A randomized trial of parenteral nutrition using a mixed lipid emulsion containing fish oil in infants of extremely low birth weight: Neurodevelopmental outcome at 12 and 24 months corrected age, a secondary outcome analysis

Abstract

Objective

To examine whether parenteral nutrition using a mixed lipid emulsion containing fish oil improves the neurodevelopmental outcome of extremely low birth weight (ELBW) infants.

Study design

The study is a secondary outcome analysis of a double-blind randomized trial of 230 ELBW infants performed at a single level IV neonatal care unit (Medical University Vienna; 06/2012-06/2015). Participants received either a mixed lipid emulsion composed of soybean oil, medium chain triglycerides, olive oil, and fish oil, or a soybean oil-based lipid emulsion for parenteral nutrition. Neurodevelopment (Bayley-III) of study participants was assessed at 12 and 24 months corrected age (08/2013-10/2017).

Results

206 out of 230 study participants were eligible at discharge. At 12 and 24 months corrected age, 174 out of 206 (85 %) and 164 out of 206 (80 %) infants were analyzed. At 12 months, there was no significant difference (median [interquartile range]) in cognitive (mixed: 95 [85-101], soybean oil: 95 [85-100]; P=.71), language (mixed: 86 [77-94], soybean oil: 89 [79-94]; P=.48) or motor scores (mixed: 88 [76-94], soybean oil: 88 [79-94]; P=.69). At 24 months, there was again no significant difference in cognitive (mixed: 95 [80-105], soybean oil: 95 [90-105]; P=.17), language (mixed: 89 [75-97], soybean oil 89 [77-100]; P=.54) and motor scores (mixed: 94 [82-103], soybean oil: 94 [85-103]; P=.53).

Conclusion

Parenteral nutrition using a mixed lipid emulsion containing fish oil did not improve neurodevelopment of ELBW infants at 12 and 24 months corrected age.

Clinical trial registration

ClinicalTrials.gov: NCT01585935.

Introduction

Infants of extremely low birth weight (ELBW) are at high risk for adverse neurodevelopment due to morbidities directly affecting the developing brain, such as intraventricular hemorrhage or cystic periventricular leucomalacia. ¹ Preterm infants have distinct nutritional requirements, ² and optimal nutrient supply supports a favorable motor and cognitive development. ³

Docosahexaenoic acid (DHA) is an ω-3 long-chain polyunsaturated fatty acid (LC-PUFA) that is considered crucial for normal brain development. ⁴ In the last trimester of pregnancy an estimated amount of 40 to 67 mg/kg/d is transferred from the mother to the fetus. ^4–7 After preterm birth, placental transfer of DHA is interrupted. Soybean oil-based lipid emulsions that are used for parenteral nutrition are almost devoid of DHA. Thus, enteral nutrition is the main source of DHA for preterm infants, but supply falls short of fetal accretion, and infants of ELBW accumulate large DHA deficits.

A mixed lipid emulsion composed of soybean oil, medium-chain triglycerides, olive oil, and fish oil is licensed for pediatric use in Europe. Fish oil provides DHA and eicosapentaenoic acid (EPA), which is a precursor of DHA. Parenteral nutrition using the mixed lipid emulsion provides infants of ELBW with amounts of DHA comparable to in utero transfer rates. ⁸ By attenuating their DHA deficit ⁶ the neurodevelopment of preterm infants of ELBW might be improved.

In a recent randomized trial, we assigned 230 infants of ELBW to receive either a mixed lipid emulsion or a soybean oil-based lipid emulsion for parenteral nutrition to evaluate its effect on liver injury. ⁸ Analysis of amplitude-integrated electroencephalography in study participants measured from birth to discharge revealed acceleration of electrophysiological brain maturation using a mixed lipid emulsion. ⁹ The aim of the present study was to analyze the effect of a mixed lipid emulsion on the neurodevelopmental outcome of study participants at 12 and 24 months corrected age.

Methods

The study is a pre-specified secondary outcome analysis of a double-blind randomized trial (ClinicalTrials.gov: NCT01585935) on the preventive effect of a mixed lipid emulsion on parenteral nutrition associated cholestasis. The study was conducted from June 2012 to June 2015 at a level IV neonatal care unit (University Children’s Hospital, Medical University of Vienna, Austria). Study design was previously described in detail, ⁸ the protocol is accessible at ClinicalTrials.gov. Inclusion criteria included birth weight <1000 grams and admission within the first 24 hours of life. Higher order multiples, infants with chromosomal aberrations, and infants with conditions associated with cholestasis were not eligible. Participants were randomized and stratified (sex and birth weight <750 grams) within their first 120 hours of life to receive parenteral nutrition using either a mixed lipid emulsion (SMOFLipid 20%; Fresenius Kabi, Bad Homburg, Germany; composed of 30% soybean oil, 30% medium chain triglycerides, 25% olive oil and 15% fish oil; ω-6:ω -3 ratio 2.5:1) or a soybean oil-based lipid emulsion (Intralipid 20%; Fresenius Kabi, Bad Homburg, Germany; ω-6: ω-3 ratio 8:1). The mixed lipid emulsion contains 2.2% DHA, 2.4% EPA, and 0.4% arachidonic acid (ARA). The soybean oil-based lipid emulsion contains 0.2 % DHA, no EPA, and 0.3 % ARA. ^{10, 11} Parenteral and enteral nutrition of participants were previously described in more detail. ^{9, 10} Lipids were dosed up to 3 g/kg/day. Serum triglycerides were measured at least weekly. Full parenteral nutrition was provided until 140-160 ml/kg/day of enteral nutrition. Parenteral vitamins (2 ml/kg Soluvit; 4 ml/kg Vitalipid N Infant; Fresenius Kabi, Bad Homburg, Germany) were provided unless enteral nutrition with own mother’s milk or donor milk was fortified at 100 ml/kg using a fat-free fortifier (Aptamil FMS; Milupa Nutricia GmbH, Frankfurt, Germany). Donor milk was used if mothers’ own milk was unavailable and switched to preterm formula at 32 weeks of postmenstrual age. This follow-up study, investigating neurodevelopment at 12 and 24 months corrected age, was performed at the hospital’s outpatient clinic from August 2013 to October 2017. Infants with congenital cerebral malformations were excluded from this analysis. Patients and observers were blinded, as were the two psychologists performing neurodevelopmental testing.

Baseline characteristics

Demographic parameters at birth are shown in Table 1. A full course of prenatal steroids was defined as two doses of betamethasone. Surfactant (Curosurf; Chiesi, Parma, Italy) was administered prophylactically in infants <28+0 weeks of gestational age ¹² or else if deficiency was suspected. Anthropometry Z-scores were calculated using growth curves by Fenton et al. ¹³ Small for gestational age was defined as birth weight below the 10^th percentile.

Table 1. Demographic parameters.

Parameter	12 months corrected age		24 months corrected age
	Mixed lipid emulsion (n=86)	Soybean oil (n=88)	Mixed lipid emulsion (n=81)	Soybean oil (n=83)
Obstetric parameters
Multiple pregnancy	19 (22)	27 (31)	19 (24)	26 (31)
Cesarean delivery	79 (92)	82 (93)	75 (93)	78 (94)
Prenatal steroids (full course)	57 (66)	52 (59) c	53 (65)	50 (60) c
Premature rupture of membranes	32 (37)	31 (35)	30 (37)	31 (37)
Preeclampsia	12 (14)	16 (18)	11 (14)	16 (19)
Highest parental education
Primary school	29 (34) d	28 (32) g	28 (35) d	26 (31) g
Secondary school	26 (30) d	23 (26) g	26 (30) d	22 (27) g
Postsecondary school	22 (26) d	20 (23) g	21 (26) d	19 (23) g
Age of mother at birth	31.5 [27-35] a	31 [27-37] b	31.5 [27-35] a	31 [27-37]
Married/in partnership	71 (82.6) d	74 (84.1) d	67 (82.7) d	70 (84.3) d
Caucasian	83 (96.5)	84 (95.5)	79 (97.5)	79 (95.2)
Neonatal parameters
Umbilical artery pH	7.31 [7.26 to 7.36] f	7.3 [7.25 to 7.35] f	7.31 [7.26 to 7.36] f	7.31 [7.25 to 7.35] e
Apgar - 5 min	8 [8 to 9]	8 [8 to 9] a	8 [8 to 9]	8 [8 to 9] a
Male sex	51 (59)	57 (65)	48 (59)	56 (68)
Surfactant	77 (90)	76 (86)	72 (89)	72 (87)
Gestational age (wk+d)	25+5 [24+6 to 27+1]	26+2 [25+0 to 28+0]	25+6 [24+6 to 27+2]	26+2 [25+0 to 28+0]
Birth weight (g)	772.5 [650 to 835]	760 [630 to 895]	775 [650 to 855]	780 [630 to 900]
Z score	-0.4 [-1.1 to 0.2]	-0.7 [-1.4 to 0.1]	-0.4 [-1.1 to 0.2]	-0.6 [-1.3 to 0.1]
Birth length (cm)	34 [31 to 35]	33 [31 to 35] b	34 [31 to 35]	33 [31 to 35]
Z score	-0.2 [-0.9 to 0.7]	-0.2 [-1 to 0.4] b	-0.1 [-1 to 0.9]	-0.2 [-1 to 0.4]
Birth head circumference (cm)	24 [23 to 25]	24 [23 to 25]	24 [23 to 25]	24 [23 to 25]
Z score	0 [-0.7 to 0.6]	-0.2 [-0.8 to 0.3]	-0.1 [-0.7 to 0.6]	-0.2 [-0.7 to 0.4]
Small for gestational age	17 (20)	27 (31)	17 (21)	24 (29)

Categorical data are presented as numbers with percentages in round parentheses and were tested using the chi2 test. Continuous data are presented as the median and interquartile range in squared parentheses and were tested using the Mann-Whitney U test.

^adata of 1 patient missing

^bdata of 2 patients missing

^cdata of 3 patients missing

^ddata of 9 patients missing

^edata of 11 patients missing

^fdata of 12 patients missing

^gdata of 17 patients missing

Neonatal morbidity and nutritional parameters of infants during hospitalization are shown in Table 2. Necrotizing enterocolitis was diagnosed clinically (Bell’s stage ≥ Ila) or after exploratory surgery. Intraventricular hemorrhage and cystic periventricular leucomalacia were diagnosed by cerebral ultrasound, retinopathy of prematurity by direct ophthalmoscopy. Chronic lung disease was defined as supplementary oxygen after 36+0 weeks post menstrual age. Data on parental education were collected at the follow-up visits and divided into three groups according to highest education of either parent (primary education – i.e. compulsory schooling until the age of 15; secondary – i.e. high school and tertiary – i.e. postsecondary education). Low parental education was defined as primary education only.

Table 2. Neonatal outcome.

Parameter	12 months corrected age			24 months corrected age
	Mixed lipid emulsion (n=86)	Soybean oil (n=88)	P	Mixed lipid emulsion (n=81)	Soybean oil (n=83)	P
Hospitalization (d)	87 [70 to 109]	83 [69 to 99]	.40	86 [70 to 109]	82 [69 to 99]	.47
Cholestasis	8 (9.3)	13 (15)	.27	8 (9.9)	12 (15)	.37
Retinopathy of prematurity (any)	55 (64)	50 (57) a	.38	50 (62)	46 (55)	.41
Highest grade (grade 1-5)	1.5 [0 to 2] a	1 [0 to 2]	.50	1 [0 to 2]	1 [0 to 2]	.54
Requiring treatment (severe ROP)	9 (11)	9 (10) a	.98	8 (9.9)	8 (9.6)	.96
Sepsis, culture proven	21 (24)	22 (25)	.93	19 (24)	20 (24)	.92
Intraventricular hemorrhage III/IV	8 (9.3)	4 (4.5)	.22	8 (9.9)	4 (4.8)	.21
Cystic periventricular leucomalacia	3 (3.5)	2 (3.4)	.98	3 (3.7)	3 (3.6)	.98
Necrotizing enterocolitis (grade ≥IIa)	3 (3.5)	5 (5.7)	.49	3 (3.7)	4 (4.8)	.72
Focal intestinal perforation	2 (2.3)	3 (3.4)	.70	2 (2.5)	3 (3.6)	.67
Abdominal surgery	8 (9.3)	10 (11)	.66	8 (9.9)	10 (12)	.66
Days on mechanical ventilation	0 [0 to 11]	1 [0 to 9]	.64	0 [0 to 10]	1 [0 to 9]	.60
Chronic lung disease	17 (20)	17 (19)	.94	15 (19)	17 (21)	.75
Steroid treatment	9 (11)	14 (16)	.29	8 (9.9)	13 (16)	.27
PDA requiring treatment	47 (55)	58 (66)	.13	43 (53)	56 (68)	.06
Number of ibuprofen cycles	1 [0 to 2]	1 [0 to 2]	.26	1 [0 to 2]	1 [0 to 2]	.15
Surgical ligation	4 (4.7)	6 (6.8)	.54	3 (3.7)	6 (7.2)	.32
Pulmonary hypertension	17 (20)	20 (23)	.63	16 (20)	18 (22)	.76
iNO/sildenafil treatment	14 (16)	19 (22)	.37	13 (16)	17 (21)	.46
Time to full enteral feeds (d)	24 [18 to 38]	24 [18 to 35]	.52	24 [18 to 38]	24 [18 to 35]	.50
Parenteral lipids (d)	22 [16 to 36]	21 [17 to 31]	.73	23 [16 to 36]	21 [17 to 31]	.57
Cumulative amount of parenteral lipids (g/kg)	43 [28 to 70]	42 [30 to 57]	.44	43 [29 to 72]	42 [30 to 57]	.36
Study lipids (g/kg/d)	2 [1.6 to 2.2]	1.9 [1.6 to 2.1]	.28	2 [1.6 to 2.2]	1.9 [1.6 to 2.1]	.29
Docosahexaenoic acid (mg/kg/d)	43 [35 to 48]	3.8 [3.2 to 4.2]	<.0001	43 [35 to 48]	3.8 [3.2 to 4.3]	<.0001
Eicosapentaenoic acid (mg/kg/d)	47 [39 to 53]	0 [0 to 0]	<.0001	47 [39 to 52]	0 [0 to 0]	<.0001

Abbreviations: iNO, inhalative nitric oxide; PDA, persistent ductus arteriosus; ROP, retinopathy of prematurity

Categorical data are presented as numbers with percentages in round parentheses and were tested using the chi2 test. Continuous data are presented as the median and interquartile range in squared parentheses and were tested using the Mann-Whitney U test. P values <.05 were considered statistically significant.

^adata of 1 patient missing

Follow-up

Study participants were invited to follow-up visits at 12 and 24 months corrected age as part of clinical routine care. Follow-up included anthropometric measurements (with Z scores and body mass index) and assessment of neurodevelopment using the Bayley Scales of Infant and Toddler Development, third edition (Bayley-III; Harcourt Assessment, San Antonio, United States, 2006.). The Bayley-III is an instrument to measure the neurodevelopment of infants aged 16 days to 42 months and consists of five subtests: cognition, receptive, and expressive communication as well as fine and gross motor function. Scaled scores are calculated for each subtest that range from 1 to 19 with a mean of 10 and a standard deviation of 3 using normative data for the infant’s age group. Scores are converted into cognitive, language, and motor composite scores with a mean of 100 and a standard deviation of 15. ¹⁴ Results were calculated using the US norms. The tests were performed and scored by two certified clinical psychologists.

Infants with a diagnosis of cerebral palsy were additionally classified using the Gross Motor Function Classification System (GMFCS) at 24 months corrected age. The GMFCS includes five levels based on functional abilities and quality of movement. ¹⁵

Statistical analysis

Differences between groups were analyzed by Mann-Whitney-U test for composite scores of cognitive, language, and motor development as well as gross motor function. Subgroup analyses were carried out by sex and increased risk for unfavorable development (birth weight below 750 g and low parental education). Sensitivity analysis was performed using a mixed model linear regression analysis with mother and child as random factor and included the confounders sex, gestational age, intraventricular hemorrhage grade III and IV, and parental education. SPSS statistical software system version 25.0 (SPSS Inc., Chicago, IL) and R version 3.5.1 (R foundation, Vienna, Austria) were used for calculations. Assessment of neurodevelopmental outcome was a pre-defined secondary outcome of the trial, confounders and subgroup analyses were defined post hoc after analysis of the whole cohort.

Ethics and registration

The study was conducted in conformance with the “Declaration of Helsinki”, “ICH GCP guidelines”, and the respective European Union directives embedded in the Austrian drug act. Written consent from one parent was sufficient due to low risk for participants. Patients were insured as legally required. The study was approved by the institution’s ethics committee (EK 2011/1030) and registered at European Clinical Trial Database (EudraCT 2011-005456-33) and clinicaltrials.gov (NCT01585935).

Results

Screening

In total, 230 infants were randomized (06/2012 to 06/2015, Figure 1, online only) and 223 infants were eligible for analysis of the primary outcome. At discharge, 206 out of 230 study participants were eligible for neurodevelopmental assessment at follow-up, with 174 (85%) infants analyzed at 12 months (loss to follow-up 15%) and 164 (80%) infants at 24 months corrected age (loss to follow-up 20%). Follow-up lasted from 08/2013 to 10/2017. Loss to follow-up did not vary significantly by group (at 12 months: 16% vs. 15%, P=.95; at 24 months: 21 vs. 20 %, P=.94).

Figure 1. Figure 1 shows the patient flow chart with the reasons for study exclusion.

Demographic parameters & neonatal morbidity

Demographic parameters of infants analyzed at 12 and 24 months corrected age are shown in Table 1.

Neonatal morbidities and nutritional parameters did not differ significantly between groups, expect for DHA and EPA, which were significantly higher in infants receiving the mixed lipid emulsion due to the intervention (Table 2). Important characteristics of dropouts (Table 3, online only) did not differ significantly. Neonatal morbidities and nutritional parameters of the subgroup of infants born < 750 g (Table 4, online only) were also not significantly different.

Table 3. online only. Important demographic and outcome parameters of dropouts at 12 and 24 months corrected age.

Parameter	12 months corrected age			24 months corrected age
	Mixed lipid emulsion (n=27)	Soybean oil (n=28)	P	Mixed lipid emulsion (n=32)	Soybean oil (n=33)	P
Male sex	14 (51.9)	19 (65.5)	.23	17 (53.1)	20 (58.8)	.54
Gestational age (wk+d)	26+6 [24+3 to 28+4]	26+0 [24+3 to 28+0]	.44	26+4 [24+3 to 28+2]	26+0 [24+3 to 28+0]	.72
Birth weight (gram)	890 [605 to 960]	737.5 [590 to 843]	.05	818.5 [612 to 956.3]	700 [582.5 to 840.5]	.07
Z score	-0.6 [-1.2 to 0]	-0.45 [-1.98 to -0.1]	.72	-0.5 [-1.1 to 0]	-0.7 [-1.9 to -0.1]	.33
Intraventricular hemorrhage III/IV	4 (14.8)	6 (20.7)	.53	4 (12.5)	6 (17.6)	.53

Categorical data are presented as numbers with percentages in round parentheses and were tested using the chi2 test. Continuous data are presented as the median and interquartile range in squared parentheses and were tested using the Mann-Whitney U test. P values <.05 were considered statistically significant.

Table 4. online only. Neonatal outcome in infants born <750g.

Parameter	12 months corrected age			24 months corrected age
	Mixed lipid emulsion (n=35)	Soybean oil (n=40)	P	Mixed lipid emulsion (n=32)	Soybean oil (n=36)	P
Hospitalization (d)	104 [85 to 115]	91.5 [82.3 to 113]	.313	104 [85.3 to 115.8]	91.5 [79.8 to 116]	.357
Cholestasis	3 (8.6)	10 (25)	.061	3 (9.4)	9 (25)	.092
Retinopathy of prematurity (any)	30 (85.7)	28 (70)	.105	27 (84.4)	24 (66.7)	.092
Requiring treatment (severe ROP)	6 (17.1)	8 (20)	.751	5 (15.6)	7 (19.4)	.68
Sepsis, culture proven	10 (28.6)	15 (37.5)	.413	9 (28.1)	13 (36.1)	.482
Intraventricular hemorrhage III/IV	5 (14.3)	3 (7.5)	.342	5 (15.6)	3 (8.3)	.352
Cystic periventricular leukomalacia	1 (2.9)	0 (0)	.282	1 (3.1)	0 (0)	.285
Necrotizing enterocolitis (grade ≥IIa)	1 (2.9)	4 (10)	.216	1 (3.1)	3 (8.3)	.362
Focal intestinal perforation	2 (5.7)	2 (5)	.891	2 (6.3)	2 (5.6)	.903
Abdominal surgery	6 (17.1)	7 (17.5)	.967	6 (18.8)	7 (19.4)	.942
Days on mechanical ventilation	6 [0 to 14]	4.5 [0 to 16.8]	.934	6.5 [0 to 13.5]	4.5 [0 to 16.8]	.904
Chronic lung disease	7 (20)	8 (20)	1	6 (18.8)	8 (22.2)	.724
Steroid treatment	6 (17.1)	10 (25)	.407	5 (15.6)	9 (25)	.34
PDA requiring treatment	23 (65.7)	29 (72.5)	.525	21 (65.6)	27 (75)	.397
Surgical ligation	3 (8.6)	4 (10)	.832	3 (9.4)	4 (11.1)	.814
Pulmonary hypertension	10 (28.6)	13 (32.5)	.713	10 (31.3)	11 (30.6)	.951
iNO/sildenafil treatment	8 (22.9)	12 (30)	.485	8 (25)	10 (27.8)	.796
Time to full enteral feeds (d)	29 [22 to 48]	28 [20.3 to 39]	.69	30 [22 to 46.3]	28 [20.3 to 39]	.627
Parenteral lipids (d)	26 [19 to 39]	25.5 [20 to 38]	.89	26 [20.5 to 41.3]	25.5 [20 to 38]	.676
Cumulative amount of parenteral lipids (g/kg)	46.5 [30.5 to 73.8]	46.92 [35.3 to 63.2]	.987	48.4 [30.7 to 76.6]	49 [35.3 to 63.2]	.868
Study lipids (g/kg/d)	1.8 [1.6 to 2.1]	1.8 [1.6 to 2.0]	.932	1.8 [1.4 to 2.1]	1.8 [1.6 to 2.0]	.966
Docosahexaenoic acid (mg/kg/d)	40 [34.2 to 46.3]	3.9 [3.2 to 4]	<.0001	40 [31.8 to 46.1]	3.9 [3.2 to 4]	<.0001
Eicosapentaenoic acid (mg/kg/d)	44.2 [37.3 to 50.5]	0 [0 to 0]	<.0001	44.1 [34.7 to 50.2]	0 [0 to 0]	<.0001

Abbreviations: iNO, inhalative nitric oxide; PDA, persistent ductus arteriosus; ROP, retinopathy of prematurity

Categorical data are presented as numbers with percentages in round parentheses and were tested using the chi2 test. Continuous data are presented as the median and interquartile range in squared parentheses and were tested using the Mann-Whitney U test. P values <.05 were considered statistically significant.

^adata of 1 patient missing

Anthropometry

Data on anthropometry at 12 and 24 months corrected age are shown in Table 5. There were no significant differences between the two groups.

Table 5. Anthropometry at 12 and 24 months corrected age.

Parameter	12 months corrected age			24 months corrected age
		Mixed lipid emulsion (n=86)	Soybean oil (n=88)	P	Mixed lipid emulsion (n=81)	Soybean oil (n=83)	P
Body Weight (kg)	8.9 [8 to 10] b	8.95 [8.1 to 9.9] d	.98	11.4 [10.2 to 12.9] a	11.6 [10.2 to 12.9] c	.84
Z score	-0.57 [-1.5 to 0.4] b	-0.44 [-1.39 to 0.41] d	.93	-0.22 [-1.09 to 0.68] a	-0.2 [-1.21 to 0.57] c	.95
Height (cm)	74 [72 to 76] a	74 [71.63 to 76] f	.90	85 [82 to 88] a	85 [83 to 88] c	.83
Z score	-0.59 [-1.25 to 0.38] a	-0.5 [-1.25 to 0.28] f	.94	-0.59 [-1.4 to 0.23] a	-0.59 [-1.29 to 0.22] c	.95
Body Mass Index	16.3 [15.1 to 17.9] a	16.3 [14.9 to 17.2] d	.51	15.7 [14.9 to 16.9] a	15.8 [14.9 to 17.1] c	.99
Head circumference (cm)	45 [43.9 to 46] c	45 [43.7 to 46.3] e	.78	47 [46 to 48.5] b	47 [46 to 48] d	.54
Z score	-1.2 [-2.06 to 0.01] c	-1.1 [-2.3 to 0.5] e	.94	-0.95 [-1.82 to 0.07] b	-0.98 [-2.49 to -0.33] d	.30

Continuous data are presented as the median and interquartile range in squared parentheses and were tested using the Mann-Whitney U test. P values <.05 were considered statistically significant.

^adata of 7 patients missing

^bdata of 8 patients missing

^cdata of 9 patients missing

^ddata of 10 patients missing

^edata of 11 patients missing

^fdata of 12 patients missing

Neurodevelopmental outcome

Bayley-III – comparison of medians

The results of neurodevelopmental follow-up (Bayley-III) are shown in Table 6. At 12 and 24 months, there was no significant difference in cognitive, language, or motor scores. Analysis of subgroups with particular risk for unfavorable outcome (birth weight <750g, Table 7, online only; low parental education, Table 8, online only) revealed no statistically significant difference between the two groups. No significant sex specific effect of the mixed lipid emulsion on neurodevelopment (Table 9, online only) was found.

Table 6. Bayley-III at 12 and 24 months corrected age.

Parameter	12 months corrected age			24 months corrected age
	Mixed lipid emulsion (n=86)	Soybean oil (n=88)	P	Mixed lipid emulsion (n=81)	Soybean oil (n=83)	P
Cognitive composite score	95 [85-101]	95 [85-100] a	.71	95 [80-105] a	95 [90-105] a	.17
Language composite score	86 [77-94] a	89 [79-95] b	.48	89 [75-97] d	89 [77-100] c	.54
Motor composite score	88 [76-94] a	88 [79-94] a	.69	94 [82-103] a	94 [85-103] b	.53

Continuous data are presented as the median and interquartile range in squared parentheses and were tested using the Mann-Whitney U test. P values <.05 were considered statistically significant.

^adata of 1 patient missing

^bdata of 2 patients missing

^cdata of 3 patients missing

^ddata of 5 patients missing

Table 7. online only. Bayley-III at 12 and 24 months corrected age in patients born <750g.

Parameter	12 months corrected age			24 months corrected age
	Mixed lipid emulsion (n=35)	Soybean oil (n=39)	P	Mixed lipid emulsion (n=32)	Soybean oil (n=36)	P
Cognitive composite score	90 [80-95]	90 [80-100] a	.43	90 [80-105] a	93 [86-105]	.20
Language composite score	86 [77-97] a	89 [77-94] a	.90	89 [77-97] a	88 [74-100] b	.78
Motor composite score	82 [76-94] a	88 [73-94] a	.36	91 [82-100]	94 [85-104] b	.36

Continuous data are presented as the median and interquartile range in squared parentheses and were tested using the Mann-Whitney U test. P values <.05 were considered statistically significant.

^adata of 1 patient missing

^bdata of 2 patients missing

Table 8. online only. Bayley-III at 12 and 24 months corrected age in patients with low parental education.

Parameter	12 months corrected age			24 months corrected age
	Mixed lipid emulsion (n=29)	Soybean oil (n=29)	P	Mixed lipid emulsion (n= 28)	Soybean oil (n=26)	P
Cognitive composite score	95 [85-100]	95 [83-104]	.75	90 [71-105]	90 [80-105]	.79
Language composite score	86 [76-95]	85 [78-91]	.84	88 [63-95] a	81 [69-96] a	.73
Motor composite score	88 [77-97]	88 [77-94]	.89	91 [82-100]	91 [85-102]	.74

Continuous data are presented as the median and interquartile range in squared parentheses and were tested using the Mann-Whitney U test. P values <.05 were considered statistically significant.

^adata of 2 patients missing

Table 9. online only. Bayley-III at 12 and 24 months corrected age in boys & girls.

Parameter	12 months corrected age			24 months corrected age
	Boys	Mixed lipid emulsion (n=51)	Soybean oil (n=57)	P	Mixed lipid emulsion (n=48)	Soybean oil (n=56)	P
Cognitive composite score	90 [80-100]	90 [85-100] a	.25	90 [79-100]	95 [85-105] a	.16
Language composite score	81 [74-91] a	86 [75-94] b	.67	83 [68-97] c	83 [73-98] b	.53
Motor composite score	86 [73-94] a	88 [79-97] a	.42	91 [82-100] a	91 [85-103] a	.41
Girls	Mixed lipid emulsion (n=35)	Soybean oil (n=31)	P	Mixed lipid emulsion (n=33)	Soybean oil (n=27)	P
Cognitive composite score	95 [95-105]	100 [95-105]	.77	100 [85-110] a	100 [90-115]	.42
Language composite score	89 [79-100]	94 [86-97]	.35	94 [81-108] a	97 [89-103] a	.62
Motor composite score	91 [82-97]	91 [88-94]	.68	100 [90-106]	97 [94-108] a	.52

Continuous data are presented as the median and interquartile range in squared parentheses and were tested using the Mann-Whitney U test. P values <.05 were considered statistically significant.

^adata of 1 patient missing

^bdata of 2 patients missing

^cdata of 4 patients missing

Gross motor function – comparison of medians

At 24 months corrected age, infants with cerebral palsy were classified using the GMFCS. In both groups, 13 infants were diagnosed with cerebral palsy without revealing a difference in severity between soybean oil (GMFCS 1: n=11, GMFCS: 2 n=1, GMFCS 3: n=0, GMFCS 4: n=1) and the mixed lipid emulsion (GMFCS: 1 n=9, GMFCS 2: n=2, GMFCS 3: n=2, GMFCS 4: n=0).

Mixed model linear regression analysis

After exclusion of patients with missing values in covariates, the sample size for the regression analysis was 153 for infants and 139 for mothers. Parenteral nutrition using the mixed lipid emulsion had no significant influence on cognitive, language, or motor scores (Tables 10 and 11, both online only). Parental education was significantly related to the cognitive and language, but not the motor scores. Intraventricular hemorrhage and male sex had a significant negative impact on all three scores. The gestational age of infants did not exert a significant influence.

Discussion

This secondary outcome analysis of a randomized trial of infants with ELBW comparing parenteral nutrition using a mixed lipid emulsion containing fish oil with a soybean oil-based lipid emulsion showed no significant difference of neurodevelopment at 12 and 24 months corrected age. Follow-up studies on neurodevelopment are crucial to demonstrate safety of novel interventions in neonatology. Studies of new lipid emulsions that provide DHA have not reported on neurodevelopmental outcome. ^{16, 17}

Docosahexaenoic acid promotes the development of neurons ^18–21 and comprises 30-40% of the brain’s grey matter. ^{19, 20, 22} In utero supply is regulated by the placenta ²² and peaks in the third trimester. ²³ Extremely preterm infants miss out this period of maximum DHA supply and cannot sufficiently convert precursor fatty acids to LC-PUFAs. ⁴ The hypothesis that additional supply with DHA to preterm infants is beneficial for their neurodevelopment ²⁴ is based on high cerebral DHA levels ^{19, 20, 22} , the subtle effects of DHA deprivation ^{25, 26} in animal studies, and more pronounced effects in vitro. ^{27, 28} Yet, clinical trials in preterm infants that aimed at improving neurodevelopmental outcome by enteral DHA supplementation did not provide clear evidence for a benefit. ²⁹ Whether parenteral DHA supply using lipid emulsions would be more efficient than enteral supply is speculative, but it might be of additional value. Parenteral lipid emulsions can be provided right after birth in adequate concentrations to close the gap in DHA supply until full enteral feeds are established – typically after a few weeks post partum. ³⁰ Studies in piglets demonstrated enrichment of DHA in the brain using a mixed lipid emulsion. ³¹ In this context, our previous finding of accelerated maturation of amplitude-integrated EEG assessed between birth and hospital discharge ⁹ may point out to cerebral DHA incorporation using the mixed lipid emulsion in preterm infants. This created considerable expectations for neurodevelopmental follow-up, but we found no measurable impact on the neurodevelopmental outcome at 12 and 24 months corrected age in this study. In vitro studies showed positive effects of DHA on synaptic protein expression ²⁸ but no influence on absolute neurite numbers. ¹⁸ It seems possible that changes of synaptic activity caused a transient impact on electrophysiological brain activity using the mixed lipid emulsion in infants of this cohort ⁹ , without significantly affecting their neurodevelopment - maybe due to a missing effect on total neurite numbers. ^{29, 32}

Besides DHA, ARA is also highly concentrated in the brain ³³ and relevant for growth and neurodevelopment. ³⁴ In enteral nutrition, ARA should meet at least DHA levels. ³⁵ Yet, ARA is five times lower than DHA in the mixed lipid emulsion, but still higher compared to a soybean oil-based lipid emulsion. ¹¹ Yet, ARA levels in the blood were shown to decrease in infants receiving the mixed lipid emulsion compared to soybean oil-based lipid emulsions. ¹⁶ This could be due to reduced conversion, ³⁶ replacement of ARA by DHA/EPA, or a shift to other tissues. ^{37, 38} While DHA supply improves, ARA blood levels decline, which may potentially outweigh positive effects of DHA. This raised concerns about long-term clinical safety concerning growth and development. ¹⁶ Yet, we did not find a statistically significant effect on neurodevelopment or growth.

Clinical trials on LC-PUFA enriched infant formula for preterm infants only inconsistently demonstrated a benefit for neurodevelopmental outcome. ^{32, 39–43} Studies that found a significant influence did so only after subgrouping: Fewtrell et al found higher mental developmental indices in boys ⁴⁰ and Makrides et al reported on a similar benefit for girls. ³⁹ Analysis of infants of this trial showed higher Bayley-III scores in girls, but no sex-specific neurodevelopmental advantage using the mixed lipid emulsion. It is furthermore well established that parental education influences children’s cognitive abilities. ⁴⁴ Lien et al proposed a benefit of LC-PUFA supplementation for disadvantaged infants with poor education. ²³ In our trial, language and cognitive scores of preterm infants were strongly influenced by parental education (Tables 10 and 11, both online only), but there was no significant effect of the mixed lipid emulsion in infants whose parents had a low educational background (Table 8, online only). Another subgroup with a particular risk for unfavorable development that might benefit from DHA supplementation are the smallest infants born <750 gram, but again there was no significant neurodevelopmental advantage of the mixed lipid emulsion (Table 4, online only).

Our study has certain limitations. Neurodevelopment was a secondary outcome and the study was not powered for assessment of neurodevelopment, with rather small numbers leading to wide confidence intervals. On the other hand, the quality of our trial was high with minimal risk for bias ⁴⁵ , loss to follow-up was low with 20%, and without a statistically significant difference in numbers and important clinical characteristics of dropouts.

The calculated parenteral supply with DHA using the mixed lipid emulsion was 43 mg/kg/d. This is at the lower end of published fetal accretion rates (40 to 67 mg/kg/d). While infants received higher amounts of DHA using the mixed lipid emulsion, even more DHA would be needed to reach in utero supply in all infants. Moreover, DHA in enteral nutrition is still far too low to meet in utero levels after weaning from parenteral nutrition, and deficiency at discharge is likely. As analysis of plasma fatty acids was not part of our study, we cannot report on blood fatty acid levels.

It is also possible that the nutritional DHA deficits were not pronounced enough in our study to exert an effect on neurodevelopment. In this context, infants of ELBW who received parenteral nutrition for more than three weeks were shown to have a poorer neurodevelopment. ³⁰ In our trial, full feeds were already reached after 23 days (interquartile range 17-37) ⁸ . We thus cannot rule out an effect in infants of ELBW who depend on parenteral nutrition for much longer.

Conclusion

The type of parenteral lipid emulsion did not significantly affect the neurodevelopmental outcome of preterm infants of ELBW at 12 and 24 months corrected age. Further information on neurodevelopmental outcome will be gathered in another follow-up analysis using the “Kaufmann Assessment Battery for Children” at 5.5 years of age. Future trials should aim at investigating the effects of a combined approach of parenteral and enteral DHA supply starting from the first day of life, and should also consider providing ARA in amounts equal to DHA.

Data Statement

The dataset generated and/or analyzed during the current study is available from the corresponding author upon request.