Outcomes of extremely low birth weight infants given early high-dose erythropoietin

Abstract

Objective

To evaluate long-term outcomes of 60 extremely low birth weight (ELBW) infants treated with or without three injections of high-dose erythropoietin (Epo).

Study Design

A retrospective analysis of anthropometric and neurodevelopmental outcome data comparing 30 ELBW infants enrolled in a phase I/II study examining the pharmacokinetics of high-dose Epo (500, 1000 and 2500 U/kg × 3 doses) administered to 30 concurrent controls.

Result

Birth characteristics and growth from 4 to 36 months were similar for untreated and Epo-treated patients. Multiple linear regression analysis of neurodevelopmental follow-up scores from 17/25 Epo-treated and 18/26 control infants identified that Epo correlated with improvement of cognitive (R = 0.22, P = 0.044) and motor (R = 0.15, P = 0.026) scores. No negative long-term effects of Epo treatment were evident.

Conclusion

Retrospective analysis of the only available long-term follow-up data from ELBW infants given high-dose Epo treatment suggests that Epo treatment is safe and correlates with modest improvement of neurodevelopmental outcomes.

Introduction

A safe intervention is needed that can provide neuroprotection for neonates at risk for brain injury and neurodevelopmental impairment (NDI) due to complications of birth and prematurity. At present, cerebral palsy (CP) is present in 10 to 15% of all preterm survivors.¹ For the extremely preterm (born <26 weeks gestation), 50% exhibit mild to severe NDI at 30 months of age,² and NDI prevalence increases to 80% by the age of 6 years.³ Erythropoietin (Epo) is a promising neuroprotective therapy being considered for such high risk neonates.⁴ Epo has anti-apoptotic^5,6 and anti-inflammatory effects,^7,8 and it also increases neurogenesis^9–11 and protects oligodendrocytes.^12–14 These effects may counteract injury and improve neurodevelopmental outcomes in vulnerable preterm infants. Pragmatically, Epo is also attractive because it is widely available, affordable and safe for promoting erythropoiesis in preterm infants.

We previously completed a prospective, dose-escalation, open-label phase I/II study of high-dose Epo given to 30 extremely low birth weight (ELBW) infants to determine pharmacokinetics and short-term safety. Short-term outcomes were compared with 30 concurrent controls.¹⁵ We now report the long-term outcomes of these infants. This report contains the only available long-term outcome data for ELBW infants treated with early high-dose Epo.

Methods

Overview

We conducted a retrospective analysis of all available follow-up data from subjects previously enrolled in the parent study described below. Prior approval was obtained from the institutional review board at the University of Washington (Seattle, WA, USA). Developmental assessments were performed at the high-risk infant follow-up clinic and archived data were collected from paper charts and the electronic medical database (Mindscape).

Parent study

The parent study¹⁵ was a prospective, dose-escalation, open-label phase I/II evaluation of the acute safety and pharmacokinetics of early high-dose Epo treatment given to ELBW infants born <1000 g and ≤28 to 6/7 weeks gestational age. Sixty subjects were enrolled from January 2006 through March 2007. Thirty concurrent controls were compared with 30 Epo-treated subjects. Epo-treated subjects, but not untreated controls, had blood samples drawn for detection of plasma Epo pharmacokinetics. Early Epo treatment involved daily intravenous Epo injections for the first 3 days of age. Three doses of recombinant human Epo (500, 1000 or 2500 U/kg) were administered in the parent study (n = 10 per dose). After the early treatment period, late Epo treatment for anemia of prematurity was available to any infant (including control) at the discretion of the attending physician. The parent study was registered with the US Food and Drug Administration (IND 12656), and results were reported in Pediatrics.¹⁵

Follow-up

Participation in the follow-up program was voluntary. All assessments were conducted from August 2006 to June 2010 by certified examiners who were blind to prior treatment status. Evaluations included a standardized neurologic examination, anthropometrics (height, weight and head circumference) and determination of motor and cognitive performance using the Bayley Scales of Infant Development 2nd Edition (BSID-II) and Bayley Scales of Infant and Toddler Development, 3rd Edition (BSID-III). BSID-II MDI (mental developmental scores) scores were transformed to BSID-III cognitive scores using the recently published conversion formula 0.59 × MDI + 52.¹⁶ Developmental delay was defined as a cognitive or motor score <85 (1 s.d. below mean) or when severity of impairment prohibited testing. The severity of NDI was assessed for all infants that survived up to 12 months corrected age and was defined as moderate if any cognitive or motor score was ≤85, or severe if any score was ≤70 or if the diagnosis of CP, blindness or deafness was made. Hearing status was determined from parental reports and supplemented with the results of available audiologic examinations. Deafness was defined as disability that required amplification. Vision status was determined from a standard eye examination (tracking, nystagmus and roving eye movement), interviews, prior ophthalmologic examinations and assessment of electronic hospital records. Blindness was defined as a corrected visual acuity <20/200. Final evaluations of retinopathy of prematurity were made at the 12-month follow-up visit based on the international classification.¹⁷

Statistics

Data were analyzed using the SPSS software (SPSS, Chicago, IL, USA). χ² test was chosen to compare proportional data. Mixed-model analysis of variance (using within- and between-group factors) or multiple linear regression for main effects was performed to identify factors associated with neurodevelopmental scores. Both the standardized correlation coefficient (R) and the unstandardized regression coefficient B (which indicates the magnitude of the score change due to each factor) are presented.¹⁸ All comparisons were two-tailed, with α≤0.05. Some data are presented as means with s.e.m. or s.d. Owing to the limited number of available scores, data from infants given different early Epo doses are collectively evaluated as Epo-treated in the current analysis.

Results

In the parent study, 30 untreated control and 30 Epo-treated ELBW infants were enrolled, and 51 infants survived up to discharge (26 control and 25 Epo-treated). Neonatal survival was not affected by treatment,¹⁵ and no additional deaths occurred after discharge. Including mortalities, follow-up outcomes were identified for 49/60 infants (82%).

Thirty-nine of the 51 surviving subjects were examined at least once in the high-risk follow-up clinic. When testing began in August 2006, tests were exclusively BSID-II. After April 2007, BSID-III tests became available. There were 35 infants evaluated with BSID-II and/or BSID-III tests: 16 infants (eight control, eight Epo-treated) were exclusively evaluated with BSID-II tests. Nine infants (four control, five Epo-treated) were exclusively evaluated with BSID-III tests, and ten infants (six control, four Epo-treated) were evaluated at early ages with BSID-II tests and then at later ages with BSID-III tests. For combined analysis, all BSID-II MDI scores were transformed into BSID-III cognitive scores using a published formula calculated by correlating BSID-II MDI with BSID-III cognitive scores in both term and preterm infants.¹⁶ Table 1 compares the group birth characteristics and Table 2 compares the group cognitive, motor and language scores for the 35 infants who underwent developmental testing from 4 to 36 months post menstrual age (PMA).

Table 1.

Birth characteristics for the population of infants that underwent neurodevelopmental testing

	Control	Epo	P
N	18	17
Mean BW (g)	795	770	0.49
Mean GA birth (weeks)	26.6	26.1	0.14
Gender (% male)	55.6	58.8	0.85
Mean APGAR scores
1 min	3.9	3.8	0.86
5 min	6.1	6.2	0.85

Abbreviations: APGAR, appearance, pulse, grimace, activity, and respiration; BW, birth weight; Epo, erythropoietin; GA, gestational age.

Table 2.

Developmental follow-up test scores (mean ± s.e.m. (N)) for infants born ELBW and given high-dose Epo treatment (N = 17) for the first 3 days of age, or for untreated control infants (N = 18)

PMA (months)
	4	8	12	18	24	36
Cognitive a
Control	100.8 ± 3.6 (16)	98.8 ± 2.7 (9)	95.2 ± 5.6 (9)	84.9 ± 5.4 (4)	94.4 ± 3.3 (7)	95 ± 0 (1)
Epo	96.3 ± 3.9 (15)	101.2 ± 2.4 (10)	102.4 ± 1.4 (10)	97.5 ± 2.5 (2)	94.6 ± 5.6 (6)	90 ± 0 (2)
Motor b
Control	81.8 ± 3.4 (16)	72.8 ± 7.1 (9)	74.2 ± 6.8 (9)	81.3 ± 2.6 (4)	83.6 ± 7.9 (5)	94 ± 0 (1)
Epo	86.5 ± 2.3 (15)	73.7 ± 6.9 (10)	86.1 ± 5.2 (10)	89.5 ± 4.5 (2)	86 ± 3.7 (6)
Language c
Control	88.5 ± 5.5 (2)	89.3 ± 4.4 (4)	95.8 ± 5.3 (4)	79.7 ± 1.8 (3)	87.3 ± 8.2 (3)	86 ± 0 (1)
Epo	95.5 ± .9 (4)	88.5 ± 5.5 (2)	86.5 ± 2.1 (4)	88.5 ± 14.5 (2)	84 ± 2.5 (4)	93 ± 4 (2)

Abbreviations: ANOVA, analysis of variance; BSID, Bayley Scales of Infant Development 2nd Edition; ELBW, extremely low birth weight; Epo, erythropoietin; PDI, Psychomotor Developmental Index; PMA, postmenstrual age.

Mixed-model ANOVA detected interaction effects for Epo × PMA for cognitive (F = 2.9, P = 0.084) and motor scores (F = 4.3, P = 0.032), but specific post hoc comparisons were not significant. There were no effects on language scores.

^aIncludes BSID II MDI scores converted to BSID-III cognitive scores using the formula: (0.59 × MDI)+52.

^bIncludes BSID II and III PDI scores evaluated together (no conversion necessary).

^cBSID III tests only.

For all subsequent analyses, cognitive scores at 4 months post menstrual age were excluded because they rely heavily on motor response and thus can overestimate early cognitive function. Next, linear regression modeling was performed to correlate neurodevelopmental scores with factors including Epo treatment, gestational age at birth and intraventricular hemorrhage (IVH). Table 3 lists the correlations (R), unstandardized regression coefficients (B) and corresponding P-values for each of these factors. As shown, Epo treatment and gestational age at birth had positive associations, whereas IVH was negatively associated with outcomes. Notably, high-dose Epo was associated with a 5.3 point increase in cognitive score and a 7.5 point increase in motor score.

Table 3.

Linear regression analysis of neurodevelopmental scores

Score	Factor	R	B	P
Cognitive
	Epo	0.22	5.3	0.044
	GA birth	0.06	0.89	0.407
	IVH	−0.43	−11.0	0.001
Motor
	Epo	0.24	7.5	0.026
	GA birth	0.34	4.3	0.002
	IVH	−0.18	−7.5	0.084
Language
	Epo	0.10	0.36	0.917
	GA birth	0.07	0.09	0.947
	IVH	−0.47	−9.2	0.015

Abbreviations: B, unstandardized regression coefficient; Epo, erythropoietin; GA, gestational age; IVH, intraventricular hemorrhage; R, correlation coefficient.

The correlation coefficient (R) and unstandardized regression coefficient (B indicates the score change due to each factor) and corresponding P-value are shown.

To evaluate the severity of NDI in each group, we combined neuropathologic diagnoses (blindness, deafness and the diagnosis of CP) with the prevalence of low cognitive or motor scores, using scores of 85 and 70 (1 and 2 s.d. below the mean, respectively). Table 4 list the proportions of low cognitive and motor scores along with the combined NDI severity score. The largest proportion of control infants exhibited severe NDI, and an equally large proportion of Epo-treated infants were only moderately impaired. We considered whether severe NDI was associated with any grade of IVH.¹⁹ In the parent study, seven infants were diagnosed with IVH (confirmed by early and late cranial ultrasound), all seven survived, and six infants participated in the follow-up. Two Epo-treated infants had grade 3 IVH and NDI scores of moderate and severe. Four control infants had grades 1, 2, 4 and 4 IVH, and their respective NDI scores were moderate, severe, severe and severe, respectively. Table 4 shows that low motor scores were the predominant source of NDI for both groups of infants.

Table 4.

Severity of neurodevelopmental impairment (NDI)

	Epo	Control	Relative risk	P
Cognitive≤85	12% (2/17)	28% (5/18)	0.43	0.24
Cognitive≤70	0% (0/17)	6% (1/18)	0	0.32
Motor≤85	65% (11/17)	72% (13/18)	0.94	0.63
Motor≤70	29% (5/17)	42% (8/18)	0.66	0.37
Severity of NDI
None	33% (6/18)	32% (7/22)	1.05	0.92
Moderate NDI	33% (6/18)	32% (7/22)	1.05	0.92
Severe NDI	33% (6/18)	36% (8/22)	0.92	0.84

Abbreviations: Epo, erythropoietin; NDI, neurodevelopmental impairment.

Proportions (n/N) of cognitive and motor follow-up scores below 85 or 70, and corresponding severity of NDI in Epo-treated and control infants. The criteria for NDI were: moderate, any neurodevelopmental score below 85; severe, any score below 70 or cerebral palsy, blindness or deafness.

Typically, baseline plasma Epo concentrations are below detection levels (0.6 mU/ml) in neonates; however, high baseline values (>50 mU/ml) may indicate that hypoxemia was present before birth.²⁰ We found that five of the Epo-treated infants for whom follow-up data were available had high baseline plasma Epo concentrations. The mean (± s.d.) plasma Epo concentration for the five high-baseline infants was 204 ± 77 mU/ml (range 112 to 322) compared with 6 ± 10 mU/ml (range 0 to 33) for the other 12 low-baseline infants. There was no difference in the proportion of infants with moderate/severe NDI for infants with high baseline Epo (80% with NDI) compared with low baseline Epo infants (82% with NDI). In addition, baseline Epo was not associated with the incidence of severe retinopathy of prematurity (stage 3 or greater). Finally, 8/17 (47%) Epo-treated infants were also treated with late Epo for anemia of prematurity, as were 9/19 (47%) control infants. There was no difference in the proportion of infants with moderate/severe NDI for infants who received late Epo treatment for anemia of prematurity (70% with NDI) compared with those without exposure to late Epo (84% with NDI).

Anthropometric data (length and occipital–frontal circumference) and daily weights were available from 96% (25/26) of control and 100% (25/25) of Epo-treated patients. There were no effects of Epo on any of these growth indices (data not shown).

Discussion

The most notable findings are that high-dose Epo treatment positively correlated with both cognitive and motor scores, and Epo treatment did not increase adverse long-term outcomes in infants. This report contains the only long-term outcome data from ELBW infants exposed to early high-dose Epo, and these data help address concerns about the long-term safety of early Epo exposure.

The principal limitations of this study are its retrospective nature and the small sample size. Fortunately, the data were balanced with near equal numbers of control and Epo-treated infants. A very recent report correlated preterm BSID-II MDI with BSID-III cognitive scores to calculate the conversion formula¹⁶ that enabled us to compare all the neurodevelopmental scores. Nevertheless, the identified associations should be considered preliminary until a prospective long-term study is conducted using more uniform developmental testing and ensuring more complete follow-up participation. In the parent study, a standardized neurologic assessment of CP was not performed, and thus the prevalence of CP is not known. In addition, data on maternal education and patient race were not available, and these variables may have influenced patient outcomes.

A clear finding from the current analysis is that Epo treatment was not associated with adverse long-term effects. The parent study examined Epo pharmacokinetics and observed that early high-dose Epo was acutely safe as it did not increase risks for neonatal retinopathy of prematurity, hypertension, polycythemia or seizures.¹⁵ This is consistent with findings from a complementary trial of Epo given to preterm infants (24 to <32 weeks gestation), which also found no increase in short-term risks.²¹ These new long-term follow-up data complement those observations.

Clinical data that address possible neuroprotective benefits of early Epo treatment are beginning to accumulate. Although doses below 200 U/kg provide no neuroprotective benefit,²² erythropoietic dosing of ELBW infants using 400 U/kg Epo (x3/week) until 35 weeks corrected age are associated with improved MDI scores.^23–24 Similarly, in LBW preterm infants treated with Epo (250 to 400 U/kg × 3/week × 6 weeks) to prevent anemia of prematurity, MDI scores correlate with cumulative Epo exposure.²⁵ Also, ELBW neonates treated for anemia of prematurity at birth (250 to 500 U/kg Epo) and evaluated at 10 years of age were more likely to exhibit normal development (55% vs 39%) and their intelligence quotient scores were higher (90.8 vs 81.3) when compared with untreated controls.²⁶ In a prospective trial of repeated Epo (300 to 500 U/kg Q48 h x14 days), NDI was decreased in term infants with moderate to severe neonatal encephalopathy.²⁷ In another prospective study of term infants with neonatal encephalopathy, daily Epo treatment (2500 U/kg × 5 days) acutely decreased serum nitric oxide concentrations and reduced neurologic and developmental abnormalities at 6 months of age.²⁸ Despite these encouraging findings, the most suitable Epo dosing strategy for effective neuroprotection in preterm neonates is still undetermined. Therefore, a large randomized controlled trial is needed to determine the neuroprotective efficacy of high-dose Epo for either term infants with neonatal encephalopathy or ELBW infants at high risk for NDI.