Proton MRS data were acquired from an 8 mL cubic volume centred on the ... carried out during recovery at ⥠7 days after the presumed ..... Floppy/rag doll. 2.
Early brain proton magnetic resonance spectroscopy and neonatal neurology related to neurodevelopmental outcome at 1 year in term infants after presumed hypoxic–ischaemic brain injury P N Amess*; Juliet Penrice, Department of Paediatrics, University College London Medical School; Marzena Wylezinska, Department of Medical Physics and Bioengineering, University College London Hospitals NHS Trust; Ann Lorek; Janice Townsend; J S Wyatt, Department of Paediatrics, University College London Medical School, London, UK. Claudine Amiel-Tison, Department of Paediatrics, PortRoyal-Baudeloque, Paris, France. E B Cady, Department of Medical Physics and Bioengineering, University College London Hospitals NHS Trust; Ann Stewart, Department of Paediatrics, University College London Medical School, London, UK. *Correspondence to first author at Department of Paediatrics, University College London Medical School, Rayne Institute, University Street, London WC1E 6JJ, UK.
This study investigated the accuracy of prediction of neurodevelopmental outcome at 1 year using cerebral proton magnetic resonance spectroscopy (MRS) and structured neonatal neurological assessment in term infants after presumed hypoxic–ischaemic brain injury. Eighteen control infants and 28 infants with presumed hypoxic–ischaemic brain injury underwent proton MRS investigation. Studies were carried out as soon as possible after the cerebral insult, most within 48 hours. Infants had an early structured neurological assessment at a median of 19 hours (range 0
436
Developmental Medicine & Child Neurology 1999, 41: 436–445
hours to 9 days) from the presumed hypoxic–ischaemic insult and a late assessment at a median of 7 days (range 3 to 25 days) during recovery. The maximum cerebral peak–area ratio lactate:N-acetylaspartate measured by proton MRS accurately predicted adverse outcome at 1 year with a specificity of 93% and positive predictive value of 92%. Neurological assessment had a tendency for false-positive predictions. However, both early and late neurological examination can be used as a reliable indicator for a favourable outcome at 1 year having negative predictive values of 100% and 91% respectively.
After perinatal hypoxic–ischaemic brain injury simple indicators of deranged intrapartum gas exchange, including arterial pH, base excess, and blood lactate, have proven unreliable indicators of late neurodevelopmental outcome (Paneth and Stark 1983, Ruth and Raivio 1988, Yudkin et al. 1994, Martin et al. 1996). With the advent of trials of neural rescue therapy (Edwards et al. 1998) there is an increasing need for early measures of outcome to aid appropriate selection of infants for treatment and to assess the efficacy of intervention. Several objective investigations have been developed to define the extent of cerebral damage and predict outcome after perinatal hypoxic–ischaemic brain injury (Wyatt 1996, Patel and Edwards 1997). One such method is proton magnetic resonance spectroscopy (MRS). Proton MRS enables non-invasive observation of intracellular cerebral metabolites, particularly lactate (Lac), total creatine (including phosphocreatine) (Cr), choline-containing compounds (Cho), and N-acetylaspartate (NAA) in newborn infants. During the development of secondary energy failure, which is characterized by the gradual depletion of high-energy phosphate metabolites (Azzopardi et al. 1989), elevated cerebral lactate signals in newborn infants have been observed using proton MRS within 18 hours of an hypoxic–ischaemic brain injury (Hanrahan et al. 1996). A raised peak–area ratio Lac:NAA has proven to be a particularly useful marker of early brain injury because cerebral lactate tends to rise while NAA falls during secondary energy failure (Penrice et al. 1996). After perinatal hypoxia–ischaemia, abnormal results from cerebral proton MRS have been associated with unfavourable neurodevelopmental outcome (Groenendaal et al. 1994, Penrice et al. 1996). Unfortunately there are obvious limitations for the general use of MRS. It is highly expensive and reliant on specialized medical physics support and is, therefore, not commonly available. In addition, investigation with MRS is a lengthy process, requiring detailed and complex monitoring procedures to allow the safe study of sick newborn infants within the environment of a super-conducting magnet. In contrast, early neurological assessment can nearly always be performed by a paediatrician in the safe environment of the neonatal intensive-care unit. Neurological examination of the most sick newborn infants may be complicated by the use of intensive-care techniques, especially mechanical ventilation. There may also be difficulty in separating the transient effects of cardiorespiratory and metabolic problems from the specific expression of brain damage (Amiel-Tison 1995). To this end a specially designed structured neurological assessment (derived from
previous studies by Saint-Anne Dargassies 1977) has been developed, gaining information from observation and simple reproducible manipulations (Amiel-Tison 1988). The importance of repeated assessment has been stressed as a means of resolving the difficulty posed by the lability of clinical signs (Amiel-Tison 1995). The aim of this study was to compare the accuracy of cerebral proton MRS and structured neonatal neurological assessment in the prediction of adverse 1-year neurodevelopmental outcome in term infants after presumed perinatal hypoxic–ischaemic brain injury. Method SUBJECTS
A total of 46 infants was studied: 18 controls and 28 with presumed hypoxic–ischaemic brain injury. The control infants were all healthy and recruited from postnatal wards. None had a clinical history compatible with perinatal hypoxic–ischaemic brain injury, neonatal neurological examinations were all normal, and their weights and head circumferences at birth were between the 3rd and 97th centiles. The recruitment of infants with presumed perinatal hypoxic–ischaemic brain injury was based on the presence of an obstetric or neonatal history consistent with an acute hypoxic–ischaemic event and one or both of the following: abnormal neurological signs including abnormal level of consciousness or seizures, and/or a base excess of more than –15 mmol/L in cord blood or in the first arterial blood sample obtained after birth or a postnatal asphyxial episode. Three infants were studied after presumed postnatal hypoxic–ischaemic brain injury, the result of sudden unexpected respiratory arrest without a preceding history of perinatal asphyxia.
NEONATAL NEUROLOGICAL ASSESSMENTS
Infants underwent both early and late neurological examinations. As for MRS investigation neonatal neurological assessment was considered ‘early’ if it was carried out within 48 hours of the presumed hypoxic–ischaemic insult. When more than one early neurological examination was performed, the assessment nearest to the earliest MRS investigation was used. Late neonatal neurological examination was carried out during recovery at ≥ 7 days after the presumed hypoxic–ischaemic insult or, if neurological recovery was rapid, before discharge from hospital. The neurological assessments were performed using a modification of a scheme described by Amiel-Tison (1988). This scheme is based on a series of structured observations and tests, including assessment of alertness, visual pursuit, motor control, and brain-stem activity. The design allows examination of a sick infant within a short time with minimal disturbance. Each part of the assessment was scored 0, 1, or 2 and the total indicates whether the infant was neurologically normal or abnormal. Abnormal results were further subdivided into mild (tone changes and excitability but no central nervous system depression), moderate (central nervous system depression), or severe (status epilepticus and coma). Details of the early and late neonatal neurological assessments and scoring are shown in Appendix 1a and 1b. When comparing the association of neonatal neurology and proton MRS with neurodevelopmental outcome at 1 year, early neonatal neurology was compared with early peak–area ratio measurements and late neonatal neurology with maximum peak–area ratio measurements (or minimum peak–area ratio measurements in the case of NAA:Cho).
PROTON MRS
A 2.4 tesla Bruker Biospec (Bruker Medizin Technik GmBH, Karlsruhe, Germany) spectrometer was used with a seriestuned, inductively coupled, 15 cm diameter Helmholtz head coil in which the infant’s head was placed (Cady 1995). Proton MRS data were acquired from an 8 mL cubic volume centred on the thalami using a sagittal scout image (Fig. 1). This brain region was chosen for study as it consists mainly of grey matter and is particularly prone to hypoxic–ischaemic damage in term infants. Spectra were acquired using pointresolved spectroscopy (PRESS) with an echo time of 270 ms, a recovery time of 1730 ms, 2048 simultaneously sampled quadrature data points, a spectral width of 1250 Hz, and 128 echoes summed as described previously (Cady 1995). Peak areas on the spectra were fitted using Lorentzian curve fitting via χ2 minimization (Cady 1991). Care of the infants while undergoing MRS investigation has been described previously (Penrice et al. 1996). The control infants were not sedated and no infants were given medication in addition to that prescribed as part of their clinical care. The aim was to carry out MRS investigation as soon as possible after the presumed perinatal hypoxic–ischaemic insult. ‘Early’ peak–area measurements have been defined, for the purposes of this study, as those obtained at
