J Neurosurg 109:583–592,

109:000–000, 2008

Intraoperative mapping of language functions: a longitudinal

neurolinguistic analysis
Clinical article

*
Josef Ilmberger, Ph.D.,1 Maximilian Ruge, M.D.,4 Friedrich-Wilhelm Kreth, M.D.,4
Josef Briegel, M.D., 2 Hans-Juergen Reulen, M.D., 3 and Joerg-Christian Tonn, M.D.4
Departments of 1Physical Medicine and Rehabilitation, 2Anesthesiology, and 3Neurosurgery, and
4
Neurosurgical Department, Ludwig Maximilians University, Munich, Germany

Object. This prospective longitudinally designed study was conducted to evaluate language functions pre- and
postoperatively in patients who underwent microsurgical treatment of tumors in close proximity to or within language
areas and to detect those patients at risk for a postoperative aphasic disturbance.
Methods. Between 1991 and 2005, 153 awake craniotomies with subsequent cortical mapping of language
functions were performed in 149 patients. Language functions were assessed using a standardized test battery. Risk
factors were obtained from multivariate logistic regression models.
Results. Language mapping was able to be performed in all patients, and complete tumor resection was achieved
in 48.4%. Within 21 days after surgery a new language deficit (aphasic disturbance) was observed in 41 (32%) of the
128 cases without preoperative deficits. There were a total of 60 cases involving postoperative aphasic disturbances,
including cases both with and without preoperative disturbances. Risk factors for postoperative aphasic disturbance
were preoperative aphasia (p < 0.0002), intraoperative complications (p < 0.02), language-positive sites within the
tumor (p < 0.001), and nonfrontal lesion location (p < 0.001). In patients without a preoperative deficit, a normal
(yet submaximal) naming performance was a powerful predictor for an early postoperative aphasic disturbance (p <
0.0003). Seven months after treatment 10.9% of the 128 cases without preoperative aphasic disturbances continued to
demonstrate new postoperative language disturbances. A total of 17.6% of all cases demonstrated new postoperative
language disturbances after 7 months. Risk factors for persistent aphasic disturbance were increased age (> 40 years,
p < 0.02) and preoperative aphasia (p < 0.001).
Conclusions. Every attempt should be undertaken to preserve language-relevant areas intraoperatively, even
when they are located within the tumor. New postoperative deficits resolve in the majority of patients, which may be
a result of cortical mapping as well as functional reorganization. (DOI: 10.3171/JNS/2008/109/10/0583)

Key Words      •      awake craniotomy      •      brain tumor      •      cortical mapping      •     

neurolinguistic analysis

I
ntraoperative stimulation mapping in the awake pa­
tient, as first described in larger patient series by Pen­
field and Boldrey26 for sensory and motor functions
and by Penfield and Rasmussen27 for language functions, is
considered the gold standard for identifying eloquent cor­
tical sites. Intraoperative stimulation mapping techniques
are usually applied with the goal of reducing the frequency
of postoperative deficits and at the same time maximizing
the extent of tumor resection. In the absence of prospective
data, however, the value of these techniques is difficult to
assess. This difficulty remains particularly true for intra­
operative mapping of language functions, which requires
a most demanding cooperation among the neurosurgical,
Abbreviations used in this paper: AAT = Aachener Aphasia Test;
fMR = functional MR; KPS = Karnofsky Performance Scale; mRS =
modified Rankin scale; WHO = World Health Organization.
* Drs. Ilmberger and Ruge contributed equally to this study.
anesthesiological, and neuropsychological teams. Even
though numerous retrospective studies have reported on
the feasibility of this complex procedure and its possibly
positive influence on clinical outcome scores,3,5–9,12,15,17,20,22
,24,28–30,34,39,43,45,48,49
detailed pre- and postoperative neurolin­
guistic data are scarce for patients undergoing operations
using this method. Moreover, uncertainties continue to ex­
ist as to the prognostic importance of obtaining linguistic
data on language functioning preoperatively, intraopera­
tively, and in the early postoperative period after micro­
surgical treatment.
The current longitudinally designed prospective
study was conducted to evaluate language functions be­
fore and after open tumor resection using a standardized
test battery and to detect prognostic factors associated
with improved/stabilized or deteriorated language func­
tion at the time of early and late follow-up evaluation.
Additionally, associations between intraoperative stimu­

J. Neurosurg. / Volume 109 / October 2008 583


lation mapping results of language functions and the ex­
tent of resection were analyzed.
Methods
Study Eligibility
Patients were considered eligible for this prospec­
tive longitudinal study (conducted from 1991 to 2005) if
they: 1) had untreated or recurrent lesions near or within
suspected language-related brain areas; 2) were between
15 and 70 years of age; 3) were fluent in speaking and
understanding the German language; 4) did not display
severe aphasia prior to surgery; 5) were not mentally dis­
oriented; and 6) did not present with risk factors for an
awake craniotomy (such as claustrophobia).
Study Protocol
Preoperative and Postoperative Neurolinguistic Eval-
uation. Formal neurolinguistic testing was performed once
preoperatively (1–3 days before the operation) and twice
postoperatively (within 21 days and 1 year after surgery)
using the AAT.11 The AAT consists of 5 subtests: 1) the
Token Test, involving pointing to and manipulating geo­
metric forms on verbal command; 2) repetition, consisting
of repeating phonemes, words, and sentences; 3) written
language, consisting of reading and writing; 4) naming,
involving naming visually presented objects, colors, and
scenes; and 5) comprehension, involving choosing the right
item in a set with distractor items after spoken/written de­
scription. In each of these subtests, disturbances are classi­
fied as none/minimal, mild, moderate, or severe according
to the norms of the AAT. For overall classification within
the framework of this study, patients were classified as hav­
ing an aphasic impairment whenever they showed at least
mild disturbance in 1 of the subtests, and severity of this
aphasic impairment was then rated according to the scores
achieved in the Token Test. If formal neurolinguistic test­
ing could not be performed at the time of follow-up evalua­
tion, language performance was classified (aphasia yes/no)
according to the clinical data.
Patient Information and Preparatory Training. Con­
frontation naming (naming of pictures of objects) was
the basic task used in our mapping procedures, because
naming is a core component in language abilities and has
proven to be a reliable and robust method for identify­
ing widespread essential language sites.20 Within 3 days
prior to surgery, patients were informed in detail about
the intraoperative stimulation monitoring procedure by
the neuropsychologist/neurolinguist also responsible for
intraoperative testing. The patient was familiarized with
the stimulus material and trained to perform the nam­
ing task. The intraoperatively applied stimulus material
was custom-tailored to the individual capabilities of each
patient; if a patient had difficulties recognizing or nam­
ing certain objects, these items were removed from the
stimulus pool.
Anesthesiological Monitoring and Management. Mon­
itoring included electrocardiography with ST segment anal­
584
J. Ilmberger et al.
ysis of leads II and V5 and pulse oximetry. Spontaneous
breathing was monitored using side-stream capnometry. A
flexible tube with a foam tip was placed in a nostril allowing
CO2 monitoring while the patient was breathing and speak­
ing. In addition, continuous transcutaneous measurement of
PO2 and PCO2 was installed. A radial catheter was placed
for invasive blood pressure monitoring and blood gas analy­
sis.
Analgesia and sedation were induced and maintained
by continuous infusion of propofol (0.5–1.2 mg/kg) and
remifentanil (0.05–0.01 μg/kg/min) with the dose titrated
according to the patients’ requirements and guided by
the parameters of the ventilation and gas exchange. Any
increase in end-tidal PCO2 > 50 mm Hg was addressed
by reduction or cessation of the remifentanil and/or the
propofol infusion. The procedure was designed to main­
tain an optimal and reproducible performance of the pa­
tient during each step of the test procedure.
In patients not yet treated with anticonvulsant drugs,
750 mg phenytoin was administered intravenously at the
beginning of surgery and then tapered during the next 2
postoperative days.
Patient Positioning. The major goal of patient posi­
tioning was to maximize the intraoperative comfort of
each patient (to reduce distress and anxiety), to allow prop­
er visualization of the screen presenting the stimuli, and
to comply with the requirements of the surgeon. On the
operating room table, the patient was placed in a comfort­
able position supported by pillows. The head was elevated
almost above the thorax level, and turned about 60° to the
right or left side. To avoid the usual head clamp, fixation of
the patient’s head and neck was achieved by using a vacu­
um pillow (Schmidt GmbH) comparable to those used for
stabilization of trauma patients. After disinfection, a local
anesthetic (60–80 ml bupivacaine hydrochloride 0.25%)
was infiltrated at the site of the planned skin incision, in­
cluding the supraorbital area and the zygomatic arc.
Drapes were fixed by sutures surrounding the opera­
tive site to avoid misplacement in case of uncontrolled head
movement. After positioning the neuropsychological mon­
itoring unit on the side contralateral to the patient’s lesion,
the drapes were arranged in such a way that the patient was
able to recognize the presented stimuli on a liquid crystal
display monitor. In case of significant refraction anomalies,
the earpiece of the patient’s glasses was unscrewed on the
side of the craniotomy, allowing the use of glasses during
surgery.
Intraoperative Cortical Stimulation
An enlarged pterional craniotomy was performed.
Numbers printed on sterilized paper (2 × 3 mm in size) were
placed on the exposed cortical surface. Prior to stimulation
the operative site was documented using photography. For
direct cortical stimulation we used a Nicolet Viking IV P
8-channel monitoring unit and bipolar stimulation forceps
(Fischer Leibinger). Electrophysiological parameters for
cortical stimulation were square wave, 50/second, for 0.2
msec. Stimulation intensities ranged from 8 to 16 mA. In
general, all sites were stimulated with the lowest intensity
first; if necessary, stimulation intensity was incrementally
J. Neurosurg. / Volume 109 / October 2008
Neurolinguistic analysis of intraoperative language mapping
in­creased by 2 mA. For validation of the mapping proce­
dure, stimulation was repeated for all sites. Moreover, a
photograph of the cortical surface with the numbers in
place was obtained immediately after the stimulation to
be able to identify sites of positive stimulation later on. In
cases in which seizures were elicited during cortex stimu­
lation, a 4°C saline solution (0.9%) was used to irrigate the
cortex. After completion of the mapping procedure, the
paper numbers were removed and the patient was sedated
for the microsurgical resection procedure. When resec­
tion was directed near language-relevant areas, sedation
was reduced, and the mapping procedure was repeated.28,29
Stimulation of subcortical pathways was performed in se­
lected patients, but results of this stimulation technique are
not described in the current report.
Intraoperative Language Testing
Stimulus presentations were performed using a com­
puter system via a liquid crystal display screen. After
placement of the stimulation forceps at a cortical site, the
stimulus presentation was begun, which triggered elec­
trical stimulation. During each stimulation (4 seconds in
duration), a line drawing of an object together with the
carrier phrase “Dies ist …” (“This is …”) was presented
to the patient. The task was to produce the carrier phrase
in addition to the name of the object. The use of a carrier
phrase allows discriminating language and speech errors.
If the carrier phrase was produced correctly, 3 types of
errors were possible: 1) an aphasic arrest (the object name
was not produced at all); 2) an aphasic disturbance (seman­
tic or phonematic paraphasia instead of the object name);
or 3) a naming delay (the object name was produced only
after a marked delay). Disorders of speech also concern
the production of the carrier phrase. If no utterance was
produced at all, it was classified as a speech arrest. If the
utterance was dysarthric or marked by a strain to speak, it
was classified as a speech disturbance. After occurrence
of an error, at least 1 stimulus presentation was performed
without stimulation to ensure that the patient was able to
perform the task correctly again.
Documentation and Analysis of Intraoperative Findings
Language and speech disturbances as well as motor
or sensory phenomena were documented by the neuropsy­
chological/neurophysiological team. Fifty intraoperative
mapping protocols randomly selected from the first half
(1991–1997, 25 protocols) and the second half (1998–2005,
25 protocols) of the study period were analyzed regarding
the number of stimulated sites, the number and intensity
of the applied stimulations, and the mode of the evoked
disturbances.
Surgical Strategy
The surgical approach to the lesion was determined
by T1- and T2-weighted MR imaging data, supported by
information from neuronavigation and/or intraoperative
ultrasonography. The site of the corticotomy procedure
was customized according to the results of the cortical
mapping procedure; the distance from the cortical inci­
sion to areas of positive motor or language response had
J. Neurosurg. / Volume 109 / October 2008
to be at least 1 cm. To ensure this distance, rectangular
cotton patties (10 × 5 mm) were placed with the short side
at the point of language response, thus indicating a line of
10 mm distant to it. Typically, resection of the tumor ad­
jacent to eloquent regions was terminated when a safety
distance of 5–10 mm was reached. The extent of tumor
resection (total vs subtotal) was determined by postopera­
tive imaging (CT scans and MR imaging).
Postoperative Adjuvant Therapy
Additional conventionally fractionated external beam
radiation (60 Gy tumor dose in 2 Gy fractions) was ad­
ministered to all patients with WHO Grade III gliomas or
glioblastomas. In patients with WHO Grade II gliomas,
radiation therapy was only initiated at the time of tumor
progression when reoperation resulted in incomplete tu­
mor resection or when the recurrent tumor was considered
ineligible for further surgical treatment. Chemotherapy
was only used as a second- or third-line treatment option
in this series.
Clinical Evaluation
Follow-up evaluation was performed within 21 days
after surgery and then 3, 6, and 12 months later using the
KPS and mRS.4,37,46 The mRS uses the following scoring
system: 0 = no symptoms at all; 1 = no significant disability
despite symptoms, able to carry out all usual duties and
activities; 2 = slight disability, unable to carry out all pre­
vious activities, but able to look after own affairs without
assistance; 3 = moderate disability, requiring some help,
but able to walk without assistance; 4 = moderately severe
disability, unable to walk without assistance and unable to
attend to own bodily needs without assistance; 5 = severe
disability, bedridden, incontinent and requiring constant
nursing care and attention; and 6 = dead. Also, speech or
language deficits that were apparent during the clinical
evaluation of the patients were documented. Severe deficits
that required logopedic treatment were distinguished from
mild disturbances that were mainly reported by the patient
(such as occasional word finding difficulties). Impairment
of speech (dysarthria) was documented separately.
Statistical Analysis for the Evaluation of Prognostic
Factors
Risk factors for postoperative aphasia were obtained
from univariate and multivariate logistic regression mod­
els. The following variables (either continuously scaled or
dichotomized) were used for prognostic evaluation: age (>
40 years or ≤ 40 years); sex; preoperative aphasic distur­
bances (yes or no); preoperative KPS score (continuously
scaled); histological results (low-grade vs high-grade tu­
mors); prior treatment; duration of initial symptoms (> 6
months or ≤ 6 months); localization of the tumor (frontal
vs other); extent of resection (complete or incomplete);
localization of language-relevant cortical areas revealed
by direct cortical stimulation (intratumoral localization
vs other); language relevant sites remote from the tumor
(> 10 mm from the tumor) versus nearby (< 10 mm); in­
traoperative complications (yes or no); and postoperative
neurological deficits (yes or no). The length of survival
585
 J. Ilmberger et al.

and time to progression for different histological entities TABLE 2

were estimated using the Kaplan–Meier method. Histopathological findings

Results
Histopathological Findings No. of Cases

neoplastic lesions (WHO grade)

Patient Characteristics pilocytic astrocytoma (I) 1
gangliglioma (I) 1
From 1991 to 2005, 153 awake craniotomies were astrocytoma (II) 63
performed in 149 patients. Four patients underwent 2 sur­ oligodendroglioma (II) 10
gical procedures because of tumor recurrence. Clinical, oligoastrocytoma (II) 2
epidemiological, and histopathological data are summa­ astrocytoma (III) 44
rized in Tables 1 and 2. Patients with a nonfrontal tu­ oligodendroglioma (III)
oligoastrocytoma (III)
2
1
mor location had significantly larger tumor volumes (p gangliglioma (III) 2
< 0.05). Low- (WHO Grade II) and high-grade (WHO glioblastoma (IV) 14
Grade III or IV) gliomas were diagnosed in 74 (49.7%) gliosarcoma (IV) 2
and 62 (41.6%) patients, respectively, and the median du­ metastasis 2
ration of symptoms prior to surgery was 7 months. The other lesions
median KPS score prior to treatment was 90, and in 125
tumorlike lesion 1
cavernoma 4
cases (81.7%) patients were able to perform all usual ac­ cortical dysplasia 3
tivities at the time of the diagnosis (mRS score = 0 or 1). epidermoid 1
Focal seizures associated with language or speech distur­
bances were observed in 69 cases (45.1%). Preoperative
neurolinguistic testing identified 25 patients with mostly
mild aphasic disturbances (Fig. 1), whereas a language
and/or speech impairment was assumed clinically in 29 in another 8 cases (5.2%). The focal seizures were short-
cases (18.9%). No patient showed severe language impair­ lasting and could be controlled by irrigating the cortex
ment. with cold saline solution; the transient brain swelling was
caused by CO2 retention and successfully treated by the
Intraoperative Stimulation anesthesiologist.
Language relevant areas were found near the tumor
Intraoperative stimulation mapping was feasible in
margin (< 1 cm) in 114 cases, within the tumor in 17 cases,
all patients and generally well tolerated. Focal seizures
and remotely from the tumor in another 12 cases (a total
during the intraoperative stimulation mapping procedure
of 143 cases with language relevant areas). In 2 patients
occurred in 7 cases (4.6%) and a transient brain swelling
with previous open resections, stimulation was restricted
to the cortex above the lesion due to massive dural/pial ad­
hesions. Inadequate size of the craniotomy was probably
TABLE 1 the cause for negative intraoperative stimulation mapping
Clinical and epidemiological data in patients results in the remaining 8 patients.
undergoing awake craniotomies and cortical mapping The detailed analysis of 50 randomly selected map­
Variable Total ping protocols showed that between 10 and 48 sites (mean
21.3 sites) had been stimulated with stimulation intensities
no. of patients 149 ranging from 6 to 20 mA (mean 13.1 mA). A total of 1967
no. of awake craniotomies 153 stimulations at 1087 sites resulted in 536 disturbances,
which were classified as aphasic arrest (184 events), aphasic
sex (no. of pts)
male 70
female 79 disturbance (126 events), pronounced delay in naming (32
patient age (yrs) events), speech arrest (159 events), and speech disturbance
mean 39 (35 events). At 311 sites, subsets of multiple stimulations
median 38 were performed with constant stimulation intensities. Re­
sults of stimulation (no disturbance/disturbance) were con­
range 15–67
tumor laterality (no. of cases)
lt hemisphere 143 sistent in 81% of the cases (number of disturbances/number
rt hemisphere 10 of stimulations × 100). Stimulations at higher intensities (>
tumor localization (no. of cases) 12 mA) resulted in only slightly more disturbances (29%)
frontal 55 than stimulations at lower intensities (< 13 mA; 25%).
frontotemporal & insular 3
frontotemporal & parietal
parietal
4
13
Extent of Resection
temporal 55 Postoperative MR imaging revealed gross-total resec­
temporomesial
temporal & insular
8
3
tion in 74 (48.4%) of 153 cases. Gross-total resection was
temporoparietal 11 achieved in 36 (48.6%) of 74 cases involving WHO Grade
insular 1 II gliomas. The permanent morbidity rate was 3.9%. There
handedness (no. of cases) were no deaths. Gross-total resection was achieved signifi­
rt 135 cantly less often in patients with language-relevant areas
lt 18
within the tumor (p < 0.05).

586 J. Neurosurg. / Volume 109 / October 2008

Neurolinguistic analysis of intraoperative language mapping

Fig. 1.  Chart showing the results of pre- and postoperative neurolinguistic testing in cases with and without apha­
sic disturbances. The classifications of mild, moderate, and severe refer to severity of aphasic disturbances in neuro­
linguistic testing. C = clinical evaluation only; T = formal neurolinguistic testing.

Neurolinguistic Outcome forming a profile for the varying language abilities as mea­
Results of postoperative neurolinguistic testing are sured by the Token Test and the naming, written language,
summarized in Fig. 1. If patients were dismissed or trans­ repetition, and comprehension tests (Fig. 2). Def­icits were
ferred very quickly or language impairments were too se­ most pronounced in the Token Test and the naming subtest
(70–80% subnormal scores), whereas performance in the
vere for formal testing, only clinical testing was performed.
subtests of written language, repetition, or comprehension
A quick transfer sometimes was performed when a bed be­
was impaired in only ~ 50% of these patients.
came unexpectedly rapidly available at the initial referring Tumor recurrence did not affect the resolution profile
hospital; if patients did not undergo a second test, it was of the deficit; that is, none of those patients with a persis­
mainly due to organizational difficulties in patients with tent new deficit developed a tumor recurrence within the
very remote places of residence. time frame of neurolinguistic longitudinal analysis.
The first testing took place after a median interval
of 7.0 days after surgery, whereas the second testing was
performed after a median interval of 29.5 weeks. At the
first testing, we found new language impairment in 41
cases (34 by formal testing, 7 by clinical report). Twenty-
five of these patients improved and had no disturbance
at the second postoperative testing (20 with formal test­
ing, 5 by clinical report); 14 patients continued to show
a language impairment (11 by formal testing, 3 by clini­
cal report), and 2 patients had missing data. Language
disturbances were mostly mild in this group of patients.
In those patients with preoperative deficits, 6 showed no
disturbance postoperatively whereas 19 continued to have
language deficits that were somewhat more pronounced
than those observed preoperatively (16 by formal testing,
3 by clinical report). Of these 19 cases, 13 continued to
show aphasic disturbances at the second postoperative
testing (8 with formal testing, 5 by clinical report), and
only 3 pa­tients were free of any disturbance at that time.
Fig. 2.  Bar graph showing the percentage of cases with mild,
Three pa­tients had missing data. modest, severe, or no aphasic disturbances on the subtests of the
In 41 patients with no preoperative deficits but with aphasia battery at the first postoperative test. COM = comprehen­
new deficits at the first postoperative testing, we analyzed sion; NAM = naming; REP = repetition; TT = Token Test; WRIT
these deficits with regard to the subtests of the AAT, thus = written language.

J. Neurosurg. / Volume 109 / October 2008 587

 J. Ilmberger et al.

Risk Factors for a Postoperative Language Disturbance

Univariate and multivariate logistic regression anal­
ysis showed that preoperative aphasia (p < 0.0002), an
intraoperative complication (p < 0.02), language-positive
stimulation sites within the lesion (p < 0.001), and non­
frontal lesion locations (p < 0.001) were risk factors for a
language disturbance in the early postoperative interval.
Fifty-three percent of the 17 patients with positive stimu­
lation sites within the tumor developed an early new post­
operative language-related deficit. Age, KPS score, sex,
duration of symptoms, symptoms at presentation, histo­
logical results of the lesion, surgery for a recurrent tu­mor,
and extent of resection were nonsignificant covariates.
Neither the number of stimulation sites outside the le­
sion nor their relationship to the lesion (remote or nearby)
gained significant influence in the prognostic models.
At the time of late neurolinguistic evaluation (7 months
after surgery) only preoperative aphasia (p < 0.001) or alter­
natively early postoperative aphasia (which was pos­itively
intercorrelated with preoperative aphasia) and in­creased age
(> 40 years of age, p < 0.02) were associated with persistent Fig. 3.  Graph showing the impact of preoperative naming ca­
language impairment. The prognostic impact of tumor lo­ pabilities on early postoperative neurolinguistic outcome. Nor­mal
cation, intraoperative complications, and positive stimula­ (but submaximal) naming performance on the naming subtest,
tion sites within the tumor was no longer significant. al­though in the normal range (111–120), is negatively correlated
A separate multivariate analysis of those patients with the percentage of patients suffering from early postoperative
harboring no language disturbance before treatment con­ aphasic disturbances.
firmed the unfavorable influence of the following inde­
pendent variables: 1) language-positive stimulation site or Time to Progression
sites within the lesion (p < 0.002); 2) intraoperative com­ The 5-year progression-free survival rate for patients
plications (p < 0.05); and 3) nonfrontal lesion locations (p with WHO Grade II and III gliomas was 63 and 57%,
= 0.003). Moreover, a submaximal performance on both respectively (Fig. 4). This difference was statistically not
the Token Test and the naming subtest (but still within the significant. Patients with glioblastomas did significantly
normal range) was significantly associated with the oc­ worse than patients with other types of tumors (median
currence of an early postoperative aphasic disturbance af­ progression free survival = 16 months, p < 0.001).
ter univariate analysis (p < 0.003 and p < 0.0007, respec­
tively); that is, the lower the performance score on either
of these subtests, the greater the risk of an early postop­ Discussion
erative language disturbance in this patient series. This Language-related brain regions have been shown to
correlation is depicted for the naming subtest in Fig. 3. be embedded in complex and highly interconnected net­
After adjustment for the effects of the other 3 covariates works,31,35 which do not permit one to establish unequivo­
noted above, the influence of the Token Test was no lon­ cal associations between a specific lesion site and aphasic
ger significant, whereas the influence of the naming sub­ symptoms.50 Accordingly, uncertainties continue to exist
test scores was maintained (p < 0.0003). After univariate as to what extent the impairment of a single intraoper-
analysis, a submaximal preoperative performance in the atively detected language-relevant area and its cortical
naming subtest also remained a statistically significant
(p < 0.02) prognostic risk factor for the late neurolinguis­
tic evaluation. TABLE 3
Clinical status evaluation as estimated using the mRS*
Clinical Evaluation
mRS Testing Point
Evaluation of the patients’ clinical status before and
after treatment, as estimated using the mRS, is summa­ Category Preop Postop 3 mos 6 mos 12 mos
rized in Table 3. Eighty-two percent of the cases presented mRS Score 1 125 (81.7) 90 (58.8) 94 (72.9) 88 (75.2) 80 (76.2)
with an mRS score of 1 before treatment. Immediately mRS Score 2 13 (8.5) 23 (15.0) 22 (17.1) 19 (16.2) 16 (15.2)
postoperatively there was a significant clinical deteriora­ mRS Score 3 15 (9.8) 28 (18.3) 10 (7.8) 5 (4.3) 7 (6.7)
tion (p < 0.0001), which continuously improved at 3, 6, and mRS Score 4
total
0
153
12 (7.8)
153
3 (2.3)
129
5 (4.3)
117
2 (1.9)
105
12 months postoperatively to near initial values (p = 0.12). lost to FU 18 8 8
One year after treatment, however, 1.9% of the cases pre­ excluded due 6 10 14
sented with an mRS score of 4. The median KPS score was to POD
90 prior to surgery as well as at 3, 6, and 12 months after * Values given are number of cases (%). Abbreviations: FU = follow-up;
treatment. POD = progression of disease.

588 J. Neurosurg. / Volume 109 / October 2008

Neurolinguistic analysis of intraoperative language mapping
Fig. 4.  Line graph showing the progression-free survival rate of patients after open tumor resection according to
WHO tumor grade.
and/or subcortical representation during tumor resection
contribute to postoperative language-related outcome
scores. Taking into account these uncertainties, every at­
tempt was undertaken to preserve those language-relevant
areas that were detected in the current patient series, and
resection of a tumor adjacent to language-relevant areas
was usually discontinued when a safety margin of 5–10
mm was reached. Using this surgical strategy, gross-total
resection was achieved in 48% of all treated patients with
tumors (and in 48% of the subpopulation of patients with
low-grade tumors as well). This approach, however, by no
means indicates that the extent of resection was optimally
adjusted in every single patient of this study. For a more
refined elucidation of structure–function relationships,
however, pre- and postsurgical functional neuroimag­
ing results and their correlation with both intraoperative
stimulation mapping and language functioning of the pa­
tient over time appears to be essential.
Functional MR imaging analysis, however, which
traditionally exploits the blood oxygen level–dependent
effect for delineation of activated cortical regions during
execution of specific tasks, has been shown to be strongly
influenced by subject-specific parameters, task generic
parameters, cortical regions analyzed, and the statisti-
cal analysis of the data (such as the choice of thresh­
old); accordingly only a moderate test–retest reliability
of the fMR imaging mapping results has been observed
so far.10,25,38,41 Moreover, the congruence of fMR imag­
ing and intraoperative stimulation mapping data remains
unsatisfying. Although very optimistic reports with sensi­
tivity values as high as 100% have been reported for fMR
imaging,32 the occurrence of missed sites (cortical sites
found to be essential by intraoperative stimulation map­
ping but not identified by preoperative imaging as report-
ed by Roux et al.40) has precluded the use of imag­ing
data for guiding resections until now. Validation of fMR
J. Neurosurg. / Volume 109 / October 2008
imaging (as compared with intraoperative stimulation
map­ping) still remains difficult. The transfer of preop­
eratively generated fMR imaging data into a neuronav-
igation system easily resolves mathematically induced
in­consistencies of the spatial resolution, which might be
fur­ther aggravated by brain-shift effects after opening of
the dura.19,36
New fMR imaging approaches with the aid of wa­
ter diffusion tensor imaging permit more direct explora­
tion of tissue metabolism (such as the study conducted by
Mori and Zhang18) than traditional blood oxygen level–
dependent monitoring and deserve further evaluation.
Ad­ditionally, white matter tracts can be visualized by
ra­diological tractography obtained during diffusion ten­
sor MR imaging. However, these data have not yet been
functionally validated and cannot be reliably integrated
into neuronavigational systems.14,16 Thus, given these and
other limitations in the fascinating field of functional im­
aging, open tumor resection in the vicinity of language-
relevant areas should still rely initially on the results of
intraoperative stimulation mapping.
In the current patient series, a highly selected patient
population was addressed. Patients were young, exhib­
ited excellent performance scores before treatment, and
the vast majority (82%) were able to perform all usual
activities at the time of the diagnosis. Open tumor resec­
tion alone (applied to all patients with WHO Grade II
gliomas) or in combination with external beam radiation
therapy (applied to all patients with high-grade tumors)
resulted in extremely favorable outcome data, which
were in the upper range of those reported in the litera­
ture (a 5-year progression-free survival rate for patients
with WHO Grade II and III gliomas of 63 and 57%, re­
spectively). Apparently, eligibility for complex surgical
procedures such as intraoperative stimulation mapping
of language function appears to be intrinsically associ­
589

ated with a favorable pattern of prognostic factors and
the obtained results should not be extrapolated to all pa­
tients with low- and high-grade gliomas. The aim of the
current prospective longitudinal study was, for the first
time, to evaluate language function before and after in­
traoperative stimulation mapping–guided surgery using a
standardized neurolinguistic test battery. The study was
designed to identify prognostic factors and patients at
risk for surgery-related language impairment. Given the
methodological limitations of intraoperative stimulation
mapping techniques of language functioning, no attempt
was made to focus on structure–function relationships in
this patient series.
Risk Factors for Early and Late Postoperative Language
Impairment
Early follow-up evaluation (within 21 days) revealed
a language-related deterioration rate of 32%. Risk factors
associated with early language impairment were preop­
erative aphasia, language positive stimulation sites within
the tumor, a nonfrontal tumor location, and intraoperative
complications. The unfavorable impact of a nonfrontal
tu­mor location—which has not yet been reported in the
literature—might be explained by its significant associa­
tion with larger tumor volumes. More data are necessary
to elucidate the impact of tumor size. Patients who pre­
sented with a suboptimal but still normal performance in
the naming test preoperatively experienced early postop­
erative disturbances significantly more often even after
adjustment for the effects of the other 4 risk factors.
Removal of positive stimulation sites within the tu­
mor, as was done were during the initial phase of this
series, was explained by the assumption of the treating
surgeons that intratumoral language positive sites might
no longer or might not relevantly be involved in the lan­
guage network. This assumption was found to be incor­
rect in the early phase of this patient series when this
was not yet common knowledge. The deficits following
resection or cortical undermining of 1 or several sites in
4 patients, even when further language-positive stimu­
lation sites were found outside the tumor, demonstrated
that such sites still bear important language functions and
are essential for favorable early neurolinguistic outcome
scores. This finding has been confirmed by other inves­
tigative groups.1,23,42,44 However, in 2 of these 4 patients
the initial new postoperative deficits resolved completely
within 12 months, and the remaining 2 patients regained
their preoperative status that had involved slight language
impairments. Thus, the resolution profile seemingly did
not differ from all other patients.
At the time of late neurolinguistic evaluation (7
months after surgery) preoperative aphasia or alternatively
early postoperative aphasia (which was positively intercor­
related with preoperative aphasia) and increased age (> 40
years) were associated with persistent language impair­
ment. Also, after univariate analysis, performance on the
naming subtest remained a prognostic factor. As naming
capabilities might be closely associated with verbal intel­
ligence, it could be assumed that patients with a high verbal
intelligence level are less prone to neurolinguistic deterio­
ration after microsurgical treatment. This hypothesis de­
590
J. Ilmberger et al.
serves further prospective evaluation and might become a
major factor for individual patient counseling and risk esti­
mation. The unfavorable prognostic impact of preoperative
aphasia has already been reported by other groups (such as
Bello et al.2) and has now been incorporated more clearly
into the perspective of an unfavorable long-term prognosis
for a considerable number of patients. The somewhat di­
vergent prognostic findings of early and late neurolinguis­
tic examinations might indicate the effects of functional
reorganization during the follow-up period, which could
involve perilesional areas and/or distant ipsilateral struc­
tures, and/or the activation of homologous contralateral
regions. One might speculate whether potentially favorable
effects of functional reorganization are less pronounced in
the older subpopulation (> 40 years old) and whether those
effects had been already exhausted in those patients pre­
senting with phasic disturbances before treatment. More
longitudinally obtained functional image-based data and
their correlation with pre- and postsurgical neurolinguistic
status are necessary to more accurately elucidate the re­
covery mechanisms of aphasia after open tumor resection
and to adjust findings of intraoperative stimulation map­
ping more precisely to long-term neurolinguistic outcome,
particularly in low-grade glioma surgery.
Clinical follow-up evaluation results in this study (as
measured by the mRS) paralleled neurolinguistic out­
come measurements: early evaluation indicated a signifi­
cant clinical deterioration, which continuously improved
at 3, 6, and 12 months postoperatively. Significant differ­
ences, as compared with the preoperative findings, were
no longer significant 3 months after surgery.
Neurolinguistic Considerations
Confrontation naming is the most widely intraopera­
tively used task and has proven to be a robust method for
identifying essential language sites also outside the clas­
sic language zones. Additional tasks such as reading or
calculation may be added in individual cases, for example
when parietal areas are to be mapped. To achieve high
reliability in mapping, we believe it is absolutely neces­
sary to use custom-tailored stimulus material that has
been selected preoperatively in a training session with the
patient. No stimulus material that the patient is unfamil­
iar with or tasks the patient has not shown to be able to
master should be used intraoperatively to avoid confusion
and delays during the mapping procedure. In our experi­
ence, a set of at least 20 items that the patient is able to
name consistently in preoperative training is necessary
to avoid too many repetitions and thus automatization of
responses during stimulation mapping.
Despite those precautions, the detailed analysis of
mapping protocols in a random sample of 50 patients
showed that results of intraoperative mapping are not al­
ways consistent at specific sites, that is, errors are not al­
ways or never evoked at a specific site during repeated
testing. Whitaker and Ojemann47 described this phenom­
enon as graded localization; Ojemann21 believes it is an
indication of “graded transition from essential to non-es­
sential cortex.” Within the connection of a network, some
nodes would have weaker connections, and stimulations
of such nodes would produce disturbances on a stochastic
J. Neurosurg. / Volume 109 / October 2008
Neurolinguistic analysis of intraoperative language mapping
basis only. An alternative explanation of these inconsis­
tencies would be that specific sites show category speci­
ficity, that is, that specific sites are differentially involved
in the processing of items from different semantic catego­
ries (such as living vs nonliving things). Category-specific
effects are well documented in functional imaging and in
lesion studies, and Ilmberger and associates13 were able to
demonstrate the influence of using items from different
semantic categories on errors evoked during stimulation
by subdural electrodes in epilepsy patients. Also, Pou­
ratian et al.33 identified a category-specific naming area
us­ing intraoperative stimulation monitoring in a single
patient. To our knowledge, however, no other studies have
focused on semantic criteria for the selection of items in
stimulation mapping, and this would be a promising vari­
able to be included in future work.
Patient Tolerance and Commitment of the Neurosurgical
Team
Earlier reports have shown that intraoperative stimu­
lation mapping is tolerated well by the vast majority of
patients and that the experience of this procedure does
not lead to psychological trauma.6,49 In our experience, it
is extremely important that the patient and the neurosur­
gical team develop some form of mutual trust in working
together in the operating room. Penfield and Boldrey26
commented on this issue in a beautiful way: “It is essen­
tial that the patient should be in sympathy with operator
and anaesthetist, and it is an interesting comment on the
bravery and fortitude of mankind that almost without ex­
ception these subjects have gone through the ordeal of op­
eration patiently and intelligently…” The neuropsycholo­
gist or neurolinguist plays a key role in the entire process,
because during the few minutes of mapping it is his or her
job to guide the patient through the procedure. Enough
time should be given to the preoperative dialogue to make
sure that all the patients’ questions are answered and that
the patient and neuropsychologist feel confident that the
situation will be mastered. A very important piece of in­
formation for the patient to know is the types of phenom­
ena elicited by the stimulations (movements, sensations,
speech, and language disturbances).
Conclusions
Awake craniotomy with intraoperative stimulation
mapping of language function for microsurgical treat­
ment of complex tumors located in close proximity to
or within speech areas is a safe and economic procedure
that is well tolerated by the patient. Given the uncertain­
ties as to what extent a single intraoperatively detected
language-relevant area contributes to language-related
outcome scores, every attempt should be undertaken to
preserve language-relevant areas intraoperatively. The
occurrence of early clinical and neurolinguistic deterio­
ration after surgery in 32% of the patients in this study
did not predict long-term outcome scores; the majority of
these patients either improved to the preoperative status
(61%) or showed only mild impairment (22%) within 3
months after treatment. The naming subtest and the Token
Test of the AAT neurolinguistic test battery were found to
J. Neurosurg. / Volume 109 / October 2008
be most sensitive for the detection of tumor-related apha­
sic disturbances and were able to identify patients at risk
before treatment; normal (but suboptimal) preoperative
naming capacities are an indication of a higher risk of
early postoperative language impairment. Further devel­
opments should include even more extended pre- and in­
traoperative testing to further elucidate the nature of risk
factors related to verbal intelligence and to strengthen the
consistency of intraoperative mapping results, such as by
systematic testing of different semantic categories.
Acknowledgments
We gratefully acknowledge the editorial and secretarial sup­
port of Ms. I. Anders during preparation of this manuscript. The
co­operative effort of Dr. W. Eisner (Innsbruck, formerly of Ludwig
Maximilians University, Munich) and Ms. E. Schröck during the ini­
tial phase of the language-mapping project is appreciated.
Hans-Juergen Reulen, M.D., is Professor Emeritus in the De­­
part­ment of Neurosurgery, Ludwig Maximilians University, Mu­nich,
Ger­many.
Disclaimer
The authors report no conflict of interest concerning the mate­
rials or methods used in this study or the findings specified in this
paper.
References
  1.  Atlas SW, Howard RS, Maldjian J, Alsop D, Detre JA, Listerud
J, et al: Functional magnetic resonance imaging of regional
brain activity in patients with intracerebral gliomas: findings
and implications for clinical management. Neurosurgery 38:
329–338, 1996
  2.  Bello L, Gallucci M, Fava M, Carrabba G, Giussani C, Acerbi F,
et al: Intraoperative subcortical language tract mapping guides
removal of gliomas involving speech areas. Neurosur­gery 60:
67–82, 2007
  3.  Berger MS, Ojemann GA: Intraoperative brain mapping tech­
niques in neuro-oncology. Stereotact Funct Neurosurg 58:
153–161, 1992
  4.  Bonita R, Beaglehole R: Modification of Rankin scale: recovery
of motor function after stroke. Stroke 19:1497–1500, 1988
  5.  Danks RA, Aglio LS, Gugino LD, Black PM: Craniotomy under
local anesthesia and monitored conscious sedation for the re­
section of tumors involving eloquent cortex. J Neurooncol 49:
131–139, 2000
  6.  Danks RA, Rogers M, Aglio LS, Gugino LD, Black PM: Pa­
tient tolerance of craniotomy performed with the patient un­
der local anesthesia and monitored conscious sedation. Neu-
rosurgery 42:28–36, 1998
  7.  Duffau H, Capelle L, Denvil D, Sichez N, Gatignol P, Tailland­
ier L, et al: Usefulness of intraoperative electrical subcortical
mapping during surgery for low-grade gliomas located within
eloquent brain regions: functional results in a consecutive se­
ries of 103 patients. J Neurosurg 98:764–778, 2003
  8.  Duffau H, Lopes M, Arthuis F, Bitar A, Sichez JP, Van Effen­
terre R, et al: Contribution of intraoperative electrical stimu­
lations in surgery of low grade gliomas: a comparative study
between two series without (1985-96) and with (1996-2003)
functional mapping in the same institution. J Neurol Neuro-
surg Psychiatry 76:845–851, 2005
  9.  Eisner W, Reulen HJ, Ilmberger J, Swozil U, Bise K: Intraop­
erative mapping of eloquent brain areas. Front Radiat Ther
Oncol 33:28–36, 1999
10.  Havel P, Braun B, Rau S, Tonn JC, Fesl G, Brückmann H, et
al: Reproducibility of activation in four motor paradigms: an
fMRI study. J Neurol 253:471–476, 2006
591

11.  Huber W, Poeck K, Weniger D, Willmes K: Aachener Aphasie
Test. Göttingen: Hogrefe, 1983
12.  Ilmberger J, Eisner W, Schmid U, Reulen HJ: Performance in
picture naming and word comprehension: evidence for com­
mon neuronal substrates from intraoperative language map­
ping. Brain Lang 76:111–118, 2001
13.  Ilmberger J, Rau S, Noachtar S, Arnold S, Winkler P: Naming
tools and animals: asymmetries observed during direct electri­
cal cortical stimulation. Neuropsychologia 40:695–700, 2002
14.  Kamada K, Todo T, Masutani Y, Aoki S, Ino K, Takano T, et
al: Combined use of tractography-integrated functional neu­
ronavigation and direct fiber stimulation. J Neurosurg 102:
664–672, 2005
15.  Keles GE, Berger MS: Advances in neurosurgical technique
in the current management of brain tumors. Semin Oncol 31:
659–665, 2004
16.  Kinoshita M, Yamada K, Hashimoto N, Kato A, Izumoto S,
Baba T, et al: Fiber-tracking does not accurately estimate size
of fiber bundle in pathological condition: initial neurosurgical
experience using neuronavigation and subcortical white mat­
ter stimulation. Neuroimage 25:424–428, 2005
17.  Meyer FB, Bates LM, Goerss SJ, Friedman JA, Windschitl
WL, Duffy JR, et al: Awake craniotomy for aggressive resec­
tion of primary gliomas located in eloquent brain. Mayo Clin
Proc 76:677–687, 2001
18.  Mori S, Zhang J: Principles of diffusion tensor imaging and its
applications to basic neuroscience research. Neuron 51:527–
539, 2006
19.  Nimsky C, Ganslandt O, Cerny S, Hastreiter P, Greiner G, Fahl­
busch R: Quantification of, visualization of, and compensation
for brain shift using intraoperative magnetic resonance imag­
ing. Neurosurgery 47:1070–1080, 2000
20.  Ojemann G, Ojemann J, Lettich E, Berger M: Cortical language
localization in left, dominant hemisphere. J Neurosurg 71:316–
326, 1989
21.  Ojemann GA: Organization of language cortex derived from in­
vestigations during neurosurgery. Semin Neurosci 2:297–305,
1990
22.  Ojemann GA, Whitaker HA: Language localization and vari­
ability. Brain Lang 6:239–260, 1978
23.  Ojemann JG, Miller JW, Silbergeld DL: Preserved function in
brain invaded by tumor. Neurosurgery 39:253–259, 1996
24.  Ojemann JG, Ojemann GA, Lettich E: Cortical stimulation
mapping of language cortex by using a verb generation task:
effects of learning and comparison to mapping based on ob­
ject naming. J Neurosurg 97:33–38, 2002
25.  Otzenberger H, Gounot D, Marrer C, Namer IJ, Metz-Lutz
MN: Reliability of individual functional MRI brain mapping
of language. Neuropsychology 19:484–493, 2005
26.  Penfield W, Boldrey E: Somatic motor and sensory represen­
tation in the cerebral cortex of man as studied by electrical
stimulation. Brain 60:389–443, 1937
27.  Penfield W, Rasmussen T: The Cerebral Cortex of Man.
New York: Macmillan Co., 1950
28.  Peraud A, Ilmberger J, Reulen HJ: Surgical resection of gli­
o­mas WHO grade II and III located in the opercular region.
Acta Neurochir (Wien) 146:9–18, 2004
29.  Peraud A, Meschede M, Eisner W, Ilmberger J, Reulen HJ:
Surgical resection of grade II astrocytomas in the superior
frontal gyrus. Neurosurgery 50:966–977, 2002
30.  Picht T, Kombos T, Gramm HJ, Brock M, Suess O: Multimod­
al protocol for awake craniotomy in language cortex tumour
surgery. Acta Neurochir (Wien) 148:127–138, 2006
31.  Poeppel D, Hickok G: Towards a new functional anatomy of
language. Cognition 92:1–12, 2004
32.  Pouratian N, Bookheimer SY, Rex DE, Martin NA, Toga AW:
Utility of preoperative functional magnetic resonance imag­
ing for identifying language cortices in patients with vascular
malformations. J Neurosurg 97:21–32, 2002
592
J. Ilmberger et al.
33.  Pouratian N, Bookheimer SY, Rubino G, Martin NA, Toga
AW: Category-specific naming deficit identified by intraop­
erative stimulation mapping and postoperative neuropsycho­
logical testing. J Neurosurg 99:170–176, 2003
34.  Pouratian N, Cannestra AF, Bookheimer SY, Martin NA,
Toga AW: Variability of intraoperative electrocortical stimu­
lation mapping parameters across and within individuals. J
Neurosurg 101:458–466, 2004
35.  Pulvermüller F: Words in the brain’s language. Behav Brain
Sci 22:253–336, 1999
36.  Quinones-Hinojosa A, Ojemann SG, Sanai N, Dillon WP,
Berger MS: Preoperative correlation of intraoperative cortical
mapping with magnetic resonance imaging landmarks to pre­
dict localization of the Broca area. J Neurosurg 99:311–318,
2003
37.  Rankin J: Cerebral vascular accidents in patients over the age
of 60. II. Prognosis. Scott Med J 2:200–215, 1957
38.  Rau S, Fesl G, Bruhns P, Havel P, Braun B, Tonn JC, et al: Re­
producibility of activations in Broca area with two language
tasks: a functional MR imaging study. AJNR Am J Neuro-
radiol 28:1346–1353, 2007
39.  Reulen HJ, Schmid DU, Ilmberger J, Eisner W, Bise K: [Tumor
surgery of the speech cortex in local anesthesia. Neuropsy­
chological and electrophysiological monitoring during opera­
tions in the dominant hemisphere.] Nervenarzt 68:813–824,
1997 (Ger)
40.  Roux FE, Boulanouar K, Lotterie JA, Mejdoubie M, LeSage JP,
Berry I: Language functional magnetic resonance im­aging in
preoperative assessment of language areas: correlation with di­
rect cortical stimulation. Neurosurgery 52:1335–1347, 2003
41.  Rutten GJ, Ramsey NF, van Rijen PC, van Veelen CW: Re­
producibility of fMRI-determined language lateralization in
individual subjects. Brain Lang 80:421–437, 2002
42.  Schiffbauer H, Ferrari P, Rowley HA, Berger MS, Roberts TP:
Functional activity within brain tumors: a magnetic source
imaging study. Neurosurgery 49:1313–1321, 2001
43.  Signorelli F, Guyotat J, Isnard J, Schneider F, Mohammedi R,
Bret P: The value of cortical stimulation applied to the surgery
of malignant gliomas in language areas. Neurol Sci 22:3–
10, 2001
44.  Skirboll SS, Ojemann GA, Berger MS, Lettich E, Winn HR:
Functional cortex and subcortical white matter located within
gliomas. Neurosurgery 38:678–685, 1996
45.  Taylor MD, Bernstein M: Awake craniotomy with brain map­
ping as the routine surgical approach to treating patients with
supratentorial intraaxial tumors: a prospective trial of 200
cases. J Neurosurg 90:35–41, 1999
46.  Van Swieten JC, Koudstaal PJ, Visser MC, Schouten HJ, van
Gijn J: Interobserver agreement for the assessment of handi­
cap in stroke patients. Stroke 19:604–607, 1988
47.  Whitaker HA, Ojemann GA: Graded localisation of naming
from electrical stimulation mapping of left cerebral cortex.
Nature 270:50–51, 1977
48.  Whittle IR, Borthwick S, Haq N: Brain dysfunction following
‘awake’ craniotomy, brain mapping and resection of glioma.
Br J Neurosurg 17:130–137, 2003
49.  Whittle IR, Midgley S, Georges H, Pringle AM, Taylor R:
Patient perceptions of “awake” brain tumour surgery. Acta
Neurochir (Wien) 147:275–277, 2005
50.  Willmes K, Poeck K: To what extent can aphasic syndromes
be localized? Brain 116:1527–1540, 1993
Manuscript submitted October 4, 2007.
Accepted February 5, 2008.
Address reprint requests to: Joerg-Christian Tonn, M.D., Neu­ro­
surgical Department, Klinikum Grosshadern, Ludwig Maximilians
University, Marchioninistrasse 15, 81377 Munich, Germany. email:
Joerg.Christian.Tonn@med.uni-muenchen.de.
J. Neurosurg. / Volume 109 / October 2008