To the Editor.—
We read, with interest, the paper in Pediatrics by Mirmiran and
colleagues (1) on the comparison of neonatal brain magnetic resonance
imaging (MRI) and cranial ultrasound (US) in predicting cerebral palsy
(CP) in very low birth weight (VLBW) infants. The authors are to be
congratulated on publishing outcome data up to 31 months relating to MRI
findings in preterm neonates.
Whilst their reported finding, that MRI at near-term in VLBW preterm
neonates is superior to US in predicting outcome, might be true one, we
question whether the study protocol allowed for a fair comparison between
the two techniques.
The authors only obtained US scans in the first 1-2 weeks after birth
except at the discretion of the attending physician. Such limited US
scanning in a study designed to assess the prognostic information
available from US is hard to understand. It is well known that localised
cystic periventricular leukomalacia (PVL) may appear beyond day 28 in
more than half of cases (2,3) and that infants with late onset cystic PVL
usually have normal initial US findings before they deteriorate later
following sepsis or necrotising enterocolitis (4). In our own study on
1460 infants under 33 weeks gestational age at birth (2) US was performed
on a weekly basis till discharge and again at 40 weeks postmenstrual age
irrespective of imaging data obtained during the first two weeks. We
found scan abnormalities in over 90% of children who developed CP, which
were major in 83%.
Whilst we are informed in detail about the MRI data in relation to
later development of CP, no data are provided about the children with
normal US findings and later development of CP. Looking at the MRI data in
table 3, it is most likely, in view of the US protocol, that the two
children with PVL were not detected with US, but this would still not
reduce the sensitivity to the values shown in figure 2.
The quality of the US images can also not be assessed by the reader
as none were provided. In a study dedicated to assessing US we are
surprised that the journal editors do not insist that the US scans from
the infants in whom lesions were missed are shown. One is left uncertain
as to whether the lesions were not detectable, or whether the scans were
either not of sufficient quality or they were not obtained at the right
time to detect the lesions. All the abnormalities pointed out in the MRI
scans shown should be seen on high quality US.
Some of the limitations of the study are given in the discussion.
Obtaining US and MRI on the same day should have been a mandatory part of
the study protocol. In our study 4 children first showed cystic PVL
lesions when examined again at 40 weeks postmenstrual age (2). It is true,
that others who performed both techniques on the same day did show that US
was not good in recognising subtle white matter lesions, but as no follow-
up data are available as yet, one does not know, whether these subtle
white matter lesions did indeed lead to subsequent development of CP.
Preliminary data (5) indicate that subtle abnormalities of white matter
may relate to poorer cognitive outcome but not CP
We are surprised by the authors’ argument that it is not useful to assess
myelination. In our experience this is one of the most useful features of
a scan at term age but necessitates that the MRI is not obtained before at
least 38 weeks. The presence/absence or asymmetry of myelination at the
level of the posterior limb of the internal capsule (6), is especially
helpful in the early prediction of hemiplegia following a unilateral
parenchymal hemorrhage and likely relevant to the case depicted in Fig 1 A
and B. The authors make no mention of assessing lesion site – cysts in the
anterior white matter will not lead to CP – as shown in Fig 1 G and H and
it is not helpful to group all abnormality together regardless of site to
predict a specific outcome.
It may well be true that more subtle abnormalities such as diffuse
excessive high signal intensity (DEHSI) in white matter together with the
use of diffusion tensor imaging and volumetric studies may be more
informative about cognitive outcomes but as yet we remain to be convinced
that MRI is superior to high quality US in predicting CP in the majority
of preterm infants. Before a recommendation can be given about performing
MRI in all VLBW infants near term age, a fair comparison between MRI and
US should be performed, using both methods optimally.
LINDA DE VRIES MD
Wilhelmina Children’s Hospital
University Medical Centre
PO box 85090
3508 AB Utrecht, Netherlands
FRANCES COWAN MRCPCH PhD
Dept of Paediatrics
Imperial College
Hammersmith Hospital
London W12 ONN, UK
REFERENCES
1. Mirmiran M, Barnes PD, Keller K, Constantinou JC, Fleisher BE,
Hintz SR, Ariagno RL Neonatal brain magnetic resonance imaging before
discharge is better than serial cranial ultrasound in predicting cerebral
palsy in very low birth weight preterm infants. Pediatrics. 2004;114:992-
8.
2 De Vries LS, van Haastert IC, Rademaker KJ, Koopman C, Groenendaal
F. Ultrasound abnormalities preceding cerebral palsy in high-risk preterm
infants. J Pediatr 2004; 144:815-20
3. Pierrat V, Duqennoy C, van Haastert IC, Ernst M, Guilley N, de
Vries LS. Ultrasound diagnosis and neurodevelopmental outcome of
localised and extensive cystic periventricular leukomalacia. Arch Dis
Child Neon Fetal Ed 2001; 84:F151-6.
4.Andre P, Thebaud B, Delavaucoupet J, Zupan V, Blanc N, d'Allest AM
et al. Late-onset cystic periventricular leukomalacia in premature
infants: a threat until term. Am J Perinatol 2001; 18:79-86.
5. Dyet L, Kennea NL, Counsell S, Duggan PJ, Allsop J, Maalouf E,
Cowan F, Rutherford M, Edwards AD. Diffuse white matter abnormalities on
magnetic resonance imaging of the brain in preterm infants and
neurodevelopmental outcome. Pediatr Research 2004;55(4):424A
6. De Vries LS, Groenendaal F, Eken P, Rademaker KJ, Meiners LC.
Asymmetrical myelination of the posterior limb of the internal capsule: an
early predictor of hemiplegia. Neuropediatrics 1999; 30:314-9
eLetters: ![[Read eLetters]](/icons/misc/sectionDOWN.gif){kind=icon}
