Experimental Neurology 261 (2014) 281–290
Contents lists available at ScienceDirect
Experimental Neurology journal homepage: www.elsevier.com/locate/yexnr
Regular Article
Detecting brain injury in neonatal hypoxic ischemic encephalopathy: Closing the gap between experimental and clinical research James D.S. Aridas a, Tamara Yawno a, Amy E. Sutherland a, Ilias Nitsos a,b, Michael Ditchfield c, Flora Y. Wong a,c, Michael C. Fahey a,c, Atul Malhotra a,c, Euan M. Wallace a,b, Graham Jenkin a,b, Suzanne L. Miller a,b,⁎ a b c
The Ritchie Centre, MIMR-PHI Institute, Clayton, Victoria, Australia Department of Obstetrics and Gynaecology, Monash University, Clayton, Victoria, Australia Monash Children's, Monash Health, Clayton, Victoria, Australia
a r t i c l e
i n f o
Article history: Received 22 April 2014 Revised 3 July 2014 Accepted 20 July 2014 Available online 28 July 2014 Keywords: Animal models Sheep Biomarker Newborn infant Hypoxia–ischemia Asphyxia Hypoxic ischemic encephalopathy Neuroprotection Magnetic resonance imaging Oxidative stress Cell death
a b s t r a c t Moderate to severe neonatal hypoxic ischemic encephalopathy remains an important cause of infant death and childhood disability. Early and accurate diagnosis of encephalopathy is difficult but critical for timely intervention. Thus, we have utilized a clinically relevant large animal model of asphyxia in-utero, followed by immediate lamb delivery, resuscitation and clinical care over the next 72 h for assessment of potential biomarkers of brain injury. In-utero asphyxia was induced in twelve near-term lambs and outcomes compared with seven controls. Asphyxia resulted in bradycardia (97 ± 12 beats/min), hypotension (12.1 ± 1 mm Hg) and metabolic acidosis (pH 6.9 ± 0.02; base-excess −13.8 ± 0.8 mmol/l). 72 h following asphyxia, cerebrospinal concentrations of malondialdehyde and S100B were elevated 2-fold and 5-fold, respectively, in asphyxic lambs compared to control lambs. Magnetic resonance spectroscopy (MRS) at 72 h showed a significant decrease in n-acetyl aspartate: choline ratio in asphyxia lambs compared to that observed at 12 h (0.56 ± 0.23 vs. 0.82 ± 0.15, respectively); lactate:choline ratio was not changed over this time. Marked neuropathology was observed in asphyxia lambs with neuronal degeneration in the hippocampus, thalamus, striatum and cortex. Astrogliosis was observed in the hippocampus and thalamus. Early blood markers of metabolic state showed limited predictive value of histological damage at 72 h. MRS outcomes at 72 h showed a modest but significant correlation with histological evidence of neuronal brain injury (lactate:N-acetyl aspartate ratio in the thalamus r2 = 0.2, p b 0.01). MRS at 72 h was best able to detect established brain injury, but a combination of biomarkers over multiple phases of injury may be able to assess the evolution of neonatal brain injury. © 2014 Elsevier Inc. All rights reserved.
Introduction Neonatal hypoxic ischemic encephalopathy (HIE) remains an important cause of perinatal death and long-term developmental disability. Clinical trial meta-analyses demonstrate that approximately 60% of untreated (normothermic) infants will die or have long-term disability after HIE (Edwards et al., 2010). For surviving infants, there is a welldescribed association between HIE and diagnosis of cerebral palsy (Badawi et al., 1998). Currently, the only effective treatment to reduce adverse outcome following term HIE is hypothermia commencing within 6 h of delivery (Jacobs et al., 2013), however 47% of treated newborns Abbreviations: aEEG, amplitude integrated electro-encephalogram; Cho, choline; DWI, diffusion-weighted imaging; GFAP, glial fibrillary acidic protein; HIE, hypoxic ischemic encephalopathy; HR, heart rate; Lac, lactate; MAP, mean arterial pressure; MDA, malondialdehyde; NAA, N-acetylaspartate; MRI, magnetic resonance imaging; MRS, magnetic resonance spectroscopy; SaO2, oxygen saturation; SEM, standard error of the mean. ⁎ Corresponding author at: The Ritchie Centre, MIMR-PHI Institute, 27-31 Wright Street, Clayton, VIC 3168, Australia. E-mail address: [email protected] (S.L. Miller).
http://dx.doi.org/10.1016/j.expneurol.2014.07.009 0014-4886/© 2014 Elsevier Inc. All rights reserved.
are still at risk of death or serious disability (Jacobs et al., 2013). It is thus recognized that adjuvant or alternative treatments for HIE are necessary (Bennet et al., 2012; Miller et al., 2012; Robertson et al., 2012). In response to asphyxia at term birth, brain injury and encephalopathic symptoms evolve from hours through to weeks (Gunn and Gluckman, 2007; Williams et al., 1991). This is characterized by a primary insult phase where neuronal degeneration begins. Delivery of the baby and, in association with resuscitation, leads to apparent recovery during the latent phase. However the secondary phase introduces biochemical cascades such as excitotoxicity, oxidative stress and inflammation, and subsequent neuropathology to neuronal populations in the cerebral cortex, hippocampus, thalamus, basal ganglia and to a lesser extent in white matter (Azzopardi and Edwards, 2010; Folkerth, 2005; Gunn and Bennet, 2009; Inder, 2000; Thayyil et al., 2010). Early identification of HIE is essential for successful initiation of neuroprotective therapies. The results of hypothermia studies highlight that treatment should begin early for maximal benefit (Gunn et al., 1999). Clinically, there is an initial focus on observational parameters of encephalopathy, first described by Sarnat and Sarnat, to reflect graded
282
J.D.S. Aridas et al. / Experimental Neurology 261 (2014) 281–290
ventilation and assisted ventilation, subsequently, ceased if the lambs began to spontaneously breath N 50% of the time. Overt clinical seizures were assessed by clinical personnel (AM, MCF, FYW, EMW) as repetitive eye movements, “smacking” of the lips, neck arching, “running” leg movements, and apneic episodes, similar to human seizures as described by Volpe (Volpe, 2001). Seizures were treated with 20 mg/kg intravenous phenobarbitone (Sigma, Australia).
abnormalities in infant reflexes, behavior and tone, and incidence of seizures (Sarnat and Sarnat, 1976). More recently with the NICHD hypothermia trial for HIE, early (b 6 h) neurological classification utilizing clinical and laboratory parameters showed good predictive value (78%) for death/disability, and therefore infants that would benefit from hypothermia (Ambalavanan et al., 2006). Magnetic resonance imaging (MRI) studies of the newborn brain confirm neuropathology after perinatal asphyxia and are predictive for long-term impairment (Azzopardi and Edwards, 2010; Thayyil et al., 2010). However MRI is usually performed 2–6 days or later after the sentinel insult; with later scans showing greater predictive value (Azzopardi and Edwards, 2010). Ideally, neonatal care would incorporate specific biomarkers of HIE severity and neuropathology to guide therapeutic intervention and prognosis. The aim of this study was to cause HIE in a clinically useful large animal model in which we could correlate physiological, biochemical, radiological and histological markers of neuronal cell degeneration. This model provides endpoints selected to assess the evolution of brain injury during clinical care, to inform timing for therapeutic interventions, and to guide future neuroprotection studies.
Arterial blood samples (250 μl) were collected in-utero, during asphyxia, and at 0.5, 1, 2, 4, 8, 12, 24, 48 and 72 h postnatal for assessment of pH, PaO2, PaCO2, lactate, glucose, bicarbonate and base-excess (ABL Blood Gas Analyzer, Radiometer, Denmark). HR, SaO2, rectal temperature, and body weight were assessed hourly. When the lambs were extubated, alert, and had a strong suckle reflex, they were offered sheep milk replacer orally every 4 h (~100–150 ml; Wombaroo Food Products, Australia). Maintenance intravenous fluid (10% dextrose; 40 ml/kg/day) was commenced if lambs were unable to feed orally. Lambs were maintained for 72 h.
Materials & methods
S100B and malondialdehyde
Experiments complied with the National Health and Medical Research Council of Australia guidelines for the care and use of animals for scientific purposes and were approved by Monash Medical Centre Animal Ethics Committee.
S100B ELISA was performed as per manufacturer's instructions (DiaSorin, Minnesota, USA) on serum (50 μl) at time 0 (in-utero), and 1, 2, 4, 8, 12, 24 and 72 h postnatal, and on CSF obtained at postmortem. Assay sensitivity was 0.03 μg/l with a run imprecision of b10% and total imprecision of b15%. Malondialdehyde (MDA) was measured in plasma (100 μl) at the same time-points as for S100B, using a thiobarbituric acid reacting substance assay (Miller et al., 2014). The assay sensitivity was 0.1 μmol/l, with 5.1% inter-assay and 3.1% intra-assay coefficients of variation.
Surgery and hemodynamic recording Near-term pregnant ewes at 139–141 days gestation (term is 145– 147 days) underwent sterile surgery under general anesthesia induced by sodium-thiopentone (20 mg/kg IV bolus; Pentothal, Boerhringer Ingelheim, Australia) and maintained with 1–2.5% isoflurane (Isoflow, Abott Pth) in oxygen/nitrous oxide (O2: 2–3 L;N2O: b 1 L). The hindquarters of the fetus were exteriorized and a femoral artery cannula inserted for continuous digital recording of mean arterial pressure (MAP) and heart rate (HR; Powerlab SP, ADInstruments), and blood sampling. A femoral vein catheter was inserted. Pulse oximetry (MasimoSet Rainbow, Radical7, Masimo) was continuously monitored via a cuff placed on the shaved tail of the lamb. Asphyxia and resuscitation Fetal sheep were randomly allocated using an envelope assignment system to control or asphyxia groups. Asphyxia was induced via complete umbilical cord occlusion. With the head and upper body remaining in-utero in the amniotic fluid, the umbilical cord was exposed and clamped. Lambs in the control group had the cord clamped and cut immediately and were delivered and resuscitated. Lambs in the asphyxia group remained in-utero until MAP decreased to 18–20 mm Hg; shown previously in late gestation fetal sheep to induce severe asphyxia and neuropathology (Castillo-Meléndez et al., 2004). The cord was then cut and lambs delivered and resuscitated. Australian and New Zealand Neonatal Resuscitation Council guidelines for resuscitation were followed. At delivery, all lambs were placed on an infant warmer, intubated (4.5 endotracheal tube; Portex), and dried with towels. Positive pressure ventilation (Neopuff™, Fisher & Paykel Healthcare; 30 cmH2O positive inspiratory pressure [PIP], 5– 8 cmH2O positive end expiratory pressure [PEEP], 10 L/min room air, 30 breaths/min) was initiated. If required, lambs were administered adrenaline (1 ml) and fluid (normal saline; 20 ml/kg). As per guidelines, oxygen saturation (SaO2) was targeted at 85–90% within 10 min. Lambs were ventilated (Babylog 8000+, Draeger; volume guarantee 5 ml/kg, PEEP 5–7 cmH2O, 30 breaths/min) as required. Ventilation parameters were decreased and changed to continuous positive airway pressure
Maintenance
Magnetic resonance imaging and spectroscopy Brain magnetic resonance imaging and spectroscopy (MRS) were performed under sedation at 12 and 72 h after birth, using a 3 T Siemens Vario (Siemens Medical Solutions, USA) in an eight-channel knee coil (400 × 420 × 310 mm). Sedation was induced using Domitor (0.1 mg/kg iv; Pfizer Australia) a synthetic alpha-2-adrenoreceptor agonist; and reversed with the antagonist, antisedan (0.1 mg/kg; Pfizer Australia). Conventional sequences were obtained with sagittal diffusion-weighted-imaging (DWI) utilizing a slice thickness of 2 mm, 20 ms echo time, and 28 ms repetition time, T1-weighted (repetition time/echo time/excitations 1900/3.92/1), and T2-weighted (5000/90/1). MRS utilized a 270 ms echo with a 2 cm3 voxel placed over the hippocampus, striatum and basal ganglia. MRS produced spectrographs of concentrations (calculated by computer algorithm as area under the curve) of lactate (Lac), choline (Cho), and n-acetylaspartate (NAA). Brain pathology Lambs were euthanased (pentobarbitone, Lethabarb Virbac Australia) immediately after the 72 h MRI. The right cerebral hemisphere was cut coronally into 5 mm slices and immersion fixed in 4% paraformaldehyde for 48 h, paraffin embedded and then coronally sliced at 10 μm. The left cerebral hemisphere was separated into anatomical locations, snap frozen in liquid nitrogen and stored at −80 °C for future assessment. Neuronal degeneration described necrotic neurons, morphologically identified with cresyl-violet/acid-fuschin staining as swelling of organelles, loss of cell and nuclear membrane integrity, pyknotic nuclei, bright eosinophilic cytoplasm, or with darkened and condensed cytoplasm (Castillo-Meléndez et al., 2013; Northington et al., 2011). Glial fibrillary acidic protein (GFAP; 1:400, Sigma) immunopositive staining was used to assess the number of astrocytes and identification of activated
J.D.S. Aridas et al. / Experimental Neurology 261 (2014) 281–290
astrocytes showing hypertrophic cell bodies and attenuated and thickened processes (Yawno et al., 2012). Six fields of view (400× magnification) over two duplicate slides per brain region were examined and averaged across all animals. This included hippocampus (CA1, CA3, dentate gyrus), midtemporal cortex (molecular, pyramidal, and pleomorphic layers), striatum (caudate nucleus, and putamen) and thalamus (paraventricular and anteroventral regions), Fig. 1. Personnel were blinded to experimental group during image capture and analysis.
283
Data are presented as mean ± standard error of the mean (SEM). Blood gas parameters were analyzed using 2-way (time and group) repeated measures ANOVA with post-hoc analysis. One-way ANOVA with post-hoc analysis was used for cell counts, biochemical assay results, and MRS. Correlation analysis utilized the Pearson correlation coefficient. Significance was considered as p ≤ 0.05.
All lambs were intubated and resuscitated according to their individual needs. Four of the 12 asphyxia lambs required both adrenaline and normal saline bolus for cardiovascular stabilization, whereas control lambs did not. Lambs received respiratory support until spontaneous regular breathing was established; the duration of ventilation was longer in asphyxia lambs compared to controls (151 ± 17 min vs. 38 ± 6 min; P b 0.001). Prior to asphyxia, fetal arterial blood measures were not different between groups (Fig. 3), and were within normal values for late gestation fetal sheep (Castillo-Meléndez et al., 2013; Sobotka et al., 2011). During asphyxia, base-excess, lactate, pCO2, pH, pO2, and SaO2 were significantly altered from control parameters (Figs. 3 A, E, F, G, H). Appropriate resuscitation following delivery returned parameters to control values within 2 h, except for lactate, which returned within 4 h. Baseexcess, bicarbonate, and lactate demonstrated significant differences in asphyxia lambs at 8, 12, and 24 h after delivery. Twenty-five percent of asphyxia lambs exhibited clinical seizures, first apparent at 8 ± 4 h after birth, and were treated with phenobarbitone. No control animals exhibited seizures.
Results
S100B and malondialdehyde
Asphyxia and resuscitation
S100B concentration in CSF collected at 72 h was elevated 5-fold in asphyxia lambs compared to control lambs (P = 0.01). CSF MDA concentration was also elevated 2-fold in asphyxia lambs compared to control lambs (P = 0.007), Fig. 4. Concentrations of serum S100B and plasma MDA were not changed over time in control or asphyxia lambs (data not shown; S100B PGROUP = 0.35; MDA: PGROUP = 0.80).
Statistics
In total, 7 control and 18 asphyxia lambs were delivered. Six (33%) of the asphyxia lambs could not be resuscitated and therefore results are presented for 19 surviving animals; 12 asphyxia (4 male, 8 female) and 7 control (5 male, 2 female). MAP (Fig. 2 A) and HR (Fig. 2 B) prior to insult were not different between groups (control values: 55 ± 3 mm Hg, 199 ± 11 beats/min, respectively). In-utero asphyxia caused transient hypertension, followed by hypotension. The asphyxic insult was terminated at 9.9 ± 0.4 min, when MAP reached ~18 mm Hg. Following delivery, MAP and HR in asphyxia lambs decreased further to a nadir of 12 ± 1 mm Hg and 97 ± 12 beats/min respectively, prior to adequate resuscitation.
MRI and MRS Qualitative assessment of the DWI, T1-, and T2-weighted imaging at 72 h demonstrated abnormal scans in 2 of 12 asphyxia lambs. One lamb had bilateral symmetrical ischemic changes affecting the basal ganglia and thalami, and the other had global hypoxic ischemic changes
Fig. 1. Representative lamb brain photomicrograph of cresyl-violet/acid-fuschin anatomical regions assessed for histology. (A) Striatum: CN — Caudate nucleus, Put — Putamen; Cortex: Assessed in three aggregated layers — pleomorphic, pyramidal, and molecular. (B) Hippocampus: CA1 — cornu ammonis 1, CA3 — cornu ammonis 3, DG — dentate gyrus; Thalamus: AV — anteroventral thalamic nuclei; PV — paraventricular thalamic nuclei.
284
J.D.S. Aridas et al. / Experimental Neurology 261 (2014) 281–290
Fig. 2. (A) Mean arterial pressure (MAP; mm Hg) and (B) hear rate (HR; BPM) recordings of 15 min of stable in-utero recordings with no difference between control (open circle) and asphyxia (closed square), during asphyxia (10 min) showing an initial hypertension followed by hypotension and bradycardia, and 30 min ex-utero during resuscitation show a return of MAP and HR within minutes of delivery and resuscitation and stable period with no difference between groups. Data are mean ± SEM *P b 0.05.
effecting cortices and deep gray matter. Representative normal MRI scans from control animals and abnormal diffusion-weighted and T2 images from 72 h asphyxia animals are presented in Fig. 5. MRS at 72 h after birth showed a significant reduction in NAA:Cho ratio in deep gray matter of asphyxia lambs compared to control lambs (Fig. 6 F; P b 0.01). Lac:NAA ratio was not changed at 12 h and 72 h in asphyxia lambs compared to control lambs (P = 0.17 and P = 0.46, respectively). The Lac:Cho ratio was not different between asphyxia lambs and control lambs at either 12 or 72 h, Figs. 6 E and F. Neuropathology Asphyxia caused significant neuronal degeneration in the hippocampus, thalamus, striatum, and pyramidal and pleomorphic zone of the cortex compared to controls (Figs. 7 C and A respectively). In response to asphyxia, astrogliotic cells were identified as morphologically
distinct from normal resting astrocytes. Astrogliosis was observed within the CA1 of the hippocampus and thalamic paraventricular nuclei (Fig. 7 D). Interestingly, these two regions with notable astrogliosis were also the regions with the greatest fold-change in neuronal degeneration in response to asphyxia. Pearson correlation analysis Correlation analysis was undertaken to determine whether associations existed between the degree of neuronal degeneration within the hippocampus, thalamus, striatum and cortex, and brain and peripheral biomarkers. ‘Early’ (before 12 h) and ‘late’ (at 72 h; Fig. 8) correlations were undertaken. Resuscitation requirements, cardiovascular support and length of ventilation did not correlate with histological damage. Early blood parameters were taken as the peak change from control from delivery to 4 h postnatal. Base-excess, bicarbonate, SaO2, lactate,
J.D.S. Aridas et al. / Experimental Neurology 261 (2014) 281–290
285
Fig. 3. Arterial blood parameters measured in-utero, during asphyxia, and ex-utero. Base-excess (A), bicarbonate (B), glucose (C), hemoglobin (D), lactate (E), partial pressure of carbon dioxide (pCO2; F), pH (G), partial pressure of oxygen (pO2; H), oxygen saturation (SaO2; I) in control (open circles) and asphyxia (closed squares). The groups showed no difference inutero prior to asphyxia (Pre). Asphyxia caused changes in base-excess, bicarbonate, lactate, pCO2, pH, pO2, and SaO2. Data are mean ± SEM *P b 0.05.
PaCO2 and pH were all shown to be strongly predictive for neuronal degeneration in all neuronal populations analyzed, Fig. 8. At the final blood sample collection at 72 h, only blood lactate maintained a correlation with hippocampal neuronal degeneration (r2 = 0.60, P b 0.01). Peripheral S100B concentration at 2 h was correlated with hippocampal (r2 = −0.60, P = 0.02) and cortical (r2 = −0.68, P = 0.01) neuronal degeneration. Blood MDA concentrations at 8 and 72 h were significantly associated with hippocampal neuronal degeneration (8 h: r2 = 0.54, P = 0.02; 72 h: r2 = − 0.46, P = 0.04). MDA within CSF at 72 h showed a significant correlation with neuronal degeneration in the striatum (r2 = 0.54, P = 0.04). S100B concentration within CSF was not predictive of neuronal degeneration. Results for the 12 h MRS
poorly predicted histological neuronal degeneration in all brain regions analyzed (data not shown). In contrast, MRS undertaken at 72 h for Lac: NAA, Lac:Cho, and NAA:Cho showed limited but statistically significant associations with deep gray matter (hippocampus, thalamus and striatum) neuronal degeneration. MRS at either time point was not correlated with neuronal degeneration within the cortex (Fig. 8). Discussion This study examined neuronal injury arising from severe asphyxia at birth and explored potential markers of brain injury using a large animal model. Results demonstrate that local biochemical assessments of the
Fig. 4. Measurement of malondialdehyde (MDA; A) and S100B (B) concentration within CSF in control (white) and asphyxia (black) collected at 72 h postnatal. MDA shows a 2-fold and S100B a 5-fold increase in asphyxia compared to control. Data are mean ± SEM *P b 0.05 compared to control.
286
J.D.S. Aridas et al. / Experimental Neurology 261 (2014) 281–290
Fig. 5. Representative MRI from control and asphyxia brains undertaken at 12 and 72 h. Modalities represented include diffusion-weighted images (DWI), T1, and T2. Arrows (D, H, L) indicate deep gray matter injury to the basal ganglia in asphyxia animals.
brain, particularly CSF or MRS markers, provide correlation with neuronal injury. However these were observed relatively late after asphyxia (72 h) and would therefore currently preclude their use for informing initiation of treatment, but may provide insight into permanent neuropathology. Early (b4 h) metabolic state (lactate and pH), unsurprisingly but importantly, was predictive for selective neuronal degeneration. We used a large animal model of HIE designed to directly reflect the clinical situation by having an in-utero asphyxic insult with delivery, resuscitation and postnatal care that adhered to clinical guidelines, while allowing for collection of physiological, biochemical and radiological assessments over time. Previous, in-utero sheep models (Castillo-Meléndez et al., 2013; Fujii et al., 2004; Williams et al., 1991) have maintained placental support during and after the insult. Neonatal rodent (Rice et al., 1981) and postnatal piglet animal models (Cady et al., 2011; Robertson et al., 2013), have not accounted for the cardiovascular and respiratory transition that occurs at birth. Non-human primates are informative, but their use is limited due to cost and housing requirements (Jacobson Misbe et al., 2011; Juul et al., 2007). Our experimental design takes into account the transition at birth as an important component in the development of HIE, although the maturity of the term sheep brain, in which myelination and gross development are advanced compared to the human, must be considered (Dobbing and Sands, 1979; Yager and Ashwal, 2009). This large animal model produced a severe asphyxia at birth, with 66% survival rate, and a spectrum of neuronal loss in surviving lambs as might be expected in the human situation following severe intrapartum insult. The asphyxic insult induced a metabolic acidosis consistent with severe birth asphyxia, with peripheral pH b 6.9 and baseexcess b − 14 mmol/l (MacLennan, 1999; Phelan et al., 2011). Histological assessment of brain injury was undertaken at 72 h after the insult, principally showing selective neuronal degeneration, apoptosis and astrogliosis within deep gray matter including hippocampus, thalamus and striatum. This distribution of neonatal brain injury is also observed in humans following birth asphyxia, and is highly predictive of long-term neurodevelopmental deficits (Barkovich et al., 1995; Leech and Alvord, 1977; Okereafor et al., 2008; Robertson and Perlman, 2006; Windle, 1963). This study therefore provided a clinically appropriate animal model of newborn HIE to examine biomarkers that are potentially translatable to the clinic or are in current clinical use. Early metabolic status in infants suspected of HIE could be used as a marker for short- and long-term outcomes. In newborn lambs, baseexcess was maximally decreased to −14.8 ± 1.5 mmol/l following asphyxia and showed a strong predictive value for hippocampal neuronal degeneration. In human infants, Ambalavanan et al. showed a base-
excess of b −22 mmol/l in the first postnatal blood sample was one of the most indicative variables for later death and disability (Ambalavanan et al., 2006). In the current study, the peak value for lactate, during or immediately after asphyxia, was strongly predictive for neuronal degeneration across all brain regions studied. Lactate remained elevated for 3 h after the insult, with a secondary increase over the secondary/tertiary phases of injury progression (8, 12 and 24 h). Our results compliment and extend a study in human infants by Mann et al., showing that lactate measured at 24 and 48 h in the peripheral circulation of 39 HIE newborns was correlated with MRI outcomes at day 5 (Mann et al., 2012). Combined, these data for the early phase following asphyxia reflect the severity of the initial insult, with acute changes predictive for neuronal degeneration. A marker that may directly reflect the response of the brain to insult is S100B. In the present study, S100B concentrations within CSF at 72 h were increased in asphyxia lambs, likely reflecting a CNS response to injury, however we did not find that CSF S100B, or circulating concentrations over time, were good predictors of neuronal degeneration. In response to asphyxia, the source of increased S100B in CSF is likely to be astrocytes (Gonçalves et al., 2008). In support of this, we found evidence for astrogliosis, with increased numbers of GFAP-positive cells with hypertrophic morphology within the hippocampus and thalamus following asphyxia. Clinically, S100B has been examined in newborns with and without asphyxia and serum levels were associated with moderate-severe HIE from birth to 24 h, but not with neurodevelopment assessment at 20 months of age (Nagdyman et al., 2001, 2003). A discrepancy in the predictive utility of S100B may reflect the balance between its neurotrophic and neurotoxic roles, and extra-cerebral sources such as adipocytes and bone marrow that may contribute to an injury response (Gonçalves et al., 2008). Following a critical asphyxic event and resuscitation, oxidative stress plays a central role in the evolution of brain injury (Azzopardi and Edwards, 2010; Ferriero, 2004). MDA, a cytotoxic product of lipid peroxidation, is used routinely as a marker of end-stage oxidative damage (Azzopardi and Edwards, 2010; Ferriero, 2004; Schmidt et al., 1996). Lipid peroxidation was increased 2-fold in cord blood of asphyxiated compared to healthy infants (Schmidt et al., 1996), and remained elevated 12–24 h later in resuscitated infants (Kumar et al., 2008). In this study, MDA in CSF at 72 h was increased following asphyxia and was associated with striatal neuronal degeneration. We did not find circulating MDA changed over time in the time-points we selected, and there was no difference between groups. In a previous study, we showed that oxygen free radical (hydroxyl) production was increased in the brain within 30 min of acute asphyxia and
J.D.S. Aridas et al. / Experimental Neurology 261 (2014) 281–290
287
Fig. 6. Representative magnetic resonance spectroscopy data (MRS) from control (A: 12 h; C: 72 h) and asphyxia (B: 12 h; D: 72 h) animals. White boxes indicate n-acetylaspartate (NAA), thatched boxes indicate lactate levels. Graphs show change in metabolic ratios of lactate:choline (Lac:Cho), NAA:Cho, and Lac:NAA at 12 h (E) and 72 h (F). Control is in white, asphyxia black. *P b 0.05.
remained elevated for ~ 90 min, with a secondary increase at 6–9 h (Miller et al., 2005). Combined, these data indicate that the brain responds to asphyxia with free radical production and lipid peroxidation that begins within 30 min and persists for at least 3 days. However, peripheral lipid peroxidation product is not strongly predictive for neuronal degeneration. MRI and MRS are considered the optimal modalities for evaluation of cerebral injury in term infants (Azzopardi and Edwards, 2010; Okereafor et al., 2008; Thayyil et al., 2010). In the present study, imaging and spectroscopy were undertaken at 12 and 72 h after birth, with the MRS region of interest covering 2 cm3 incorporating the basal ganglia, thalamus and hippocampus, as per clinical evaluation. No
significant differences were observed in brain metabolites at 12 h. At 72 h, 2 of 12 asphyxia lambs showed qualitative MRI abnormalities. In contrast, MRS at 72 h provided most reliable for neuronal degeneration, with NAA:Cho significantly altered in asphyxia lambs compared to controls. NAA is found almost exclusively in neurons and reduced NAA: Cho is indicative of degeneration of functional neurons (Cady et al., 2003). This loss was confirmed in the current study with a significant correlation between MRS-derived NAA:Cho and neuronal degeneration in hippocampus, thalamus and striatum regions. Furthermore, Lac:Cho metabolite ratio was also correlated to neuronal degeneration, which highlights the association between brain metabolic disturbance, increased lactate by anaerobic glycolysis, and neuronal degeneration.
288
J.D.S. Aridas et al. / Experimental Neurology 261 (2014) 281–290
Fig. 7. Representative photomicrographs of immunohistochemistry for neuronal degeneration (cresyl-violet/acid-fuschin; control: A; asphyxia: C) and astrogliosis (glial fibrillary acidic protein; control: B; HI: D) from the CA1 region of the hippocampus collected at 72 h postnatal. Table shows mean cell counts (cells/mm2). Counts in bold and underlined indicate significantly different data. Scale bar = 50 μm.
Clinically, Lac:NAA ratio is an effective HIE biomarker to predict death and disability at 1 year (Amess et al., 1999; Azzopardi and Edwards, 2010). It was, therefore, not surprising that lactate:NAA ratios in our lambs provided the ability to predict histological evidence of neuronal degeneration in deep gray matter. These data support that MRI and MRS are reliable tools in predicting term asphyxia brain injury, but are optimal at 2+ days after birth, reflecting the evolution of neuropathology. Whilst this is useful for long-term prediction of brain injury, it is essential to detect babies at high risk of encephalopathy earlier so that safe neuroprotective treatments, if indicated, can be commenced as soon as possible. In the present study the most useful and predictive measures for neuronal degeneration were early circulating lactate and base-excess, and late (72 h) brain MRS metabolite ratios, and CSF concentrations of MDA and S100B. We did not sample CSF at earlier time-points although other evidence supports that CSF may be examined for biomarker analysis (Aly et al., 2006; García-Alix et al., 1994; Oygür et al., 1998). This study strongly supports the utility of MRS at 72 h to detect histological neuronal degeneration once established. It should be noted that in the current study, asphyxia lambs were not cooled, which may be considered a limitation given that cooling could alter some physiological and/or biochemical parameters that were measured, although hypothermia does not affect MRI assessment (Neil, 2010). The decision not to include hypothermia in this study is based upon the need to first characterize fundamental responses to asphyxia, and we recognize that hypothermia treatment should be addressed in future studies.
Conclusion This study reinforces, in a clinically appropriate model, that no individual ‘gold standard’ biomarker is currently available to predict or detect the progression of neuronal degeneration following birth asphyxia. Our goal is to bring together a set of biomarkers that together could quickly and reliably detect the severity and timing of an asphyxial insult. We have shown that early markers such as blood lactate and base-excess, and robust secondary phase cell death markers in the brain (MRS) and CSF, can provide assessments of neuronal injury. Based on these observations, a combination of early, latent and secondary insult phase makers could be developed to measure the evolution of brain injury, to guide ongoing clinical care and options for therapeutic interventions, and to test new treatment strategies. Acknowledgments The authors gratefully acknowledge the expert technical assistance of Jan Loose, Yen Pham, Monique Mortale, Courtney McDonald, Madison Paton, Lesley Wiadrowski, David Shipp, Richard McIntyre, and Patricia Heidmann. Funding This study was supported by the National Health and Medical Research Council (NHMRC) of Australia Grant No: 1048039 and the Victorian Government's Operational Infrastructure Support Program.
J.D.S. Aridas et al. / Experimental Neurology 261 (2014) 281–290
Hippocampus r2
p
Thalamus
Striatum
r2
r2
P
289
Cortex
p
r2
p
Early (< 4 h) blood parameters Base-excess
-.63
Contents lists available at ScienceDirect
Experimental Neurology journal homepage: www.elsevier.com/locate/yexnr
Regular Article
Detecting brain injury in neonatal hypoxic ischemic encephalopathy: Closing the gap between experimental and clinical research James D.S. Aridas a, Tamara Yawno a, Amy E. Sutherland a, Ilias Nitsos a,b, Michael Ditchfield c, Flora Y. Wong a,c, Michael C. Fahey a,c, Atul Malhotra a,c, Euan M. Wallace a,b, Graham Jenkin a,b, Suzanne L. Miller a,b,⁎ a b c
The Ritchie Centre, MIMR-PHI Institute, Clayton, Victoria, Australia Department of Obstetrics and Gynaecology, Monash University, Clayton, Victoria, Australia Monash Children's, Monash Health, Clayton, Victoria, Australia
a r t i c l e
i n f o
Article history: Received 22 April 2014 Revised 3 July 2014 Accepted 20 July 2014 Available online 28 July 2014 Keywords: Animal models Sheep Biomarker Newborn infant Hypoxia–ischemia Asphyxia Hypoxic ischemic encephalopathy Neuroprotection Magnetic resonance imaging Oxidative stress Cell death
a b s t r a c t Moderate to severe neonatal hypoxic ischemic encephalopathy remains an important cause of infant death and childhood disability. Early and accurate diagnosis of encephalopathy is difficult but critical for timely intervention. Thus, we have utilized a clinically relevant large animal model of asphyxia in-utero, followed by immediate lamb delivery, resuscitation and clinical care over the next 72 h for assessment of potential biomarkers of brain injury. In-utero asphyxia was induced in twelve near-term lambs and outcomes compared with seven controls. Asphyxia resulted in bradycardia (97 ± 12 beats/min), hypotension (12.1 ± 1 mm Hg) and metabolic acidosis (pH 6.9 ± 0.02; base-excess −13.8 ± 0.8 mmol/l). 72 h following asphyxia, cerebrospinal concentrations of malondialdehyde and S100B were elevated 2-fold and 5-fold, respectively, in asphyxic lambs compared to control lambs. Magnetic resonance spectroscopy (MRS) at 72 h showed a significant decrease in n-acetyl aspartate: choline ratio in asphyxia lambs compared to that observed at 12 h (0.56 ± 0.23 vs. 0.82 ± 0.15, respectively); lactate:choline ratio was not changed over this time. Marked neuropathology was observed in asphyxia lambs with neuronal degeneration in the hippocampus, thalamus, striatum and cortex. Astrogliosis was observed in the hippocampus and thalamus. Early blood markers of metabolic state showed limited predictive value of histological damage at 72 h. MRS outcomes at 72 h showed a modest but significant correlation with histological evidence of neuronal brain injury (lactate:N-acetyl aspartate ratio in the thalamus r2 = 0.2, p b 0.01). MRS at 72 h was best able to detect established brain injury, but a combination of biomarkers over multiple phases of injury may be able to assess the evolution of neonatal brain injury. © 2014 Elsevier Inc. All rights reserved.
Introduction Neonatal hypoxic ischemic encephalopathy (HIE) remains an important cause of perinatal death and long-term developmental disability. Clinical trial meta-analyses demonstrate that approximately 60% of untreated (normothermic) infants will die or have long-term disability after HIE (Edwards et al., 2010). For surviving infants, there is a welldescribed association between HIE and diagnosis of cerebral palsy (Badawi et al., 1998). Currently, the only effective treatment to reduce adverse outcome following term HIE is hypothermia commencing within 6 h of delivery (Jacobs et al., 2013), however 47% of treated newborns Abbreviations: aEEG, amplitude integrated electro-encephalogram; Cho, choline; DWI, diffusion-weighted imaging; GFAP, glial fibrillary acidic protein; HIE, hypoxic ischemic encephalopathy; HR, heart rate; Lac, lactate; MAP, mean arterial pressure; MDA, malondialdehyde; NAA, N-acetylaspartate; MRI, magnetic resonance imaging; MRS, magnetic resonance spectroscopy; SaO2, oxygen saturation; SEM, standard error of the mean. ⁎ Corresponding author at: The Ritchie Centre, MIMR-PHI Institute, 27-31 Wright Street, Clayton, VIC 3168, Australia. E-mail address: [email protected] (S.L. Miller).
http://dx.doi.org/10.1016/j.expneurol.2014.07.009 0014-4886/© 2014 Elsevier Inc. All rights reserved.
are still at risk of death or serious disability (Jacobs et al., 2013). It is thus recognized that adjuvant or alternative treatments for HIE are necessary (Bennet et al., 2012; Miller et al., 2012; Robertson et al., 2012). In response to asphyxia at term birth, brain injury and encephalopathic symptoms evolve from hours through to weeks (Gunn and Gluckman, 2007; Williams et al., 1991). This is characterized by a primary insult phase where neuronal degeneration begins. Delivery of the baby and, in association with resuscitation, leads to apparent recovery during the latent phase. However the secondary phase introduces biochemical cascades such as excitotoxicity, oxidative stress and inflammation, and subsequent neuropathology to neuronal populations in the cerebral cortex, hippocampus, thalamus, basal ganglia and to a lesser extent in white matter (Azzopardi and Edwards, 2010; Folkerth, 2005; Gunn and Bennet, 2009; Inder, 2000; Thayyil et al., 2010). Early identification of HIE is essential for successful initiation of neuroprotective therapies. The results of hypothermia studies highlight that treatment should begin early for maximal benefit (Gunn et al., 1999). Clinically, there is an initial focus on observational parameters of encephalopathy, first described by Sarnat and Sarnat, to reflect graded
282
J.D.S. Aridas et al. / Experimental Neurology 261 (2014) 281–290
ventilation and assisted ventilation, subsequently, ceased if the lambs began to spontaneously breath N 50% of the time. Overt clinical seizures were assessed by clinical personnel (AM, MCF, FYW, EMW) as repetitive eye movements, “smacking” of the lips, neck arching, “running” leg movements, and apneic episodes, similar to human seizures as described by Volpe (Volpe, 2001). Seizures were treated with 20 mg/kg intravenous phenobarbitone (Sigma, Australia).
abnormalities in infant reflexes, behavior and tone, and incidence of seizures (Sarnat and Sarnat, 1976). More recently with the NICHD hypothermia trial for HIE, early (b 6 h) neurological classification utilizing clinical and laboratory parameters showed good predictive value (78%) for death/disability, and therefore infants that would benefit from hypothermia (Ambalavanan et al., 2006). Magnetic resonance imaging (MRI) studies of the newborn brain confirm neuropathology after perinatal asphyxia and are predictive for long-term impairment (Azzopardi and Edwards, 2010; Thayyil et al., 2010). However MRI is usually performed 2–6 days or later after the sentinel insult; with later scans showing greater predictive value (Azzopardi and Edwards, 2010). Ideally, neonatal care would incorporate specific biomarkers of HIE severity and neuropathology to guide therapeutic intervention and prognosis. The aim of this study was to cause HIE in a clinically useful large animal model in which we could correlate physiological, biochemical, radiological and histological markers of neuronal cell degeneration. This model provides endpoints selected to assess the evolution of brain injury during clinical care, to inform timing for therapeutic interventions, and to guide future neuroprotection studies.
Arterial blood samples (250 μl) were collected in-utero, during asphyxia, and at 0.5, 1, 2, 4, 8, 12, 24, 48 and 72 h postnatal for assessment of pH, PaO2, PaCO2, lactate, glucose, bicarbonate and base-excess (ABL Blood Gas Analyzer, Radiometer, Denmark). HR, SaO2, rectal temperature, and body weight were assessed hourly. When the lambs were extubated, alert, and had a strong suckle reflex, they were offered sheep milk replacer orally every 4 h (~100–150 ml; Wombaroo Food Products, Australia). Maintenance intravenous fluid (10% dextrose; 40 ml/kg/day) was commenced if lambs were unable to feed orally. Lambs were maintained for 72 h.
Materials & methods
S100B and malondialdehyde
Experiments complied with the National Health and Medical Research Council of Australia guidelines for the care and use of animals for scientific purposes and were approved by Monash Medical Centre Animal Ethics Committee.
S100B ELISA was performed as per manufacturer's instructions (DiaSorin, Minnesota, USA) on serum (50 μl) at time 0 (in-utero), and 1, 2, 4, 8, 12, 24 and 72 h postnatal, and on CSF obtained at postmortem. Assay sensitivity was 0.03 μg/l with a run imprecision of b10% and total imprecision of b15%. Malondialdehyde (MDA) was measured in plasma (100 μl) at the same time-points as for S100B, using a thiobarbituric acid reacting substance assay (Miller et al., 2014). The assay sensitivity was 0.1 μmol/l, with 5.1% inter-assay and 3.1% intra-assay coefficients of variation.
Surgery and hemodynamic recording Near-term pregnant ewes at 139–141 days gestation (term is 145– 147 days) underwent sterile surgery under general anesthesia induced by sodium-thiopentone (20 mg/kg IV bolus; Pentothal, Boerhringer Ingelheim, Australia) and maintained with 1–2.5% isoflurane (Isoflow, Abott Pth) in oxygen/nitrous oxide (O2: 2–3 L;N2O: b 1 L). The hindquarters of the fetus were exteriorized and a femoral artery cannula inserted for continuous digital recording of mean arterial pressure (MAP) and heart rate (HR; Powerlab SP, ADInstruments), and blood sampling. A femoral vein catheter was inserted. Pulse oximetry (MasimoSet Rainbow, Radical7, Masimo) was continuously monitored via a cuff placed on the shaved tail of the lamb. Asphyxia and resuscitation Fetal sheep were randomly allocated using an envelope assignment system to control or asphyxia groups. Asphyxia was induced via complete umbilical cord occlusion. With the head and upper body remaining in-utero in the amniotic fluid, the umbilical cord was exposed and clamped. Lambs in the control group had the cord clamped and cut immediately and were delivered and resuscitated. Lambs in the asphyxia group remained in-utero until MAP decreased to 18–20 mm Hg; shown previously in late gestation fetal sheep to induce severe asphyxia and neuropathology (Castillo-Meléndez et al., 2004). The cord was then cut and lambs delivered and resuscitated. Australian and New Zealand Neonatal Resuscitation Council guidelines for resuscitation were followed. At delivery, all lambs were placed on an infant warmer, intubated (4.5 endotracheal tube; Portex), and dried with towels. Positive pressure ventilation (Neopuff™, Fisher & Paykel Healthcare; 30 cmH2O positive inspiratory pressure [PIP], 5– 8 cmH2O positive end expiratory pressure [PEEP], 10 L/min room air, 30 breaths/min) was initiated. If required, lambs were administered adrenaline (1 ml) and fluid (normal saline; 20 ml/kg). As per guidelines, oxygen saturation (SaO2) was targeted at 85–90% within 10 min. Lambs were ventilated (Babylog 8000+, Draeger; volume guarantee 5 ml/kg, PEEP 5–7 cmH2O, 30 breaths/min) as required. Ventilation parameters were decreased and changed to continuous positive airway pressure
Maintenance
Magnetic resonance imaging and spectroscopy Brain magnetic resonance imaging and spectroscopy (MRS) were performed under sedation at 12 and 72 h after birth, using a 3 T Siemens Vario (Siemens Medical Solutions, USA) in an eight-channel knee coil (400 × 420 × 310 mm). Sedation was induced using Domitor (0.1 mg/kg iv; Pfizer Australia) a synthetic alpha-2-adrenoreceptor agonist; and reversed with the antagonist, antisedan (0.1 mg/kg; Pfizer Australia). Conventional sequences were obtained with sagittal diffusion-weighted-imaging (DWI) utilizing a slice thickness of 2 mm, 20 ms echo time, and 28 ms repetition time, T1-weighted (repetition time/echo time/excitations 1900/3.92/1), and T2-weighted (5000/90/1). MRS utilized a 270 ms echo with a 2 cm3 voxel placed over the hippocampus, striatum and basal ganglia. MRS produced spectrographs of concentrations (calculated by computer algorithm as area under the curve) of lactate (Lac), choline (Cho), and n-acetylaspartate (NAA). Brain pathology Lambs were euthanased (pentobarbitone, Lethabarb Virbac Australia) immediately after the 72 h MRI. The right cerebral hemisphere was cut coronally into 5 mm slices and immersion fixed in 4% paraformaldehyde for 48 h, paraffin embedded and then coronally sliced at 10 μm. The left cerebral hemisphere was separated into anatomical locations, snap frozen in liquid nitrogen and stored at −80 °C for future assessment. Neuronal degeneration described necrotic neurons, morphologically identified with cresyl-violet/acid-fuschin staining as swelling of organelles, loss of cell and nuclear membrane integrity, pyknotic nuclei, bright eosinophilic cytoplasm, or with darkened and condensed cytoplasm (Castillo-Meléndez et al., 2013; Northington et al., 2011). Glial fibrillary acidic protein (GFAP; 1:400, Sigma) immunopositive staining was used to assess the number of astrocytes and identification of activated
J.D.S. Aridas et al. / Experimental Neurology 261 (2014) 281–290
astrocytes showing hypertrophic cell bodies and attenuated and thickened processes (Yawno et al., 2012). Six fields of view (400× magnification) over two duplicate slides per brain region were examined and averaged across all animals. This included hippocampus (CA1, CA3, dentate gyrus), midtemporal cortex (molecular, pyramidal, and pleomorphic layers), striatum (caudate nucleus, and putamen) and thalamus (paraventricular and anteroventral regions), Fig. 1. Personnel were blinded to experimental group during image capture and analysis.
283
Data are presented as mean ± standard error of the mean (SEM). Blood gas parameters were analyzed using 2-way (time and group) repeated measures ANOVA with post-hoc analysis. One-way ANOVA with post-hoc analysis was used for cell counts, biochemical assay results, and MRS. Correlation analysis utilized the Pearson correlation coefficient. Significance was considered as p ≤ 0.05.
All lambs were intubated and resuscitated according to their individual needs. Four of the 12 asphyxia lambs required both adrenaline and normal saline bolus for cardiovascular stabilization, whereas control lambs did not. Lambs received respiratory support until spontaneous regular breathing was established; the duration of ventilation was longer in asphyxia lambs compared to controls (151 ± 17 min vs. 38 ± 6 min; P b 0.001). Prior to asphyxia, fetal arterial blood measures were not different between groups (Fig. 3), and were within normal values for late gestation fetal sheep (Castillo-Meléndez et al., 2013; Sobotka et al., 2011). During asphyxia, base-excess, lactate, pCO2, pH, pO2, and SaO2 were significantly altered from control parameters (Figs. 3 A, E, F, G, H). Appropriate resuscitation following delivery returned parameters to control values within 2 h, except for lactate, which returned within 4 h. Baseexcess, bicarbonate, and lactate demonstrated significant differences in asphyxia lambs at 8, 12, and 24 h after delivery. Twenty-five percent of asphyxia lambs exhibited clinical seizures, first apparent at 8 ± 4 h after birth, and were treated with phenobarbitone. No control animals exhibited seizures.
Results
S100B and malondialdehyde
Asphyxia and resuscitation
S100B concentration in CSF collected at 72 h was elevated 5-fold in asphyxia lambs compared to control lambs (P = 0.01). CSF MDA concentration was also elevated 2-fold in asphyxia lambs compared to control lambs (P = 0.007), Fig. 4. Concentrations of serum S100B and plasma MDA were not changed over time in control or asphyxia lambs (data not shown; S100B PGROUP = 0.35; MDA: PGROUP = 0.80).
Statistics
In total, 7 control and 18 asphyxia lambs were delivered. Six (33%) of the asphyxia lambs could not be resuscitated and therefore results are presented for 19 surviving animals; 12 asphyxia (4 male, 8 female) and 7 control (5 male, 2 female). MAP (Fig. 2 A) and HR (Fig. 2 B) prior to insult were not different between groups (control values: 55 ± 3 mm Hg, 199 ± 11 beats/min, respectively). In-utero asphyxia caused transient hypertension, followed by hypotension. The asphyxic insult was terminated at 9.9 ± 0.4 min, when MAP reached ~18 mm Hg. Following delivery, MAP and HR in asphyxia lambs decreased further to a nadir of 12 ± 1 mm Hg and 97 ± 12 beats/min respectively, prior to adequate resuscitation.
MRI and MRS Qualitative assessment of the DWI, T1-, and T2-weighted imaging at 72 h demonstrated abnormal scans in 2 of 12 asphyxia lambs. One lamb had bilateral symmetrical ischemic changes affecting the basal ganglia and thalami, and the other had global hypoxic ischemic changes
Fig. 1. Representative lamb brain photomicrograph of cresyl-violet/acid-fuschin anatomical regions assessed for histology. (A) Striatum: CN — Caudate nucleus, Put — Putamen; Cortex: Assessed in three aggregated layers — pleomorphic, pyramidal, and molecular. (B) Hippocampus: CA1 — cornu ammonis 1, CA3 — cornu ammonis 3, DG — dentate gyrus; Thalamus: AV — anteroventral thalamic nuclei; PV — paraventricular thalamic nuclei.
284
J.D.S. Aridas et al. / Experimental Neurology 261 (2014) 281–290
Fig. 2. (A) Mean arterial pressure (MAP; mm Hg) and (B) hear rate (HR; BPM) recordings of 15 min of stable in-utero recordings with no difference between control (open circle) and asphyxia (closed square), during asphyxia (10 min) showing an initial hypertension followed by hypotension and bradycardia, and 30 min ex-utero during resuscitation show a return of MAP and HR within minutes of delivery and resuscitation and stable period with no difference between groups. Data are mean ± SEM *P b 0.05.
effecting cortices and deep gray matter. Representative normal MRI scans from control animals and abnormal diffusion-weighted and T2 images from 72 h asphyxia animals are presented in Fig. 5. MRS at 72 h after birth showed a significant reduction in NAA:Cho ratio in deep gray matter of asphyxia lambs compared to control lambs (Fig. 6 F; P b 0.01). Lac:NAA ratio was not changed at 12 h and 72 h in asphyxia lambs compared to control lambs (P = 0.17 and P = 0.46, respectively). The Lac:Cho ratio was not different between asphyxia lambs and control lambs at either 12 or 72 h, Figs. 6 E and F. Neuropathology Asphyxia caused significant neuronal degeneration in the hippocampus, thalamus, striatum, and pyramidal and pleomorphic zone of the cortex compared to controls (Figs. 7 C and A respectively). In response to asphyxia, astrogliotic cells were identified as morphologically
distinct from normal resting astrocytes. Astrogliosis was observed within the CA1 of the hippocampus and thalamic paraventricular nuclei (Fig. 7 D). Interestingly, these two regions with notable astrogliosis were also the regions with the greatest fold-change in neuronal degeneration in response to asphyxia. Pearson correlation analysis Correlation analysis was undertaken to determine whether associations existed between the degree of neuronal degeneration within the hippocampus, thalamus, striatum and cortex, and brain and peripheral biomarkers. ‘Early’ (before 12 h) and ‘late’ (at 72 h; Fig. 8) correlations were undertaken. Resuscitation requirements, cardiovascular support and length of ventilation did not correlate with histological damage. Early blood parameters were taken as the peak change from control from delivery to 4 h postnatal. Base-excess, bicarbonate, SaO2, lactate,
J.D.S. Aridas et al. / Experimental Neurology 261 (2014) 281–290
285
Fig. 3. Arterial blood parameters measured in-utero, during asphyxia, and ex-utero. Base-excess (A), bicarbonate (B), glucose (C), hemoglobin (D), lactate (E), partial pressure of carbon dioxide (pCO2; F), pH (G), partial pressure of oxygen (pO2; H), oxygen saturation (SaO2; I) in control (open circles) and asphyxia (closed squares). The groups showed no difference inutero prior to asphyxia (Pre). Asphyxia caused changes in base-excess, bicarbonate, lactate, pCO2, pH, pO2, and SaO2. Data are mean ± SEM *P b 0.05.
PaCO2 and pH were all shown to be strongly predictive for neuronal degeneration in all neuronal populations analyzed, Fig. 8. At the final blood sample collection at 72 h, only blood lactate maintained a correlation with hippocampal neuronal degeneration (r2 = 0.60, P b 0.01). Peripheral S100B concentration at 2 h was correlated with hippocampal (r2 = −0.60, P = 0.02) and cortical (r2 = −0.68, P = 0.01) neuronal degeneration. Blood MDA concentrations at 8 and 72 h were significantly associated with hippocampal neuronal degeneration (8 h: r2 = 0.54, P = 0.02; 72 h: r2 = − 0.46, P = 0.04). MDA within CSF at 72 h showed a significant correlation with neuronal degeneration in the striatum (r2 = 0.54, P = 0.04). S100B concentration within CSF was not predictive of neuronal degeneration. Results for the 12 h MRS
poorly predicted histological neuronal degeneration in all brain regions analyzed (data not shown). In contrast, MRS undertaken at 72 h for Lac: NAA, Lac:Cho, and NAA:Cho showed limited but statistically significant associations with deep gray matter (hippocampus, thalamus and striatum) neuronal degeneration. MRS at either time point was not correlated with neuronal degeneration within the cortex (Fig. 8). Discussion This study examined neuronal injury arising from severe asphyxia at birth and explored potential markers of brain injury using a large animal model. Results demonstrate that local biochemical assessments of the
Fig. 4. Measurement of malondialdehyde (MDA; A) and S100B (B) concentration within CSF in control (white) and asphyxia (black) collected at 72 h postnatal. MDA shows a 2-fold and S100B a 5-fold increase in asphyxia compared to control. Data are mean ± SEM *P b 0.05 compared to control.
286
J.D.S. Aridas et al. / Experimental Neurology 261 (2014) 281–290
Fig. 5. Representative MRI from control and asphyxia brains undertaken at 12 and 72 h. Modalities represented include diffusion-weighted images (DWI), T1, and T2. Arrows (D, H, L) indicate deep gray matter injury to the basal ganglia in asphyxia animals.
brain, particularly CSF or MRS markers, provide correlation with neuronal injury. However these were observed relatively late after asphyxia (72 h) and would therefore currently preclude their use for informing initiation of treatment, but may provide insight into permanent neuropathology. Early (b4 h) metabolic state (lactate and pH), unsurprisingly but importantly, was predictive for selective neuronal degeneration. We used a large animal model of HIE designed to directly reflect the clinical situation by having an in-utero asphyxic insult with delivery, resuscitation and postnatal care that adhered to clinical guidelines, while allowing for collection of physiological, biochemical and radiological assessments over time. Previous, in-utero sheep models (Castillo-Meléndez et al., 2013; Fujii et al., 2004; Williams et al., 1991) have maintained placental support during and after the insult. Neonatal rodent (Rice et al., 1981) and postnatal piglet animal models (Cady et al., 2011; Robertson et al., 2013), have not accounted for the cardiovascular and respiratory transition that occurs at birth. Non-human primates are informative, but their use is limited due to cost and housing requirements (Jacobson Misbe et al., 2011; Juul et al., 2007). Our experimental design takes into account the transition at birth as an important component in the development of HIE, although the maturity of the term sheep brain, in which myelination and gross development are advanced compared to the human, must be considered (Dobbing and Sands, 1979; Yager and Ashwal, 2009). This large animal model produced a severe asphyxia at birth, with 66% survival rate, and a spectrum of neuronal loss in surviving lambs as might be expected in the human situation following severe intrapartum insult. The asphyxic insult induced a metabolic acidosis consistent with severe birth asphyxia, with peripheral pH b 6.9 and baseexcess b − 14 mmol/l (MacLennan, 1999; Phelan et al., 2011). Histological assessment of brain injury was undertaken at 72 h after the insult, principally showing selective neuronal degeneration, apoptosis and astrogliosis within deep gray matter including hippocampus, thalamus and striatum. This distribution of neonatal brain injury is also observed in humans following birth asphyxia, and is highly predictive of long-term neurodevelopmental deficits (Barkovich et al., 1995; Leech and Alvord, 1977; Okereafor et al., 2008; Robertson and Perlman, 2006; Windle, 1963). This study therefore provided a clinically appropriate animal model of newborn HIE to examine biomarkers that are potentially translatable to the clinic or are in current clinical use. Early metabolic status in infants suspected of HIE could be used as a marker for short- and long-term outcomes. In newborn lambs, baseexcess was maximally decreased to −14.8 ± 1.5 mmol/l following asphyxia and showed a strong predictive value for hippocampal neuronal degeneration. In human infants, Ambalavanan et al. showed a base-
excess of b −22 mmol/l in the first postnatal blood sample was one of the most indicative variables for later death and disability (Ambalavanan et al., 2006). In the current study, the peak value for lactate, during or immediately after asphyxia, was strongly predictive for neuronal degeneration across all brain regions studied. Lactate remained elevated for 3 h after the insult, with a secondary increase over the secondary/tertiary phases of injury progression (8, 12 and 24 h). Our results compliment and extend a study in human infants by Mann et al., showing that lactate measured at 24 and 48 h in the peripheral circulation of 39 HIE newborns was correlated with MRI outcomes at day 5 (Mann et al., 2012). Combined, these data for the early phase following asphyxia reflect the severity of the initial insult, with acute changes predictive for neuronal degeneration. A marker that may directly reflect the response of the brain to insult is S100B. In the present study, S100B concentrations within CSF at 72 h were increased in asphyxia lambs, likely reflecting a CNS response to injury, however we did not find that CSF S100B, or circulating concentrations over time, were good predictors of neuronal degeneration. In response to asphyxia, the source of increased S100B in CSF is likely to be astrocytes (Gonçalves et al., 2008). In support of this, we found evidence for astrogliosis, with increased numbers of GFAP-positive cells with hypertrophic morphology within the hippocampus and thalamus following asphyxia. Clinically, S100B has been examined in newborns with and without asphyxia and serum levels were associated with moderate-severe HIE from birth to 24 h, but not with neurodevelopment assessment at 20 months of age (Nagdyman et al., 2001, 2003). A discrepancy in the predictive utility of S100B may reflect the balance between its neurotrophic and neurotoxic roles, and extra-cerebral sources such as adipocytes and bone marrow that may contribute to an injury response (Gonçalves et al., 2008). Following a critical asphyxic event and resuscitation, oxidative stress plays a central role in the evolution of brain injury (Azzopardi and Edwards, 2010; Ferriero, 2004). MDA, a cytotoxic product of lipid peroxidation, is used routinely as a marker of end-stage oxidative damage (Azzopardi and Edwards, 2010; Ferriero, 2004; Schmidt et al., 1996). Lipid peroxidation was increased 2-fold in cord blood of asphyxiated compared to healthy infants (Schmidt et al., 1996), and remained elevated 12–24 h later in resuscitated infants (Kumar et al., 2008). In this study, MDA in CSF at 72 h was increased following asphyxia and was associated with striatal neuronal degeneration. We did not find circulating MDA changed over time in the time-points we selected, and there was no difference between groups. In a previous study, we showed that oxygen free radical (hydroxyl) production was increased in the brain within 30 min of acute asphyxia and
J.D.S. Aridas et al. / Experimental Neurology 261 (2014) 281–290
287
Fig. 6. Representative magnetic resonance spectroscopy data (MRS) from control (A: 12 h; C: 72 h) and asphyxia (B: 12 h; D: 72 h) animals. White boxes indicate n-acetylaspartate (NAA), thatched boxes indicate lactate levels. Graphs show change in metabolic ratios of lactate:choline (Lac:Cho), NAA:Cho, and Lac:NAA at 12 h (E) and 72 h (F). Control is in white, asphyxia black. *P b 0.05.
remained elevated for ~ 90 min, with a secondary increase at 6–9 h (Miller et al., 2005). Combined, these data indicate that the brain responds to asphyxia with free radical production and lipid peroxidation that begins within 30 min and persists for at least 3 days. However, peripheral lipid peroxidation product is not strongly predictive for neuronal degeneration. MRI and MRS are considered the optimal modalities for evaluation of cerebral injury in term infants (Azzopardi and Edwards, 2010; Okereafor et al., 2008; Thayyil et al., 2010). In the present study, imaging and spectroscopy were undertaken at 12 and 72 h after birth, with the MRS region of interest covering 2 cm3 incorporating the basal ganglia, thalamus and hippocampus, as per clinical evaluation. No
significant differences were observed in brain metabolites at 12 h. At 72 h, 2 of 12 asphyxia lambs showed qualitative MRI abnormalities. In contrast, MRS at 72 h provided most reliable for neuronal degeneration, with NAA:Cho significantly altered in asphyxia lambs compared to controls. NAA is found almost exclusively in neurons and reduced NAA: Cho is indicative of degeneration of functional neurons (Cady et al., 2003). This loss was confirmed in the current study with a significant correlation between MRS-derived NAA:Cho and neuronal degeneration in hippocampus, thalamus and striatum regions. Furthermore, Lac:Cho metabolite ratio was also correlated to neuronal degeneration, which highlights the association between brain metabolic disturbance, increased lactate by anaerobic glycolysis, and neuronal degeneration.
288
J.D.S. Aridas et al. / Experimental Neurology 261 (2014) 281–290
Fig. 7. Representative photomicrographs of immunohistochemistry for neuronal degeneration (cresyl-violet/acid-fuschin; control: A; asphyxia: C) and astrogliosis (glial fibrillary acidic protein; control: B; HI: D) from the CA1 region of the hippocampus collected at 72 h postnatal. Table shows mean cell counts (cells/mm2). Counts in bold and underlined indicate significantly different data. Scale bar = 50 μm.
Clinically, Lac:NAA ratio is an effective HIE biomarker to predict death and disability at 1 year (Amess et al., 1999; Azzopardi and Edwards, 2010). It was, therefore, not surprising that lactate:NAA ratios in our lambs provided the ability to predict histological evidence of neuronal degeneration in deep gray matter. These data support that MRI and MRS are reliable tools in predicting term asphyxia brain injury, but are optimal at 2+ days after birth, reflecting the evolution of neuropathology. Whilst this is useful for long-term prediction of brain injury, it is essential to detect babies at high risk of encephalopathy earlier so that safe neuroprotective treatments, if indicated, can be commenced as soon as possible. In the present study the most useful and predictive measures for neuronal degeneration were early circulating lactate and base-excess, and late (72 h) brain MRS metabolite ratios, and CSF concentrations of MDA and S100B. We did not sample CSF at earlier time-points although other evidence supports that CSF may be examined for biomarker analysis (Aly et al., 2006; García-Alix et al., 1994; Oygür et al., 1998). This study strongly supports the utility of MRS at 72 h to detect histological neuronal degeneration once established. It should be noted that in the current study, asphyxia lambs were not cooled, which may be considered a limitation given that cooling could alter some physiological and/or biochemical parameters that were measured, although hypothermia does not affect MRI assessment (Neil, 2010). The decision not to include hypothermia in this study is based upon the need to first characterize fundamental responses to asphyxia, and we recognize that hypothermia treatment should be addressed in future studies.
Conclusion This study reinforces, in a clinically appropriate model, that no individual ‘gold standard’ biomarker is currently available to predict or detect the progression of neuronal degeneration following birth asphyxia. Our goal is to bring together a set of biomarkers that together could quickly and reliably detect the severity and timing of an asphyxial insult. We have shown that early markers such as blood lactate and base-excess, and robust secondary phase cell death markers in the brain (MRS) and CSF, can provide assessments of neuronal injury. Based on these observations, a combination of early, latent and secondary insult phase makers could be developed to measure the evolution of brain injury, to guide ongoing clinical care and options for therapeutic interventions, and to test new treatment strategies. Acknowledgments The authors gratefully acknowledge the expert technical assistance of Jan Loose, Yen Pham, Monique Mortale, Courtney McDonald, Madison Paton, Lesley Wiadrowski, David Shipp, Richard McIntyre, and Patricia Heidmann. Funding This study was supported by the National Health and Medical Research Council (NHMRC) of Australia Grant No: 1048039 and the Victorian Government's Operational Infrastructure Support Program.
J.D.S. Aridas et al. / Experimental Neurology 261 (2014) 281–290
Hippocampus r2
p
Thalamus
Striatum
r2
r2
P
289
Cortex
p
r2
p
Early (< 4 h) blood parameters Base-excess
-.63
