DEVELOPMENTAL MEDICINE & CHILD NEUROLOGY

ORIGINAL ARTICLE

Topography of brain damage in metabolic hypoglycaemia is

determined by age at which hypoglycaemia occurred
SVETLANA GATAULLINA 1 | PASCALE DE LONLAY 2 | GEORGES DELLATOLAS 3 | VASSILI VALAYANNAPOULOS 4
SILVIA NAPURI 5 | LNA DAMAJ 5 | GUY TOUATI 2 | CECILA ALTUZARRA 6 | OLIVIER DULAC 7
NATHALIE BODDAERT 8

1 Neuropediatrics Department, Necker-Enfants Malades Hospital, AP-HP, Paris; 2 Centre for Inherited Metabolic Disorders, Necker-Enfants Malades Hospital AP-HP, Universit
Paris Descartes, Paris; 3 Inserm U669, Universit Paris Descartes, Universit Paris-Sud, Paris; 4 Reference Centre for Inherited Metabolic Disorders, Necker-Enfants Malades
Hospital AP-HP, Paris; 5 Paediatric Department, CHU Hpital Sud, Rennes; 6 Paediatric Department, CHU Hpital Saint-Jacques, Besanon; 7 INSERM UMR663, Universit Paris
Descartes, Paris; 8 Radiologic Department, Necker-Enfants Malades Hospital, AP-HP, Universit Paris Descartes, Paris, France.
Correspondence to Dr Nathalie Boddaert at Service de Radiologie pdiatrique, Hpital Necker-Enfants Malades, 149 rue de Svres, 75015 Paris, France. E-mail: nathalie.boddaert@nck.aphp.fr

PUBLICATION DATA

AIM Having previously shown that comorbidity is a major determinant of neurological sequelae

Accepted for publication 3rd September 2012.

Published online 4th December 2012.

in hypoglycaemia, our aim was to describe the neuroimaging patterns of brain damage in different
hypoglycaemic situations and to elucidate the factors that determine lesion topography.
METHOD We reviewed 50 patients (31 females, 19 males) with symptomatic hypoglycaemia
(<2.8mmol L) occurring between 1 day and 5 years of age (median 4d) who had undergone
magnetic resonance imaging (MRI; at least axial T2-weighted, sagittal T1-weighted, and coronal
fluid-attenuated inversion recovery [FLAIR]-weighted imaging). MRI was performed during the
follow-up examination at least 1 month after the occurrence of symptomatic hypoglycaemia, i.e.
between 1 month and 5 years of age (median 3mo). Hypoglycaemia resulted from three inborn
errors of metabolism: congenital hyperinsulinism (33 patients), fatty acid b-oxidation disorders
(13 patients), or glycogen storage disease type I (four patients). We selected the patients with clear
MRI abnormalities and analysed their topography according to aetiology and age at occurrence of
the lesion.
RESULTS The topography of the brain lesions depended on age: from the neonatal period to
6 months of age, lesions predominantly involved the posterior white matter; between 6 and
22 months the basal ganglia, and after 22 months the parietotemporal cortex (p=0.04).
INTERPRETATION The relationship between brain lesions and age could reflect the maturation
sequence of the brain.

ABBREVIATIONS

FAOD Fatty acid b-oxidation defect

GSDI Glycogen storage disease type I

Hypoglycaemia is a cause of brain damage that particularly

affects the neonate and infant.1,2 Damage is usually considered
to involve the white matter of the posterior regions,3 but the
basal ganglia and cortex may also be affected.4 The factors
determining the topography of brain injury remain unknown.
In a previous study focusing on inborn errors of metabolism
that prevent the production of alternative sources of energy,
we showed that, in addition to aetiology and prolonged convulsions, additional factors, namely hypoxiaischaemia and
infection, are the main determinants of neurological sequelae,
whatever the age at the time of occurrence of the hypoglycaemia.5 We now investigated how age, although not the main
factor influencing neurological sequelae, affects the topography of brain lesions produced by hypoglycaemia.

magnetic resonance imaging (MRI). Hypoglycaemia was

defined as a plasma glucose level below 2.8mmol L (Dextrostix followed by venous measure). Severe sequelae were considered to be global psychomotor delay, microcephaly, motor
deficit, lack of visual contact, and or pharmacoresistant
epilepsy. Mild sequelae consisted of speech delay, learning
difficulties, and pharmacosensitive epilepsy.

METHOD
Of 169 patients who experienced hypoglycaemia due to three
inborn errors of metabolism, all 50 with symptomatic hypoglycaemia with or without neurological sequelae underwent

Participants
This cohort study was a retrospective analysis of data from
files, for which no additional investigation was performed.
Therefore, under French law, no ethical approval was
required.
There were four patients with glycogen storage disease type
I (GSDI), 13 with fatty acid b-oxidation defect (FAOD), and
33 with congenital hyperinsulinism presenting with hypoglycaemia that occurred between 1 day and 5 years of age
(median 4d). During the hypoglycaemic event, additional
factors were present in 27 out of 50 patients and consisted of

162 DOI: 10.1111/dmcn.12045

The Authors. Developmental Medicine & Child Neurology 2012 Mac Keith Press

fever (over 38.5C), infection, and or hypoxiaischaemia

(shock, cardiorespiratory arrest, or severe bradycardia). In
addition, status epilepticus with or without coma was observed
in 25 patients and brief seizures in 18, while seven patients had
neither seizures nor coma. Clinical sequelae were severe in 22
children and mild in 13 others; 15 children experienced no
sequelae.

What this paper adds

Topography of brain damage is mainly determined by age at occurrence of
complicated hypoglycaemia.
This age relationship is consistent with the physiological sequence of brain
maturation.

hypoglycaemia differed between those with and without brain

lesions.

Imaging and analysis of images

In all patients, MRI was performed during the follow-up
examination more than 1 month after the symptomatic hypoglycaemia, at between 1 month and 5 years of age (median
3mo), and comprised at least axial T2-weighted, sagittal T1weighted, and coronal fluid-attenuated inversion recovery
(FLAIR)-weighted imaging. We carried out a visual analysis
to evaluate the anatomical abnormalities of the posterior fossa,
cortical grey matter, periventricular, juxtaventricular, subcortical white matter, and basal ganglia. The cortical and white
matter abnormalities are seen as a hypointensity on T1 and
hyperintensity on T2 and FLAIR sequences. In addition, in
the case of 11 patients who also underwent MRI within
48 hours of severe hypoglycaemia, axial diffusion-weighted
imaging sequences (b=1000mm2 s) were carried out and the
diffusion-weighted imaging abnormalities corresponding to
abnormal apparent diffusion coefficient values were measured
when possible. Assessment of MRI anomalies and disease
course were carried out independently by a radiologist and a
clinician respectively.

Topography of the brain lesions

Three types of brain lesion dominated the results.
First, white matter lesions, seen as T2 hyperintensity
(19 28), affected the parieto-occipital area of one (three
patients; Fig. 1a) or both (16 patients) hemispheres. Lesions
extended to the frontal white matter in four patients and had a
multicystic aspect. In two patients a haemorrhagic component
was described. In nine patients, white matter hyperintensity
reached the overlying posterior cortex.
Second, basal ganglia lesions (i.e. putamen, caudate, and
eventually pallidal nucleus abnormalities), either unilateral
(one patient) or bilateral (four patients) (Fig. 1b), appeared as a
hypersignal on T2- and FLAIR-weighted sequences. The
thalamus and brainstem were also affected in one patient each.
Basal ganglia lesions occurred in isolation in four patients
and in association with unilateral cortical hypersignal in the
other.
Thirdly, cortical hypersignal on T2 and diffusion sequences
(Fig. 1c) predominating in the temporoparietal areas on one
side (four patients) with extension into the hippocampus was
seen in three patients. Involvement was confined to the cortex
in three patients, and associated with basal ganglia damage in
the other.
An isolated unilateral left hippocampal T2 hypersignal was
noted in one patient.

Statistical analysis
We compared patients with and without lesions and we
explored associations within the lesion group according to the
pattern of brain damage. We used Fishers exact test to compare proportions, and a t-test or Wilcoxon rank test to compare a continuous variable between groups.

Topography of brain lesions according to age and aetiology

Age at discovery of hypoglycaemia (Fig. 2)
Parieto-occipital white matter lesions were seen following
symptomatic hypoglycaemia with seizures in 19 neonates and
infants under 6 months (within the first 5d in 17 patients and
at 4mo and 6mo in one patient each; median 2d). Basal ganglia
were affected in five infants aged between 6 and 22 months
(median 12mo), and temporoparietal cortical damage occurred
in four patients at the ages of 20 and 22 months and 3 and

RESULTS
MRI revealed brain lesions in 28 patients. Compared with the
22 patients without lesions, among patients with lesions the
mean glucose level was lower, the incidence of long-lasting
seizures coma was higher, the incidence of brief seizures was
lower, and more additional factors were present (Table I).
There was no evidence that the median age at discovery of

Table I: Association of brain neuroimaging pattern with glucose level, seizures, and additional factors
Lesion (2), n=28

Glucose level (mmol L), mean (SD)

Long-lasting seizures coma, n (%)
Brief seizures, n (%)
Comorbidityc, n (%)

No lesion (1),
n=22

White matter (3),

n=19

Basal ganglia
cortex (4), n=9

Total lesion,
n=28

p (1) (2)a

p (3) (4)b

1.20 (0.80)
2 (9)
13 (59)
7 (32)

0.69 (0.59)
14 (74)
5 (26)
12 (63)

0.43 (0.36)
9 (100)
0
8 (89)d

0.61 (0.53)
23 (82)
5 (18)
20 (71)d

0.006
<0.001
0.004
0.01

0.17
0.14
0.14
0.17

Comparisons between patients with lesions and patients without lesions; t-test for means and Fischers exact test for proportions. bComparisons
within the lesion group between patients with white matter lesions and patients with cortex basal ganglia lesions: t-test for means and Fischers
exact test for proportions. cFever, infection, hypoxiaischaemia. dMainly fever infection as comorbidity.

Brain Lesion Topography in Hypoglycaemia Svetlana Gataullina et al. 163

c1

c2

Figure 1: (a) Axial T1-weighted sequence showing left parieto-occipital white matter anomalies in a 2-month-old infant with fatty acid b-oxidation defect
who at 3 days had hypoglycaemia, status epilepticus, and respiratory distress, extending to the overlying cortex. (b) Coronal fluid-attenuated inversion recovery-weighted sequence showing bilateral putamen and caudate atrophy with hypersignal in a female with medium-chain acyl-CoA dehydrogenase deficiency
who at 9 months had fever, coma, seizures, collapse, and blood level below 0.5 mmol L. (c) A male with glycogen storage disease type I who presented at
5 years with prolonged left-side convulsions and hemiparesis with fever: axial diffusion-weighted sequence showing cortical hyperintensity in right temporal
areas 48 hours following status epilepticus (c1) and coronal T2-weighted sequence showing the atrophy of right hemisphere 1 month later (c2).

5 years (median 2y 5mo). In other words, when age at discovery of hypoglycaemia was under 6 months, all patients had
white matter lesions (except one with a hippocampal lesion
that had occurred at 3mo), and none had basal ganglia or cortical damage (p<0.001; Wilcoxon rank test). At 6 months, one
patient had white matter and one basal ganglia lesions.
Between 6 and 12 months, all three patients had basal ganglia
lesions. Among the three patients with hypoglycaemia
between 20 and 22 months, one had basal ganglia damage,
one cortical damage, and one both basal ganglia and cortical
damage. Finally, the two patients with hypoglycaemia after
2 years had cortical damage. Considering the seven patients
with either basal ganglia lesions only or cortical damage only,
age at hypoglycaemia was lower in the four with only basal
ganglia lesions than in the three with only cortical damage
(Wilcoxon unilateral signed-rank test).

Aetiology
All 15 patients with hyperinsulinism had posterior white
matter damage, and hyperinsulinism appeared to be an
164 Developmental Medicine & Child Neurology 2013, 55: 162166

aetiological determinant of this damage; however, it is not specific and is also shared by both the other aetiologies.
Only one patient with GSDI, whose seizures had occurred
in the first days of life, had white matter damage whereas
another patient acquired temporoparietal cortical damage at
the age of 5 years. Two others had no lesion.
Regarding FAOD, three of the 10 patients had white matter
lesions alone; in five others the basal ganglia was affected
(together with the hippocampus and cortex in one case), and
two patients had predominantly parietotemporal cortical damage, requiring in one case a left hemispherotomy. There were
five patients with medium-chain acyl-CoA dehydrogenase
deficiency, three with long-chain 3-hydroxyacyl CoA dehydrogenase, and one each with, respectively, glutaric acid type II
and carnitine palmitoyltransferase II, with no relation to
topography.
Of the seven patients with neither seizures nor coma, none
had brain lesions. Of the 19 with brief seizures, six had brain
lesions; in all six, the seizure occurred early, in the first
6 months of life (median age 5d, range 4h6mo). Following

Aetiology
GSDI
FAOD
HI
Age (months)
1
White matter

12

24

36

Basal ganglia

48

60

Cortex

Figure 2: Topography of brain lesions according to age at onset of symptomatic hypoglycaemia and aetiology (white matter damage vs basal ganglia, with or without cortex vs cortex). GSDI, glycogen storage disease
type I; FAOD, fatty acid b-oxidation defect; HI, hyperintensity.

coma status epilepticus, all but one patient had a brain

lesion.
Plasma glucose level, coma status epilepticus, brief seizures,
and associated factors were not significantly associated with
the pattern of brain damage (Table I). There was no evidence
of a difference in the mean plasma glucose level between
patients with lesions in the basal ganglia and those with lesions
in the cortex. However, with the small sample sizes for these
comparisons, power to detect differences is expected to be
poor for all but extreme differences.

DISCUSSION
This study shows that (1) the neurological sequelae of symptomatic hypoglycaemia in childhood are due to brain damage
that affects white matter, basal ganglia, or the parietal cortex;
and (2) topography is determined by age, but not by glucose
level or aetiology.
Indication, timing, and reliability of MRI
Investigation was mostly performed in patients in whom mild
to severe neurological sequelae remained following seizures or
coma. In a few children with neither seizures nor sequelae,
MRI was requested following transient problems, including
focal deficit; in all such cases the results were normal.
Although subclinical lesions might have been overlooked, this
is unlikely, and, in any case, the prevalence of lesions was not
the subject of our research. Anomalies disclosed by MRI may
disappear within a few weeks following injury, either because
of reversible oedema or because of the collapse of cystic
lesions, but since such anomalies could be disclosed only by
neuropathology, we decided to base our study on chronic
lesions visible on MRI performed at least 1 month after the
injury.
Type of brain lesions
Our findings are consistent with reports from the literature of
hyperintense lesions, encephalomalacia, and cerebral atrophy
following hypoglycaemia.6 In newborn infants, lesions affect
the occipital white matter in up to 82% of cases.7 Lesions in
older children are rarely reported. One individual with basal

ganglia8 and two with cortical lesions have been described.9

We encountered one child with a brainstem lesion not previously reported, to our knowledge, but encountered fewer cases
of brain haemorrhage, which is reported to affect up to 30%
of patients with neonatal hypoglycaemia.10,11 Our choice of
patients, with exclusively inborn errors of metabolism, could
explain this divergence. The interpretation of images may be
challenging since the injuries are often considered as middle
cerebral artery territory infarctions10 whereas we viewed such
lesions as the result of stroke-like episodes, as is the case in disorders affecting energy metabolism. The possible prognostic
value of diffusion sequence performed early has been pointed
out12 but this needs to be validated.

The role of aetiology

Typical hypoglycaemic parieto-occipital white matter lesions
affect mainly hyperinsulinaemic newborn infants.3 Occasional
extension to frontal areas in some neonates with severe hypoglycaemia involves distal field infarctions, a clearly ischaemic
condition.13 The overlying occipital cortex is occasionally
affected. Such images correlate with symptomatic neonatal
hypoglycaemia not only due to congenital hyperinsulinism but
also as a result of FAOD and in rare cases GSDI, i.e. mainly
hypoglycaemic conditions associated with an inability to use
alternative fuels and accompanied by additional factors.5
In addition to this posterior location, the basal ganglia as
well as the temporoparietal cortex and hippocampus were
affected. However, we did not encounter a combination of
posterior white matter and parietotemporal cortex involvement. Hypoglycaemia is said to spare the cerebellum and
brainstem,14 and, indeed, no patient in this series had cerebellar involvement and only one had a brainstem lesion.
In contrast to hyperinsulinism, which was associated with
severe episodes of hypoglycaemia occurring over a narrow age
range and was related to a single type of brain lesion, in FAOD,
acute episodes occurred over a wide age range and resulted in
the whole range of brain lesions, whereas GSDI occurred at
the two extremes of the age range. As many types of biochemical anomalies were included in FAOD without correlation to
age at occurrence of lesions, it is unlikely that any toxic substrate could account for a particular type of brain lesion.
Severity of hypoglycaemia and occurrence of convulsions
Basal ganglia and cortical lesions were associated with lower
values of glucose levels than white matter lesions but this difference was not significant, which reflects conflicting reports
in the literature.15,16 No topography was common to all
patients, which should be the case if the main determinant of
brain lesion topography was glucose level.
It has been reported that the duration of hypoglycaemia in
preterm infants followed prospectively from birth is correlated
with sequelae in combination with many associated factors.16
In our series, the impact of hypoglycaemia duration could
obviously not be determined as action was taken to correct
hypoglycaemia as soon as it was identified. Furthermore, it has
been shown in the rat that varying the duration of fasting does
not alter the severity of regional cell injury.17
Brain Lesion Topography in Hypoglycaemia Svetlana Gataullina et al. 165

In our series, neither the presence nor duration of seizures

was associated with topography of lesions.

The role of age

We found an association between the topography of brain
lesions and age at the occurrence of hypoglycaemia, with
involvement of the posterior white matter in neonates, the
basal ganglia in infants, and the parietotemporal cortex in children. The only patient in whom a combination of basal ganglia and cortical lesions was observed was at the border
between infancy and childhood.
Similar findings have been reported elsewhere: in individuals with carnitine deficiency, hypoglycaemia affects the white
matter in neonates, but the parietal cortex from the age of
16 months.11
This regional pattern follows the maturational characteristics of other age-related conditions determining energy failure:
parieto-occipital white matter injury characteristic of the

term-born infant with hypoxicischaemic encephalopathy,

basal ganglia in Leigh syndrome after 3 months of age, and
the cortex later in childhood in children with mitochondrial
pseudo-stroke.18 Similarly, age-related regional vulnerability
has been shown in animal models, for example rats submitted
to deep hypoglycaemia in late postnatal age showed more pronounced susceptibility to grey matter damage.17

CONCLUSION
Although the sample size is limited and there may be many
confounders, it appears that the age of severe hypoglycaemia
determines the topography of the brain injury more than aetiology, glycaemia, convulsions, or additional factors.
ACKNOWLEDGEMENTS
We are grateful to C Chiron, R Nabbout, and G Sebire, who
reviewed the manuscript, and the French Neuropediatric Society
(SFNP), which supported this study.

REFERENCES
13. Belden JR, Caplan LR, Pessin MS, Kwan E. Mechanisms

1. Udani V, Munot P, Ursekar M, Gupta S. Neonatal hypogly-

7. Alkalay AL, Flores-Sarnat L, Sarnat HB, Moser FG, Sim-

cemic brain injury a common cause of infantile onset remote

mons CF. Brain imaging findings in neonatal hypoglycemia:

and clinical features of posterior border-zone infarcts. Neu-

symptomatic epilepsy. Indian Pediatr 2009; 46: 12732.

case report and review of 23 cases. Clin Pediatr (Phila) 2005;

rology 1999; 53: 13128.

2. Lucas A, Morley R, Cole TJ. Adverse neurodevelopmental

outcome of moderate neonatal hypoglycemia. BMJ 1988;
297: 13048.
3. Barkovich AJ, Ali FA, Rowley HA, Bass N. Imaging patterns
of neonatal hypoglycemia. AJNR Am J Neuroradiol 1998;
19: 5238.
4. Ma JH, Kim YJ, Yoo WJ, et al. MR imaging of hypoglyce-

44: 78390.
8. Kara C, Aydin OF, Aslan B, Gurer YK. Bilateral basal ganglia
lesions after hypoglycemic coma in a 6-year-old child. J Child
Neurol 2007; 22: 22831.
9. Christiaens FJ, Mewasingh LD, Christophe C, Goldman S,
Dan B. Unilateral cortical necrosis following status epilepticus with hypoglycemia. Brain Dev 2003; 25: 10712.

14. Auer RN. Hypoglycemic brain damage. Metab Brain Dis

2004; 19: 16975.
15. Alkalay AL, Flores-Sarnat L, Sarnat HB, Farber SJ, Simmons
CF. Plasma glucose concentrations in profound neonatal
hypoglycemia. Clin Pediatr (Phila) 2006; 45: 5508.
16. Montassir H, Maegaki Y, Ogura K, et al. Associated factors
in neonatal hypoglycaemic brain injury. Brain Dev 2009; 31:

mic encephalopathy: lesion distribution and prognosis pre-

10. Burns CM, Rutherford MA, Boardman JP, Cowan FM. Pat-

diction by diffusion-weighted imaging. Neuroradiology 2009;

terns of cerebral injury and neurodevelopmental outcomes

17. Ennis K, Tran PV, Seaquist ER, Rao R. Postnatal age influ-

51: 6419.

after symptomatic neonatal hypoglycemia. Pediatrics 2008;

ences hypoglycemia-induced neuronal injury in the rat brain.

5. Gataullina S, Dellatolas G, Perdry H, et al. Comorbidity and

122: 6574.

64956.

Brain Res 2008; 1224: 11926.

metabolic context are crucial factors determining neurologi-

11. Thompson JE, Smith M, Castillo M, Barrow M, Mukherji

18. Dinopoulos A, Cecil KM, Schapiro MB, et al. Brain MRI

cal sequelae of hypoglycaemia. Dev Med Child Neurol 2012;

SK. MR in children with L-carnitine deficiency. AJNR Am J

and proton MRS findings in infants and children with respi-

54: 10127.

Neuroradiol 1996; 17: 15858.

ratory chain defects. Neuropediatrics 2005; 36: 290301.

6. Caksen H, Guven AS, Yilmaz C, et al. Clinical outcome and

12. Lo L, Tan AC, Umapathi T, Lim CC. Diffusion-weighted

magnetic resonance imaging findings in infants with hypo-

MR imaging in early diagnosis and prognosis of hypoglyce-

glycemia. J Child Neurol 2011; 26: 2530.

mia. AJNR Am J Neuroradiol 2006; 27: 12224.

166 Developmental Medicine & Child Neurology 2013, 55: 162166