A Mixed Lipid Emulsion Containing Fish Oil for Prevention of Parenteral Nutrition Associated Cholestasis in Extremely Low Birth Weight Infants – a Randomized Clinical Trial

To the Editor,

We appreciate the comments from Dr. Tauber concerning our study on prevention of parenteral nutrition associated cholestasis (PNAC) using a mixed lipid emulsion based on soybean-oil, medium chain triglycerides, olive oil and fish oil (SMOF-LE) in extremely low birth weight infants¹.

We agree, that exposure to soybean oil based lipid emulsion before randomization is not ideal. As recruitment prior to birth did not seem practicable to us and parenteral nutrition with lipids right after birth is part of our policy², we could not avoid exposure to soybean oil. We thus tried to keep the time from birth until randomization as short as possible (both groups: day of life 3 [2-4]; median [IQR]; P = .23). It is unclear whether this short time exposure predisposes to PNAC or would attenuate a potentially protective effect of SMOF-LE.

Although it is tempting to assume that SGA infants have a better lipid tolerance due to higher maturity, a reduced activity of endothelial lipoprotein lipase and more hypertriglyceridemia were described in these infants^{3, 4}, and intrauterine growth retardation is clearly a risk factor for PNAC^{2, 5–7}. Thus, the higher proportion of SGA infants in our standard group was potentially confounding in favor of SMOF-LE. We included the Z-scores of birth weight in our multivariate analysis, but they did not remain in the final model because of restriction of covariates¹. However, neither post hoc inclusion of the Z-Score of birth weight nor SGA as categorical variable significantly changes the effect of SMOF-LE in the model (risk of PNAC using SMOF-LE: 0.501 [0.176-1.427] and 0.547 [0.190-1.581]; adjusted ORs [95% CI]).

As pointed out by Tauber, a subgroup analysis of infants born < 750 g was not part our statistical analysis plan. We here provide a post hoc comparison by birth weight (Table 1) and a multivariate analysis of infants < 750 g (Table 2) with confounders selected as described in the methods section¹. Although the difference of PNAC is more pronounced between groups in the subgroup < 750 g (Table 1), statistical analyses again failed to demonstrate a significant effect of SMOF-LE for prevention of PNAC (Table 1 and 2).

Table 1. Subgroup analysis according to birth weight.

PNAC	SMOF-LE	S-LE	P
ELBW infants	11/110 (10%)	18/113 (16%)	.23
Subgroup < 750 gram	5/43 (12%)	15/53 (28%)	.08
Subgroup ≥750 gram	6/67 (9%)	3/60 (5%)	.39

Analysis by intention to treat. Categorical data were tested using the Fisher-exact-test. P values < .05 were considered statistically significant. ELBW, extremely low birth weight (< 1000 gram); PNAC, parenteral nutrition associated cholestasis; S-LE, soybean oil based lipid emulsion; SMOF-LE, lipid emulsion based on a mixture of soybean oil, medium-chain triglycerides, olive oil and fish oil.

Table 2. Multivariable analysis on the risk for PNAC, infants < 750 g.

	Intention to treat (n=96)
	adjusted OR	95% CI	P
SMOF-LE	0.243	0.054-1.086	.06
Male Sex	3.836	0.836-17.612	.08
NEC/ FIP/ GI surgery	2.840	0.752-10.726	.12
Time on parenteral nutrition (days)	1.046	1.014-1.078	.01
Z-Score of birth weight	0.444	0.190-1.040	.06

Analysis by intention to treat. Binary logistic regression analysis showing the adjusted odds ratio for the effect of SMOF-LE on PNAC compared to S-LE. Adjusted for male sex, “compound outcome of necrotizing enterocolitis (NEC), focal intestinal perforation (FIP) and gastrointestinal (GI) surgery”, time on parenteral nutrition and the Z-Score of birth weight. P values < .05 were considered statistically significant. CI=95% confidence interval, OR=odds ratio.