Original article Early development destructive brain lesions and their relationship to epilepsy and hippocampal damage Ricardo A. Teixeira a , Li M. Li a , Sergio L.M. Santos b , Veronica A. Zanardi b , Donizeti C. Honorato a , Carlos A.M. Guerreiro a , Fernando Cendes a, * a Department of Neurology, University of Campinas (UNICAMP), Campinas, SP, CEP 13083-970, Brazil b Department of Radiology, University of Campinas (UNICAMP), Campinas, SP, CEP 13083-970, Brazil Received 7 August 2002; received in revised form 17 March 2003; accepted 19 March 2003 Abstract Fifty-one consecutive adult patients with epilepsy and early development destructive brain lesions were divided into three main groups according to the topographic distribution of the lesion on magnetic resonance imaging: hemispheric (H) ðn ¼ 9Þ; main arterial territory (AT) ðn ¼ 25Þ and arterial borderzone (Bdz) ðn ¼ 17Þ. Eight (89%) patients from group H presented status epilepticus in the first 5 years of life, five of them associated with fever. Seventeen of the 25 patients from group AT (76%) had an obvious hemiparesis observed early in life. In addition, major prenatal events were significantly more common in the group AT compared with the other two groups. Among patients from group Bdz, prenatal or postnatal events were not identified, except for one patient. Conversely, nine patients from group Bdz (60%) showed a history of perinatal complications. Hippocampal atrophy (HA) was determined by visual analysis in 74.5% of all patients and by volumetry in 92%. The frequency of HA was comparable among groups, but patients from group H presented the most severe atrophy and more frequent hyperintense T2 hippocampal signal. In conclusion, these three groups of patients with early destructive lesions and epilepsy (H, AT and Bdz), appear to have distinct pathogenic mechanisms. Our data show that there is a striking association of HA with different patterns of neocortical destructive lesions of early development. This association seems to be related to a common and synchronic pathogenic mechanism. The recognition of the pattern and degree of HA among these patients with intractable seizures may influence the surgical rationale. q 2003 Elsevier B.V. All rights reserved. Keywords: Epilepsy; Hippocampal atrophy; Hemiatrophy; Arterial borderzone; Infarct 1. Introduction Destructive brain lesions of early development include a wide variety of congenital, perinatal and postnatal acquired neuropathological conditions that have in common tissue necrosis of a previously normally formed brain, and constitutes one of the most important causes of neurological morbidity acquired in this period of development [1–3]. Different topographic and morphological patterns of brain lesions are recognized depending on the nature of the insult, its severity and the period of development in which it occurs [4,5]. It has long been known that certain regions of the brain are more vulnerable than others when the whole brain is submitted to an insult (e.g. diffuse hypoperfusion, status epilepticus) and this can be connoted by the term regional selective vulnerability [6]. The hippocampus is considered to be one of these vulnerable regions, a concept very well supported by experimental and clinical data [7–10]. Epilepsy is a common long-term sequel of those precocious destructive lesions, often presenting with intractable seizures [3]. The recognition of hippocampal atrophy (HA) on the magnetic resonance imaging (MRI) of Brain & Development 25 (2003) 560–570 www.elsevier.com/locate/braindev 0387-7604/03/$ - see front matter q 2003 Elsevier B.V. All rights reserved. doi:10.1016/S0387-7604(03)00065-2 * Corresponding author. Fax: þ 55-19-3289-3801. E-mail address: [email protected] (F. Cendes). Abbreviations: ACA, anterior cerebral artery; AI, asymmetry index; AT, arterial territory; Bdz, arterial borderzone; EEG, electroencephalogram; FSE, fast spin-echo; FOV, field of view; GRE, gradient echo; H, hemispheric; HA, hippocampal atrophy; HS, hippocampal sclerosis; HV, hippocampal volume; ICVC, intracranial volume-control group; ICVI, intracranial volume-individual; IR, inversion recovery; LH, left hippocampus; MCA, middle cerebral artery; MRI, magnetic resonance imaging; PCA, posterior cerebral artery; RH, right hippocampus; SE, status epilepticus; TE, time of echo; TI, time of inversion; TR, time of repetition.
11
Embed
Early development destructive brain lesions and their relationship to epilepsy and hippocampal damage
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Original article
Early development destructive brain lesions and their relationship to
epilepsy and hippocampal damage
Ricardo A. Teixeiraa, Li M. Lia, Sergio L.M. Santosb, Veronica A. Zanardib,Donizeti C. Honoratoa, Carlos A.M. Guerreiroa, Fernando Cendesa,*
aDepartment of Neurology, University of Campinas (UNICAMP), Campinas, SP, CEP 13083-970, BrazilbDepartment of Radiology, University of Campinas (UNICAMP), Campinas, SP, CEP 13083-970, Brazil
Received 7 August 2002; received in revised form 17 March 2003; accepted 19 March 2003
Abstract
Fifty-one consecutive adult patients with epilepsy and early development destructive brain lesions were divided into three main groups
according to the topographic distribution of the lesion on magnetic resonance imaging: hemispheric (H) ðn ¼ 9Þ; main arterial territory (AT)
ðn ¼ 25Þ and arterial borderzone (Bdz) ðn ¼ 17Þ. Eight (89%) patients from group H presented status epilepticus in the first 5 years of life,
five of them associated with fever. Seventeen of the 25 patients from group AT (76%) had an obvious hemiparesis observed early in life. In
addition, major prenatal events were significantly more common in the group AT compared with the other two groups. Among patients from
group Bdz, prenatal or postnatal events were not identified, except for one patient. Conversely, nine patients from group Bdz (60%) showed a
history of perinatal complications. Hippocampal atrophy (HA) was determined by visual analysis in 74.5% of all patients and by volumetry in
92%. The frequency of HA was comparable among groups, but patients from group H presented the most severe atrophy and more frequent
hyperintense T2 hippocampal signal. In conclusion, these three groups of patients with early destructive lesions and epilepsy (H, AT and
Bdz), appear to have distinct pathogenic mechanisms. Our data show that there is a striking association of HA with different patterns of
neocortical destructive lesions of early development. This association seems to be related to a common and synchronic pathogenic
mechanism. The recognition of the pattern and degree of HA among these patients with intractable seizures may influence the surgical
Visual analysis of MRI and multiplanar reconstruction
were systematically performed on a workstation (O2 Silicon
Graphic) using Omnipro software (Elscint Prestige, Haifa,
Israel). Curvilinear reconstruction of 3D volumetric images
was performed for all patients in a Power Macintosh using
the Brainsight software (Rogue Research, Montreal, Que-
bec, Canada). Curvilinear reconstruction was very helpful
for a clear visualization of the extent of cortical involvement
and assisted in the classification of the patients into three
different groups, according to the topographical distribution
of the lesion (Fig. 1): Hemispheric lesions (H) – patients in
this group had homogeneous atrophy of an entire hemi-
sphere without loss of tissue continuity; Arterial territory
lesions (AT) – this group had lesions limited to a main
arterial territory, constituted by cavitations or a localized
retraction of the cerebral tissue with an heterogeneous
appearance suggestive of a substantial gliotic scar; Arterial
borderzone lesions (Bdz) – these patients had lesions
consisting of atrophy between the main cerebral arterial
territories, with the aspect of ulegyria [15,16]. Eleven
patients showed more than one of these lesional patterns and
they were classified according to the most exuberant pattern.
We performed volumetric measurements of the hippo-
campi in all patients using 3 mm thick coronal IR images
with a semiautomatic software developed by the National
Institute of Health (NIH-image). Anatomic boundaries were
as described by a specific protocol [17]. Hippocampal
volumes (HV) were compared with a normal control group
consisting of 25 healthy volunteers. In order to check for
bilateral atrophy, HV were normalized to the individual
intracranial volume and mean intracranial volume of the
control group using the following formula: Corrected
HV ¼ HV £ ICVC=ICVI, where ICVC is the mean intra-
cranial volume of the control group and ICVI is the
individual intracranial volume. To determine the degree of
asymmetry between left and right normalized HV (LH, RH),
we used an asymmetry index (AI) defined by the following
formula: AI ¼ LH 2 RH=½ðLH þ RHÞ=2�. Values more than
2 standard deviations (SD) from the mean of normal
controls were considered abnormal.
In addition, to evaluate the volume distribution along the
major axis of the hippocampus, the mean hippocampal
cross-sectional areas of each group of patients were plotted
in a graph as a function of slice position. In the same graph,
we plotted the mean values of control group and lines
representing 2 SD below these values. The graphs were
analyzed qualitatively by visual assessment and comparison
between the ipsilateral and contralateral hippocampal
graphs. Patients with non-lateralized main destructive
lesions had their atrophic hippocampus assumed as the
‘ipsilateral hippocampus’. The profile of HA was classified
as diffuse or segmental.
We used the Pearson’s chi-square (x 2) or Fisher’s exact
tests for comparing proportions among groups. ANOVA test
and Tukey post hoc pairwise comparisons were applied for
comparison on continuous variables among the three
R.A. Teixeira et al. / Brain & Development 25 (2003) 560–570 561
groups. Pearson’s correlation coefficient (r) was used where
indicated. We considered the significance level of 0.05.
3. Results
Fifty-one patients were distributed as follows: Groups H,
AT and Bdz with 9, 25 and 17 patients, respectively. The
clinical and EEG findings are summarized in Table 1.
3.1. Clinical features
Direct interview with mothers was possible for 48
patients (94.1%). It is worth emphasizing that they could
recall and provide clear detailed histories concerning the
period of gestation and patients’ first years of life.
Among patients from group H, eight (89%) presented
status epilepticus (SE) in the first 5 years of life, five of them
associated with fever. In six patients the SE was the first
manifestation of their disease. Two patients presented SE
after experiencing uncomplicated seizures, one of them
since the neonatal period. The patient without the
antecedent of SE had the first symptoms at the age of 2
years with fever and conscience impairment for 2 days, so
that SE cannot be completely discarded in this case. SE was
by far more common in patients from group H (89%) than in
groups Bdz (6.6%) and AT (4%) (x 2, P , 0:001). All
patients who had SE experienced it only once and developed
Fig. 1. Curvilinear reconstruction of 3D T1-weighted and axial T2-weighted images illustrating the three groups. (A) Diffuse atrophy of the whole right
cerebral hemisphere without loss of tissue continuity. (B) Bilateral atrophy in the borderzone between the three main arterial territories with a left
predominance; a discrete left parasagittal atrophy can also be observed on the borderzone between the anterior cerebral artery and middle cerebral artery
(MCA) (white arrow). (C) Large cavity in the territory of the right MCA. R, right; L, left.
R.A. Teixeira et al. / Brain & Development 25 (2003) 560–570562
Table 1
Clinical and MRI features
Patient/Sex/Age (y) Group Early life events Lesion distribution on MRI S HT2 R HV L HV AI H HT2
1/M/41 H Age 2 y: fever þ conscience alteration for 2 days L Hem þ R Bdz (ACA–MCA) 2 24.5 26 6.0 þ
2/F/39 H Age 2 y: first sz (SE) þ fever þ motor deficit R Hem 2 27.6 21 22.5 þ
3/F/45 H Age 2 y: first sz (SE) R Hem 2 25.9 1.2 18.9 þ
4/M/17 11 Age 6 y: second sz (SE) þ motor deficit R Hem 2 26.8 21.5 17.5 þ
5/F/44 H Age 11 m: first sz (SE) þ exantema þ fever þ motor deficit R Hem þ 29.2 0.4 32.8 þ
6/M/40 H Neonatal sz. Age 6 y: (SE) R Hem þ 29.8 0 36.0 þ
7/F/48 H Age 11 m:first sz (SE) þ fever þ motor deficit R Hem þL Bdz (ACA–MCA) þ 211.7 23.8 50.0 NA
8/M/37 H Age 4 y: first sz (SE) þ fever þ motor deficit R Hem þ bilateral Bdz (ACA–MCA–PCA) þ 21.4 0.8 4.9 þ
9/F/34 H Age 2 y: first sz (SE) þ fever þ motor deficit R Hem þ 28.2 2.9 30.8 þ
44/M/27 AT Vaginal bleeding during pregnancy R MCA (whole territory) þ 26 20.8 15.6 þ
(continued on next page)
R.A
.T
eixeiraet
al.
/B
rain
&D
evelop
men
t2
5(2
00
3)
56
0–
57
05
63
a permanent motor deficit, clinically matching the diagnosis
of Hemiconvulsion–Hemiplegia–Epilepsy syndrome [18].
Patients from group AT exhibited quite a different
history. Seventeen of the 25 patients (76%) had an obvious
hemiparesis associated with hemiatrophy, which was
noticed by the parents within the first 2 years of life in all
of them. None of these 17 patients had any evidence of
postnatal events that could be related to the deficit (except
for patient 27 who had a complex febrile convulsion at the
age of 5 months followed by hemiparesis). Five patients
(20%) presented a discrete hemiparesis that was identified
by a neurologist in early childhood (for patient 31 we do not
have this information). Visual field deficits were identified
in five patients (20%). In addition, major prenatal events
such as abortion attempt and severe maternal trauma were
observed in six patients (25%) of group AT and in none from
the other groups, reaching statistical significance (Fisher’s
exact test, P ¼ 0:036).
Among patients from group Bdz, prenatal or postnatal
events were not identified, except for patient 17 who had the
antecedent of SE at the age of 18 months. Conversely, nine
patients (60%) showed a history of perinatal complications,
referred by the mother as fetal distress and neonatal
respiratory problems culminating with delay in hospital
discharge. This antecedent was significantly more frequent
in this group than in group H (11%) and group AT (33.3%)
(x 2, P ¼ 0:043). Four patients exhibited an evident motor
deficit that was detected by the parents in the first 2 years of
life without any postnatal potential precipitating insult. Two
other patients presented a severe visual acuity deficit and
another one had a visual hemifield deficit.
Duration of epilepsy was similar among groups
(ANOVA, P ¼ 0:14) and so was seizure frequency (x 2,
P ¼ 0:82). Epileptiform activity was present in routine
EEGs of 38 patients (74.5%); 26 patients (51%) showed
epileptiform activity restricted to the temporal region while
in eight other patients (15%) it was more widespread but
also involving temporal regions.
3.2. MRI studies
The MRI findings are summarized in Table 1.
Patients from group H revealed a uniform pattern of
atrophy involving the entire cerebral hemisphere. Five
patients (55%) showed a diffuse enhanced subcortical T2
signal in the affected hemisphere, and these were the
patients with the most severe hemiatrophy.
Fourteen patients from group Bdz (82%) exhibited
bilateral lesions, and in nine of them they were asymmetric.
Associated hyperintense T2 signal was present in 11
patients (65%), and it was more diffuse among patients
with the most extensive lesions. In 12 patients (70%), the
lesion was posteriorly distributed on the watershed between
the three main arteries. Five patients had their lesions more
anteriorly, on the watershed between the anterior cerebral
artery (ACA) and middle cerebral artery (MCA).Tab
le1
(co
nti
nued
)
Pat
ient/
Sex
/Age
(y)
Gro
up
Ear
lyli
feev
ents
Les
ion
dis
trib
uti
on
on
MR
IS
HT
2R
HV
LH
VA
IH
HT
2
45
/F/1
8A
TV
agin
alb
leed
ing
,fe
tal
dis
tres
sL
MC
A(w
ho
lete
rrit
ory
)þ
NA
NA
NA
NA
46
/M/2
2A
TF
etal
dis
tres
sL
PC
Aþ
21
.42
0.4
2.3
2
47
/M/1
8A
TT
rau
ma
du
ring
pre
gn
ancy
,fe
tal
dis
tres
sL
PC
A2
20
.82
0.1
1.6
2
48
/F/3
5A
TF
etal
dis
tres
sL
PC
Aþ
21
.82
5.3
10
.8þ
49
/F/2
5A
TV
agin
alb
leed
ing
du
ring
pre
gn
ancy
RP
CA
þ2
2.4
1.9
9.4
2
50
/M/2
2A
TT
win
ges
tati
on
RA
CA
þ2
11
.14
.52
51
/F/2
0A
TN
on
eR
AC
Aþ
0.5
0.2
0.6
2
L,l
eft;
R,ri
gh
t;m
,mo
nth
;y
,yea
r;N
A,n
ot
avai
lab
le;
sz,se
izu
re;
SE
,st
atu
sep
ilep
ticu
s;H
em,h
emis
ph
eric
;A
CA
,an
teri
or
cere
bra
lar
tery
;M
CA
,m
iddle
cere
bra
lar
tery
;P
CA
,post
erio
rce
rebra
lar
tery
;H
V,z
sco
reo
fh
ipp
oca
mp
alv
olu
me;
AI,
zsc
ore
of
asy
mm
etry
ind
ex(a
bso
lute
val
ues
);S
HT
2,
sub
cort
ical
hy
per
inte
nse
T2
sign
al;
HH
T2
,h
ipp
oca
mp
alh
yp
erin
ten
seT
2si
gn
al;
‘þ
’p
rese
nt;
‘2
’n
ot
pre
sen
t.
R.A. Teixeira et al. / Brain & Development 25 (2003) 560–570564
Nineteen patients from group AT (76%) presented lesions
on the territory of the MCA, four on the posterior cerebral
artery (PCA) and two on the ACA. Twenty patients (80%)
exhibited hyperintense T2 signal extending beyond the area
of the lesion. Five patients did not exhibit hyperintense T2
signal associated with the lesion, and all these lesions were
of cystic aspect. Other patients with cystic lesions showed
associated enhanced T2 signal, but more discrete when
compared with the patients with non-cystic lesions (Fig. 2).
Thirty-eight patients presented HA (74.5%) on visual
analysis of MRI (Fig. 3). Volumetric study showed HA in 47
patients (92%). All patients from group H, 16/17 (94%) of
group Bdz and 22/25 (88%) of group AT had abnormal
hippocampal volumes with a similar frequency distribution
(x 2, P ¼ 0:48). Moreover, in two patients (patients 7 and
15) who had a visual diagnosis of unilateral HA, the
volumetric study revealed bilateral HA. The HA was
unilateral in 41 patients and it was concordant with the
side of the main lesion, except in two (patients 13 and 46).
The HA was bilateral in six patients and it was more severely
ipsilateral to the main lesion in all six. Hippocampal
hyperintense T2 signal was more frequent among patients
from group H (100%) than from groups AT (54%) and Bdz
(65%) (x 2, P ¼ 0:01). In addition, presence of hyperintense
T2 signal was associated with the severity of atrophy of the
ipsilateral hippocampus (ANOVA, P ¼ 0:023), but not to
the contralateral hippocampus (ANOVA, P ¼ 0:85).
The three groups failed to show any difference in the
frequency distribution of unilateral (x 2, P ¼ 0:53) or
bilateral (x 2, P ¼ 0:48) HA. ANOVA demonstrated that
the normalized mean volume of the hippocampus ipsilateral
to the main lesion was different among the groups
(ANOVA, P , 0:001) and pairwise post hoc comparisons
showed that group H had smaller hippocampal volumes than
groups Bdz and AT. Contralateral normalized hippocampal
volume was not different among the groups (ANOVA,
P ¼ 0:82).
All patients from group H had temporal lobe atrophy
(visually assessed) associated with the HA. In 18 of 22 patients
(82%) of group AT with HA, the atrophy also involved
the temporal lobe. In contrast, only five of 16 patients (31%)
of group Bdz with HA had temporal lobe atrophy, and it
was different from the other two groups (x 2, P ¼ 0:001).
The graphs with the distribution of the hippocampal
cross-sectional areas of the different groups showed some
interesting features. The three groups presented a diffuse
volume loss of the ipsilateral hippocampus but it was
unequivocally more pronounced on the anterior sections.
The long axis of the ipsilateral hippocampus in group H was
shortened in three slices (9 mm) compared with the other
groups (Fig. 4). The patients without HA presented normal
graphs individually.
3.3. Correlation of MRI, clinical, and EEG data
Epilepsy duration was weakly correlated to the volume
of the hippocampus ipsilateral to the main lesion
(r ¼ 20:29, P ¼ 0:046) but no correlation was found with
the contralateral hippocampus (r ¼ 20:23, P ¼ 0:115). In
contrast, hippocampal volumes were not significantly
different among patients with rare seizures, patients with
weekly or monthly seizures and those with daily seizures
ðP . 0:4Þ. History of SE, in turn, was strongly associated
with the severity of the hippocampal volume loss ipsilat-
erally to the main lesion (ANOVA, P , 0:001) but not
contralateral (ANOVA, P ¼ 0:78).
3.4. Surgical results
Nine patients (one H, one Bdz, seven AT), all with HA,
were submitted to a standard anterior temporal lobe removal
including the amygdala and anterior portion of the
hippocampus for control of refractory seizures. This
approach was proposed to the patients as a first step of an
escalated strategy where other interventions could become
necessary. Those patients were selected on the basis that
they presented electroclinical data indicating that the
affected temporal region was the most epileptogenic, and
that a limited temporal resection could have an impact on
seizure control.
Five patients achieved good seizure control with the
temporal resection after a follow-up of at least 12 months. In
two of them (patients 48 and 49), pre-operative MRI
revealed an old infarct lesion with cystic aspect in the PCA
territory (Engel’s classes I and II [19]); one (patient 39) had
an old infarct lesion in the whole territory of the MCA with
a cystic appearance (Engel’s class II); one (patient 3) had
hemiatrophy without hyperintense T2 signal (Engel’s class
II), and the remaining patient (patient 27) had a small non-
cavitated infarct on the MCA territory (Engel’s class I).
The temporal resection was not successful in three
individuals (patients 32, 33 and 44). They presented infarcts
in the MCA territory with a diffuse enhanced T2 signal
extending beyond the area of the lesion. They had
hemiplegia prior to surgery, with neither fine finger
movements nor foot tapping and underwent reoperation
Fig. 2. Coronal fast spin-echo T2-weighted images showing different
degrees of subcortical enhanced signal. Left: Patient 9. Right hemiatrophy
associated with a diffuse subcortical enhanced signal. Right: Patient 49.
Cystic lesion in the territory of the PCA, contiguous to the lateral ventricle.
A discrete area of hyperintense T2 signal can be identified adjacent to the
lesion (arrow). R, right; L, left.
R.A. Teixeira et al. / Brain & Development 25 (2003) 560–570 565
with extension of the initial resection guided by electro-
corticography. After this second procedure, their complex
partial seizures were controlled but the other seizure types
remained unchanged. Patient 44 persisted only with a
monthly discrete sensitive experience over the jaw region
after a follow-up of 20 months (Engel’s class I). A
functional hemispherectomy was performed in patients 32
and 33, and both are seizure-free after a period of 9 and 30
months, respectively.
The ninth patient who underwent surgery (patient 24)
had a bilateral arterial borderzone lesion associated with left
HA and an enhanced T2 signal extending beyond the lesion.
Before surgery, this patient presented daily complex partial
seizures and ictal and interictal EEG activity well localized
over the left temporal region. After surgery, the patient
continued to present monthly seizures (right hemiconvul-
sion) and is now being reevaluated for possible
reintervention.
Fig. 3. Coronal IR T1-weighted images showing unilateral HA (arrows) in patients of the three different groups studied: hemispheric (left), arterial territory
(center) and arterial borderzone on the right (the borderzone lesion is not shown). R, right; L, left.
Fig. 4. Graphs of the hippocampal cross-sectional areas as a function of slice position. Three lines are shown on each plot: (1) the mean of the control group
(broken line); (2) 2 SD below the mean of the control group (dotted line); (3) the mean of the group of interest (continuous line). The unity shown in the Y axis
(0–200) represents the cross-sectional area of each hippocampal slice in mm2. The contralateral hippocampal graphs (not shown here) were very similar to the
graphs of the control group. The three patient groups presented a diffuse hippocampal volume loss ipsilaterally to the main lesion, more pronounced on the
anterior sections. Group H (hemispheric) showed lower ipsilateral hippocampal values and a shortening of its long axis not observed in the other groups,
reflecting a more severe insult rather than a higher frequency of HA in this group.
R.A. Teixeira et al. / Brain & Development 25 (2003) 560–570566
Enhanced diffuse T2 signal was absent in the five patients
who achieved good seizure control (Engel’s classes I and II)
after temporal resection, and present in all three patients
with poor seizure control (x 2, P ¼ 0:003).
4. Discussion
HS is one of the most frequent neuropathological
substrates associated with epilepsy and is characterized by
hippocampal cell loss and reorganization of the dentate
gyrus [8]. HA on MRI, frequently associated with
hyperintense T2 signal, has been shown to correlate with
HS and to the degree of hippocampal cell loss [8,20–22].
Identification of HA in patients with lesional epilepsies is
important for surgical treatment planning and for prognosis
of postoperative seizure outcome [13,20–22].
The etiology of HS is still a matter of controversy and it
is now assumed that it can be the end result of different
pathogenetic conditions [23,24]. In the present series, all
patients had destructive neocortical brain lesions acquired in
early life and the division into three groups was made on the
basis of the topographical distribution of these lesions.
However, we observed that the natural histories within each
group were also distinct, tending to reflect three different
kinds of insults known to be potential causes of HS.
Most of the patients with cerebral hemiatrophy presented
the SE antecedent in the first years of life. SE is consistently
associated with neuronal necrosis in vulnerable regions of
the brain, as proven by neuropathological studies in humans
and experimental studies in animal models [25–30]. Brain
damage primarily affects the hippocampus (hilus and CA1
and CA3 segments), amygdala, piriform cortex and to a
lesser extent the cerebellar cortex, thalamus and cerebral
cortex [25–28]. It has been well documented that the same
neuropathological patterns can be strictly unilateral when
the SE is lateralized [29,30].
Arterial borderzone lesions were significantly associated
with perinatal complications. There is large clinical and
experimental evidence that perinatal hypoxic–ischemic
insults are strongly related to this pattern of lesion [31,
32]. The end fields of the major cerebral arteries are
expected to be the first regions to experience perfusional
insufficiency after a drop in systemic blood pressure. The
classical borderzone lesion is a bilateral wedge-shaped
infarct at the parasagittal high convexity, between the
territories of the anterior and middle cerebral arteries.
However, the parieto-occipital region is more commonly
involved and it lays at the junction of the territorial
borderzones of the three major arteries; although the lesions
are usually bilateral they can be fairly asymmetric or
unilateral [3,33]. The hippocampus itself is also considered
to be in a watershed region (between the PCA and the MCA)
[34–36]. Experimental studies with primates demonstrated
hippocampal damage commonly associated with arterial
borderzone lesions after hypotension or hypoxia, but there
are many instances where the hippocampi are spared [37,
38]. This inconsistency may reflect different vulnerabilities
of the hippocampi and the parasagittal region to different
insults.
Patients with lesions on a main arterial territory had
histories of risk factors, indicating that the lesions were
presumably of prenatal or perinatal origin. Perinatal
complications, such as prolonged delivery and fetal distress,
were not uncommon among these patients but it is well
known that they can be just a consequence of a prenatal
insult [39]. The arterial supply of the hippocampus is made
by the anterior choroidal artery (MCA branch) and by the
trunk of PCA or its branches with certain variability [40,41].
Thus, insults that lead to damage in the MCA or PCA
territory can also involve the hippocampus. These insults
can represent arterial obstruction, but in the context of an
abnormal or immature anastomotic supply, a systemic
circulatory failure can also lead to localized damage in the
territory of individual arteries [38,42].
There is evidence that the formation of a brain glial scar
in response to a destructive insult is very discrete before the
30th week of gestation, and that insults before this period
usually produce cystic lesions [3,5,43]. Hyperintense T2
signal (suggestive of gliosis) over and around the main
lesion was commonly identified among patients of the three
groups studied. Among patients with lesions in arterial
territories, only patients with cystic lesions did not exhibit
hyperintense T2 signals. However, there were also patients
with cystic lesions who presented hyperintense T2 signal
adjacent to the limits of the cavity, confirming that gliosis
can coexist with cystic lesions of early development.
Pathological studies in humans and animals show that
besides the cavity (which is the result of a complete infarct),
there are neighboring areas in which astroglia survive and
most neurons are selectively destroyed, constituting an
‘incomplete infarct’ [44,45]. Thus, the hyperintense T2
signal discussed earlier could represent areas of incomplete
infarcts, and its extension can be related not only to the time
the lesion occurred but also to the nature and severity of the
insult, as well as the efficiency of the collateral circulation
[4,46]. The relevance of the severity of the insult is well
illustrated in this series by the patients with hemiatrophy
and borderzone lesions, as we observed that the hyper-
intense T2 signal was more evident among patients with the
most severe lesions. This is in agreement with pathological
observations which demonstrated that reactive gliosis is
greater in patients with more severe atrophy [47].
We identified a striking high frequency of HA among the
three groups studied. This coexistence of HA with
extrahippocampal lesions has been referred to as dual
pathology [11,48]. Previous studies have shown that this
association occurs in 5–30% of patients with refractory
epilepsy and that the degree of HA was not affected by the
duration or severity of seizures [11,48]. This suggests that a
common pathogenic mechanism may cause concomitant
HA and extrahippocampal lesions in these patients.
R.A. Teixeira et al. / Brain & Development 25 (2003) 560–570 567
However, studies in animals and in humans have shown that
repetitive seizures can produce pathological and MRI
features of HS [49,50]. Our results suggest that habitual
repetitive seizures (excluding, of course, SE) may have only
a minor role in the progression of HA in these patients, since
severity of HA showed a weak correlation with duration of
epilepsy.
The nature of the extrahippocampal lesion is related to
the frequency of associated HA [12,51,52]. An earlier MRI
volumetric study found that HA was more common in
patients with cortical dysplasias (25%), porencephalic
(31%) and gliotic lesions (23.5%) than in those with tumors
or vascular malformations (2 and 9%) [11,51]. Compared
with the patients in this previous work with destructive
lesions (porencephalic and gliotic lesions), the present series
showed a much higher frequency of HA, probably reflecting
different patient selections. Ho et al. [52] identified HA in
95% of their series of patients with congenital porence-
phaly, a result similar to ours. Thus, HA seems to be much
more associated with large destructive lesions than to other
types of lesions. It is worth emphasizing that all these
studies, including the present one, analyzed highly selected
patients from epilepsy surgery programs, and this can
account for the high frequency of HA.
Most of the patients analyzed by us (particularly
from groups H and AT) had the HA ‘contiguous’ to the
neocortical lesion and one could even suggest that the term
single pathology seems more adequate. However, the
hippocampus and neocortex are anatomically separated
areas and also have distinct embryological origins. More-
over, a good surgical result after temporal resection in some
of these patients indicates that the lesions have at least
different epileptogenic roles. The term dual pathology was
coined and disseminated in the epilepsy literature to denote
that two distinct potential epileptogenic lesions are present
in the same patient. It does not have the connotation that the
two lesions have neuropathologic substrates of distinct
natures, neither does it imply that the lesions occurred at
different times. Thus, we consider the term dual pathology
appropriate in these cases.
The frequency of HA was comparable among the groups,
but patients from group H presented the most severe
hippocampal volume loss, with a shortening of its long axis
and also exhibited T2 hyperintense signal in the atrophic
hippocampus more frequently. This suggests that the
hippocampi of patients from group H were submitted to a
more caustic insult than the ones from groups AT and Bdz.
The antecedent of SE, most commonly identified among
patients from group H, seems to play a major role on the
severity of HA. This is supported by studies of ‘pure’ HA, in
which patients with the antecedent of prolonged febrile
convulsions had smaller hippocampal volumes and more
hippocampal neuron losses than those without [53–55].
Earlier studies have addressed the question of hippo-
campal volume loss along its major axis in patients with
‘pure’ HA. Some studies showed that diffuse atrophy is
the most common pattern identified and that segmental
anterior atrophy is also identified but less frequently
[56–59]. However, other studies found that the majority
of patients had segmental anterior atrophy [60,61]. In the
present series, the HA in the three groups was diffuse but
relatively more severe in the anterior segment. This is in
keeping with pathological studies of HS that show a diffuse
but greater neuronal loss in the anterior segment of
hippocampus [62,63]. The diagnostic sensitivity for HA in
our series was not increased by segmental volumetric
analysis of hippocampus.
The high frequency of interictal epileptiform activity
over the temporal lobe and the striking frequency of
associated HA suggest that the hippocampus can play an
important role in seizure generation in these patients.
Few studies have addressed the problem of surgical
approach for patients with dual pathology. Cascino et al.
[64] reported three patients with temporal lesions (two
with vascular malformation and one with a ganglioglioma)
who achieved seizure control only after a second operation
in which the atrophic hippocampus was removed. Li et al.
[21] compared the results between lesionectomy, resection
of the atrophic hippocampus and the combination of both
procedures in patients with diverse pathologies. Although
the sample was too small, this study [21] suggested that
the combination of the two procedures offers the best
seizure control. The same authors expanded the number of
patients in a second report and confirmed their first
impressions that resection of both lesions should be
considered whenever possible [13]. In selected cases,
however, the removal of the atrophic hippocampus can be
sufficient to achieve seizure control as demonstrated by Ho
et al. [52] and by our study.
All patients operated in the present series exhibited
extensive hemispheric lesions and the HA was presumably
the most epileptogenic lesion, according to electroclinical
data. Resection of the extrahippocampal lesion in these
cases meant hemispherectomy or multilobar resection. In
order to avoid the risks and new deficits potentially
associated with such procedures, we proposed a standard
temporal resection as a first step of an escalated approach.
We could observe that patients with diffuse enhanced T2
signal tended to present poor outcomes after temporal
resection and achieved seizure control only after resection
of the extrahippocampal lesion. These results, albeit
preliminary, led us to begin considering the possibility of
resection of both the atrophied hippocampus and the
extrahippocampal lesion in a single step whenever there
are signs of diffuse gliosis, even if the HA seems to be the
most epileptogenic lesion. The extent of the resection must
naturally be dictated by clinical-EEG findings and the
principle of avoidance of new neurological deficits. On the
other hand, we are encouraged to propose a standard temporal
resection to patients with electroclinical localization over
the temporal lobe and lesions without diffuse gliotic signs,
especially for those with cystic lesions.
R.A. Teixeira et al. / Brain & Development 25 (2003) 560–570568
In conclusion, our data show that there is a striking
frequency of HA associated with different patterns of
neocortical destructive lesions of early development. This
association seems to be related to a common and synchronic
pathogenic mechanism. The recognition of HA among these
patients is of major importance since it can influence the
surgical rationale in those who present intractable seizures.
Acknowledgements
R.A.T. was supported by a grant from FAPESP
(# 98/13101-8).
The authors are grateful to Dr Susana Mori, Eliane
Kobayashi, Tania Cardoso and Alberto Costa, who assisted
in the management of these patients and to Marcelo
Brunnini, Pablo Rios and Fabricio Ramos for their technical
support.
References
[1] Raybaud C. Destructive lesions of the brain. Neuroradiology 1983;25: