Concurrent erythropoietin and hypothermia treatment improve outcomes in a term nonhuman primate model of perinatal asphyxia

Abstract

Background

Up to 65% of untreated infants suffering from moderate to severe hypoxic-ischemic encephalopathy (HIE) are at risk of death or major disability. Therapeutic hypothermia (HT) reduces this risk to approximately 50% (number needed to treat 7-9). Erythropoietin (Epo) is a neuroprotective treatment that is promising as an adjunctive therapy to decrease HIE-induced injury because Epo decreases apoptosis, inflammation, and oxidative injury, and promotes glial cell survival, and angiogenesis. We hypothesized that HT and concurrent Epo will be safe, effective, improve survival and reduce moderate-severe cerebral palsy (CP) in a term nonhuman primate model of perinatal asphyxia.

Methodology

35 Macaca nemestrina were delivered after 15-18 min of umbilical cord occlusion (UCO) and randomized to saline (n=14), HT only (n=9) or HT+Epo (n=12). There were 12 unasphyxiated controls. Epo (3500 U/kg × 1 followed by 3 doses of 2500 U/Kg, or Epo 1000 U/kg/d × 4 doses) was given on days 1, 2, 3, and 7. Timed blood samples were collected to measure plasma Epo concentrations. Animals underwent MRI/MRS and diffusion tensor imaging (DTI) at < 72 hours of age and again at 9 months. A battery of weekly developmental assessments was performed.

Results

UCO resulted in death or moderate-severe CP in 43% of saline, 44% of HT, and 0% of HT+Epo treated animals. Compared to non-UCO control animals, UCO animals exhibit poor weight gain, behavioral impairment, poor cerebellar growth and abnormal brain DTI. Compared to UCO saline, UCO HT+Epo improved motor and cognitive responses, cerebellar growth, DTI measures, and produced a death/disability relative risk reduction of 0.911 (95% CI −0.429 to 0.994), an absolute risk reduction of 0.395 (95% CI 0.072 to 0.635), and a number needed to treat of 2 (95% CI 14 to 2). HT+Epo effects on DTI included improved mode of anisotropy, fractional anisotropy, relative anisotropy and volume ratio as compared to UCO saline treated infants. No adverse drug reactions were noted in animals receiving Epo, nor were there any hematology, liver, or kidney laboratory effects.

Conclusions/Significance

HT+Epo treatment improved outcomes in nonhuman primates exposed to UCO. Adjunctive use of Epo combined with HT may improve the outcomes of term human infants with HIE and clinical trials are warranted.

Introduction

Hypoxic-ischemic (HI) brain injury at birth remains a significant problem, affecting 1.3-1.7/1000 term liveborn infants in the US, and contributing to 23% of neonatal deaths globally.^{1, 2} Untreated, moderate to severe hypoxic ischemic encephalopathy (HIE) carries a 60-65% risk of either death or major neurodevelopmental disability, including mental retardation, cerebral palsy (CP), hydrocephalus, and seizures.^{3, 4} Therapeutic hypothermia (HT) in term infants reduces this risk to approximately 50%, with a number needed to treat ranging from 7 to 9.³ The efficacy of hypothermia provides proof-of-concept that hypoxic-ischemic brain injury can be mitigated; however, additional treatments are needed to further improve outcomes.

HIE evolves over more than a week.^5-7 Therapeutic hypothermia targets early mechanisms of HI injury by reducing cerebral metabolism, excitotoxic neurotransmitter accumulation, ATP depletion, oxygen and nitrogen free radical release, and lipid peroxidation of cell membranes.^8-10 Ideally, supplementary HIE treatments will complement mechanisms of HT neuroprotection.^{11, 12} Erythropoietin (Epo) is a potentially promising supplemental therapy to augment brain repair and improve neurodevelopment outcomes. Small animal models of HIE have demonstrated that high dose Epo treatment decreases neuronal apoptosis, oligodendrocyte injury, inflammation, oxidative injury, nitric oxide toxicity, and glutamate toxicity while increasing neurogenesis, oligodendrogenesis, glial cell proliferation, and angiogenesis.^{13, 14} We hypothesize that Epo plus HT will be a safe and effective combination therapy that will improve survival and prevent the development of moderate-severe CP after HIE in a term nonhuman primate model of perinatal asphyxia.¹⁵

Methods

Animals

The Animal Care and Use Committees at the University of Washington in accordance with U.S. NIH guidelines approved all experimental protocols. Fifty-six Macaca nemestrina (pigtailed macaques) were delivered for the purposes of this study. Forty-two animals were delivered after umbilical cord occlusion (UCO) for either 15 or 18 min. UCO subjects were assigned to one of four treatment groups: saline, Epo only, HT only or HT+Epo. Nine animals of similar gestational age were delivered by Cesarean section to serve as normal controls. The first 25 treatment assignments were simple randomization with the order of treatments determined by sorting a sequence of random numbers, and subsequent assignments followed adaptive randomization such that the sex of the animal at birth determined the treatment assignment until the groups were balanced, this was done in an effort to minimize gender effects.¹⁶ Longitudinal neurodevelopment, anthropometric data, and spectroscopy data were also collected from concurrent controls (N=5).

Delivery and resuscitation

The use of UCO to model perinatal asphyxia in M. nemestrina has previously been reported and is summarized in Figure 1.^{15, 17} UCO was performed on animals in utero 1 - 8d prior to term (173d). The uterus was incised and the umbilical cord was externalized and clamped for 15 or 18 minutes (asphyxia group). While the clamp was in place, the uterus was supported with saline-soaked towels and an umbilical artery catheter was installed. After the UCO, the fetuses were delivered by hysterotomy. Control animals were also delivered by hysterotomy after intrauterine installation of an umbilical arterial catheter (2-3 min procedure). Fetuses were delivered and stabilized by a team of neonatologists using standardized neonatal resuscitation practice. Resuscitations included endotracheal intubation, positive pressure ventilation, chest compressions, and bolus epinephrine as indicated. Apgar scores were assigned at 1, 5, 10, and 20 min. Monitoring included a pulse oximeter, rectal thermometer, and amplitude-integrated EEG (aEEG) (BrainZ BRM3, Natus Medical Incorp., San Carlos, California). A covered heating pad, radiant warmer, and polyethylene sheet were used to provide thermal support during stabilization, then the animals were moved to a thermal-neutral incubator.

Figure 1.

Schedule of procedures and assessments for the nonhuman primate model of perinatal asphyxia. Abbreviations: UCO, umbilical cord occlusion; MRI, magnetic resonance and diffustion tensor imaging and spectroscopy; PT, physical therapy; EEG, amplitude-integrated electroencephalography. * Note, the first MRI scan was performed at either 24 or 72 hours of age.

Treatment groups

UCO animals were treated with saline, Epo, HT or Epo+HT. Epo treatment was either 3500 U/kg × 1 dose i.v. followed by 3 i.v. doses of 2500 U/Kg given at 30 min, 24h, 48h and 7d or Epo 1000 U/kg/d i.v. × 4 dosages at 30 min, 24h, 48h and 7d (Epogen®, Epoetin Alfa Recombinant, Amgen).¹⁵ Initial Epo dosing was based on 2500 U/kg used in the phase I/II clinical trial¹⁸ (with the addition of a 3500 U/kg loading dose) but the dosing was lowered to 1000 U/kg because initial pharmacokinetics parameters were 25% higher than expected possibly due to organ dysfunction, or to HT. To produce HT, the animals were not actively warmed at delivery. Active cooling was begun after resuscitation of the infants and always occurred by the third hour of life. To maintain HT, a water blanket was applied to the animal’s head and thermal support from the incubator was adjusted to achieve a rectal temperature of 33.5°C for 72h (Olympic Medical Cool Care System, Olympic Medical, Seattle, WA, USA). Rectal temperature was maintained by adjusting the incubator temperature. Rewarming was done slowly, raising the rectal temperature by 0.5°C per hour.

Animal care

Post-resuscitation care was conducted as previously reported.¹⁵ Briefly, arterial blood gas, lactate, and electrolytes (iSTAT®, HESKA Corp., Loveland, Colorado) were measured at multiple, scheduled intervals. Study animals were maintained for a minimum of 3d on parenteral fluids adjusted to maintain euglycemia and hydration. Enteral feedings were started on postnatal day 4, after rewarming for infants who were treated with HT, when the abdominal exam was normal and stooling was established. Weight was followed daily and standardized anthropometric measurements were done every week, including crown-rump length and head circumference.

Pharmacokinetic analysis

Blood samples (0.1 mL/sample) were collected to measure plasma Epo concentrations at timed intervals: 0 (predose baseline), 3, 6, 12, 24, 48, and 72 hours as previously described.¹⁸ Data analysis was conducted using noncompartmental pharmacokinetic techniques. The trapezoidal rule was used to compute the area under plasma concentration versus time curve (AUC) until the last measured value at 24 hours. The AUC was extended to infinity by taking the quotient of the 24 hour concentration and the elimination rate constant. Cmax is the maximum plasma concentration observed after the first dose.

The t_1/2 for the plasma Epo concentration versus time curve was computed by dividing 0.693 by the elimination rate constant. Clearance (Cl), volume of distribution (using the steady-state (Vss) and area (Varea) methods), and mean residence time (MRT) were calculated using the following formulas: Cl=D/AUC, Vss=[D(AUMC)]/AUC², Varea=D/[k(AUC)], MRT= AUMC/AUC, where D is the rEpo dose and AUMC is the area under the first moment curve.

Developmental evaluations

Developmental assessment was performed by the Infant Primate Research Laboratory at the University of Washington, as previously described.¹⁵ Briefly, age at self-feeding and temperature stability, newborn reflexes, muscle tone, behavioral state, and neurological responses were assessed 5d per week for the first 20d, using tests based on the Brazelton Neonatal Behavioral Assessment Scale. Neonatal activity was recorded every 4h. Evaluators who were blinded to treatment assigned an overall assessment of motor dysfunction. Object permanence testing paradigm, screening for deficits in visual acuity, novelty preference, social and motor behavior in mixed-sex play groups, and a standard series of learning and memory problems utilizing the Wisconsin General Test Apparatus were administered by trained evaluators.

To assess for the presence or absence of CP, animals were assessed by a physical therapist with a special interest in neonates. Sequential exams were done at one week, one month, and 8 months to document any evidence of motor abnormalities and contractures. Assessment included evaluation of their ability to control active movement. Muscle tone at each joint was graded on the Ashford scale of 0 (normal) to 4 (affected parts rigid in flexion or extension).¹⁹

MRI acquisition and analyses

All surviving asphyxiated animals, control animals, and five colony animals underwent sedated MR imaging as previously described at 24 or 72h and again at 9 months.¹⁵ Total scan time was approximately 2h and consisted of magnetization prepared rapid gradient echo (MPRAGE) high resolution T1 weighted imaging, diffusion tensor imaging (DTI), and single-voxel proton spectroscopy (MRS) acquired on a Philips Achieva 3.0 Tesla magnet with X-series Quasar Dual gradient system. Two 8-channel array head coils were custom-made to fit neonatal and juvenile macaques. Details of the sequence acquisition were performed as previously reported.¹⁵

Volumetric analysis

Manual 3D tracings of the whole brain, cerebellum, and caudate were created using the interactive semi-automated tools within RView software (http://rview.colinstudholme.net/) and the corresponding volumes were calculated.²⁰ Tracings were performed by three separate observers who were blind to the treatment. Specific boundaries for total brain, caudate and cerebellum were defined. Total brain volume included the cerebral hemispheres, superior sagittal sinus, diencephalon, brainstem, and cerebellum. Ventricular and extra-axial cerebrospinal fluid, optic chiasm, and pituitary stalk were excluded, and the brainstem was truncated at the foramen magnum. The cerebellum was outlined separately with the peduncles and brainstem excluded. Individuals masked to treatment group and outcome performed outlining in duplicate. Interrater reliability of brain segmentation was determined by Dice similarity coefficients. The Dice similarity coefficients between the three tracers ranged from 0.885-0.921 for caudate, 0.9255-0.9862 for cerebellum and 0.9135-0.9738 for cortex. Final images were rendered in 3D and inspected for accuracy before structural volumes were computed.

DTI

Voxelwise statistical analysis of the FA data was carried out using TBSS (Tract-Based Spatial Statistics²¹), part of FSL.²² First, FA images were created by fitting a tensor model to the raw diffusion data using FDT, and then brain-extracted using BET.²³ All subjects’ FA data were then aligned into a common space using the nonlinear registration tool FNIRT,^{24, 25} which uses a b-spline representation of the registration warp field.²⁶ Representative control animal at 3 days and 9 month scans were used to align animals instead of the human normal space. Next, the mean FA image was created and thinned to create a mean FA skeleton, which represents the centers of all tracts common to the group. Each subject’s aligned FA data was then projected onto this skeleton, and the resulting data fed into voxelwise cross-subject statistics. Voxelwise cross-subject statistics on non-FA DTI data (first eigenvalue, second eigenvalue, third eigenvalue, relative anisotropy, mean diffusivity, mode of anisotropy, volume ratio, radial diffusivity, and raw T2 signal without diffusion weighting) was performed using the tbss_non_FA command. Threshold-Free Cluster Enhancement with correction for multiple comparisons results were used to determine significance.

MR Spectroscopy

Single-voxel MRS was acquired using the point-resolved spectroscopy (PRESS) pulse sequence centered on a 10 × 10 × 10 mm voxel on the right thalamus in order to calculate the absolute concentrations of N-acetyl aspartate, creatine, choline, myo-inositol, glycine, glutamate, glutamine, as previously described using LCModel.^{15, 27, 28}

TE phase lactate

Lactate was analyzed separately from the other brain metabolites in order to isolate it from macromolecules/proteins as previously described.¹⁵

J-Resolved Spectroscopy

For each subject, a 2D J-resolved acquisition sequence was acquired from a 2 × 2 × 2 cm voxel centered on the thalamus as previously described.¹⁵ Custom software was used to perform a 2D Fourier transform and then the resulting signals were combined with LCmodel to estimate metabolite concentrations. With software called GAVA (www.briansoher.com), simulated JRES data were used to find the exact position of the J-coupled metabolites gamma-aminobutyric acid and glutamine. Gaussian filter adjustments were made to maximize separation of each metabolite.

Statistics

DTI was compared by voxelwise statistical analysis and can compare only two groups at once.²¹ Levene’s test evaluated homogeneity. Parametric comparisons are presented as means with standard error (SEM) and number of animals. All comparisons were two-tailed, with α ≤ 0.05. Two analyses were conducted. First, validation that the model produces injury was determined by comparing control animals to any animal that received UCO was performed. Second, determination of neuroprotection of treatments was performed with the exclusion of the control group. Post hoc testing of treatment effects were compared to UCO saline as the control group. Death or moderate-severe CP and epinephrine, sodium bicarbonate, and phenobarbital administration were evaluated by crosstabs with the binary output as the row and asphyxia (yes or no) as one column (Pearson Chi-Square) and treatment (Saline, HT, HT+Epo) as the other column (Fisher’s Exact Test) using SPSS (SPSS Inc., Chicago, Ill., USA). All other comparisons were made using univariate ANOVA analysis with fixed factors of asphyxia (yes or no) and treatment (saline, HT, HT+Epo) with appropriate post hoc comparisons to UCO saline using SPSS.

Results

Fifty-six M. nemestrina non-human primates were included in this study. Forty-two were delivered after umbilical cord occlusion (UCO) for either 15 (N=19) or 18 min (N=23) and randomized to: Saline (N=14), Epo (N=6), HT (N=10), or HT+Epo (N=12). Controls (N=14 total) included age-matched surgically delivered animals (N=9) and concurrent colony controls (N=5). Two of the 6 Epo-only treated animals were disqualified: one due to a difficult resuscitation with prolonged hypoxia, and one due to a cardiac arrest while undergoing MRI. During the study period, HT became the clinical standard of care in humans. No further animals were enrolled in the Epo-only group, which excluded this group from analysis due to low numbers. Three additional animals were excluded from the final analysis: 1 animal randomized to HT (could not be resuscitated at birth) and 2 control animals (1 required intubation at birth with evidence of intrauterine compromise,²⁹ and 1 had unexpected idiopathic hydrocephalus by MRI).

Of the 35 asphyxiated animals included in the analysis, 5 (14%) died prior to planned necropsy, with deaths distributed by treatment as shown in Figure 2. Three deaths occurred early (2 were euthanized at 3 days of age, and one at 6 days of age for severe neurological compromise), and 2 were due to late complications (one due to aspiration pneumonia, one died unexpectedly at 5 months after unexplained tachypnea since the first week of age). Three deaths were in animals exposed to 15 min of UCO, 2 were in animals exposed to 18 min of UCO. The gestational age of study animals at birth was not different between controls and UCO (169±1 vs. 168±1 days, p=0.34). Birth weight (547±16 versus 547±13 g, p=0.99), head circumference (20.6±0.2 versus 20.5±0.1 cm, p=0.57 and crown-rump length (20.6±0.2 versus 20.3±0.1 cm, p=0.28) were not different between groups. Epinephrine was given in 32% of resuscitations, and 20% of animals received sodium bicarbonate in the first 3 hours after birth. The initial neurologic exam of all UCO animals prior to study treatments was profoundly abnormal, with no spontaneous respirations or movement, no withdrawal to painful stimuli, and flaccidity. All UCO animals required intubation and mechanical ventilation before study treatments were administered. UCO animals differed from controls in initial pH (6.94±0.03 versus 7.19±0.04, p<0.001), base deficit (19.4±0.7 versus 9.3±1.3, p<0.001) and serum lactate (13.2±0.4 versus 5.9±0.9, p<0.001). Apgar scores in animals with UCO of 0, 15, and 18 min are shown in Figure 3. An early aEEG in all UCO animals was consistent with the obtunded physical exam, with lower margins generally <2 microvolts, upper margins <10 microvolts, and a burst suppression pattern. Slow improvement in voltage occurred over the first 6 hours of life. Phenobarbital was given to 49% of UCO infants for documented clinical seizures (7 UCO saline, 5 HT, 5 HT+Epo, Pearson Chi-Square p=0.81).

Figure 2.

Death or severity of cerebral palsy (CP) after umbilical cord occlusion (UCO). Top panel shows the number of animals with death or moderate/severe CP, and the bottom panel shows the individual severity scores for all animals by treatment group. Severity of CP was based upon four physical therapy exams performed at one week, and at 1, 4, and 8 months of age. UCO increased the rate of death or moderate/severe CP as compared to controls (ANOVA * = p<0.05). Treatment with hypothermia plus erythropoietin (HT+Epo) improved outcomes (Dunnett’s test, † = p<.05).

Figure 3.

Initial response to umbilical cord occlusion (UCO). Data are mean ± SEM Apgar scores for neonatal animals separated by the duration of UCO. Controls are gray boxes, 15 min UCO as black boxes (dashed line), and 18 min UCO as black diamond (solid line).

Growth

Figure 4 illustrates the daily weights of animals over 38 postnatal weeks to show that the average weights for UCO-exposed animals remained below the 50^th percentile compared to control animals despite adjustments to the caloric density of formula (up to 24 kilocalories/ounce) to foster weight gain. Inset in Fig 4 is a comparison of the mean time to regain birthweight and shows that this time was increased for both the UCO saline and UCO HT groups. UCO was associated with decreased head circumference growth rate (p <0.05), and this parameter improved with all treatments (data not shown).

Figure 4.

Longitudinal growth of nonhuman primates. The main graph plots the average weight of UCO-exposed animals (lines) superimposed upon the individual daily weights of control animals (gray squares) with an inset graph plotting the mean (± SEM) days to regain birthweight. Mean daily weights for UCO saline (black line), UCO HT (blue line) and UCO HT+Epo (red line) groups are shown separately. The inset shows that the time to regain birthweight was significantly elevated for UCO saline and UCO HT animals (Dunnett’s test, † = p<.01). These graphs illustrates that UCO reduces the growth of animals despite caloric adjustments to foster weight gain.

Behavioral Results

UCO was associated with delayed achievement of early developmental milestones, including the ability to regulate temperature, as reflected by the number of days required to wean from an incubator and maintain euthermia without a heating pad (Table 1). UCO also delayed the animal’s ability to feed themselves (Table 1). Time to resolution of primitive reflexes (rooting, righting, snout, sucking, and placing) was also delayed by UCO (Table 1). Study treatments did not significantly affect these parameters, however, infants treated with HT+EPO trended towards improved scores. Compared to controls, UCO animals exhibited delays in multiple behavioral indices including the capacity to sit, stand, walk, climb, and climb off a diaper roll (MTR); to leave the security of their “sleep diaper” in order to explore the playroom (diaper trip); to release their grasp and explore when challenged to hang from a support (release grasp); to touch another animal (initiate contact); and to climb up a chain (Table 2). Concurrent HT+Epo treatment significantly decreased the age at which infants reached for a toy, exhibited hand over hand chain climbing (Table 2) and object permanence (Table 3) compared to UCO saline treated infants.

Table 1.

Neonatal Behavioral Measures. Data are mean age (± SEM) at which a task was successfully performed for each separate treatment group. Abbreviations: UCO, umbilical cord occlusion; HT, hypothermia; Epo, erythropoietin.

Index	Controls	UCO All	UCO Saline	UCO HT	UCO HT +Epo
N	12	32	12	8	12
Out of incubator	5.4±0.4	9.7±0.8 *	11.0±2.2	9.6±0.5	8.5±0.4
Off heating pad	11.8±0.4	16.1±1.0 *	17.7±2.5	15.6±0.8	183.0±0.8
Self-feed	12.7±0.6	23.3±2.7 *	27.6±4.0	28.5±8.6	16.5±0.8
Primitive Reflexes	2.5±1.6	5.7±0.7 *	7.4±1.1	5.5±1.5	4.1±0.7

^*Significant ANOVA post hoc tests are indicated as = p<.05 compared to Controls.

Table 2.

Developmental behavioral measures. Data are mean age (± SEM) at which a task was successfully performed for each separate treatment group. Abbreviations: UCO, umbilical cord occlusion; HT, hypothermia; Epo, erythropoietin.

Index	Controls	UCO All	UCO Saline	UCO HT	UCO HT +Epo
N	12	32	12	8	12
MTR	9.8±1.4	11.3±0.8 *	13.5±0.4	11.0±1.9	9.4±1.5
Diaper Trip	18.2±2.1	23.3±1.7 *	27.1±2.1	22.4±4.5	20.1±2.3
Release Grasp	41.5±4.0	54.2±6.3 *	67.2±9.1	59.4±19.4	37.8±3.4
Initiate Contact	19.1±2.4	25.8±1.7 *	28.3±2.4	25.4±4.6	23.5±2.4
Climb Up Chain	96.8±6.2	125.5±9.0 *	149.7±13.7	132.5±24.9	96.6±4.5 †
Climb Down Chain	96.8±6.2	126.7±8.9	152.6±13.2	132.5±24.9	97.0±4.6 †
Reach for toy	27.3±2.9	29.5±1.9	34.3±2.9	31.4±4.1	23.6±2.3 †

^*Significant ANOVA post hoc tests are indicated as = p<.05 compared to Controls,

^†=p<.05 compared to UCO saline.

Table 3.

Object Permanence. Data are mean age (± SEM) at which a hidden object was recognized for each separate treatment group. Abbreviations: UCO, umbilical cord occlusion; HT, hypothermia; Epo, erythropoietin.

Index	Controls	UCO All	UCO Saline	UCO HT	UCO HT +Epo
N	12	32	12	8	12
No screen	35.8±2.8	39.4±1.9 *	45.3±2.5	38.4±4.1	34.3±3.0 †
With screen	36.2±2.8	42.1±2.3 *	47.3±2.3	41.8±6.3	37.1±3.8
Item in well	38.3±2.5	40.4±2.2	47.1±2.8	38.9±4.1	34.7±3.8 †

^*Significant ANOVA post hoc tests are indicated as = p<.05 compared to Controls,

^†=p<.05 compared to UCO saline.

The Visual Paired-Comparison test methodology was used to study emerging recognition memory skills in the first months of postnatal life (for review, see Burbacher and Grant, 2012).³⁰ On problems using simple geometric forms, UCO saline infants performed poorly, while infants receiving HT+Epo provided significant evidence of recognition memory (p<0.05). This finding suggests a disruption in the early memory processing of UCO infants but also, a sparing of some adverse memory effects in animals receiving treatment. When more challenging test problems were administered, both UCO saline and treated animals failed to provide evidence of memory when compared to control infants. Analysis of overall performance across experimental groups, independent of problem difficulty, revealed that the highest memory scores were observed in control and HT+Epo infants (p<0.0001). Animals in HT treatment group were able to provide some evidence of memory on this task (p<0.05) but control and HT+Epo infants provided the strongest psychometric profile on this early measure of cognition and information processing.

Magnetic Resonance Imaging and Spectroscopy

No significant effects of UCO or treatment were detected on gross brain structure evaluated either early (day 1 to 3) or late (6 or 9 months). To evaluate brain growth, brain volume at 9 months was compared to brain volume within 72h of birth (N = 35, 7 controls, 9 UCO saline, 6 HT, 8 HT+Epo). There were no effects of UCO on total brain, cortical or basal ganglia volumes. In contrast, cerebellar volume was decreased by UCO compared to controls, and combined HT+Epo treatment improved cerebellar growth after UCO (Figure 5). There were no left-right differences in brain volumes or specific MRI lesions among the study groups.

Figure 5.

Cerebellar growth from birth to 9 months of age. Box and whisker plots (median and quartile range) illustrate the percentage increase in cerebellar volume for animals in each treatment group. Cerebellar growth was decreased by umbilical cord occlusion (UCO) compared to controls. Significant differences are indicated (ANOVA * = p≤.05). Treatment with hypothermia (HT) plus erythropoietin (Epo) is associated with improved growth after UCO when compared to UCO saline treated animals (Dunnett’s t, bar p=0.019, N=9 UCO saline, UCO 6 HT, and 8 UCO HT+Epo).

We used fractional anisotropy (FA) to evaluate fiber density, axonal diameter, and myelination. FMRIB Software Library (FSL) tract-based spatial statistics revealed changes in the mode of anisotropy between UCO saline and control animals (Figure 6).³¹ FA, relative anisotropy (Figure 6), and volume ratio differences between HT+Epo versus UCO saline were detected in animals scanned at 72h. At nine months, the raw T2 signal was increased in injured animals, with UCO saline having the highest signal and control the lowest signal (Figure 4). In animals that underwent UCO for 18 min, early scans (24 and 72h) of HT+Epo animals showed improved FA, relative anisotropy, and improved volume ratios compared to UCO saline (Figure 7). Mode of anisotropy differentiated between UCO saline and HT+Epo animals within the first 72h after an UCO duration of 18 min. At nine months, the raw T2 signal was similar to the combined 15 and 18 min UCO results.

Figure 6.

Diffusion tensor imaging after umbilical cord occlusion (UCO). TBSS analysis of mode of anisotropy, relative anisotropy and raw T2 signal maps with red indicating areas that had increased signal compared to UCO saline and blue areas indicating lower signals. Highlighted (red or blue) regions p ≤ 0.05 using Threshold-Free Cluster Enhancement with correction for multiple comparisons. Higher values of mode of anisotropy and relative anisotropy are associated with healthy white matter. Lower values of T2 signal are associated with healthy white matter. Top two panels show validation of the perinatal asphyxia model and the bottom panel demonstrating that therapeutic hypothermia (HT) plus erythropoietin (Epo) improves diffusion tensor imaging.

Figure 7.

Diffusion tensor imaging (DTI) after 18 min of umbilical cord occlusion (UCO) .TBSS analysis of fractional anisotropy, relative anisotropy and volume ratio between UCO saline and hypothermia plus erythropoietin (HT+Epo) after 18 min of UCO is shown. DTI scans occurred within the first 72h. Red areas indicate where HT+Epo group had higher values than UCO saline while blue signifies lower values in the HT+Epo group. Highlighted regions (red or blue) p ≤ 0.05 using Threshold-Free Cluster Enhancement with correction for multiple comparisons. Higher values of fractional anisotropy and relative anisotropy are indicative of healthy white matter. Lower values of volume ratio are indicative of healthy white matter.

Spectroscopy results are shown in Table 4. The N-acetyl aspartate (NAA)/Creatine ratio in the first 72 hours and the GABA level at 9 months were altered by UCO. While the GABA levels decreased with treatment, this change was not statistically significant. Treatment did affect choline level and the Choline/Creatine, NAA/Choline, and NAA/Creatine ratios, so that animals treated with HT+Epo demonstrated improved values. Choline levels correlated with death or moderate-severe CP in animals that underwent UCO (Pearson correlation 0.375, p=0.029, N=34). Analysis of phosphorous data did not reveal any effects of UCO or treatment. However, the overall energy available to cells increased (as measured by lower inorganic phosphorous levels) to control values over time in the HT and HT+Epo groups.

Table 4.

Magnetic Resonance Spectroscopy (MRS) Findings. Data are mean (± SEM) values determined by MRS at two different ages separated by treatment group. Abbreviations: UCO, umbilical cord occlusion; HT, hypothermia; Epo, erythropoietin; NAA, N-acetylaspartate; GABA, gamma-aminobutyric acid.

	Controls	UCO All	UCO Saline	UCO HT	UCO HT +Epo
First 72 hours
N	8-12	27-34	11-13	7-9	9-12
Choline (mM/kg)	1.60±0.05	1.47±0.04	1.63±0.07	1.44±0.03 †	1.32±0.05 †
Choline/Creatine	0.29±0.01	0.27±0.01	0.30±0.01	0.26±0.01 †	0.25±0.01 †
NAA/Choline	2.79±0.08	2.79±0.09	2.50±0.13	2.81±0.12	3.13±0.11 †
NAA/Creatine	0.79±0.02	0.74±0.01 *	0.73±0.02	0.72±0.03	0.78±0.02
9 Month
N	8	24	9	6	9
GABA (mM/kg)	0.91±0.08	0.76±0.04 *	0.67±0.08	0.80±0.05	0.82±0.08

^*Significant ANOVA post hoc tests are indicated as = p<.05 compared to Controls,

^†=p<.05 compared to UCO saline.

Safety parameters

Blood urea nitrogen levels were elevated 24 hours after UCO compared to controls (20.7±1.0 versus 16.3±2.4 mg/dL), but this did not persist, and no treatment effects were noted. Serum creatinine levels were not different between control and UCO animals. Animals undergoing HT therapy had lower serum creatinine values at 24 hours than infants receiving UCO saline (UCO saline 1.20±0.07 mg/dL; HT 0.96±0.06 mg/dL, p=0.071; HT+Epo 0.92±0.08 mg/dL, p=0.017). Liver function tests (aspartate aminotransferase, alanine aminotransferase, serum bilirubin, and cholesterol) did not differ at any time point. Although not affected by treatment, white blood cell counts were decreased in UCO infants compared to control animals at one week of age (8.2±1.1 versus 5.1±0.3 thou/μL, p=0.003), but not at any other time point (Table 5). Hematocrit and platelet count were unchanged by UCO or treatment group at any time point (Table 5).

Table 5.

CBC Results. Data are mean values (± SEM) of laboratory measures for separate treatment groups.

Age (d)	Control	UCO All	UCO Saline	UCO HT	UCO HT+Epo
Hematocrit (%)
1	40.5 ± 1.4	42.4±0.9	44.0 ± 1.3	41.1 ± 2.0	42.9 ± 1.5
7	35.0 ± 1.3	37.3±0.8	37.1 ± 1.2	36.2 ± 1.0	38.4 ± 1.5
270	41.0 ± 1.3	40.4±0.4	40.4 ± 0.7	40.5 ± 0.8	40.1 ± 0.6
Platelet count (1000/μL)
1	360 ± 31	326±13	333 ± 26	295 ± 25	353 ± 26
7	512 ± 36	477±28	501 ± 65	419 ± 25	498 ± 24
270	500 ± 22	462±20	495 ± 33	404 ± 37	451 ± 38
White blood cell count (1000/μL)
1	7.6 ± 1.3	6.7±0.4	7.5 ± 0.4	5.8 ± 0.8	7.0 ± 0.7
7	8.2 ± 1.1	5.1±0.3 *	5.2 ± 0.5	4.0 ± 1.2	5.6 ± 0.3
270	7.3 ± 1.2	5.8±0.5	5.8 ± 0.6	6.0 ± 1.7	4.9 ± 0.3

^*UCO significantly lowered the white blood cell counts on day 7 compared to controls as indicated (ANOVA = p<.05). Abbreviations: UCO, umbilical cord occlusion; HT, hypothermia; Epo, erythropoietin.

Epo Pharmacokinetics

Peak Epo concentration (Cmax) for animals receiving 3,500 U/kg Epo+HT was 60,996 ± 25,635 mU/mL (N=6) compared to 16,290 ± 3,678 mU/mL for those who received 1,000 U/kg Epo+HT (N=4). Note that insufficient samples were available from some animals to conduct a complete pharmacokinetic analysis and that one of the higher dose infants received 3,000 U/kg due to a dosage administration error. Of the 4 animals receiving 1,000 U/kg, 3 had no CP and 1 had mild CP. Of the 6 receiving higher dose Epo, 2 had no CP and 4 had mild CP. Figure 8a and 8b show the relationship of Cmax and area under the curve (AUC) to outcome. The Cmax in animals without CP was 31,889 mU/mL with a mean AUC of 215,520 U*h/mL. In comparison, the Cmax for animals with mild CP was 54,337 mU/mL with a mean AUC of 356,719 U*h/mL. No animals receiving either dose of Epo in conjunction with HT died or had moderate or severe CP. A summary of all pharmacokinetic parameters are shown in Table 6. As noted previously, Epo dosing results in non-linear kinetics with the higher dose resulting in > 3.5 fold increase in AUC.^{18, 32} Of interest, endogenous Epo production was not increased by up to 18 min of UCO (Figure 8c).

Figure 8.

Pharmacokinetics of erythropoietin (Epo) in the setting of hypothermia. Panel a shows the peak erythropoietin concentrations (C_max) of the ten animals treated with hypothermia and Epo (HT+Epo) divided into outcomes of no cerebral palsy (CP) or mild CP. Panel b illustrates the paramater area under the curve (AUC). Panel c shows mean (± SEM) plasma concentrations of Epo in the first 72 hours. Epo concentrations of non-Epo-treated asphyxiated animals are shown by white diamonds. Pharmacokinetics of high dose Epo are depicted as black circles while the lower dose Epo is shown as gray squares. Epo dosing occurred at times 0, 24 and 48 hours.

Table 6.

Epo pharmacokinetics. Data are mean ± SEM values for two separate Epo doses. Abbreviations: rEpo, recombinant erythropoietin; AUC, area under the plasma concentration-time curve; C_max, maximum plasma concentration; V_D, volume of distribution.

Epo Dose (U/kg)	1000	3500
N	4	6
AUC (U*h)/L	94,377 ± 5,801	413,948 ± 39,605
Cmax (U/L)	16,290 ± 1,839	60,996 ± 10,465
Clearance (ml/h/kg)	10.7 ± 0.7	8.6 ± 0.8
Half-life (h)	6.1 ± 0.6	6.8 ± 1.0
Mean residence time (h)	7.2 ± 0.7	8.1 ± 0.5
VD, steady state (ml/kg)	76 ± 4	70 ± 6.9
VD, area (ml/kg)	93 ± 7	85 ± 15.1

Death and Cerebral Palsy

In keeping with clinical trials of hypothermia, our primary outcome was the presence of death (prior to planned necropsy) or CP (moderate-severe). Figure 2 shows the number of animals with early death or moderate-severe CP by treatment group. Outcomes for animals that underwent UCO were significantly different than control animal outcomes (Pearson Chi-Square p=0.046). Treatment changed the outcome of death or moderate-severe CP (Fisher’s Exact Test p=0.46). Comparing the UCO saline and combined HT+Epo animals gave a relative risk reduction of 0.911 (95% CI −0.429 to 0.994), an absolute risk reduction of 0.395 (95% CI 0.072 to 0.635), and a number needed to treat of 2 (95% CI 14 to 2).

Discussion

In agreement with our prior report,¹⁵ 15 to 18 min of UCO in M. nemestrina produces features that accurately model human perinatal asphyxia including abnormal physical exam, acidosis, elevated lactate, large base deficits, and Apgar scores that are all comparable to data from human infants enrolled in studies of moderate to severe HIE. Because hypothermia is now the standard of care for the treatment of neonatal HIE,³ the safety and efficacy of new therapies for HIE must be evaluated in concert with HT. This study is the first to evaluate the safety and efficacy of Epo for perinatal asphyxia in combination with HT therapy in primates. The major findings of this study are that although 72h of HT alone did not reduce the risk of death or moderate-severe CP after UCO, HT combined with 4 doses of Epo decreased risk of death or moderate-severe CP to 0%, and preserved normal motor functions. The improvements in motor function are consistent with the finding that combined HT+Epo therapy also preserved cerebellar growth.

The optimal dose and regimen for human Epo neuroprotection is still not known. During the course of this study, we observed that peak Epo concentrations in hypothermic neonatal primates given 2500 U/kg were 25% greater than expected when compared to data from normothermic infants given 2500 U/kg¹² and, therefore, we lowered the Epo dose to 1000 U/kg as a safety precaution. At 1000 U/kg, the pharmacokinetic parameters obtained from hypothermic primates (C_max ~16,000 U/L, AUC ~94,000) were comparable to those obtained from normothermic human neonates (C_max ~ 14,000 U/L, AUC ~81,000).^{18, 32} This dose also results in pharmacokinetic parameters similar to those obtained with 5000 U/kg in rats (Cmax 6,000-10,000 U/L, AUC 120,000-140,000).³³ Collectively, given the evidence in rodent models that 5,000 U/kg produces neuroprotection¹³ and that multiple injections and late one-week dosing are most effective,^34-36 and given that no harmful drug effects have been noted clinically, we are satisfied that repeated 1000 U/kg i.v. Epo is a safe and effective dose, suitable for future neuroprotection studies.

Several minor observations about the model deserve comment. The initial pH of control animals was lower than what would be considered normal in humans, but quickly normalized. Perhaps the lower pH is secondary to surgical delivery, general anesthesia or umbilical catheter placement produce brief HI and subsequent mild acidosis. As seen previously,¹⁵ UCO slightly reduced the growth of animals despite nutritional adjustments and it is possible that delayed somatic growth influenced neurodevelopmental outcomes. Also, it is interesting to note that endogenous Epo concentrations were not increased in control animals, nor in animals exposed to 15-18 min of UCO. Apparently, this duration of UCO is insufficient to acutely induce HIF-1 and elevate endogenous Epo release. The lack of an endogenous neonatal response strengthens the rationale for early use of exogenous Epo in neonates after HI.

The DTI findings in this study changed over time after UCO, consistent with pseudo-normalization as brain injury evolves.³⁷ Mode of anisotropy correlated with HI injury within the first 24h, and we speculate that this index might be interesting as an early diagnostic biomarker of HIE. HT affected the natural progression of DTI changes after HIE. HT+Epo animals tended to have DTI results most similar to controls. The elevated T2 signal in UCO saline animals is similar to diffuse excessive high signal intensity (DEHSI) seen in preterm infants, however the neurodevelopmental outcomes of DEHSI are currently debated.^{38, 39}

The optimal timing of MRI following injury is variable, depending on whether maximizing sensitivity or specificity is the goal. The known timing of injury in this model provides an opportunity for further delineation of best practice for timing of MRI as the brain injury evolves over time. Magnetic resonance spectroscopy has been used to demonstrate injury to the brain during hypoxic-ischemia.^{40, 41} We expected to see changes in several metabolites especially the NAA/lactate ratio, but the only differences between controls and UCO were NAA/Creatine ratio in the first 72 hours, and GABA at 9 months. However, HT and HT+EPO decreased the amount of choline and resulted in better ratios of Choline/Creatine and NAA/Choline. Increased choline levels have been correlated with increased risk of mortality following HIE.⁴² HT+EPO treatment had no deaths and had the lowest choline levels. Choline may be a biomarker of response to treatment and/or a prognostic biomarker. We speculate that choline might be used to determine length of hypothermia to improve outcomes.

Cerebellar hypoplasia is seen in human infants after HIE.⁴³ UCO saline decreased the rate of growth in the cerebellum, and treatment with HT+Epo normalized cerebellar growth. The motor problems associated with cerebellar abnormalities are well documented, but there is mounting evidence that the cerebellum is also involved in cognitive performance and social interactions.^{44, 45} Poor neurocognitive outcomes of cerebellar injury/hypoplasia are increasingly being seen in preterm infants without cortical injuries.⁴⁶ Social interaction is impaired in humans with autism, and autopsy reports show that 95% of autistic children have hypoplastic cerebella.⁴⁷ While some of the animals that underwent UCO manifested autistic-like repetitive behaviors like self-clasping and rocking, no formal testing was done in this cohort.

This model of UCO has several strengths. Brain development and complexity is similar to humans, allowing for complex neurocognitive testing over time.⁴⁸ This study measured developmental indices until 9 months of age, which is comparable to 3 years of human development. The clinical and laboratory findings of asphyxiated animals are very similar to humans with moderate to severe HIE. Unlike clinical scenarios in human neonates, however, the timing and severity of initial injury is known in this model allowing for uniform initiation of treatments across groups.^49-52 This allows for detection of biomarkers that differentiate timing and severity of injury.

Limitations of this study include the small number of animals in each group, the refinement of dosing regimens (discussed above) and the change from 15 to 18 min of UCO. This increase in UCO duration was effected because as developmental and long-term MRI data were collected, it became apparent that 15 min of UCO was insufficient to reliably produce significant disability and lasting brain injury needed to test neuroprotective strategies. The repeated developmental testing may also have served as unintended physical therapy, fostering improved outcomes. Work with nonhuman primates requires teams of clinicians and technicians, sophisticated diagnostic procedures, intensive 24h care and longitudinal behavioral testing, so large studies can be prohibitively expensive. Despite the small number of animals, we are confident that the data presented is informative. Like their human counterparts, these animals exhibit variability in the degree of brain injury sustained, providing an excellent opportunity to identify potential biomarkers of severity of brain injury, response to therapy, and prognosis.

In conclusion, HT+Epo improved survival without moderate-severe CP, preserved cerebellar growth rates, and improved many neurocognitive behavioral scores after perinatal asphyxia. There were no adverse events attributable to Epo, suggesting that Epo use is safe in the setting of HIE and HT. This translational study sets the stage for further studies of HT+Epo in human neonates with HIE. Future treatment for HIE will likely be multi-modal, with accumulating data advocating that Epo be included as one of these treatments.