Key Points
Question
In infants with moderate or severe hypoxic-ischemic encephalopathy (HIE) receiving therapeutic hypothermia, does high-dose erythropoietin administered in 5 doses over the first week after birth improve survival without neurodevelopmental disability?
Findings
In a multicenter randomized clinical trial conducted in 24 neonatal intensive care units in Australia, New Zealand, and Singapore that included 313 participants, death or moderate or severe disability occurred in 47 of 138 participants (34.1%) who received erythropoietin and in 41 of 143 participants (28.7%) who received placebo.
Meaning
In infants with moderate or severe HIE, high-dose erythropoietin administered over the first week after birth did not reduce death or disability, but there was no definite signal for harm.
Abstract
Importance
Therapeutic hypothermia (TH) is the standard of care to improve outcomes following neonatal hypoxic-ischemic encephalopathy (HIE), but mortality and morbidity remain high, prompting research into treatment adjuncts.
Objective
To assess whether erythropoietin (EPO) plus TH improves outcomes in infants with moderate or severe HIE.
Design, Setting, and Participants
The PAEAN (Preventing Adverse Outcomes of Neonatal Hypoxic Ischaemic Encephalopathy with Erythropoietin) study was a phase 3, multicenter, double-blinded, placebo-controlled randomized clinical trial conducted in 24 neonatal intensive care units in Australia, New Zealand, and Singapore. Infants 35 weeks’ gestation or older, less than 23 hours old, with perinatal depression (umbilical cord arterial pH <7.0 or base excess ≥12 mmol/L or, at 10 minutes, Apgar score ≤5 or still receiving resuscitation), and moderate or severe HIE (modified Sarnat criteria) who had commenced TH within 6 hours of birth were eligible for inclusion. Study recruitment was conducted between May 2016 and March 2021, with data collection completed in September 2024. Data analysis was performed from September 2024 to October 2025.
Intervention
Intravenous EPO, 1000 IU/kg (5 doses over the first week), or placebo.
Main Outcomes and Measures
The primary outcome was a composite of death or moderate or severe motor or cognitive disability using standardized neurological and developmental assessments at 2 years. Moderate or severe disability comprised motor deficit (diagnosis of cerebral palsy with Gross Motor Function Classification Scale score ≥2) or moderate or severe cognitive deficit using Bayley Scales of Infant Development, Third Edition. Secondary outcomes included death, disabilities alone, and safety.
Results
From 2016 to 2021, 313 infants were randomized (EPO: n = 156; placebo: n = 157). Baseline characteristics (including mean [SD] gestational age at birth: 39.4 [1.7] weeks vs 39.3 [1.6] weeks and ratio of moderate:severe encephalopathy: 78%:22% vs 77%:23%) and loss to follow-up (10.2% overall) were similar across both groups. There were 67 (42.9%) and 59 (37.6%) female infants, respectively. There was no significant difference in the primary outcome (EPO: 47 of 138 [34.1%] vs placebo: 41 of 143 [28.7%]; relative risk, 1.19; 95% CI, 0.84-1.68; P = .33), in secondary outcomes (death: 22 of 146 [15.1%] vs 18 of 149 [12.1%]; P = .45; cerebral palsy: 22 of 120 [18.3%] vs 22 of 127 [17.3%]; motor deficit: 13 of 120 [10.8%] vs 15 of 127 [11.8%]; cognitive deficit: 20 of 113 [17.7%] vs 16 of 118 [13.6%]), or in any safety outcomes.
Conclusions and Relevance
Per the results of this multicenter randomized clinical trial, EPO did not demonstrate any adjunctive effect to hypothermia, but no safety concerns were evident.
Trial Registration
anzctr.org.au Identifier: ACTRN12614000669695
This multicenter randomized clinical trial conducted in Australia, New Zealand, and Singapore assess whether erythropoietin plus therapeutic hypothermia improves outcomes in infants with moderate or severe hypoxic-ischemic encephalopathy.
Introduction
Neonatal hypoxic-ischemic encephalopathy (HIE) is a syndrome of neurological dysfunction ensuing from insufficient oxygen or blood flow to the fetal brain shortly before birth. In New Zealand and Australia, moderate or severe HIE is diagnosed in 1.05 per 1000 term and near-term infants, a low rate compared to many countries. Causes include acute intrapartum sentinel events, such as cord prolapse or placental abruption. Even with rapid intervention, interruption of placental gas exchange leading to hypoxic impairment of cardiac function and bradycardia may cause critical cerebral hypoxia and ischemia. Where there is no apparent sentinel event, other causes of placental dysfunction are implicated but may be difficult to prevent. For some infants, the cause remains enigmatic, and some causes may be genetic.
Because prevention is not always achievable, methods to mitigate HIE after birth are esssential. Together with skilled resuscitation and neonatal intensive care, therapeutic hypothermia (TH) has a strong evidence base for improving outcomes in moderate or severe HIE. Despite these measures, the risk of death or long-term neurodevelopmental impairments remains high, establishing an imperative to find therapies that enhance the effect of TH.
The glycoprotein cytokine erythropoietin (EPO) is produced in response to hypoxia, recruits erythropoietic progenitors to hematopoiesis, and promotes their survival, increasing red cell mass and oxygen-carrying capacity. EPO receptors have been reported in most tissues, albeit at low levels compared to bone marrow, indicating potential for effects throughout the body. EPO and EPO receptors are expressed in the central nervous system, especially in the developing brain, with brain EPO expression induced by hypoxia. Because of supportive preclinical research, various forms of recombinant EPO and biosimilar erythropoietic stimulating agents (ESAs) have been investigated for neuroprotection in newborns.
The PAEAN (Preventing Adverse Outcomes of Neonatal Hypoxic Ischaemic Encephalopathy with Erythropoietin) randomized clinical trial was designed in collaboration with investigators from the HEAL (High-Dose Erythropoietin for Asphyxia and Encephalopathy) trial to have almost identical eligibility criteria and identical study drug and dosing regimens. PAEAN was a phase 3, double-blinded, placebo-controlled, parallel-group, multicenter, superiority randomized clinical trial to determine, in late-preterm and term newborns receiving TH for HIE, whether repeated high doses of EPO alfa given over the first week reduced death and disability at 2 years of age. The hypothesis of the PAEAN study was that EPO given to these infants would improve neurodevelopmental outcome at 2 years of age, without significant adverse effects, compared to TH alone.
Methods
Setting, Ethical Approval, and Consent
Participants were recruited in 24 neonatal intensive care units (NICUs) in Australia, New Zealand, and Singapore between May 2016 and March 2021. The study was approved by Human Research Ethics Committees for all sites (Supplement 1) and was registered with the Australian and New Zealand Clinical Trials registry (ACTRN12614000669695) and on ClinicalTrials.gov (NCT03079167). Written parental consent was obtained for all participants. The study is reported according to Consolidated Standards of Reporting Trials (CONSORT) reporting guidelines.
Participants
The detailed protocol is appended (Supplement 1), and the statistical analysis plan is avilable in Supplement 2. Eligibility criteria were: inborn or outborn infants at 35+0 weeks’ gestation or greater who had 1 or more of the following indicators of depression at birth: Apgar score of 5 or less or receiving ongoing resuscitation (positive pressure ventilation or continuous positive airway pressure) at 10 minutes, or pH less than 7.00 or base deficit of 12 mmol/L or greater plus evidence of moderate or severe encephalopathy, defined at 1 to 6 hours after birth as at least 3 of 6 modified Sarnat criteria (Supplement 1) scoring as moderate or severe (or 2 criteria plus seizures at any time before randomization).
Additional criteria were TH (standardized 72-hour course) initiated by 6 hours and ability to start study treatment within 24 hours of birth, a parent aged 18 years or older, and anticipated availability for follow-up to 2 years of age. Exclusion criteria were contraindications to EPO, plan to administer EPO or other ESA within 2 weeks after birth, birth weight less than 1800 g, suspected major chromosomal or congenital anomalies, microcephaly (head circumference less than third centile), and infants for whom imminent withdrawal of life support was under consideration.
Screening and Recruitment
Newborn infants at risk for HIE were screened for eligibility by clinicians or research teams at participating hospitals or by transport or retrieval services before NICU admission.
Randomization and Blinding
Infants who met inclusion criteria and whose parents consented were centrally randomized through the National Health and Medical Research Council Clinical Trials Centre. Treatment allocation was 1:1, with randomization stratified by study site and severity of encephalopathy (moderate or severe by modified Sarnat score at 1-6 hours). Recruiting clinicians entered baseline details on an online form and then received a study number corresponding to numbered vials of EPO or placebo, both repackaged in identical vials before distribution to sites. Parents and all clinical and study personnel were blinded to group allocation until retrieval of all treatment and follow-up data.
Trial Interventions
Infants received EPO, 1000 units/kg (epoeitin alfa [Amgen]), or equivalent volume of sodium chloride, 0.9%, intravenously (IV). Five doses commenced within the first day from birth and then as close as possible to 24, 48, 96, and 144 hours after the first dose or until death, decision to withdraw life-sustaining treatment, or discharge from the NICU, if before the fifth dose. A protocol amendment allowed the last dose to be given 24 or more hours early if removal of IV access, discharge, or transfer was imminent. All sites used TH protocols comprising 72 hours whole-body cooling to a target temperature of 33.5 °C, with controlled rewarming.
Safety Monitoring
An independent data safety monitoring committee established a charter at trial commencement and then conducted interim analyses to assess 30-day safety data at 25%, 50%, and 75% of planned recruitment (Supplement 3). Stopping rules were predefined but not enacted.
Primary Outcome
The primary outcome was a composite at 2 years of all-cause death or survival with moderate or severe developmental deficit (motor or cognitive) based on standardized neurological and developmental assessments. Moderate or severe developmental deficit consisted of any of: motor deficit (any diagnosis of cerebral palsy [CP] plus any level of functioning ≥2 using the Gross Motor Function Classification Scale [GMFCS] score) or moderate or severe cognitive deficit (cognitive score <85 [ie, 1 standard deviation below the mean], assessed by the Bayley Scales of Infant Development, Third Edition [BSID-III]). Because of the COVID-19 pandemic, the permitted window for neurodevelopmental assessment was extended from 22 to 26 months to 36 months of age.
Secondary Outcomes
Secondary outcomes were death (from randomization to 2 years of age), and among survivors: CP; motor deficit (CP plus GMFCS); cognitive deficit (BSID-III or -IV cognitive score); respiratory support; nutritional support; cortical visual impairment; hearing impairment; epilepsy; and costs of health care and/or service utilization (Medicare, Pharmaceutical Benefits Scheme), all up to 2 years. Selected adverse events were recorded at 30 days after last study treatment (eTable 5 in Supplement 4).
Tertiary Outcomes
Tertiary outcomes included Global Dubowitz Optimality score before hospital discharge; death or moderate or severe encephalopathy at last Sarnat assessment; and shift in distribution of overall severity across the following 4 domains: normal, mild motor or cognitive deficit, moderate or severe motor or cognitive deficit, and death.
Statistical Analysis
Prior sample size calculation determined that 150 participants per treatment group would detect a 19% absolute risk reduction in the combined end point of death or severe or moderate motor or cognitive deficit assuming a control event rate of 46% (decrease from 46% to 27%), allowing for a 10% noncompliance or lost-to-follow-up rate with 90% power and a 2-sided type I error of .05. This effect was comparable to that seen in a pilot study and was considered to be clinically important. Analyses were by intention to treat except safety analyses, which included infants who received at least 1 study treatment dose. Where CP designation and GMFCS or BSID-III at 2 to 3 years were not available, other information collected up to 43 months was used by an outcome adjudication committee, comprising 4 investigators blinded to study group allocation, to impute outcomes where possible (eAppendix 1 in Supplement 4). This included review of pediatric and neurodevelopmental assessments, consideration of the Warner Initial Developmental Evaluation of Adaptive and Functional Skills, and parent questionnaires.
Baseline and treatment adherence data were summarized as counts and percentages for categorical data or as mean and SD or median and range or interquartile range for continuous data. Continuous variables were compared using t tests or the nonparametric Wilcoxon rank sum test, and categorical data were compared using χ2 tests or the conditional binomial exact test when expected data cell counts were small.
For dichotomous outcomes, relative risk (RR) and 95% confidence intervals using log-binomial regression were calculated. Proportional-odds regression was used to compare the ordered categorical tertiary long-term disability outcome (normal, mild, moderate or severe, and death). For continuous outcomes, the mean difference with 95% confidence intervals was calculated using t tests. All analyses were conducted without adjustment for covariates. A sensitivity analysis was performed for mortality using time-to-event analysis. Kaplan-Meier curves were compared between groups with the use of the log-rank test, and hazard ratios with 95% confidence intervals calculated by means of a Cox regression model.
A tipping point analysis was performed post hoc as a sensitivity analysis to evaluate the impact of departures from the missing-at-random assumption on the study conclusions (eAppendix 3 in Supplement 4).
A 2-sided significance level of .05 was used for the prespecified hypothesis test of the primary outcome and comparisons of predefined safety outcomes between the 2 groups. Analyses of secondary and tertiary outcomes were exploratory, and no adjustment for multiple comparisons was made; therefore, 95% confidence intervals for estimates of RRs, odds ratios, and mean differences should be interpreted cautiously. All analyses were conducted in SAS version 9.4 (SAS Institute).
Results
Participants
Between May 2016 and March 2021, 313 infants were enrolled at 18 sites in Australia, 5 in New Zealand, and 1 in Singapore (75%, 23%, and 2% of enrolled infants, respectively) (Figure 1; eTable 1 in Supplement 4). At 2 years, mortality status was unknown for 5.8% of infants, and the primary outcome data were unavailable for 10.2% (similar between groups).
Figure 1. Flowchart of Trial Profile.
aTwo infants with Warner Initial Developmental Evaluation of Adaptive and Functional Skills scores available at 19 and 20 months that did not indicate moderate or severe disability were imputed as alive at 2 years.
Baseline characteristics for the treatment groups were similar, including for the stratification vaiables of severity of HIE at enrollment (EPO: 78% moderate, 22% severe vs placebo: 77% moderate, 23% severe) and site (Table 1). For EPO and placebo groups, mean (SD) gestational ages at birth were 39.4 (1.7) weeks and 39.3 (1.6) weeks, and there were 67 (42.9%) and 59 (37.6%) female infants, respectively. EPO and placebo groups were similar for mean (SD) birth weight (3457 [575] g vs 3327 [552] g), inborn vs outborn status, onset of labor, mode of delivery, and presence of any predefined intrapartum sentinel event (46% vs 47%).
Table 1. Delivery and Postnatal Characteristics at Baseline by Study Groups.
| Characteristic | No./total No. (%)a | |
|---|---|---|
| Erythropoietin (n = 156) | Placebo (n = 157) | |
| Outborn | 93/156 (59.6) | 90/157 (57.3) |
| Onset of labor | ||
| Spontaneous | 84/156 (53.8) | 84/157 (53.5) |
| Induced | 42/156 (26.9) | 33/157 (21.0) |
| None (cesarean delivery) | 30/156 (19.2) | 40/157 (25.5) |
| Presentation | ||
| Vertex | 142/156 (91.0) | 135/156 (86.5) |
| Breech | 9/156 (5.8) | 18/156 (11.5) |
| Face or brow | 1/156 (0.6) | 0/156 |
| Transverse or shoulder | 4/156 (2.6) | 0/156 |
| Other | 0/156 | 3/156 (1.9) |
| Mode of delivery | ||
| Vaginal noninstrumental | 55/156 (35.3) | 43/157 (27.4) |
| Vaginal instrumental | 27/156 (17.3) | 26/157 (16.6) |
| Cesarean delivery | 72/156 (46.2) | 85/157 (54.1) |
| Other | 2/156 (1.3) | 3/157 (1.9) |
| Maternal temperature ≥38 °C during labor or <6 h before birth | 6/109 (5.5) | 7/104 (6.7) |
| CTG <6 h before delivery | ||
| Reassuring | 26/123 (21.1) | 26/117 (22.2) |
| Nonreassuring | 97/123 (78.9) | 91/117 (77.8) |
| Acute intrapartum events | ||
| Umbilical cord prolapse | 2/156 (1.3) | 6/157 (3.8) |
| Shoulder dystocia | 22/156 (14.1) | 15/156 (9.6) |
| Uterine rupture | 9/156 (5.8) | 9/157 (5.7) |
| Intrapartum hemorrhage | 13/154 (8.4) | 21/156 (13.5) |
| Umbilical cord around neck | 13/151 (8.6) | 20/151 (13.2) |
| Amniotic fluid embolism | 0/156 | 0/157 |
| Maternal cardiopulmonary arrest | 2/156 (1.3) | 1/157 (0.6) |
| Second stage of labor >2 h | 12/146 (8.2) | 9/148 (6.1) |
| Other | 49/155 (31.6) | 49/156 (31.4) |
| Perinatal details | ||
| Gestational age at birth, mean (SD), wk | 39.4 (1.7) | 39.3 (1.6) |
| Sex | ||
| Female | 67/156 (42.9) | 59/157 (37.6) |
| Male | 89/156 (57.1) | 98/157 (62.4) |
| Twin | 1/156 (0.6) | 4/156 (2.6) |
| Birth weight, mean (SD), g | 3457 (575) [available for n = 156] | 3327 (552) [available for n = 157] |
| Head circumference, mean (SD), cm | 34.7 (1.6) [available for n = 151] | 34.8 (1.9) [available for n = 154] |
| Apgar score <5 at 10 min | 63/141 (44.7) | 56/140 (40.0) |
| Ongoing resuscitation at 10 min after birth | 134/155 (86.5) | 134/153 (87.6) |
| Blood pH (umbilical cord or infant <1 h after birth), mean (SD) | 6.94 (0.18) [available for n = 146] | 6.92 (0.18) [available for n = 146] |
| Base excess (umbilical cord or infant <1 h after birth), mean (SD), mmol/L | 16.8 (6.1) [available for n = 146] | 17.5 (5.9) [available for n = 140] |
| TH | ||
| Time from birth to commencing TH, median (IQR), h | 1.5 (0.7-2.8) [available for n = 156] | 1.8 (0.9-3.5) [available for n = 156] |
| Time to reach target temperature after TH commenced, median (IQR), h | 2.0 (1.0-3.5) [available for n = 156] | 1.5 (0.8-3.0) [available for n = 153] |
| Manual control (with or without cold packs) | 23/156 (14.7) | 29/157 (18.5) |
| Servo-controlled device | 133/156 (85.3) | 128/157 (81.5) |
| Level of encephalopathy at randomization | ||
| Moderate | 121/156 (77.6) | 121/157 (77.1) |
| Severe | 35/156 (22.4) | 36/157 (22.9) |
Abbreviations: CTG, cardiotocography; TH, therapeutic hypothermia.
Percentages may not total 100 because of rounding.
Mean (SD) 10-minute Apgar scores were 4.6 (2.4) vs 4.8 (2.2) for EPO and placebo groups, respectively, and mean (SD) pHs (cord blood or neonatal sample within 60 minutes) were 6.94 (0.18) vs 6.92 (0.18) in the respective groups. Of multiple births (EPO: n = 1; placebo: n = 4), only 1 infant from each multiple was included in the study. TH was commenced a median (IQR) of 1.5 (0.7-2.8) hours vs 1.8 (0.9-3.5) hours from birth, respectively, and time to target temperature and use of manual vs servo-controlled device were also similar (Table 1).
Adherence to Treatment Protocol
In both groups, more than 96% of participants received the first and second doses, and more than 93%, 85%, and 56% received the third, fourth, and fifth doses, respectively, while 49% and 52% received all 5 doses in the EPO and placebo groups, respectively (Table 2). Almost all doses (>99%) were within 90% of the dose specified in the protocol. No doses were withheld for suspected treatment-related adverse events. Nonadherence was most commonly due to decisions to remove intravascular cannulas (due to clinical recovery), death, or withdrawal of intensive life support, with no difference in distribution of causes by study arm (eTable 2 in Supplement 4).
Table 2. Study Treatment Adherence by Randomized Treatmenta.
| Adherence criterion | No./total No. (%) | |
|---|---|---|
| Erythropoietin (n = 156) | Placebo (n = 157) | |
| Day 1 | ||
| Treatment given | 154/156 (98.7) | 157/157 (100) |
| Time since birth treatment administered, mean (range), h | 18.3 (4.2-24.5) | 18.7 (5.8-28.0) |
| Dose received within 90% of protocol | ||
| Yes | 153/153 (100) | 155/156 (99.4) |
| Unknown | 1 | 1 |
| Dose given, mean (range), IU | 3392 (1860-4000) | 3281 (2100-4000) |
| Day 2 | ||
| Treatment given | 151/156 (96.8) | 154/155 (99.4) |
| Time since first dose treatment administered, mean (range), h | 24.1 (19.9-31.2) | 24.0 (20.0-26.7) |
| Dose received within 90% of protocol | ||
| Yes | 150/150 (100) | 152/153 (99.3) |
| Unknown | 1 | 1 |
| Dose given, mean (range), IU | 3393 (1860-4000) | 3275 (2100-4000) |
| Day 3 | ||
| Treatment given | 144/154 (93.5) | 149/153 (97.4) |
| Time since first dose treatment administered, mean (range), h | 48.0 (43.7-52.1) | 48.1 (43.7-50.8) |
| Dose received within 90% of protocol | ||
| Yes | 143/143 (100) | 148/148 (100) |
| Unknown | 1 | 1 |
| Dose given, mean (range), IU | 3404 (1860-4000) | 3291 (2100-4000) |
| Day 5 | ||
| Treatment given | 126/147 (85.7) | 133/149 (89.3) |
| Time since first dose treatment administered, mean (range), h | 96.0 (71.8-109.9) | 96.1 (72.0-119.8) |
| Dose received within 90% of protocol | ||
| Yes | 125/125 (100) | 132/132 (100) |
| Unknown | 1 | 1 |
| Dose given, mean (range), IU | 3385 (1860-4000) | 3317 (2100-4000) |
| Day 7 | ||
| Treatment given | 80/142 (56.3) | 82/146 (56.2) |
| Time since first dose treatment administered, mean (range), h | 142.4 (119.0-152.1) | 141.8 (118.3-168.0) |
| Dose received within 90% of protocol | ||
| Yes | 79/79 (100) | 81/81 (100) |
| Unknown | 1 | 1 |
| Dose given, mean (range), IU | 3408 (2400-4000) | 3278 (2100-4000) |
| Received all 5 doses | 76/156 (48.7) | 81/157 (51.6) |
Treatment was to be administered within 24 hours after birth, then as close as possible to 24, 48, 96, and 144 hours after the first dose. Infants who died are excluded from the denominator for later doses.
Mean (range) time to first study drug administration was 18.3 (4.2-24.5) hours for EPO vs 18.7 (5.8-28.0) hours for placebo.
Primary Outcome: Death or Survival With Moderate or Severe Developmental Deficit
The primary outcome was ascertained for 281 study participants (89.8%) and occurred in 47 of 138 participants (34.1%) in the EPO group vs 41 of 143 participants (28.7%) in the placebo group (RR, 1.19; 95% CI, 0.84-1.68; P = .33) (Table 3).
Table 3. Comparison of Primary and Secondary Outcomes at 2 Years of Agea.
| Outcome | No./total No. (%) | Group comparison, relative risk or difference in means (95% CI)b | |
|---|---|---|---|
| Erythropoietin | Placebo | ||
| Primary outcomec | |||
| Death or moderate or severe development deficitd,e | 47/138 (34.1) | 41/143 (28.7) | 1.19 (0.84 to 1.68) |
| Secondary outcomes | |||
| Deathf | 22/146 (15.1) | 18/149 (12.1) | 1.25 (0.70 to 2.23) |
| Cerebral palsy (yes or suspected)g | 22/120 (18.3) | 22/127 (17.3) | 1.06 (0.62 to 1.81) |
| Motor deficitd | 13/120 (10.8) | 15/127 (11.8) | 0.92 (0.46 to 1.85) |
| Cognitive deficite | 20/113 (17.7) | 16/118 (13.6) | 1.31 (0.71 to 2.39) |
| Respiratory supporth | 0/112 | 0/115 | NA |
| Nutritional supporti | 7/112 (6.3) | 7/115 (6.1) | 1.03 (0.37 to 2.83) |
| Major cortical visual impairment | 3/111 (2.7) | 3/115 (2.6) | 1.04 (0.21 to 5.02) |
| Hearing impairmentj | 6/111 (5.4) | 4/113 (3.5) | 1.53 (0.44 to 5.26) |
| Epilepsyk | 6/112 (5.4) | 6/115 (5.2) | 1.03 (0.34 to 3.09) |
| BSID-III composite score, mean (SD) | |||
| Cognitive | 99.1 (17.0) [available for n = 92] | 97.8 (16.8) [available for n = 94] | 1.4 (−3.5 to 6.3) |
| Language | 96.4 (19.7) [available for n = 86] | 94.0 (18.8) [available for n = 91] | 2.5 (−3.3 to 8.2) |
| Motor | 97.3 (18.8) [available for n = 89] | 96.0 (19.1) [available for n = 95] | 1.3 (−4.2 to 6.8) |
Abbreviations: BSID-III, Bayley Scales of Infant Development, Third Edition; NA, not applicable.
Developmental and health outcomes were assessed at 2 years only.
For binary outcomes, relative risks were estimated using log-binomial regression without covariate adjustment. For continuous outcomes, differences in means were estimated with t tests.
P = .33.
Moderate or severe motor deficit is defined as any case of cerebral palsy plus any level of functional impairment using the Gross Motor Function Classification Scale (GMFCS) ≥2. Includes 1 case of suspected cerebral palsy (GMFCS level 4) and 1 probable and 2 definite cases based on clinical review.
Moderate or severe cognitive deficit is defined as a cognitive score <85 as assessed by the BSID-III. Includes 7 probable and 1 definite event based on clinical review.
A 2-month window was used for 2-year survival (ie, babies known to be alive at 22 months are counted as 2-year survivors). An additional 2 infants with Warner Initial Developmental Evaluation of Adaptive and Functional Skills scores available at 19 and 20 months that did not indicate moderate or severe disability were imputed as alive at 2 years.
Seven cases of suspected cerebral palsy.
Includes tracheostomy, ventilator, high flow nasal cannula, continuous positive airway pressure, or oxygen dependency.
Includes gastrostomy, nasogastric feeds, parenteral feeds, and liquid diet.
Defined as the requirement for hearing aids (either diagnosis of hears well or with only a little difficulty with a hearing aid, has severe difficulty even with a hearing aid, or hearing is not helped with an aid).
Defined by a history of ≥2 afebrile, unprovoked seizures since discharge from the neonatal unit where study treatment was provided and up to 2 years of age, or use of anticonvulsants at 2 years of age.
Secondary Outcomes
The proportion of infants who died was 22 of 146 (15.1%) in the EPO group vs 18 of 149 (12.1%) in the placebo group (RR, 1.25; 95% CI, 0.70-2.23) (Table 3). Kaplan-Meier analysis showed no difference in timing of mortality by treatment group (Figure 2). Among survivors, there were no differences for the EPO vs placebo group in CP (22 of 120 [18.3%] vs 22 of 127 [17.3%]; RR, 1.06; 95% CI, 0.62-1.81), motor deficit (13 of 120 [10.8%] vs 15 of 127 [11.8%]; RR, 0.92; 95% CI, 0.46-1.85), or cognitive deficit (20 of 113 [17.7%] vs 16 of 118 [13.6%]; RR, 1.31; 95% CI, 0.71-2.39) (Table 3). No infants required supplementary respiratory support, and there were no differences in nutritional support (7 of 112 [6.3%] vs 7 of 115 [6.1%]), cortical visual impairment (3 of 111 [2.7%] vs 3 of 115 [2.6%]), hearing impairment (6 of 111 [5.4%] vs 4 of 113 [3.5%]), or epilepsy (6 of 112 [5.4%] vs 6 of 115 [5.2%]). The mean (SD) BSID-III (or IV) cognitive composite score was 99.1 (17.0) for the EPO group among 92 infants tested and 97.8 (16.8) for the placebo group among 94 infants tested (mean difference, 1.4; 95% CI, −3.5 to 6.3). There were no differences in language or motor composite scores.
Figure 2. Kaplan-Meier Estimates of Survival Until 2 Years in the Full Intention-to-Treat Population.
Data for all infants who did not die were censored; censoring is indicated by tick marks. HR indicates hazard ratio.
Tertiary Outcomes
Among children assessed, there was a small shift in the ordinal distribution of 2-year outcomes (no, mild motor or cognitive deficit, moderate or severe motor or cognitive deficit, or death), with increased odds of a worst outcome category in the EPO group; however, the estimate was imprecise (odds ratio, 1.38; 95% CI, 0.87-2.21). No material differences were observed in the combined outcome of death within 7 days, moderate or severe encephalopathy at last Sarnat assessment, or in Global Dubowitz Optimality Score assessed before hospital discharge (eTable 3 in Supplement 4). Other results, including those of magnetic resonance imaging, amplitude-integrated electroencephalography (aEEG), and General Movements Assessment substudies, will be reported separately.
Safety Outcomes
There was no difference between EPO and placebo groups in predefined safety outcomes within 30 days from last study treatment, including deaths (20 of 152 [13.2%] vs 17 of 155 [11.0%]; P = .56), number still in hospital (14 of 130 [10.8%] vs 10 of 137 [7.3%]; P = .32), or number still receiving oxygen (3 of 130 [2.3%] vs 3 of 135 [2.2%]; P = .94). There were no episodes of severe cardiorespiratory event within 2 hours of a study dose or pure red cell aplasia, polycythemia requiring treatment, or other adverse events resulting in treatment interruption or discontinuation. New thromboses of a major vessel occurred in 4 of 155 EPO participants (2.6%) vs 1 of 157 placebo participants (0.6%) within 7 days of dosing (P = .19), and in 1 of 155 (0.6%) vs 0 of 157 more than 7 days after last study dose (P = .46) (eTable 5 in Supplement 4).
Post Hoc Meta-Analyses for Major Thrombosis
In a post hoc meta-analysis, the results of the PAEAN, HEAL, and NEATO studies for major thrombosis were combined (eAppendix 2 in Supplement 4). Major thromboses occurred in 10 of 435 EPO participants (2.3%) vs 5 of 425 placebo participants (1.2%) (RR, 1.96; 95% CI, 0.68-5.70; P = .21; I2 = 0%).
Other Characteristics and Treatments of Infants During the Study Treatment Interval
No clinically important differences were seen between EPO and placebo groups in highest fraction of inspired oxygen (Fio2), respiratory support, anticonvulsant medication, or inotropic medication during study treatment. Peak hemoglobin, creatinine, prothrombin time, and activated partial thromboplastin time and lowest platelets also did not differ (eTable 4 in Supplement 4). The distribution of severity of encephalopathy by treatment day appeared similar between treatment arms (eFigure 3 in Supplement 4).
Discussion
Consistent with the HEAL trial, the PAEAN randomized clinical trial found that high-dose EPO administered early in the evolution of neonatal HIE to infants receiving TH did not improve the primary or any secondary outcome compared to placebo.
The reasons why EPO was ineffective are unclear. The dosing regimens for HEAL and PAEAN were based on a pharmacokinetic study, and a nested analysis within the HEAL study confirmed that target plasma levels were reached. It remains uncertain whether EPO achieved sufficient levels within the central nervous system or what those target levels should be. However, the dose regimen was selected to achieve plasma levels seen in animal models, where EPO was effective when administered systemically. Possibly, due to overlapping mechanisms of action, there was no incremental neuroprotection from EPO achievable beyond that of TH alone.
The diverse timing of hypoxic-ischemic brain injury might have meant that the optimal treatment window for EPO was missed. However, approximately half the enrolled infants had an intrapartum sentinel event, suggesting the study sample was not dominated by infants with slowly evolving or chronic injury.
Strengths and Limitations
Strengths of this study include its rigorous, preregistered study protocol, central randomization, (innocuous) placebo control, and blinding at all levels of the study. Limitations include that recruitment varied between regions and sites. The highest-recruiting site had a validated protocol to identify all eligible infants early. The COVID-19 pandemic contributed to a higher proportion than the anticipated 10% who missed standardized testing at 22 to 26 months. A post hoc tipping point analysis (eAppendix 3 in Supplement 4) assessing the impact of departures from the missing-at-random assumption indicated that the conclusions of the study were unlikely to change. Diligent efforts by study and site personnel meant that sufficient information was obtained for blinded adjudication of the primary outcome for 89.8% of participants, and of survival status for 94%, similar in the 2 arms. The primary outcome occurred in 28.7% of the placebo group, lower than the 46% estimate on which the sample size was based. While the observed event rate was lower than expected and limited precision, the combined evidence from HEAL and PAEAN rules out even moderate treatment benefits. Severity of HIE at baseline is a profound determinant of outcomes. Improved recognition of moderate HIE could partly explain the improvement in outcomes, as could more standardized TH and intensive care. These results may be helpful in counseling families whose infants have HIE requiring TH.
An individual patient meta-analysis combining similar trials might determine whether subgroups of infants who benefited can be recognized before deciding that no further clinical trials in infants receiving TH are needed. The HEAL and PAEAN studies confirm the value of large, high-quality, multicenter randomized clinical trials to confirm or refute efficacy, despite the results of promising preclinical and early phase trials. Conducting the 2 trials may have provided a more rapid path to a robust result than a single larger trial.
Conclusions
In conclusion, per the results of this randomized clinical trial, in infants with moderate or severe HIE undergoing TH, high-dose EPO administered over the first week did not reduce death or disability. There was no definite signal for harm, but adverse event rates were sufficiently low that clinical benefit or harm cannot be excluded.
Trial Protocol
Statistical Analysis Plan
Independent Data and Safety Monitoring Committee
eTable 1. Stratification of Enrollment (by Study Site of Recruitment and Severity of Encephalopathy at Enrollment)
eTable 2. Completion of Treatment and Reasons for Cessation of Treatment
eTable 3. Comparison of Tertiary Outcomes
eTable 4. Laboratory Assessments During the Study Treatment Period (Days 1-7), by Randomized Group
eTable 5. Selected Safety Outcomes
eFigure. Changes in Encephalopathy Severity From Days 1-7, by Randomized Treatment Group
eAppendix 1. Outcome Assessment Committee
eAppendix 2. Post Hoc Meta-Analyses for Major Thrombosis
eAppendix 3. Post Hoc Tipping Point Analysis
eAppendix 4. Acknowledgement of Gratitude to Study Site Research Nurses and Coordinators
PAEAN Study Investigators
Data Sharing Statement
References
- 1.Branagan A, Hurley T, Quirke F, et al. ; Steering Group for DEFiNE (Definition of Neonatal Encephalopathy) . Consensus definition and diagnostic criteria for neonatal encephalopathy-study protocol for a real-time modified Delphi study. Pediatr Res. 2025;97(1):430-436. doi: 10.1038/s41390-024-03303-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Chow SSW, Creighton P, Holberton JR, Chambers GM, Lui K. Report of the Australian and New Zealand Neonatal Network 2022. Accessed January 23, 2026. http://www.anznn.net
- 3.McIntyre S, Nelson KB, Mulkey SB, Lechpammer M, Molloy E, Badawi N; Newborn Brain Society Guidelines and Publications Committee . Neonatal encephalopathy: focus on epidemiology and underexplored aspects of etiology. Semin Fetal Neonatal Med. 2021;26(4):101265. doi: 10.1016/j.siny.2021.101265 [DOI] [PubMed] [Google Scholar]
- 4.Badawi N, Kurinczuk JJ, Keogh JM, et al. Intrapartum risk factors for newborn encephalopathy: the Western Australian case-control study. BMJ. 1998;317(7172):1554-1558. doi: 10.1136/bmj.317.7172.1554 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Turner JM, Mitchell MD, Kumar SS. The physiology of intrapartum fetal compromise at term. Am J Obstet Gynecol. 2020;222(1):17-26. doi: 10.1016/j.ajog.2019.07.032 [DOI] [PubMed] [Google Scholar]
- 6.Wang Y, Xu Y, Zhou C, et al. Exome sequencing reveals genetic heterogeneity and clinically actionable findings in children with cerebral palsy. Nat Med. 2024;30(5):1395-1405. doi: 10.1038/s41591-024-02912-z [DOI] [PubMed] [Google Scholar]
- 7.Jacobs SE, Berg M, Hunt R, Tarnow-Mordi WO, Inder TE, Davis PG. Cooling for newborns with hypoxic ischaemic encephalopathy. Cochrane Database Syst Rev. 2013;2013(1):CD003311. doi: 10.1002/14651858.CD003311.pub3 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Lee HC, Costa-Nobre DT, Katheria AC, et al. ; International Liaison Committee on Resuscitation Neonatal Life Support Task Force . Therapeutic hypothermia in low- and middle-income countries: a systematic review and meta-analysis. J Pediatr. 2026;288:114793. doi: 10.1016/j.jpeds.2025.114793 [DOI] [PubMed] [Google Scholar]
- 9.Suzuki N, Nakai T, Iwamura Y, Kato K. Hypoxia signaling in the cell type-specific regulation of erythropoietin production throughout mammalian development. Mol Cell Biol. 2025;45(9):386-394. doi: 10.1080/10985549.2025.2522720 [DOI] [PubMed] [Google Scholar]
- 10.Elliott S, Pham E, Macdougall IC. Erythropoietins: a common mechanism of action. Exp Hematol. 2008;36(12):1573-1584. doi: 10.1016/j.exphem.2008.08.003 [DOI] [PubMed] [Google Scholar]
- 11.Kowalczyk M, Banach M, Mikhailidis DP, Rysz J. Erythropoietin update 2011. Med Sci Monit. 2011;17(11):RA240-RA247. doi: 10.12659/MSM.882037 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Bernaudin M, Nedelec AS, Divoux D, MacKenzie ET, Petit E, Schumann-Bard P. Normobaric hypoxia induces tolerance to focal permanent cerebral ischemia in association with an increased expression of hypoxia-inducible factor-1 and its target genes, erythropoietin and VEGF, in the adult mouse brain. J Cereb Blood Flow Metab. 2002;22(4):393-403. doi: 10.1097/00004647-200204000-00003 [DOI] [PubMed] [Google Scholar]
- 13.Bernaudin M, Tang Y, Reilly M, Petit E, Sharp FR. Brain genomic response following hypoxia and re-oxygenation in the neonatal rat. identification of genes that might contribute to hypoxia-induced ischemic tolerance. J Biol Chem. 2002;277(42):39728-39738. doi: 10.1074/jbc.M204619200 [DOI] [PubMed] [Google Scholar]
- 14.Perrone S, Lembo C, Gironi F, et al. Erythropoietin as a neuroprotective drug for newborn infants: ten years after the first use. Antioxidants (Basel). 2022;11(4):652. doi: 10.3390/antiox11040652 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Wu YW, Goodman AM, Chang T, et al. Placental pathology and neonatal brain MRI in a randomized trial of erythropoietin for hypoxic-ischemic encephalopathy. Pediatr Res. 2020;87(5):879-884. doi: 10.1038/s41390-019-0493-6 [DOI] [PubMed] [Google Scholar]
- 16.Juul SE, Comstock BA, Heagerty PJ, et al. High-Dose Erythropoietin for Asphyxia and Encephalopathy (HEAL): a randomized controlled trial - background, aims, and study protocol. Neonatology. 2018;113(4):331-338. doi: 10.1159/000486820 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Hopewell S, Chan AW, Collins GS, et al. CONSORT 2025 Statement: Updated Guideline for Reporting Randomized Trials. JAMA. 2025;333(22):1998-2005. doi: 10.1001/jama.2025.4347 [DOI] [PubMed] [Google Scholar]
- 18.Palisano R, Rosenbaum P, Walter S, Russell D, Wood E, Galuppi B. Development and reliability of a system to classify gross motor function in children with cerebral palsy. Dev Med Child Neurol. 1997;39(4):214-223. doi: 10.1111/j.1469-8749.1997.tb07414.x [DOI] [PubMed] [Google Scholar]
- 19.Haataja L, Mercuri E, Regev R, et al. Optimality score for the neurologic examination of the infant at 12 and 18 months of age. J Pediatr. 1999;135(2 Pt 1):153-161. doi: 10.1016/S0022-3476(99)70016-8 [DOI] [PubMed] [Google Scholar]
- 20.Wu YW, Mathur AM, Chang T, et al. High-dose erythropoietin and hypothermia for hypoxic-ischemic encephalopathy: a phase II trial. Pediatrics. 2016;137(6):e20160191. doi: 10.1542/peds.2016-0191 [DOI] [PubMed] [Google Scholar]
- 21.Peyton C, Wroblewski K, Park J, et al. ; WIDEA-FS Network . Validity of the Warner Initial Developmental Evaluation of Adaptive and Functional Skills (WIDEA-FS): a daily activity criterion checklist for infants and toddlers. Pediatr Res. 2021;90(5):1052-1057. doi: 10.1038/s41390-020-01342-0 [DOI] [PubMed] [Google Scholar]
- 22.Wu YW, Comstock BA, Gonzalez FF, et al. ; HEAL Consortium . Trial of erythropoietin for hypoxic-ischemic encephalopathy in newborns. N Engl J Med. 2022;387(2):148-159. doi: 10.1056/NEJMoa2119660 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Frymoyer A, Juul SE, Massaro AN, Bammler TK, Wu YW. High-dose erythropoietin population pharmacokinetics in neonates with hypoxic-ischemic encephalopathy receiving hypothermia. Pediatr Res. 2017;81(6):865-872. doi: 10.1038/pr.2017.15 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Frymoyer A, Vasconcelos AG, Juul SE, Comstock BA, Heagerty PJ, Wu YW. On target dosing: erythropoietin exposure in neonates with hypoxic-ischemic encephalopathy in the HEAL trial. Pediatr Res. 2025;98(1):218-223. doi: 10.1038/s41390-024-03709-z [DOI] [PubMed] [Google Scholar]
- 25.Kellert BA, McPherson RJ, Juul SE. A comparison of high-dose recombinant erythropoietin treatment regimens in brain-injured neonatal rats. Pediatr Res. 2007;61(4):451-455. doi: 10.1203/pdr.0b013e3180332cec [DOI] [PubMed] [Google Scholar]
- 26.Wassink G, Davidson JO, Crisostomo A, et al. Recombinant erythropoietin does not augment hypothermic white matter protection after global cerebral ischaemia in near-term fetal sheep. Brain Commun. 2021;3(3):fcab172. doi: 10.1093/braincomms/fcab172 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Haakons K, Hocking K, Mausling R, Liley HG. Strategy to identify infants with hypoxic ischemic encephalopathy for therapeutic hypothermia-a retrospective audit. Children (Basel). 2025;12(7):892. doi: 10.3390/children12070892 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Natarajan G, McDonald SA, Shankaran S, et al. ; NICHD Neonatal Research Network . Prediction of 18 to 22 month neurodevelopmental outcomes using the numerical Sarnat Score compared with modified Sarnat staging in infants with moderate to severe hypoxic-ischemic encephalopathy. J Pediatr. 2025;281:114522. doi: 10.1016/j.jpeds.2025.114522 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Trial Protocol
Statistical Analysis Plan
Independent Data and Safety Monitoring Committee
eTable 1. Stratification of Enrollment (by Study Site of Recruitment and Severity of Encephalopathy at Enrollment)
eTable 2. Completion of Treatment and Reasons for Cessation of Treatment
eTable 3. Comparison of Tertiary Outcomes
eTable 4. Laboratory Assessments During the Study Treatment Period (Days 1-7), by Randomized Group
eTable 5. Selected Safety Outcomes
eFigure. Changes in Encephalopathy Severity From Days 1-7, by Randomized Treatment Group
eAppendix 1. Outcome Assessment Committee
eAppendix 2. Post Hoc Meta-Analyses for Major Thrombosis
eAppendix 3. Post Hoc Tipping Point Analysis
eAppendix 4. Acknowledgement of Gratitude to Study Site Research Nurses and Coordinators
PAEAN Study Investigators
Data Sharing Statement
