This article appeared in a journal published by Elsevier.

The attached
copy is furnished to the author for internal non-commercial research
and education use, including for instruction at the authors institution
and sharing with colleagues.
Other uses, including reproduction and distribution, or selling or
licensing copies, or posting to personal, institutional or third party
websites are prohibited.
In most cases authors are permitted to post their version of the
article (e.g. in Word or Tex form) to their personal website or
institutional repository. Authors requiring further information
regarding Elseviers archiving and manuscript policies are
encouraged to visit:
http://www.elsevier.com/copyright

Author's personal copy

Epilepsy & Behavior 15 (2009) 367371

Contents lists available at ScienceDirect

Epilepsy & Behavior

journal homepage: www.elsevier.com/locate/yebeh

Emotional content does not interfere with verbal memory in patients

with temporal lobe epilepsy
Notger G. Mller a,b,*, Bernd Wohlrath b, Ute A. Kopp c, Ulrike Lengler b
a

Clinic for Neurology, Otto von Guericke University, Magdeburg, Germany

Clinic for Neurology, Johann Wolfgang Goethe University, Frankfurt/M, Germany
c
Clinic for Neurology, CharitUniversity Medicine, Berlin, Germany
b

a r t i c l e

i n f o

Article history:
Received 21 April 2009
Revised 8 May 2009
Accepted 11 May 2009
Available online 13 June 2009
Keywords:
Temporal lobe epilepsy
Memory
Emotion

a b s t r a c t
In healthy humans, memory for words with emotional valence is better than memory for neutral words.
At the same time, the word preceding the emotional word in a word list learning task is remembered less
often than other neutral words. Both effects, enhanced memory for emotional words and retrograde
amnesia for preceding words, are dependent on intactness of the amygdala. In this study we asked
whether patients with temporal lobe epilepsy (TLE), a disease that often involves the amygdala as well,
show altered memory for emotional words and the words presented in close temporal proximity.
Whereas we found enhanced memory performance for emotional and decreased recognition performance for the preceding and successive word in our 19 control subjects, both effects were strongly
reduced in our 21 patients. No group differences occurred in memory for perceptually deviant words.
The lack of emotion effects on memory in the patients cannot simply be attributed to altered perception
of emotions as the patients rated the emotionality of the words no different than control subjects. Hence,
we conclude that patients with TLE have a specic decit in the emotion-driven encoding enhancement
mediated by the amygdalahippocampus loop.
2009 Elsevier Inc. All rights reserved.

1. Introduction
It is an everyday experience that we remember outstanding,
emotionally arousing events better than routine situations. In the
psychology laboratory, it was demonstrated as early as 1933 [1]
that perceptually deviant words (also called oddballs) are remembered better than those presented in the standard format (von
Restorff effect). Oddballs conveying a strong, usually aversive emotional content lead to even better recollection performance [2]. At
the same time they impair memory for words that have been presented just before the emotional oddball. This has been taken as
evidence that emotions interfere with the process of encoding
and/or consolidation of preceding events. It could be demonstrated
that these effects are indeed emotion specic and are not simply a
more pronounced von Restorff effect: When Strange and colleagues blocked the aymgdalas cholinergic output by administering the b-adrenergic antagonist propranolol to their subjects, the
emotion-related effects (enhancement for the emotional word,
anterograde amnesia for the preceding word) were abolished,
whereas the von Restorff effect for the perceptually deviant word

* Corresponding author. Address: Clinic for Neurology, Otto-von-Guericke-University, Leipziger Strasse 44, Magdeburg 39120, Germany. Fax: +49 (0)391
6715233.
E-mail address: notger.mueller@med.ovgu.de (N.G. Mller).
1525-5050/$ - see front matter 2009 Elsevier Inc. All rights reserved.
doi:10.1016/j.yebeh.2009.05.006

remained unaffected [3]. Hence, only the emotion effect on memory depends on integrity of the amygdalae, which are known to
play a key role in emotion processing. This assumption is further
supported by the nding that a patient with bilateral amygdala lesions behaved like pharmacologically treated subjects, that is, did
not show any emotion-related memory alteration [36]. The
amygdala is intensively connected with other structures in the
mesiotemporal lobe, including the hippocampus, which is critical
for encoding and recollection within episodic memory [7]. Regarding enhanced memory for emotional content, it has been speculated that the amygdala releases noradrenergic neurotransmitters
in response to emotional input which, via connections to the hippocampus, enhances encoding of emotional stimuli into episodic
memory by the hippocampus [3,810].
Tulving was the rst to show that emotional arousal also impairs memory for preceding events [11] and he speculated that this
was due to a premature termination of the encoding process
induced by the emotional event. Later it was speculated that this
effect is driven by the amygdalas release of neurotransmitter in response to emotional stimuli which may disrupt either the encoding
or consolidation of preceding items into memory [3].
Temporal lobe epilepsy (TLE) usually involves neural changes in
the hippocampus and surrounding areas, including to a variable
degree the amygdala, and is often accompanied by hippocampal
sclerosis [6,12,13]. Memory decits in these patients are a common

Author's personal copy

368

N.G. Mller et al. / Epilepsy & Behavior 15 (2009) 367371

nding, and usually verbal memory decits are linked to left-sided

pathology whereas visual memory decits are more common
when the right hemisphere is affected [14]. Moreover, recognition
of facial expressions can be impaired, pointing to a decit in the
judgment of emotional value [15].
It was our aim in the present study to shed further light on the
memory alterations in patients with TLE. More specically, we
asked whether the emotion-related enhancement and impairment
for word recollection are altered in a way comparable to those of
subjects with pharmacologically blocked or lesioned amygdalae.
If that were the case, it would indicate a clear dysfunction in the
patients amygdalahippocampus loop that normally enhances
memory for emotional words and, at the same time, interference
with the encoding of stimuli that are presented in close temporal
proximity.
2. Materials and methods
2.1. Subjects
We tested 31 patients with proven TLE recruited from the specialized outpatient clinic for epilepsy at the University Clinic for
Neurology in Frankfurt, Germany. The inclusion criterion was a
denite diagnosis of TLE by an expert who rated the patient based
on semiology and, where applicable, MRI and EEG ndings (hippocampal sclerosis and temporal interictal EEG focus, respectively) as
documented in the patients charts. Exclusion criteria were current
psychiatric symptoms (e.g., sadness, delusion) as assessed in the
clinical interview and documented brain pathology other than hippocampal sclerosis. From the initial cohort, 9 patients had to be excluded from the nal analysis because either a psychiatric
comorbidity was detected that was initially unknown to us
(n = 4) or performance in the memory task was near chance level
(hit rate for control words below 60%, n = 5), in which case it can
be assumed that the subject was guessing. Another patient aborted
testing after the rst list. The remaining 21 patients had the following characteristics: 13 females, 8 males, age range 2267 (mean
39) years, 15.5 years of education. For detailed data, see Table 1.
The patients performance was compared with that of an age-,
sex-, and education-matched control group that initially comprised
21 neurologically and psychiatrically healthy persons who were recruited from hospital staff or were acquaintances of the authors.

Table 1
Characteristics of patients.
Patient

Sex

Age

Years of
education

Hippocampal
sclerosis

Interictal
focus

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21

M
F
M
F
F
F
f
M
F
F
M
F
F
M
M
F
F
M
F
F
M

32
32
23
50
61
34
37
54
28
35
67
22
43
31
34
38
53
33
38
47
35

12
11.5
17
12.5
12
14
21
18.5
10
20
16.5
15
9
12
21
17.5
15
18
22
12
19.5

No
Yes
Yes
No
No
No
Yes
Yes
No
No
No
Yes
No
No
No
No
No
Yes
Yes
Yes
No

Bilat
Left
Right
Left
Left
Unknown
Right
Left
Unknown
Right
Right
Right
Unknown
Left
Left
Left
Left
Right
Left
Left
Left

Two had to be excluded from nal analysis because of performance

near chance level in the experiment. The remaining control group
comprised 19 subjects (11 females, 8 males), aged 2171 (mean
41) years with 14.2 years of education.
2.2. Selection of stimuli
To create word lists with emotional and neutral words we selected 294 nouns which were presented in alphabetical order to
20 healthy controls (12 females, age range 2076 years). Subjects
had to rate the emotional valence of each word on the scale 2,
1, 0, +1, +2, where 2 indicates a strong negative, +2 a strong positive, and 0 a neutral word.
We then calculated the mean rating for each word as well as the
absolute sum by ignoring the sign of the numbers. The latter procedure was aimed at preventing words considered by some subjects as very positive and by others as very negative from being
classied as neutral. On the basis of these measures we selected
a total of 240 words. Ten of them were considered emotionally
negative (range: 0.85 to 1.90), and 230 were considered neutral.
From these 240 words, we created 10 lists of 24 semantically related words to use in the experiment. Among the 24 words, 23
were neutral and 1 was emotional. This procedure was followed
so that emotional deviance and not semantic deviance, was driving
the effect of interest.
2.3. Experimental procedure
Our experimental procedure was designed similarly to those
used in earlier experiments to make the results comparable (see
Fig. 1). Of the 24 nouns in a list, 16 were presented during the
learning phase; the remaining 8 were new words used during
the following recognition phase. Each of the 10 experimental
blocks started with the words new list (in German); presentation
of the learning list commenced after the subject pressed the space
bar on the notebook keyboard. Words were presented for 1 s each
with an interstimulus interval of 2 s; they were white uppercase
letters presented on a black 15-in notebook screen running the
Presentation software package. For each learning list, 14 words
were of the same semantic category, were emotionally neutral,
and were presented in the same font (Times New Roman, font size
95). These words are referred to as standard words. An additional
word was rendered perceptually deviant (P-word) by presenting
it in one of the following fonts: Comic Sans MS, Alba, Kidprint, or
Modern. The P-word was emotionally neutral and of the same
semantic category and font size as the standard words. The other
oddball word was emotionally aversive in content (E-word) but
of the same semantic category and perceptually identical to the
standards. Two of the standard nouns were chosen as control
words (C-words). C-words were selected randomly, with the
exception that these words were never presented directly before
or after the E- and P-word. Other constraints regarding the sequence of words were that the E-word and P-word had to be separated by a minimum of two words and that the rst ve words of
each block were always standard words (to avoid primacy effects).
In addition to being instructed to encode the words into memory, the subjects were told to indicate by a mouse button press
whether the rst letter of each word had an enclosed space. This
procedure had been proven in a prior study to induce a strong
emotion effect on memory [3]. The button had to be pressed before
the next word was ashed.
After a learning list was presented, subjects performed a 30-s
distraction task during which they had to count backward in steps
of three. To ease understanding of the task, the rst three numbers,
for example, 90, 87, 84, were presented on the computer screen. In

Author's personal copy

N.G. Mller et al. / Epilepsy & Behavior 15 (2009) 367371

369

Fig. 1. Schematic experimental design. Note that no colors were used in the actual experiment.

the next block the starting number was one less, that is, 89 and so
forth, to prevent the numbers from being memorized.
After the distraction task, a list of 16 nouns was presented to
test recognition performance. Subjects had to indicate by a mouse
button press (counterbalanced across subjects) whether they believed that the presented word had been part of the preceding
learning list or not. They were told that reaction speed also mattered. Each word remained on the screen until button press or after
4 s had passed without a button press. Of the 16 nouns, 8 were old
(E, P, E 1, E + 1, P 1, P + 1, C1, C2). The eight new words were
those remaining from the initial 24-word list. Sequence of stimuli
was random, and they were all presented in the same standard font
(Times New Roman 95).
About a month after the patients had completed the experiment
they were sent a letter in which they were asked to rate the emotionality of the words in the same manner as described before for
the controls. This was done to ensure that the patients did not differ in rating of emotional valence per se.

the results of neuropsychological testing, patients performed

somewhat worse (65% correct answers vs 71% in the recognition
task) and more slowly (1.44 s vs 1.27 s); however, none of these
differences reached signicance in the respective t tests (P > 0.3)
with the exception of reaction times, which yielded a trend
(P = 0.071).
3.3. Oddball effects
The groups were rst compared using an ANOVA with the within-subject factors oddball type (emotional, perceptual) and position relative to the oddball ( 1, 0, +1). This analysis revealed a
main effect for position (F[2, 76] = 22.14, P < 0.005) and for the position  group (F[2, 76] = 4.27, P < 0.05) and type  position
(F[2, 76] = 6.04, P < 0.05) interactions. Most importantly, there
was a type  position  group interaction (F[2, 76] = 6.12,
P < 0.05), indicating that emotional words and their neighbors
were remembered differently by the patients. We therefore analyzed the data separately for controls and patients.

2.4. Analysis
3.4. Oddball effects in controls
The main question of the study was whether the oddball nouns
and their predecessors and successors (hence called nouns of interest) would be remembered more often than neutral control words.
To do so, a factor was calculated that indicated whether the word
of interest was remembered more or less often than control words,
where a value of 1 means that the specic noun (e.g., E-word) was
remembered as often as the control words, values >1 indicate
superior memory performance, and values <1 indicated inferior
memory performance for the noun of interest compared with control words. To calculate this value, for each word of interest the
absolute number of correctly remembered words was divided by
half of the number of remembered control words (as there were
a total of 20 control words, but only 10 in the other categories).
If, for example a subject remembered 8 of the 10 E-words and 14
of the 20 C-words, this would yield a factor of 8/(14/2) = 1.14, indicating that E-words were remembered 1.14 times more often than
control words.
With these factors a t test was calculated across all subjects of a
group that tested whether the calculated factor for a noun of interest differed from 1 within a 95% condence interval.
Group comparisons involved an ANOVA with the factors oddball
type (emotional, perceptual), position relative to the oddball ( 1,
0, +1), and group. Because of the small sample sizes, subgroup
analyses were performed nonparametrically with two-sample KolmogorovSmirnov tests. All statistical analyses were performed
with SPSS software.

Fig. 2 shows that control subjects recognized the E-words signicantly more often (factor of 1.16) than control words; the words
preceding and following the E-word, however, were remembered
less often (E 1: 0.88, E + 1: 0.85). All effects reached signicance
at P < 0.005. Although P-words were also remembered more often
(1.04) and surrounding words less often (P 1: 0.96, P + 1: 0.98),
none of these effects reached signicance (P > 0.2 for P 1,
P > 0.6 for P, P > 0.5 for P + 1).
A type (emotional, perceptual oddball)  position ( 1, 0, +1)
ANOVA revealed a main effect for position (F[2, 36] = 18.61,
P < 0.005) and the type  position interaction (F[2, 36] = 6.83,
P < 0.05), conrming that only emotional and nearby nouns were
remembered better or worse, respectively, compared with neutral
control nouns.
3.5. Oddball effects in patients
Patients did not remember E-words signicantly better than Cwords (factor 1.05, t = 1.34, P = 0.19) (see Fig. 3). Memory for the

3. Results
3.1. Emotional rating
The emotional words were not rated differently by the patients
relative to the normal controls who had participated in our pilot
study (t = 0.16, P > 0.87; note, however, that only 14 of the 21 patients had completed the questionnaire).
3.2. Memory for control nouns
We rst assessed memory for the neutral control nouns to reveal a possible general memory decit. As could be expected from

Fig. 2. Oddball effects during recognition in the control group.

Author's personal copy

370

N.G. Mller et al. / Epilepsy & Behavior 15 (2009) 367371

Fig. 3. Oddball effects during recognition in the patient group.

preceding word was also not altered (factor 0.98, t = 0.39,

P = 0.709); only for the subsequent word could a trend toward
memory impairment be observed (factor 0.88, t = 1.91, P = 0.08).
As for controls, P-words and nearby words were not remembered
differently from standard words (P: 1.03, P 1: 0.99, P + 1:1.00, P
all >0.3).
The type  position ANOVA revealed a main effect only for position (F[2, 40] = 5.40, P < 0.05) but no type  position interaction
as observed for controls.
3.6. Reaction times
Reaction times for correct responses during recognition were
assessed as indicators of response condence. As with the accuracy
data, a factor was calculated (mean reaction time to noun of interest divided by mean reaction time to control words) that indicated
whether responses to nouns of interest were faster or slower than
responses to control words (see Fig. 4).
Control subjects were faster when they responded to E-words
(factor 0.85, t = 4.5 P < 0.005), whereas the responses to words
that had been presented before or after the emotional word during
learning were relatively slowed (E 1: 1.09, t = 0.89, P = 0.39;
E + 1: 1.14, t = 3.73, P < 0.005). Patients did not show any signicant reaction time effects to any of the oddballs or their neighbors.
This is conrmed by the ANOVA that revealed a position  group
interaction (F[2, 76] = 3.6, P < 0.05) as well as a main effect of position (F[2, 76] = 6.18, P < 0.05) and a type  position interaction
(F[2, 76] = 5.08, P < 0.05).
3.7. Subgroup analyses
We tested whether any of the following factors had a signicant
effect on patients performance: sex, lateralization of the epileptic
focus (where applicable), frequency of seizures, presence of hippocampal sclerosis, number of drugs, verbal and working memory
performance as assessed in a neuropsychological test battery. None
of these factors had a signicant effect in the KolmogorovSmirnov
tests. Note, however, that interpretation of this observation is limited by the small sizes of the subgroups.
4. Discussion
In this study, patients with TLE did not show the emotion-driven memory effects that were observed in the healthy control

group, namely, enhanced memory for emotional words and retroand anterograde amnesia for preceding and subsequent words,
respectively. The lack of an emotion effect was independent of
the specic form of TLE.
Taken together, the results in our patient group are very similar
to those previously seen in healthy subjects who had received a
drug that interfered with the amygdala-driven noradrenergic response to emotional stimuli and those seen in a patient with bilateral sclerosis of the amygdalae [3]. As in the present study, in this
earlier work no effects on memory for perceptually deviant oddball
words were induced by amygdala dysfunction. Our patients memory for the perceptual deviant word was comparable to that of controls; so was their memory for control words. Also, they did not
rate the emotional valence of the words differently from the controls, speaking against a general decit in emotion processing. This
is in accordance with previous studies that have shown normal
skin conductance responses to emotional stimuli in patients with
amygdala damage [16]. As emotional memory critically relies on
amygdalahippocampus interaction [6,9,10], our results lead us
to speculate that this interaction is disturbed in patients with
TLE. In demonstrating that emotional memory is altered even in
unoperated patients, this nding extends earlier results on emotional memory in patients after unilateral temporal lobectomy
[17].
Patients with TLE suffer from structural damage within temporal regions that to a varying degree can also affect the amygdalae
[12,18]. To what extent the amygdalae were structurally affected
in our patient group could not be reliably assessed from their
MRI scans (often we had to rely on written reports) and no volumetric analyses were performed. Nevertheless, the behavioral deficit observed in emotion memory indicates that there is at least a
functional decit in the amygdalahippocampus interaction during memory formation for emotional information.
This study had been designed to closely match earlier studies on
emotion-related memory effects [3,10]. Although we replicated
many of the ndings of these earlier studies, some ndings differ
and came as a surprise: First, we did not observe a von Restorff effect for the perceptually deviant word. Apart from issues of power
we believe this was due to the rather subtle perceptual changes
introduced by the font changes and the limited number of deviant
fonts we used so that they had to be presented repeatedly. The latter may have reduced the novelty effect of the P-word, and it has
been shown that novelty is associated with memory improvement
as well [19]. Thus, one can speculate that if we had used stronger
and more frequent perceptual changes, we would have observed
the usual von Restorff effect. Most crucial, however, there was no
difference between patients and controls in memory for the Pwords.
Another somewhat unexpected nding was the reduced recognition performance in controls for words following the emotional
word in the sense of anterograde amnesia. Tulving originally reported a mere retrograde amnesia induced by emotional content
[11]. However, other studies have reported anterograde amnesia
as well, suggesting this effect to be susceptible to the specic
experimental design. Hurlemann and colleagues found that emotional arousal rather than valence correlates with anterograde
amnesia so that both stimulus types, those with negative and those
with positive content, can cause anterograde amnesia as long as
they induce high arousal [20]. We cannot test this hypothesis
based on our data as we neither assessed arousal nor had positive
words, yet it is easily conceivable that the negative words induced
more arousal than neutral words and hence interfered with encoding of the subsequent word.
In this study, we did not observe a signicant difference in
memory performance between patients with a left- and those with
a right-sided epileptic focus. A verbal memory decit is more often

Author's personal copy

N.G. Mller et al. / Epilepsy & Behavior 15 (2009) 367371

371

Fig. 4. Reaction time data for the control (left) and patient (right) groups.

found in patients with a left-sided focus. Neuropsychological

assessment of the patients (not reported here) revealed that of
the six patients with impaired verbal memory, ve had their focus
on the left and one had a bilateral focus. On the other hand, when
we tested whether these patients performed worse in our experiment compared with patients with verbal memory in the normal
range, we found no signicant effect. Yet, because of the small
sample size of these subgroups, this should not be overinterpreted.
In conclusion, in this study we found strong evidence that patients with TLE have a functional decit in amygdalahippocampus
interaction that normally underlies emotion-driven effects on
memory. As emotions, however, induce not only hypermnesia for
the emotional stimulus, but also amnesia for stimuli in close proximity, this may not exclusively be seen as a decit. It remains a
question for future research whether these laboratory ndings extend to events in everyday life, that is, whether life events of strong
emotional valence (and preceding or subsequent events) are
remembered differently by patients with TLE.
References
[1] Von Restorff H. ber die Wirkung von Bereichsbildungen im Spurenfeld.
Psychol Forsch 1933;18:299334.
[2] Gallagher M, Chiba AA. The amygdala and emotion. Curr Opin Neurobiol
1996;6:2217.
[3] Strange BA, Hurlemann R, Dolan RJ. An emotion-induced retrograde amnesia in
humans is amygdala- and beta-adrenergic-dependent. Proc Natl Acad Sci USA
2003;100:1362631.
[4] Cahill L, McGaugh JL. A novel demonstration of enhanced memory associated
with emotional arousal. Conscious Cogn 1995;4:41021.
[5] Phelps EA, LaBar KS, Anderson AK, OConnor KJ. Specifying the contributions of
the human amygdala to emotional memory: a case study. NeuroCase
1998;4:52740.

[6] Richardson MP, Strange BA, Dolan RJ. Encoding of emotional memories
depends on amygdala and hippocampus and their interactions. Nat Neurosci
2004;7:27885.
[7] Pitknen A, Pikkarainen M, Nurminen N, Ylinen A. Reciprocal connections
between the amygdala and the hippocampal formation, perirhinal cortex, and
postrhinal cortex in rat: a review. Ann NY Acad Sci 2000;911:36991.
[8] Cahill L, Prins B, Weber M, McGaugh JL. Beta-adrenergic activation and
memory for emotional events. Nature 1994;371:7024.
[9] Phelps EA. Human emotion and memory: interactions of the amygdala and
hippocampal complex. Curr Opin Neurobiol 2004;14:198202.
[10] Strange BA, Dolan RJ. Beta-adrenergic modulation of emotional memoryevoked human amygdala and hippocampal responses. Proc Natl Acad Sci USA
2004;101:114548.
[11] Tulving E. Retrograde amnesia in free recall. Science 1969;164:
8890.
[12] Bernasconi N, Bernasconi A, Caramanos Z, Antel SB, Andermann F, Arnold DL.
Mesial temporal damage in temporal lobe epilepsy: a volumetric MRI study of
the hippocampus, amygdala and parahippocampal region. Brain 2003;126(Pt.
2):4629.
[13] Bonilha L, Kobayashi E, Rorden C, Cendes F, Li LM. Medial temporal lobe
atrophy in patients with refractory temporal lobe epilepsy. J Neurol Neurosurg
Psychiatry 2003;74:162730.
[14] Hermann BP, Seidenberg M, Schoenfeld J, Davies K. Neuropsychological
characteristics of the syndrome of mesial temporal lobe epilepsy. Arch
Neurol 1997;54:36976.
[15] Benuzzi F, Meletti S, Zamboni G, et al. Impaired fear processing in right mesial
temporal sclerosis: a fMRI study. Brain Res Bull 2004;63:26981.
[16] Tranel D, Damasio H. Intact electrodermal skin conductance responses after
bilateral amygdala damage. Neuropsychologia 1989;27:38190.
[17] LaBar KS, LeDoux JE, Spencer DD, Phelps EA. Impaired fear conditioning
following unilateral temporal lobectomy in humans. J Neurosci
1995;15:684655.
[18] Bonilha L, Rorden C, Kobayashi E, et al. Voxel based morphometry study of
partial epilepsies. Arq Neuropsiquiatr 2003;61(Suppl. 1):937.
[19] Bunzeck N, Dzel E. Absolute coding of stimulus novelty in the human
substantia nigra/VTA. Neuron 2006;51:36979.
[20] Hurlemann R, Hawellek B, Matusch A, et al. Noradrenergic modulation of
emotion-induced forgetting and remembering. J Neurosci 2005;25:
63439.