The prognostic value of multivoxel magnetic resonance spectroscopy determined metabolite levels in white and grey matter brain tissue for adverse outcome in term newborns following perinatal asphyxia

This retrospective study was performed at the neonatal intensive care unit of the University Medical Center Groningen. From our medical records, we identified all term infants who were admitted because of clinical signs of hypoxic ischemic encephalopathy (HIE) and in whom MRI including MRS was performed, between January 2001 and July 2007. Perinatal asphyxia was defined by at least two of the following criteria: 1) signs of fetal distress, consisting of decelerations on fetal monitoring by cardiotocograph or meconium-stained amniotic fluid; 2) umbilical cord arterial pH less than 7.1; 3) Apgar scores less than 5 at 5 min after birth; and within days 4) multi organ failure syndrome; 5) neonatal HIE as described by Sarnat [19]. Exclusion criteria were: a gestational age below 36 weeks, presence of severe congenital malformations or diseases with neurological complications, and lack of clinical follow up examination.

Cerebral damage was evaluated by a combination of neurological examination, and amplitude integrated EEG. If outcome could not be predicted on these diagnostic tests, an additional MRI was performed to identify the extent of the cerebral damage following HIE. In the worst cases withdrawal of care was not based on the results of MRI and MRS, indicating that the selection of survivors/non-survivors was not biased by this. None of the infants were treated with therapeutic hypothermia, because this had not been introduced in our unit at the time of the study. The study was approved by the review board of the University Medical Center Groningen.

An 8-channel transmit/receive head coil of a 1.5T Magnetom Sonata system (Siemens AG, Erlangen, Germany) was used for proton (1H) MRS. 2D-chemical shift imaging (CSI) point resolved spectroscopy measurements were performed with a repetition time (TR) of 1500 ms and an TE of 135 ms resulting in T1- and T2-weighting of spectra [20]. A transverse T2-weighted fast spin-echo series was used as guidance for examining an approximately 5 × 5 × 2 cm³ supraventricular volume of interest, located cranial to the lateral ventricles, in a 16 × 16 phase encoded field of view of 16 × 16 cm² resulting in MRS voxels of 1 × 1 × 2 cm³ (7 min acquisition time) (Fig. 1). Since 2001, the standard brain MRI measurement protocol for infants following developmental delay and perinatal complications such as asphyxia has included MRS measurement according to the above specifications [21]. To obtain good quality spectra not affected by partial volume effects or by artefacts caused by movements near the eyes and ears we chose this particular volume of interest superior to the corpus callosum to measure a large transverse brain area containing both white and grey matter and little cerebrospinal fluid [22].

The standardised postprocessing protocol consisted of water reference processing, Hanning filtering (center 0 ms, width 512 ms), zero filling from 512 data points to 1024 data points, Fourier transformation, frequency shift correction, sixth order polynomial baseline correction, phase correction and frequency domain curve fitting. The curve fitting was set to fit peaks to Gaussian lineshapes, including the chemical shift ranges of 3.1–3.3 ppm for Cho, 2.9–3.1 for Cr, 1.9–2.1 for NAA, and 1.2–1.4 for Lact. The raw data were thus processed automatically, allowing for operator-independent quantification. The CSI voxels on the edge of the volume of interest (subject to signal drop-off) were deducted from the total data matrix. The inner 24 voxels were analysed and separated into the two central columns mainly containing grey matter (GM) (12 voxels) and the remainder of 12 voxels filled with white matter (WM) (Fig. 1), as described elsewhere [23]. The quantifications, also those of the small lactate peak in the summarised blocks of 12 or 24 spectra, were reliable and, therefore, no manual adjustments were performed. To limit patient examination times, absolute quantification requiring additional CSI measurements without water suppression was not performed. To facilitate tissue signal comparison, we expressed the GM and WM metabolites in percent (%) of the summed peak areas of Cho, Cr and NAA in the 24 analysed voxels, a method used before [22, 24].

Neurodevelopmental follow-up was performed at a minimum age of 18 months. Long term outcome was divided into four groups: 1) normal neurological outcome without complications; 2) minor neurological dysfunction; 3) severe neurological dysfunction; and 4) deceased. Severe neurological dysfunction was defined by a gross motor function classification system (GMFCS) [25] of three or more (not being able to walk independently). These children were treated in specialised rehabilitation centres, which performed systematic Bayleys developmental tests [26]. Children who were not treated systematically in specialised rehabilitation centres, were invited for follow-up examination. The neurological examination for toddler age by Hempel [27] was used till the age of 5 years. In two children who were older than 5 years we used the neurological examination by Touwen [28]. The examination was filmed and classified by a specialised paediatrician (C. v.d. V.) blinded for the clinical condition after birth and for the MRI and MRS results.

Statistical analysis was performed using SPSS version 17.0. A Kruskal Wallis test for non-parametric parameters was used to investigate significance between the four groups. Group-to-group differences were subsequently assessed by Mann–Whitney U-test. Differences between survivors (n = 19) and non-survivors (n = 5) were assessed by Mann–Whitney U-test. A P value of less than 0.05 was considered significant.