Presented at the Neonatal Society 2008 Autumn Meeting.
Kendall G1, Bainbridge A2, De Vita E2, Hagmann C1, Kapetanakis A1, Cady E2, Robertson NJ1
1 Academic Neonatology, EGA UCL Institute for Women’s Health, University College London, UK
2 Medical Physics & Bio-Engineering, UCL Hospitals NHS Foundation Trust, UK
Background: Experimental studies have suggested a correlation between deep brain and rectal temperatures (TDB and Trec respectively) under both normoxia and hypoxia (1). Indeed, current therapeutic hypothermia treatment protocols for term encephalopathic infants use Trec monitoring as a proxy for brain temperature during cooling. However, our knowledge of regional brain temperature in neonatal encephalopathy is limited despite the fact that small temperature differences can critically influence neuropathological outcome during the post-ischaemic period. We have demonstrated that intrinsic factors such as body size may influence temperature gradients in un-injured brain (2). Knowledge of the temporal and regional variation in brain temperature in infants of different birth weights and injury patterns together with an assessment of brain perfusion may assist in refining and optimising therapeutic hypothemia protocols.
Aims: To assess the relation between TDB (thalamic) and Trec using localised proton magnetic resonance spectroscopy thermometry (MRSt) in normothermic and hypothermic human neonates.
Methods: Twenty one neonates had Trec measurement and thalamic MRSt in addition to a standard clinical imaging protocol. Of these, 12 had suspected hypoxic-ischaemic encephalopathy (HIE) (Table 1). Four neonates with suspected HIE were cooled, 3 were cooled at MRSt. Proton spectra were acquired on a Siemens 1.5 Tesla Avanto scanner using point-resolved spectroscopy (PRESS) without water-suppression with a 1.5×1.5×1.5 cm3 cubic voxel centred on the left thalamus (repetition time 1370 ms, echo time 288 ms, 392 summed echoes, 2048 complex datapoints). Spectra were analysed using AMARES (3) (jMRUI software) (4). The water frequency was determined by fitting a single component and then the water signal was removed using HLSVD (5) (jMRUI software) (4): the N-acetyl-aspartate (Naa) frequency was also determined using a single component. TDB was estimated using the chemical-shift difference between water and Naa (6). TDB was plotted against Trec and linear regression was performed. The correlation between TDB and Trec was tested using the Pearson product moment.
Results: TDB is plotted against Trec in Fig. 1. There was a strong linear correlation between TDB and Trec(slope 0.84; P < 0.0001).
Discussion: These data suggest that we can accurately measure deep brain temperature in neonates using MRSt over a broad range of temperature values. Further work is required to refine the systematic errors in the measurement of MRSt. Future application of MRSt to other brain regions together with magnetic resonance imaging of regional brain perfusion may allow cooling protocols to be tailored to specific infants.
References
1. Laptook AR et al, Pediatrics. 2001; 108:1103-10.
2. Iwata S et al. Ann Neurol, 2006; 60:578–585
3. Vanhamme L et al, Journal of Magnetic Resonance 1997; 129: 35-43.
4. Naressi A et al, MAGMA, 2001;12:141-52
5. Pijnappel WWF et al, Journal of Magnetic Resonance 1992; 97: 122-134.
6. Cady EB et al, Magnetic Resonance in Medicine 1995; 33:862-867