Is Erythropoietin Combining with Therapeutic Hypothermia an Ecient and Safe Therapy in Neonatal Hypoxic Ischemic Encephalopathy: A Prospective and Randomized Clinical Trial


 BackgroundTo evaluate the efficacy and safety of erythropoietin (Epo) combined with therapeutic hypothermia (TH) in neonatal hypoxic-ischemic encephalopathy (HIE).MethodsA total of 78 term infants with HIE were assigned randomly to receive Epo (n = 40) or placebo (n = 38). All infants received TH. Blood samples before TH, after TH and after Epo/placebo were collected for measuring TH associated adverse events, Epo associated factors and potential neural biomarkers. Basal ganglia/ watershed (BG/W) scoring system was used to assess brain injury in MRI. Neurodevelopmental evaluations were performed at 18 months by using BayleyScales of Infant Development II (Bayley II).ResultsEpo-treated group tend to have lower serum creatine kinase (CK) concentration (114 vs 202, P = .04) and higher serum K+, Mg2+ concentration (5.0 vs 4.5, P = .03; 1.0 vs 0.9, P = .02) than control group after intervention. Brain MRI was performed in 65 (83%) neonatal. Totally brain injury score was in even distribution between two groups (median, 0 vs 0, P = .61), but injury region in cortex plus basal nuclei comparing with in basal nuclei solely was less common in the Epo than in the control group (21% vs 31%, P = .046). Only forty patients (40/78, 51%) succeeded in achieving 18-month follow up data. The totally adverse outcomes were trend to decline in the Epo group (35% vs 60%, P = .21). No adverse events were ascribed to Epo treatment.ConclusionsThe combination of Epo and TH is proved to be feasible, safe and potential effective.Trial registration: ChiCTR-TRC-14004532, date of registration: April 18th, 2014.


Background
Hypoxic-ischemic(HI) induced brain injury at newborn remains a concernful problem, affecting 26 per 1,000 live births in developing countries, and contributing to an estimated 22% of deaths in neonatal period. (1) In the vast territory of China, there are different level of hygiene between urban and rural areas, caused hypoxic-ischemic encephalopathy (HIE) is still a main cause of neonatal death. The traditional treatment for HIE is limited to supportive methods, to maintain physiologic signs, which played restricted role in decreasing mortality and improving neurodevelopmental prognosis. Multicenter clinical trials assessing the e cacy of therapeutic hypothermia (TH) demonstrated improved outcomes for term neonates with moderate HIE, and now TH is the standard of care in many countries. (1,2) Although TH is clearly bene cial in reducing the risk of death or major neurodevelopmental disability from 60 ~ 65-50%, (2,3) over 40% infants with moderate/ severe HIE received this therapy still experienced disabilities, including cerebral palsy, cognitive impairment and epilepsy. (2,4) Adjuvant therapies to further inspiring outcomes of HIE are imperatively needed.
Epo, a glycoprotein used widely in treating anemia, is a rmed to be neuroprotective medicine for HIE. As Epo binds to receptor, activation of intracellular signaling pathways may trigger a function in preventing apoptosis of cell.(5) Other short-term neuroprotective mechanisms contained neurotrophic, anti-oxidant and anti-in ammatory. Epo also may be responsible for brain repair through erythropoiesis, angiogenesis, oligodendrogenesis and neurogenesis. (6,7) A phase trail demonstrated Epo at 1000U/kg per dose given intravenously in conjunction with TH is safe and produces plasma concentrations that are similar neuroprotective in animals. (8) Continuously phase trail concluded that Epo combing TH for HIE infant may result in less MRI brain injury and improved motor function at 12-month, but when the researchers removed two patients who had been met with excluded criterias, the improvement in motor function attributed by Epo were not statistically signi cant. (9) Thus, the long-term function of Epo is still need further study. Rarely study mentioned the change level of potential neurological prognosis associated serum biochemical markers. Based on our previcous Chinacooling RCT study(10), we had gone further to conduct a prospective, double-blinded, placebo-controlled, phase trial, exploring easy-measuring potential neurological prognosis associated biomarkers and using a new kind of MRI scoring system, to determine whether Epo as an adjuvant therapy to TH is effective and safe.

Methods
In this prospective, randomized, double-blinded, placebo-controlled trial, we enrolled 78 newborns at > 36 weeks gestational age diagnosed HIE according to standard from pediatrics section of the Chinese Medical Association, which was similar to our previous RCT study protocol. The randomization codes were generated by computer and supplied in numbered, sealed envelopes. Participants were from hospitalized during May 1, 2014 to May 1, 2018 on Department of Neonatology, Children's Hospital of Fudan University. This study was approved by the Ethics Committee. Written informed consent was obtained from the parents.

Hypothesis
We hypothesized that high doses of Epo (1000 U/kg per dose) given combined with TH will release the level of biomarkers associated with poor prognosis, decrease the brain injury in MRI, improve the longterm neurological outcome and be safe.
Withdrawal standards: (1) intractable pulmonary hypertension; (2) hypotension (MAP ≤ 30 mmHg) insensitive to treat; (3) severe arrhythmia; (4) platelet count < 50 × 10 9/L ; (5) oliguria (urine volume < 0.5 mL/kg/h continuous for 24 hours and progressive aggravated azotemia); (6) severe infections with clinical and laboratory evidence; (7) scleroderma; (8) parents/legal guardians refuse to continue. Intervention Within 6 hours after birth, participants began TH continuously for 72 hours. Epo group were given 1000U/kg per dose human recombinant erythropoietin (r-Hu-Epo)intravenously every other day for two weeks and the placebo group given an equal volume of physiological saline. The basal treatment and supportive care were identical between two groups: intravenous nutrition to meet the energy requirement; when seizures occurred, phenobarbital was used rst at 20 mg/kg, with maintenance dose of 5 mg/kg.d ; dopamine, fentanyl or antibiotics were utilized when necessary.

Potential neurological biomarkers
Participants will have at least two blood samples (before TH and after TH) for potential neurological biomarkers serum level detecting. Reviewed previous study, we picked up several potential biomarkers which were common detected in clinical uses including: creatine kinase (CK), lactate dehydrogenase (LDH), Na +, K + , Mg 2+ and Ca 2+ . CK, usually used in judging myocardial ischemic injury, have also been found that high level was associated with serious brain damage. (14) LDH, one of the important enzyme participating in anaerobic glycolysis and gluconeogenesis, have been demonstrated the level was signi cantly higher in the HIE group. (15) Besides, HI triggers the disruption of ionic homeostasis, characterizing by enhanced K + e ux and Na + , Ca 2+ in ux, which has been considered as the most important alteration which eventually leads to cell death or injury. (16,17) Evidences showed Mg 2+ may keep cell membrane stabilization, inhibits free radical production and decreases secondary in ammation. (18,19) The laboratory technician was blinded to the group distribution. Neuroimaging Neonatal brain MRI was performed within 14 days of age. The BG/W score system, obtained basing on combining damage region, with the rst-echo T2-weighted sequence, was the useful score for predicting motor outcome at 3 and 12 months and cognitive outcome at 12 months. The score ranged from 0 to 4, and higher scores were given for more extensive damage. Detailed scoring criteria as below: 0 = normal; 1 = abnormal signal in basal nuclei (BG or thalamus); 2 = abnormal signal in cortex; 3 = abnormal signal in cortex and basal nuclei (BG or thalamus); 4 = abnormal signal in entire cortex and basal nuclei. (20) Two central reviewers after training were blinded to treatment allocation, independently scored brain MRI outcome.

Neurodevelopmental Outcomes
Forty patients (40/78, 51%) succeeded in achieving 18-month follow up data. In the Epo group, three infants died after discharged from hospital at the age around three days to one month. In the placebo group, three patients died during their hospitalization (two of persistent pulmonary hypertension, one of multiple organ failure) and one died after left hospital for one week. The Bayley II developmental scale results showed that the Epo group were likely to perform better in MDI score (mean, 88 vs 81, P = .23) and PDI score (mean, 84 vs 78, P = .29) The totally adverse outcomes were trend to decline in the Epo group (35% vs 60%, P = .21) ( Table 4).

Safety Observation
Participants will have at least three blood samples (prior to TH, end of TH approximately 4d after birth and nished time of Epo/placebo transfusion). We detected serum concentrations of sodium, potassium, calcium, blood sugar, aspartate aminotransferase (AST), alanine aminotransferase (ALT), creatinine (Cr), blood urea nitrogen (BUN), prothrombin time (PT), activated partial thromboplastin time (APTT) and serum total bilirubin. Complete blood counts (CBC) were collected through peripheral blood. Continuous noninvasive blood pressure monitoring and anal temperature data were collected. Hyponatremia is de ned as serum sodium lower than 130 mmol/L; hypokalemia is de ned as serum potassium lower than 3.5 mmol/L; hypocalcemia is de ned as blood calcium lower than 1.8 mmol/L or free calcium lower than 0.9 mmol/L; hypoglycemia is de ned as blood sugar lower than 2.6 mmol/L; hyperbilirubinemia is de ned as serum total bilirubin over than 221umol/L; liver dysfunction is de ned as level of AST over than 200 IU/L or ALT over than 100 IU/L; renal dysfunction is de ned as level of Cr over than 88ummol/L or BUN over than 7.5 mmol/L; coagulation disorder is de ned as PT delay over than 3 s or APTT delay over than 10 s; thrombocytopenia is de ned as platelet lower than 100 000 per ul; polycythemia is de ned as hematocrit over than 60%; hypotension is de ned as mean arterial pressure (MAP) lower than 40 mmHg or need vasoactive drugs containing; hypothermia is de ned as rectal temperature lower than 33.5℃ over 1 h; hyperthermia is de ned as rectal temperature higher than 37.5℃ over 1 h. The laboratory technician was blinded to the study group.

Statistical analysis
For accurately analysis of treatment effect, intention-to-treat strategy was used. We analyzed the data by using SPSS software (SPSS Inc, Chicago, IL). χ 2 test or Fisher's exact test (when the count was ≤ 5) were used when compared categorical variables. If continuous variables meet normal distribution, data were expressed as mean ± SD and counted in t test, otherwise median and inter-quartile range (IQR) were used and used Wilcoxon rank sum test for analyzing. Potential biomarkers comparison between two groups were used 2-sided T test or Mann-Whitney U test as appropriate. P values of < .05 (2-sided) were considered statistically signi cant.

Results
A total of 137 neonates were assessed and 78 (57%) approached for consent. Inclusion participants randomly divided to receive Epo with hypothermia (n = 40) or hypothermia alone (n = 38) (Fig. 1). The two groups were similar in demographic and clinical characteristics (Table 1). Blood gas values and clinical seizures occurring rates were similar at enrollment. All neonates were used whole body cooling for 72 hours, and received cooling at a mean age of 4.8 hours (SD, 4.6) after birth. The rst study drug was administered at a mean age of 7.5 hours (IQR, 5.0-16.4) after delivery. Infants received an average of 3.8 doses (SD, 1.9) of Epo. Four patients completed all 7 doses; the remaining did not caused by withdrawn from the trial (n = 3), lost intravenous access (n = 8), and discharged to home before 14 days (n = 25).

Changes Of Potential Biomarkers
We obtained blood samples at before hypothermia (an average age of 7 hours) and after hypothermia (an average age of 82 hours) from participants. Average serum levels in the rst available sample from patients of CK and LDH were equal between two groups, but when come to the second sample the CK level was lower in patients with Epo when compared with the controls (P = 0.04). As to ionic storm, the level of Na + , K + , Mg 2+ and Ca 2+ were equally distributed between two groups at enrollment, while after intervention, K + and Mg 2+ levels of Epo group were higher than control group (

Neuroimaging Results
Brain MRI was performed in 65 (83%) neonatal at a mean age of 7.6 days (SD, 4.8). In the Epo group, 33 infants had a brain MRI examination and all had implemented at least 3 doses of the study drug. BG/W score on image results ranged from 0 to 4 (IQR, 0-3). Totally brain injury score was in even distribution between two groups (median, 0 vs 0, P = .33) ( Table 3). There was a trend toward brain injury more localized in basal nuclei (not extend to cortex) among infants who received Epo. Injury to cortex plus basal nuclei comparing with in basal nuclei solely was less common in the Epo than in the control group (21% vs 31%, P = .046).
a P values based on 2-sided T test.

Disscussion
This is the rst clinical trial exploring potential biomarkers including CK, LDH, Na + , K + , Mg 2+ , Ca 2+ and a new MRI score system in Epo combined with TH. Our results add to the evidence that Epo in conjunction with TH was associated with hopeful biochemical markers', neuroimagic and neurodevelopmental improvements, including decreased the serum level of CK, increased serum concentrations of K + and Mg 2+ , localized MRI injury and potentially decreased adverse outcomes of neurodevelopment in long term.
Epo receptor are expressed on numerous cell types of the central nerve system, including neurons, astrocytes, microglia, and oligodendrocytes. (22) Epo gene expression in the brain tissue is regulated by hypoxia-inducible factor-1 which can activated by ischemia. (23) As Epo binds to receptor, intracellular signaling pathways are triggered. Activation of the Janus kinase-2 (Jak2), nuclear factor kappa B (NF-B), and mitogen-activated protein kinase (MAPK) pathway may play a role in preventing apoptosis of cell. (5,24) Evidence suggested that multiple doses of Epo, 250/500/1000/2500 U/kg per dose was tolerable.
(25) Our study was consistent with previous trial that no adverse events were attributed to Epo in 1000 U/kg per dose through intravenously. There is a physiological RBC destroyed in neonates during the several days after birth, which caused hemoglobin decreased. We found that epo group improved in RBC and hemoglobin count indicates that the decrease can be compensated by Epo. Previous trail showed one patient developed deep vein thrombosis though not ascribed to Epo nally.(9) During our study, no patients developed polycythemia and venous thromboses.
Unlike previous study, we focused on several potential neurological outcome related biomarkers which were widely detected in routine clinical work. Evidence revealed high serum level of CK and LDH was contributed to poor outcome in HIE neonates. (14,26) Our results showed higher level of CK decreased obviously in both group after treatment, while average level of CK in Epo group was closer to normal value than placebo group. Evidently, Epo combined with hypothermia plays an augment effect in reducing tissue damage. While LDH level decreased signi cantly after intervention in both group, but the percentage change of LDH made no difference between two groups. The widespread distribution in tissues of LDH caused low speci city of LDH when brain damaging may be responsible for the indifference.
Besides biomarkers in serum, HI triggers ironic storm which may contributes to cell death or injury and is characterized by enhanced K + e ux and Na + , Ca 2+ in ux. (16) Actually, comparing with the harmful role of Ca 2+ and Na + overload in HI, the function of K + e ux is complicated and unclear. Previous study thought the enhanced K + e ux was an adaptive mechanism by reducing cell excitability during energy depletion. (27) In our results, the serum K + level in Epo group, no matter comparing within group or with the control group, had increased signi cantly. The serum level of extracellular Ca 2+ increased after Epo implemented. We supposed the Epo may play a role in releasing intracellular Ca 2+ and K + overload, but the mechanism need further research. Mg 2+ , regarded as neuroprotective strategy, was ascended in the Epo group, which may predict a positive neuroprotection function.
We used a MRI scoring system which were different from those used in other studies. (9,28,29)In our study, the Epo group 33 infants had a brain MRI examination and all had implemented at least 3 doses of the study drug before examination. Although totally brain injury was in even distribution between two groups, injury to cortex plus basal nuclei comparing with in basal nuclei solely was less common in the Epo than in the control group (21% vs 31%, P = .046). Commonly, there are two categories of term infant HIE brain damage in MRI image. One is characterized as BG (ganglia or thalamus) damage, which is usually due to acute HI and relates to the worst outcomes of motor and cognitive function. The other is watershed region damaging in dominantly, which is most occurring after long-term incomplete asphyxia and is relates to the defect of cognitive function.(30) Epo may play a role in decreasing damage spread to cortex through improving arteries' perfusion.
The loss of follow-up is the most important limitation in our study. Despite our best efforts, only forty patients (40/78, 51%) succeed in achieving 18-month follow up in the end. The remaining did not because of parents refused gone back to hospital for assessment at 18-month (n = 16), children resided out of Shanghai (n = 10), parents' reserved phone number was invalid(n = 8), and chose follow up in other hospital (n = 4). Although had not reached a signi cant difference between two groups, the results of Bayley II developmental scale showed that the Epo group were likely to perform better in MDI score (mean, 88 vs 81, P = .23) and PDI (mean, 84 vs 78, P = .29) score and totally adverse outcomes were trend to decline in the Epo group (35% vs 60%, P = .21).
Besides the defect of follow-up data, the relatively small scale of this trial is another important limitation. Our study need authentication in a bigger trial with an adequate sample size and multicenter to lesson bias, with a higher completion rate of follow-up data in long-term, with extender biomarkers for an accurately predicting outcomes in short term and with standard electrophysiological data collection.

Conclusions
Among HIE neonates receiving TH, high doses of erythropoietin (1000U/kg) administrated intravenously every other day over two weeks may result in improvement of tissue damage and ionic disequilibrium, less MRI brain injury and may lead to potential improved long-term neurological outcomes. Trial pro le. 3 died during hospitalization: 2 died of persistent pulmonary hypertension, 1 died of multiple organ dysfunction syndrome. Dropped out, parents requested participants be withdrawn from the trial.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.