Introduction

Preterm infants often require parenteral feeding to meet energy requirements and provide essential nutrients, optimising postnatal growth and development [15, 24]. Total parenteral nutrition (TPN) supplies nutrients, caloric requirements, as well as essential and non-essential fatty acids vital for development [9, 16, 27].

Intravenous fat emulsions provide necessary lipids required to form key structural components in nearly all tissues and serve as precursors for prostaglandins, prostacyclins and other eicosanoids. These eicosanoids are regulators of various cellular processes, including inflammation, immune response and platelet aggregation [17, 27]. Parenteral nutrition can present with adverse neonatal outcomes that may be related to the type of fat emulsion, rate of infusion and total duration of administration [17]. Hence, there has been a widespread interest in developing alternate lipid formulations with goals of optimising neonatal outcomes.

Intralipid (Baxter Australia, NSW), a pure soy-based emulsion, has been the traditional lipid emulsion used in neonatal intensive care units (NICUs) around the world. Pure soy-based lipid preparations contain minimal long-chain polyunsaturated fatty acid (LCPUFA) derivatives of ω-6 and ω-3 fatty acids (arachidonic acid (ARA), docosahexaenoic acid (DHA) and eicosapentaenoic acid (EPA)). EPA and ARA play an important immunomodulatory role in activation of pro-inflammatory (mainly ARA derived) and anti-inflammatory pathways (mainly EPA derived) [30]. Soybean emulsions contain a high ratio of ω-6: ω-3 fatty acids, which in turn leads to lesser proportions of ω-3 fatty acid derivatives, DHA and EPA. DHA and EPA are important substances for neural and retinal development with key anti-inflammatory properties [23]. As the preterm infant is unable to convert and generate a sufficient supply of vital LCPUFA themselves, the impact of fatty acid deficits upon neonatal growth and clinical outcomes can be significant [15, 27].

Alternate lipid emulsions are being used in a number of neonatal units—including combinations of clinoleic acid, fish oil, medium and long chain triglycerides. In the last decade, SMOFlipid (Fresenius Kabi Australia Pty Ltd), a formula consisting of soybean, medium chain triglycerols (MCT), olive oil and fish oil has become a popular lipid emulsion as it provides immunomodulatory benefits and a balanced ω-6:ω-3 fatty acid ratio [30]. MCTs are commonly derived from coconut oil and are shown to lower ω-6 fatty acid levels. MCT and soy oil mixtures are also less susceptible to lipid peroxidation, thus associated with reduced systematic inflammatory response and positive effect on leukocyte and reticuloendothelial function [30].

Another alternative source, olive oil, has been found to reduce high content of ω-6 fatty acid and provide additional monosaturated fatty acids and vitamin E, important to prevent cellular damage [30, 31]. Fish oil is believed to modulate various anti-inflammatory pathways as it contains a high proportion of ω-3 fatty acid which is rarely found in other plant oil-based emulsions [23, 30]. Hence, due to its lesser phytosterol content [8, 10] and perceived anti-inflammatory benefits, SMOFlipid has been conventionally used in surgical patients and infants with intestinal failure-associated liver disease [19]. SMOFlipid has been found to have a positive effect in lowering total bilirubin, with the fish oil component associated to the reversal of parenteral nutrition-associated liver disease (PNALD) in some infants [7, 14, 18, 20].

SMOFlipid has shown to be safe and is associated with increased levels of DHA and EPA, with conversely lower arachidonic acid levels in both erythrocyte membranes and plasma of premature infants as compared to pure soy emulsions. Studies have compared biochemical markers in premature infants receiving a pure soy-based against alternate lipid emulsions with statistically significant differences seen in serum composition [6, 21, 28, 29, 32]. However, there is limited evidence to suggest any clinically significant differences amongst infants receiving SMOFlipid versus a soy-based lipid emulsion when used routinely in preterm infants [13].

Despite many previous intervention studies comparing SMOFlipid to alternate lipid emulsions, there is insufficient data upon the clinical impact. Our aim was to conduct an epoch study to evaluate the incidence of common neonatal outcomes between two time periods in order to directly compare SMOFlipid with Intralipid.

Methods

Study design

This was a retrospective cohort study analysing data from the neonatal unit (Monash Newborn) at Monash Children’s Hospital, Melbourne. Monash Newborn is a tertiary combined medical and surgical neonatal unit with approximately 4000 births per annum. Appropriate ethics approval was obtained from the Hospital Human Research Ethics Committee to review data. In July 2015, Monash Newborn moved from using Intralipid to SMOFlipid as the primary lipid emulsion for parenteral nutrition in preterm neonatal infants. We thus outlined two epoch groups to help compare the time periods in which SMOFlipid and Intralipid were used.

The two epoch periods were defined as:

  • Epoch 1: 20 months period prior to the introduction of SMOFlipid, where Intralipid was primarily used: October 2013–June 2015

  • Epoch 2: 20 months period after the introduction of SMOFlipid: July 2015–March 2017

We collaborated with the hospital pharmacy to acquire TPN data of the preterm infants (< 32 weeks) that required either Intralipid or SMOFlipid for at least 14 days from both epoch periods. Hospital electronic databases (medical records, pathology and imaging) were then accessed to retrieve clinical data from the two cohorts. Chart reviews of individual patients were also conducted to ensure accuracy and completeness of information. Baseline characteristics of each epoch group were compared in respect to gestational age at birth, maternal age, sex, birth weight, length, head circumference, antenatal and immediate postnatal characteristics.

Inclusion criteria

The inclusion criteria were preterm infants born prior to 32 weeks of gestation and requiring a single parenteral lipid emulsion formula for a minimum of 14 consecutive days.

Exclusion criteria

The exclusion criteria were infants with severe congenital malformation, metabolic disorders or medical records with insufficient/missing data. Infants with insufficient medical records (15 in total) were typically preterms that had been transferred to an alternative health network at parents’ request.

Intervention

Infants requiring TPN were administered either SMOFlipid or Intralipid emulsion for duration of at least 14 days starting within 24 h of birth. Central venous access was used to administer TPN in most cases and either lipid emulsion was titrated up to a rate of 3–4 g/kg/day. Both lipid emulsions were administered in a separate syringe alongside standardised solution containing protein, glucose, minerals and trace elements. Parenteral nutrition continued in both epochs until full enteral feeds were tolerated with no difference in standard cardio-respiratory care between the epochs.

Outcomes

Primary outcomes compared between the epoch groups included mortality, rates of PNALD (defined via abnormal biochemical markers), patent ductus arteriosus requiring treatment, bronchopulmonary dysplasia (defined as need for oxygen or respiratory support at 36 weeks postconceptional age), severe intraventricular haemorrhage (grade 3–4), cystic periventricular leukomalacia, retinopathy of prematurity requiring laser treatment, necrotising enterocolitis (all stages) and late-onset culture-proven sepsis. Composite outcome of mortality or any of the above serious morbidities was also compared. Secondary outcomes compared included duration of invasive ventilation, duration of total pressure support, duration of oxygen requirement, need for home oxygen, any stage of retinopathy of prematurity, any grade of intraventricular haemorrhage, periventricular leukomalacia, days till enteral feeding tolerated, length of stay and growth parameters at 36 weeks.

Based on an estimated 80% incidence of composite outcome (death or major morbidity) in this infant population, to expect a 20% improvement in composite outcomes (with SMOFlipid) with an alpha error of 0.05 at 0.8 power, a minimum of 81 infants would be needed in each epoch.

All statistical analysis was performed using STATA 14 (StataCorp LP, TX, USA). Continuous variables were expressed as mean (standard deviation (SD)) or median (interquartile range (IQR)). Normally distributed data were analysed by Students t test and skewed data by Wilcoxon rank-sum test. Categorical variables were analysed by chi-squared test. Multivariate and univariate logistic regression analysis adjusted for baseline variables was undertaken to determine the significant risk factors for mortality and PNALD, expressed as adjusted odds ratio (OR) and 95% confidence (CI). Statistical significance was set at p < 0.05.

Results

During the study period, 1113 preterm infants (less than 32 weeks’ gestation) were admitted to the NICU. Overall, 222 infants were enrolled in the study who had received TPN lipid for 14 or more days, with 123 in epoch 1 receiving Intralipid and 99 infants in epoch 2 receiving SMOFlipid (Fig. 1). The mean gestational age (SD) (26.7 (2.1) versus 26.7 (2.0) weeks; p = 0.9) and birth weight (935.6 (342) versus 935.5 (25) grams; p = 1.0) were very similar in the two epoch groups. Both groups had similar baseline characteristics (Table 1), with no statistically significant disparity between sex, duration of lipid administration and rates of IUGR. There was a slightly significant difference seen in rates (%) of premature rupture of membranes (PROM) (42 versus 21%; p < 0.05) and antepartum haemorrhage (APH) (10 versus 20%; p < 0.05) between the epochs.

Fig. 1
figure 1

Flow diagram of included infants

Full size image
Table 1 Baseline maternal and infant demographics in epoch 1 and 2
Full size table

Table 2 shows the results of the primary outcomes in both epochs. There was no significant difference in the rates of bronchopulmonary dysplasia (57 versus 67%; p = 0.3, risk difference = 0.1 (− 0.1, 0.2)), PNALD (7 versus 11%; p = 0.3, risk difference = 0.03 (− 0.03, 0.1)), necrotising enterocolitis (22 versus 19%; p = 0.6, risk difference = − 0.02 (− 0.1, 0.1)) or mortality (6 versus 4%; p = 0.4, risk difference = − 0.01 (− 0.1, 0.03)) between the two epoch groups. A noticeably larger incidence of late onset sepsis was observed in epoch 1 as compared to epoch 2 (56 versus 30%; p < 0.005, risk difference = − 0.3 (− 0.4, − 0.1)), yet composite outcomes (death and/or any major morbidity) (85 versus 80%; p = 0.3, risk difference = − 0.1 (− 0.2, 0.04)) remained similar between groups.

Table 2 Comparison of primary outcomes in epoch 1 (Intralipid) versus epoch 2 (SMOFlipid)
Full size table

Secondary outcomes are presented in Table 3. Retinopathy of prematurity (ROP) (any stage) was observed less frequent in epoch 1 versus epoch 2 which was statistically significant (39 versus 54%; p = 0.03, risk difference = 0.1 (0.01, 0.3)), yet there was no difference amongst rates of infants requiring laser therapy for management of ROP. On the other hand, rates of intraventricular haemorrhage (any grade) were shown to be statistically greater in epoch 1 versus epoch 2 (44 versus 30%; p = 0.03, risk difference = − 0.1 (− 0.3, − 0.01), yet no significant difference was seen in rates of intraventricular haemorrhage grades 3–4. There was a trend towards higher use of postnatal steroids (24 versus 18%, p = 0.06, risk difference = 0.1 (− 0.007, 0.2)) in epoch 2, concurrent with the slightly higher rate of BPD. When comparing growth outcomes, infants receiving SMOFlipid were found to have a greater weight at 36 weeks postconception (2016 versus 2141 g; p < 0.05, risk difference = 125.0 (3.0, 247.1), alongside a significant increase in weight at 36 weeks.

Table 3 Comparison of secondary outcomes in epoch 1 (Intralipid) versus epoch 2 (SMOFlipid)
Full size table

Univariate and multivariate logistic regression analyses were performed to investigate the association between clinical risk factors and two outcomes of interest, mortality and PNALD (Table 4). Both regression models found that the type of lipid did not influence rates of death (univariate odds ratio = 1.4 (CI 0.4, 5.0); multivariate odds ratio = 0.6 (CI 0.4, 51.3)) or PNALD (univariate odds ratio = 0.6 (CI 0.3, 1.6); multivariate odds ratio = 0.6 (CI 0.2, 1.9)). The univariate model established intrauterine growth restriction as the only significant risk factor for mortality associated with a sixfold increased risk of death (odds ratio = 6.4 (CI 1.9, 22.3)). Univariate analysis identified total lipid duration and necrotising enterocolitis as significant risk factors for development of PNALD. There was found to be a 10% increased risk of PNALD with every additional day of lipid administration (odds ratio = 1.1 (CI 1.05–1.1)) and a near sixfold increase in rates of PNALD associated with necrotising enterocolitis (odds ratio = 5.8 (2.2, 15.1). Total lipid duration was the only significant risk factor identified for PNALD (odds ratio = 1.1 (CI 1.04–1.1)) in the multivariate analysis. In addition, multivariate regression identified gestational age to be the only significant risk for mortality with a threefold increased risk of death with every week decrease in gestation (odds ratio = 3.0 (CI 1.2, 7.9)).

Table 4 Risk factors for death and PNALD in logistic regression models
Full size table

Discussion

We compared clinically relevant outcomes in preterm infants receiving SMOFlipid as the primary lipid emulsion to those receiving Intralipid emulsion in this epoch-based study. SMOFlipid is increasingly being used as the primary lipid component in parenteral nutrition, albeit with limited data evaluating its clinical impact. Thus, this being one of the first studies to primarily compare clinical outcomes between these two commonly used lipid emulsions in preterm infants.

A recent Cochrane review [13] concentrating on clinical outcomes in infants receiving SMOFlipid showed no notable difference in significant outcomes including mortality, bronchopulmonary dysplasia (BPD), sepsis, growth rates and parenteral nutrition-associated liver disease (PNALD). However, most studies included in the systematic review [13] focused on biochemical markers as primary outcomes, with the largest study consisting of only 96 participants [29]. All studies attested SMOFlipid to be safe and well tolerated in the neonatal population as compared to other lipid formulations. However, it was concluded that larger studies were required to investigate clinically relevant outcomes in targeted neonatal population groups.

This study investigated the incidence of important clinical outcomes in very preterm infants born less than 32 weeks receiving Intralipid versus SMOFlipid. Infants received minimum of 14 days’ continuous lipid emulsion as a component of parenteral nutrition to ensure adequate exposure to the lipid formulations. Previous trials show an influence on biochemical parameters in infants receiving lipid emulsion for at least 7–14 days [5, 6, 21, 29, 32], thus we postulated this to be an adequate period to produce meaningful impacts on clinical outcomes. Participants from both epoch periods were well matched with no significant differences in nearly all demographic features. The difference seen in rates of PROM and APH between cohorts is most likely coincidental as there is no recognisable difference in diagnosis of either complication. Prenatal and postnatal care remained largely consistent between both epochs, although there was a change in central line care practice in late 2014 based upon a quasi-experimental study at the study site [26]. The new checklist initiative was associated with reduced central line-associated bloodstream infections which most likely contributed towards lesser rates of late-onset sepsis seen in the SMOFlipid cohort.

No previous trials have recognised any mortality benefit associated with any alternate lipid emulsion formula [13]. Intrauterine growth restriction and gestational age were the only predictors of mortality identified in this study, which remains consistent with published data.

An interesting trend was seen in the respiratory outcomes between the groups, with infants receiving SMOFlipid observed to have slightly higher rates of bronchopulmonary dysplasia, require longer duration of invasive ventilation, total days of pressure support and oxygen use (none of them however statistically significant). With significantly lesser rates of culture-proven sepsis in the SMOFlipid cohort, it may have been expected to see better rates of bronchopulmonary dysplasia and related respiratory outcomes. These figures stand similar to findings of the recently concluded N3RO trial [4] and may warrant larger randomised control trials directly comparing respiratory outcomes in SMOFlipid versus Intralipid. N3RO trial specifically focused on rates of bronchopulmonary dysplasia in neonates receiving enteral supplemental docosahexaenoic acid (DHA) versus control soy lipid emulsion. The results of the study challenged traditional perspectives and found DHA supplementation to possibly increase risk of bronchopulmonary dysplasia in preterm infants. The N3RO trial and this study have both focused upon vulnerable preterm infants requiring longer durations of parenteral nutrition. Hence, the population groups investigated have been at higher risk for developing bronchopulmonary dysplasia as well as other adverse clinical outcomes. This data may raise concern regarding the potential impact of SMOFlipid on respiratory outcomes in a susceptible preterm population.

Lesser rates of retinopathy of prematurity (ROP) (any stage) were seen in the group receiving Intralipid, yet no difference was noted in rates of ROP requiring laser therapy. One study has previously reported lower rates of ROP (stage 1–2) in association with use of SMOFlipid [2], yet no evidence exists to favour either lipid emulsion in the progression of ROP to advanced stages [13].

Differences were also recognised in rates of intraventricular haemorrhage (any grade), with lesser incidence associated to SMOFlipid. Despite having potential neurodevelopment benefits [12, 23], SMOFlipid did not impact on the progression of haemorrhage as rates of severe IVH (grade 3–4) and periventricular leukomalacia remained similar. These results stand consistent with previous trials [13].

PNALD is one of the most well-reported adverse effects associated to parenteral nutrition, with evidence suggesting SMOFlipid may aid in delaying progression or even reversal of the complication [7, 18]. Additionally, SMOFlipid has been associated with increased levels of eicosapentaenoic acid (EPA) and DHA, whilst decreasing plasma bilirubin [6, 32]. Both our univariate and multivariate regression models identified lipid duration to be a significant risk factor in the development of PNALD with the choice of lipid being inconsequential. In our univariate model, however, necrotising enterocolitis was associated with a notable sixfold increase in the rate of PNALD. Further investigations focusing on at risk populations would be needed to assess the possible influence of alternate lipid emulsions and necrotising enterocolitis on PNALD rates.

Average birthweights in both epochs were near identical, yet infants receiving SMOFlipid had a significantly greater body weight at 36 weeks’ postconception. Another randomised control trial has associated SMOFlipid with a greater rate of weight gain [29], whilst majority shows inconsequential difference [13, 21, 22]. It is difficult to attribute better growth outcomes to either lipid emulsion as the greater incidence of late-onset sepsis may be confounding results. As mentioned, a number of studies have associated infants receiving SMOFlipid with increased levels of EPA and DHA, yet concurrently having lower levels of plasma arachidonic acid (ARA) [32]. Interestingly, previous data has correlated ARA levels with infantile growth [3], whilst other data suggests that ω-3 supplementation in pregnancy may be associated with greater head circumference [25]. European guidelines currently recommend formulations with balanced components of ω-3 and ω-6 fatty acids [1, 11, 15], as further research is necessary to comprehend the role of fatty acid components in growth [12].

It is important to emphasise that our primary focus on clinical outcomes increases the relevance of our study to clinical practice. Furthermore, the study targeted an at-risk population which further raises the clinical relevance as a comprehensive range of common neonatal complications was measured. Each of the infantile outcomes had been predefined, with definitions remaining constant through the two cohorts to minimise variation in detection.

As a retrospective study, limitations include potential selection bias and inability to accurately control exposure to lipid formulations. The study did not focus on biochemical parameters which may have added some significance in correlation to clinical outcomes. Long-term follow-up may add further value to assess development of infantile complications as data from the study is limited to discharge date.

Conclusions

SMOFlipid is fast becoming the primary lipid emulsion of choice replacing soybean-based formulations in neonatal units around the word. With evidence suggesting it may provide a more balanced nutritional supply, there is still a need to further evaluate its clinical impact and particular effect on respiratory outcomes, parenteral nutrition-associated liver disease and growth.