square test (Table 1). For this purpose, all qualitative
variables were dichotomized and recorded as either absent
or present. Receiver operator characteristic (ROC) was
performed for Ki-67 LI, to define cut off values and for
dichotomization. Threshold value for patients age was
defined at 65 years (65 years=0; >65 years=1). Cox
proportional hazard modeling techniques were used to
identify independent predictors of recurrence. Age adjusted
risk estimations were calculated for (1) the entire cohort and
(2) only WHO grade I and II meningiomas, excluding
anaplastic meningiomas.
Results
Based on strict histological criteria according to the
current WHO grading system [14], 227 tumors were
classified as WHO grade I, 30 tumors as WHO grade II,
and 8 tumors as WHO grade III meningiomas.
There were no significant differences in the follow-up
periods for different WHO grades or brain-invasive meningi-
omas: We revealed a mean follow-up period of 44 months in
the cohort of WHO grade I meningiomas, 46 months in grade
II meningiomas, and 44 months in grade III meningiomas.
The mean follow-up period for the subset of brain-invasive
meningiomas was 55 months and in detail, 57 months for
WHO grade I brain-invasive meningiomas and 51 months for
WHO grade II or III brain-invasive meningiomas.
A gross total resection (GTR) was reached in 240 cases,
in 25 tumors only a subtotal resection (STR) was achieved.
Without exception, the latter were assigned to WHO grade I
tumors. The majority of these tumors, which underwent
STR was located at the base of the skull (18/25).
The overall recurrence rate was 11% (n=31). The
relative risk of recurrence (RR) was significantly higher
for meningiomas with STR (9/25, RR=36%) than for those
having undergone GTR (22/240, RR=9%). Relapses were
evident after a mean period of 45 months; 48 months after
GTR and 36 months after STR, respectively. Tumor
location at convexity was significantly associated with loss
of ALPL activity and with 1p or hyperdiploidy. In the
subset of WHO grade I meningiomas, convexity meningi-
omas exhibited a significantly higher risk of recurrence than
meningiomas in other locations (p=0.0152).
The recurrence rates in the WHO grade I meningiomas
was 5% regarding only the GTR subset (11/204), 9% (20/
229) regarding also the STR subset. Recurrences were
recorded in 18% (5/28) of WHO grade II meningiomas and
75% (6/8) of anaplastic meningiomas. Brain invasions were
observed in 14 tumors (9 WHO grade I, 2 grade II, and 3
grade III meningiomas) within 42 cases where central
nervous system parenchyma was assessable. Recurrence
rate among the few brain-invasive meningiomas was 28%
(4/14). All of these recurrent brain-invasive tumors were
either atypical or anaplastic.
ROC revealed a Ki-67 LI >8% as a cut off point with a
convenient sensitivity/specificity ratio (0.63 to 0.65),
Fig. 2 Recurrence-free survival
based on Ki-67 labeling index
(LI) in the subset of gross totally
resected meningiomas, n=181
Virchows Arch (2006) 449:529538 533
however, sufficient specificity was reached at a threshold
value up to LI 12% (specificity >98%). The relative area
under the ROC-curve was 0.708. The MannWhitney U
test for testing the discriminatory power of Ki-67 LI was
significant (p<0.05).
The results of the univariate analyses of clinicopatho-
logical parameters and cytogenetic findings are summarized
in Table 1.
Independent prognostic factors for recurrence-free sur-
vival within the entire cohort (GTR and STR) as revealed
by Cox regression included (1) anaplasia (p<0.001), (2)
almost always in combination with MI 20/10 HPF
(p=0.0062), (3) 1p (p=0.0077), (4) Ki-67 LI >12%
(p=0.0181), and (5) STR (p=0.0037) as further indepen-
dent predictors. These variables (excluding STR) provided
prognostic information also within the subset of tumors that
underwent GTR and adjusted for age, tumor site, and
histomorphologic parameters. Ki-67 LI >8% was found to
be univariately associated with recurrence, but did not reach
significance in the multivariate modeling. However, signif-
icance in the multivariate setting was reached when the cut
off point was scaled up to 12%. Recurrence-free survival
Fig. 3 Recurrence-free survival
based on cytogenetic findings
in the subset of gross totally
resected meningiomas, n=179;
part of the cytogenetic meningi-
oma data base, Institute of
Human Genetics, Saarland Uni-
versity, described in detail by
Zang [31] and Ketter et al. [10]
Table 2 Karyotype and pathologic findings
Cytogenetic subsets (n=197)
a
Normal (n=95) Monosomy 22 (n=45) Hypodiploid (n=19) 1p (n=31) Hyperdiploid (n=7)
Tumorsite
Convexity 32 (33%) 22 (48%) 11 (57%) 25 (80%)
b
5 (71%)
WHO grade
I 87 (92%) 43 (96%) 15 (79%) 14 (45%) 5 (71%)
II 8 (8%) 2 (4%) 4 (21%) 11 (35%) 1 (14%)
III 0 (0%) 0 (0%) 0 (0%) 6 (19%) 1 (14%)
Histologic subtype
Meningothelial 40 (41%) 8 (18%) 3 (16%) 3 (10%) 0 (0%)
Fibroblastic 10 (10%) 26 (57%)
b
7 (37%) 8 (26%) 2 (29%)
Transitional 6 (6%) 8 (18%) 4 (21%) 1 (3%) 0 (0%)
Median Ki-67 LI [%] range 3.5 (024) 4.8 (0.715.3) 5.8 (013.3) 6.9 (035) 4 (09.4)
Recurrence
c
(recurrence rate) 3/71 (4%) 0/29 (0%) 2/18 (11%) 10/22 (45%) 1/5 (20%)
Ki-67 LI=Ki-67 labeling index
a
Part of the cytogenetic meningioma data base, Institute of Human Genetics, Saarland University, founded and characterized by Zang [31] and
described in detail by Ketter et al. [10]
b
Significant chi-square test statistics
c
Only cases followed-up at least 6 months after surgery, n=154
534 Virchows Arch (2006) 449:529538
based on Ki-67 LI is displayed in Fig. 2. Cox regression
modeling regarding only the subset of totally resected
WHO grade I meningiomas revealed 1p as the only
independent predictor of recurrence (p=0.0077). Recur-
rence-free survival based on cytogenetic findings is
displayed in Fig. 3. Among specific histologic WHO grade
I variants, there was no evidence of statistically significant
associations with recurrence.
Cytogenetically, fibroblastic meningiomas had a
significantly higher rate of monosomy 22 as compared to
the other subsets (p<0.05) (Table 2). Meningiomas with
1p showed a higher proliferative potential in terms of Ki-
67 LI as well as significantly higher rate of histologic
dedifferentiation.
Partial or complete loss of ALPL was strongly
associated with a higher WHO grade (II or III) and
higher Ki-67 LI (p<0.025). A partial or complete loss of
ALPL expression indicated 1p with a sensitivity of 90%
and a positive predictive value of 90%, however, speci-
ficity was low (44%). Loss of enzyme activity was
significantly associated with tumor recurrence in the
univariate analysis (Table 1), whereas no significant
contribution was assessable in the multivariate modeling.
Associations of ALPL expression pattern to tumor site,
histopathologic parameters, and cytogenetic findings are
displayed in Table 3.
Discussion
The appropriate risk stratification in individual meningioma
patients based on pathological features without overgrading
or undergrading is difficult at best [5]. Among histologic
parameters, high cellularity, invasion of the brain, large
number of mitoses, prominent nucleoli and cell necrosis
appear to correlate with aggressive behavior [9]. The three-
step grading system of meningiomas, distinguishing well-
differentiated, atypical, or anaplastic meningiomas along
the guidelines suggested by Perry et al. [22] and adopted by
the WHO [14] admits a risk estimation based on histo-
morphologic criteria. However, the precise grading of
meningiomas is not always possible and treatment planning
will consider clinical variables [5] and likewise, the post-
operative treatment regime will consider further prognostic
factors, which may predict the individual clinical course.
Among these potential risk factors, incomplete removal is
well-known as a prognostic unfavorable finding, as it
could be confirmed in the multivariate statistics presented
here. Subtotal tumor resection was reported as a prognos-
tic indicator, independent from the histologic grade
already used by Simpson [28] and subsequently, by others
[8, 15, 22].
However, on the other hand, independent from the extent
of resection, meningiomas with histologic anaplasia and/or
high mitotic activity (20/10 HPF) exhibit a very high risk
of recurrence. The recurrence rate of 75% in our series of
anaplastic meningiomas is comparable with the recurrence
rates of 78% reported by Jskelinen et al. [8] or 72% by
Maier et al. [15].
Brain invasion suggests a greater likelihood of recur-
rence and as reported by Perry et al. [22], brain-invasive
histologically common-type meningiomas behave similarly
to atypical meningiomas in general. We also found a
significantly higher rate of recurrence in brain-invasive
Table 3 ALPL expression pattern and pathologic and genetic factors
ALPL expression (n=265)
Positive (n=97) Partial loss (n=98) Complete loss n=70) p
Tumorsite
Convexity 21 (21%) 62 (63%) 50 (71%) <0.001
a
WHO grade
I 94 (97%) 43 (96%) 15 (79%) <0.001
II 2 (2%) 2 (4%) 4 (21%)
III 1 (1%) 0 (0%) 0 (0%)
Median Ki-67 LI [%] (range) 3 (0.218.1) 5 (0.525) 5 (0.424)
Cases with Ki-67 LI >8% 8 (8%) 15 (15%) 17 (24%) <0.025
Cytogenetics (n=197) <0.001
Normal karyotype 54 (71%) 25 (35%) 16 (33%)
Monosomy 22 16 (21%) 15 (15%) 7 (14%)
Marked hypodiploid 3 (4%) 8 (11%) 8 (16%)
1p 3 (4%) 13 (18%) 15 (31%)
Hyperdiploidy 0 (0%) 4 (6%) 3 (6%)
ALPL=tissue non-specific alkaline phosphatase; Ki-67 LI=Ki-67 labeling index; 1p=partial or complete deletion of short arm of chromosome 1;
p=p value of chi-square test
a
Partial or complete loss compared to positive ALPL cases, respectively
Virchows Arch (2006) 449:529538 535
meningiomas (relative risk: 28%), but in contrast, all of the
recurrent meningiomas with brain invasion were of atypical
or anaplastic type whereas none of the brain-invasive
WHO grade I meningiomas recurred. Due to the confound-
ing in histologic grade, brain invasion did not reach
significance in the multivariate modeling. It has to be
mentioned, that the number of recorded brain invasion
among the investigated cohort was small, thus, limiting the
significance of these findings. However, in our series,
atypical and anaplastic meningiomas are significantly over-
represented in the subset of brain-invasive meningiomas,
suggesting, that brain invasion rate may increase with the
histologic grade of dedifferentiation.
Patients age at diagnosis is another reported prognostic
factor. Statistic analyses of over 9,000 patients with
meningioma based on the National Cancer Data Base
revealed patients age at diagnosis 65 years [16] as a
multivariately significant negative prognostic factor. Con-
cordantly, in our study, age at diagnosis showed a highly
significant association with shorter recurrence-free survival
in univariate analysis, although age did not reach signifi-
cance in the multivariate modeling.
Among the histologic features, mitotic activity is
known to exhibit the most powerful prognostic impact.
Perry et al. [22] have found a notable increase in the rate
of recurrence among totally resected meningiomas with
four or more mitoses per 10 high-power fields. This
mitotic threshold is adopted as a grading criterion for
atypical meningiomas in the current WHO classification.
Despite a univariate association with recurrence, the
multivariate analysis presented here did not reveal any
independent predictive capability at a mitotic threshold
4/10 HPF. However, there was no doubt in the prognostic
impact of apparently increased mitotic activity with the
excess of MI 20. The overlap in the ranges of mitotic
indices among different studies are well-known and may
be due to different counting techniques, variation in field
sizes, institutional bias, and inter-observer variability [5,
22]. Thus, these data reflect the need for careful
interpretation of MI in meningioma grading, in particular,
if mitotic thresholds are suggested as single grading
criterion only.
A quantitative approach in the assessment of cell
proliferation is the determination of the Ki-67 LI. The
prognostic impact of Ki-67 LI is indisputable, since many
studies have demonstrated the correlation between prolif-
eration index and tumor grade and/or tumor recurrence in
meningiomas [1, 7, 12, 21] and finally, since the
prognostic role of Ki-67 is advocated in the current
WHO classification [14]. We have found a Ki-67 LI
>12% as independent predictor of recurrence adjusted for
age, tumor site, and histologic parameters, including
mitotic index. The proposed threshold value is comparable
with the threshold values of Ki-67 LI >10%, proposed by
Ho et al. [7] and recently by Torp et al. [30] for
distinguishing recurrent vs non-recurrent meningiomas.
In atypical and anaplastic meningiomas, Ki-67 LI allows
to identify meningiomas at high risk of recurrence
independently from histomorphology or mitotic activity.
However, since levels of Ki-67 LI exceeding 10% are
implausible in meningiomas of conventional type, this
criterion is not suited to distinguish recurrent vs non-
recurrent WHO grade I meningiomas. This is supported by
the findings of Roser et al. [24] who investigated the
prognostic significance of Ki-67 LI in 600 meningiomas.
They came to the same conclusion, that Ki-67 is not a
statistically significant predictor of survival time in totally
resected WHO grade I meningiomas. The determination of
Ki-67 LI is a subjective task and the counting techniques
vary, which leads to a remarkable overlap in the values
recorded for typical, atypical, and anaplastic lesions, as well
as for recurring vs non-recurring tumors among reported
data [1, 7, 12, 19, 21, 30]. Therefore, it remains problematic
to suggest universal Ki-67 LI values for determining
recurrence risk [14]. The standardization of the Ki-67 LI
assessment and computed image analysis approach may lead
to more robust and comparable data.
In contrast to the mentioned prognostic parameters, we
found deletion of 1p as the only independent predictor of
recurrence also among WHO grade I meningiomas. This is
supported by Al-Mefty et al. [2] who demonstrated a
complex karyotype, including loss of 1p present ab initio in
histologically lower-grade tumors, which later on recurred
and upgraded to a higher-grade tumor contradicting the
popular model of stepwise clonal evolution. These and the
presented data confirm our previously reported findings
[10], which emphasize the prognostic evidence of loss of 1p
independently from histologic grade. Our data also suggest,
that the histochemical detection of ALPL expression is an
appropriate, fast, and low-cost as well as specific and
sensitive screening method for identification of those cases
with 1p. We therefore could confirm our previously
reported thesis [20] and the recent findings of Bouvier et
al. [4], that ALPL, which is encoded in 1p36.1p34, could
be applied as enzyme marker for 1p. Niedermayer et al.
suggested one or more putative tumor suppressor genes
located on the distal, ALPL encoding region of chromo-
some 1 [20]. The strong correlation between ALPL activity
loss and monosomy 1p led also to the presumption that the
ALPL gene itself might act as a tumor suppressor gene,
with one allele being deleted and the remaining ALPL
allele inactivated by deletion, mutation, or genomic
imprinting [18]. Updated genomic sequencing studies
support the hypothesis of a two-step inactivation of the
ALPL gene during progression of meningiomas by deletion
plus promoter methylation [23]. Recent studies have
536 Virchows Arch (2006) 449:529538
demonstrated consistent regions of deletion on 1p36.21
[17] and on 1p33p34 [29] as well as on 1p34p32 [3],
suggesting that putative suppressor genes are actually
located on 1p36p32. Apart from the significant correlation
to 1p, our data also suggest that loss of ALPL is
associated with a higher histologic grade, higher prolifer-
ation potential, and higher rate of recurrence. However, in
contrast to the univariate modeling as mentioned by
previous reports [4, 20], loss of ALPL activity did not
contribute significantly to the prediction of recurrence in
the multivariate modeling. On the one hand, this may be
founded in the remarkable loss of histochemical reaction
due to the susceptibility of this histochemical reaction to the
state of preservation of tumor tissue, which leads to an
inferior predictive potential compared to confounding
predictors. On the other hand, equivocal representativeness
of cells growing in cell culture, low percentage of cells with
1p and also small deletions on chromosome 1p cause low
sensitivity of classic cytogenetic methods. This results in a
misleading discrepancy between ALPL reaction and cyto-
genetic findings and in individual cases the loss of ALPL
activity do not correspond to a deletion of 1p, because the
presumptive deletion was not detected by means of classic
metaphase cytogenetic methods. Interphase techniques such
as fluorescence in situ hybridization (FISH) are more
sensitive and accurate in the detection of the deletion of
1p and may result in a higher concordance between ALPL
expression and genetic findings. For the comparability of
data, we applied the cut off point of at least 50% enzyme
negative tumor cells for heterogenous pattern of ALPL
expression (ALPL+/) as suggested by Bouvier et al. [4].
Regarding the arbitrary assessment of this cut off point on
the one hand and the low sensitivity of the metaphase
cytogentic method on the other, the moderate concordance
between ALPL+/ and detectable loss of 1p has to be
reconceived. Despite its minor role as independent risk
factor itself, we suggest that ALPL expression keeps a key
role as a sensitive screening method for 1p, in particular if
the enzyme activity is lacking in a histologically common-
type meningioma.
Apart from 1p, allelic losses of other chromosomal
arms, e.g. 10q or 14q, are further cytogenetic aberrations
suggested to be progression-associated [27]. In the pre-
sented study, these aberrations were subsumed in the group
markedly hypodiploid meningiomas with intact 1p. A
more detailed and differentiated analysis of specific
chromosomal aberrations is given by Zang [31]. A multi-
parametric analysis in more than 600 cytogenetically
characterized cases is the subject of current studies and
will provide a more instructive insight into the genetics and
biology of meningiomas (Ketter et al., preliminary data
presented at the Meeting of the Neurooncology Section of
the German Society of Neurosurgery, October 2005).
Conclusions
Anaplasia, incomplete tumor resection, Ki-67 LI >12%, and
1p are independent predictors of recurrence in meningio-
mas. Independent from mitotic activity, a Ki-67 LI >12%
indicates a high risk of recurrence among atypical or
anaplastic meningiomas. Deletion of the short arm of
chromosome 1 also indicates histologically common-type
meningiomas with higher risk of recurrence. The fast and
low-cost histochemical detection of ALPL expression is
suggested as a highly specific screening method for
potential 1p. In particular, recurrence prone meningiomas
of histologically benign or borderline phenotype could be
therefore identified sensitively by ALPL detection followed
by 1p in situ hybridization.
Acknowledgement The authors gratefullythankMr. CraigWashington
for the editorial help.
References
1. Abramovich CM, Prayson RA (1999) Histopathologic features
and MIB-1 labeling indices in recurrent and nonrecurrent
meningiomas. Arch Pathol Lab Med 123:793800
2. Al-Mefty O, Kadri PA, Pravdenkova S, Sawyer JR, Stangeby
C, Husain M (2004) Malignant progression in meningioma:
documentation of a series and analysis of cytogenetic findings.
J Neurosurg 101:210218
3. Bello MJ, de Campos JM, Vaquero J, Kusak ME, Sarasa JL, Rey
JA (2000) High-resolution analysis of chromosome 1p alterations
in meningioma. Cancer Genet Cytogenet 120:3036
4. Bouvier C, Liprandi A, Colin C, Giorgi R, Quilichini B, Metellus
P, Figarella-Branger D (2005) Lack of alkaline phosphatase
activity predicts meningioma recurrence. Am J Clin Pathol
124:252258
5. Burger PC, Scheithauer BW, Vogel FS (eds) (2002) Surgical
pathology of the nervous system and its coverings. Churchill
Livingstone, New York
6. Cai DX, Banerjee R, Scheithauer BW, Lohse CM, Kleinschmidt-
Demasters BK, Perry A (2001) Chromosome 1p and 14q FISH
analysis in clinicopathologic subsets of meningioma: diagnostic
and prognostic implications. J Neuropathol Exp Neurol 60:
628636
7. Ho DM, Hsu CY, Ting LT, Chiang H (2002) Histopathology and
MIB-1 labeling index predicted recurrence of meningiomas. A
proposal of diagnostic criteria for patients with atypical meningi-
omas. Cancer 94:15381547
8. Jskelinen J, Haltia M, Servo A (1986) Atypical and anaplastic
meningiomas: radiology, surgery, radiotherapy, and outcome. Surg
Neurol 25:233242
9. Kepes JJ (1986) Presidential address: the histopathology of
meningiomas. A reflection of origins and expected behavior?
J Neuropathol Exp Neurol 45:95107
10. Ketter R, Henn W, Niedermayer I, Steilen-Gimbel H, Konig J,
Zang KD, Steudel WI (2001) Predictive value of progression-
associated chromosomal aberrations for the prognosis of meningi-
omas: a retrospective study of 198 cases. J Neurosurg 95:601607
11. Kim YJ, Romeike BFM, Uszkoreit J, Feiden W (2006) Automated
nuclear segmentation in the determination of the Ki-67 labeling-
index in meningiomas. Clin Neuropathol 25:6773
Virchows Arch (2006) 449:529538 537
12. Kolles H, Niedermayer I, Schmitt C, Henn W, Feld R, Steudel WI,
Zang KD, Feiden W (1995) Triple approach for diagnosis and
grading of meningiomas: histology, morphometry of Ki-67/
Feulgen stainings, and cytogenetics. Acta Neurochir (Wien)
137:174181
13. Limon J, Dal Cin P, Sandberg AA (1986) Application of long-
term collagenase disaggregation for the cytogenetic analysis of
human solid tumors. Cancer Genet Cytogenet 23: 305313
14. Kleihues P, Cavenee WK (eds) (2000) Pathology and Genetics,
Tumours of the nervous system. IARC Press, Lyon
15. Maier H, Ofner D, Hittmair A, Kitz K, Budka H (1992) Classical,
atypical, and anaplastic meningioma: three histopathological
subtypes of clinical relevance. J Neurosurg 7:616623
16. McCarthy BJ, Davis FG, Freels S, Surawicz TS, Damek DM,
Grutsch J, Menck HR, Laws ER Jr (1998) Factors associated
with survival in patients with meningioma. J Neurosurg 88:
831839
17. Murakami M, Hashimoto N, Takahashi Y, Hosokawa Y, Inazawa
J, Mineura K (2003) A consistent region of deletion on 1p36 in
meningiomas: identification and relation to malignant progression.
Cancer Genet Cytogenet 140:99106
18. Mller P, Henn W, Niedermayer I, Ketter R, Feiden W,
Steudel WI, Zang KD, Steilen-Gimbel H (1999) Deletion of
chromosome 1p and loss of expression of alkaline phosphatase
indicate progression of meningiomas. Clin Cancer Res 5:
35693577
19. Nakasu S, Li DH, Okabe H, Nakajima M, Matsuda M (2001)
Significance of MIB-1 staining indices in meningiomas. Compar-
ison of two counting methods. Am J Surg Pathol 5:472478
20. Niedermayer I, Feiden W, Henn W, Steilen-Gimbel H, Steudel
WI, Zang KD (1997) Loss of alkaline phosphatase activity in
meningiomas: a rapid histochemical technique indicating progres-
sion-associated deletion of a putative tumor suppressor gene on
the distal part of the short arm of chromosome 1. J Neuropathol
Exp Neurol 56:879886
21. Perry A, Stafford SL, Scheithauer BW, Suman VJ, Lohse CM
(1998) The prognostic significance of MIB-1, p53, and DNA flow
cytometry in completely resected primary meningiomas. Cancer
82:22622269
22. Perry A, Stafford SL, Scheithauer BW, Suman VJ, Lohse CM
(1997) Meningioma grading. An analysis of histological param-
eters. Am J Surg Pathol 21:14551465
23. Prowald A, Ketter R, Zang K, Steudel WI (2003) Two-step
inactivation of the tissue non-specific alkaline phosphatase
(ALPL) gene during progression of meningiomas by deletion
plus promoter methylation. Genes Chromosomes Cancer 9:81
24. Roser F, Samii M, Ostertag H, Bellinzona M (2004) The Ki-67
proliferation antigen in meningiomas. Experience in 600 cases.
Acta Neurochir 146:3744
25. Rosenblum MK, Bilbao JM, Ang LC (2004) Meningiomas. In:
Rosai J (ed) Rosai and Ackermans surgical pathology. Mosby,
New York, pp 25642572
26. Russell DS, Rubinstein LJ (eds) (1990) Pathology of tumors of
the nervous system. Lippincott Williams and Wilkins, Baltimore
27. Simon M, von Deimling A, Larson JJ, Wellenreuther R, Kaskel P,
Waha A, Warnick RE, Tew JM Jr, Menon AG (1995) Allelic
losses of chromosomes 14, 10, and 1 in atypical and malignant
meningiomas: a genetic model of meningioma progression.
Cancer Res 55:46964701
28. Simpson D (1957) The recurrence of intracranial meningiomas
after surgical treatment. J Neurol Neurosurg Psychiatry 20:2239
29. Sulman EP, White PS, Brodeur GM (2004) Genomic annotation
of the meningioma tumor suppressor locus on chromosome 1p34.
Oncogene 23:10141020
30. Torp SH, Lindboe CF, Gronberg BH, Lydersen S, Sundstrom S
(2005) Prognostic significance of Ki-67/MIB-1 proliferation index
in meningiomas. Clin Neuropathol 24:170174
31. Zang KD (2001) Meningioma: a cytogenetic model of a complex
benign human tumor, including data on 394 karyotyped cases.
Cytogenet Cell Genet 93:207220
32. Zang KD (1982) Cytological and cytogenetical studies on human
meningioma. Cancer Genet Cytogenet 6:249274
33. Zang KD, Singer H (1967) Chromosomal constitution of
meningiomas. Nature 216:8485
538 Virchows Arch (2006) 449:529538