Choline supplementation for preterm infants: metabolism of four Deuterium-labeled choline compounds

The supply of the essential nutrient choline is often insufficient, especially in preterm infants [1‐3]. Choline is required as constituent of phosphatidylcholine (PC) and sphingomyelin (SPH) in all cell membranes, and in many secretions like bile and surfactant. In the form of PC, it is also required for the transport of the long-chain poly-unsaturated fatty acids (PUFA), namely docosahexaenoic acid (DHA) and arachidonic acid (ARA), via lipoproteins. Choline is required for neurodevelopment via synthesis of the neurotransmitter acetylcholine (ACh) [1, 4], and, via its oxidation product betaine, it is an important methyl group donor for the regeneration of methionine from homocysteine and essential methylation processes [5]. When choline is not completely absorbed in the small intestine, trimethylamine (TMA) is formed as a bacterial degradation product in the colon, depending on the intestinal microbiota. Following absorption, trimethylamine is oxidized by the liver to trimethylamine oxide (TMAO), which may be associated with cardiovascular events [3, 6‐8]. In preterm infants, this seems to be less important, because enzymatic oxidation of trimethylamine is reduced due to immaturity [8].

Prenatally, choline is actively transported to the fetus via the placenta, resulting in high plasma-free choline concentrations of ⁓ 40 µmol/L [9]. We showed that plasma levels rapidly decrease by 50% or even more of fetal values after preterm birth [10]. This reduction in plasma concentrations is slower and occurs at a later developmental stage in term infants, characterized by a lower growth rate [11]. Due to preterm birth, the high level of continuous active trans-placental choline supply is interrupted in a phase of exponential general and brain growth and development (24–32 week postmenstrual age). Moreover, the fetus is exposed to and swallows ⁓ 400 mL/d-amniotic fluid comprising ⁓ 90 µM choline compounds, mainly PC, choline, SPH and α-glycerophosphorylcholine (GPC) ([25] and unpublished own analyses).

During the third trimester, volume of parenchyma, being rich in PC/SPH, increases. Particularly, brain gray matter becomes enriched in PUFA-PC [12, 13]. However, when choline levels in plasma and hepatic PC pools are low, choline is mobilized from other organs, particularly the lung [14]. In cystic fibrosis (CF) patients, low free choline levels were associated with impaired liver and lung function [3, 15]. Hence, due to this lung-to-liver drainage of choline in choline-deficient states, insufficient choline supply could also be a contributor to the multi-factorial pathogenesis of chronic lung disease of prematurity.

The US Food and Drug Administration defined choline as an essential nutrient in 1998 and adequate intakes (AI) were defined by the National Academy of Medicine of the USA (NAM) and the European Food Safety Authority (EFSA). The requirements are defined by age, sex, life conditions and genetics, with tightly defined AI and high upper tolerable limit (UL) values [16‐18]. The European Society of the Paediatric Gastroenterology, Hepatology, and Nutrition Committee on Nutrition (ESPGHAN) recommended (2010) an enteral choline supply of 8–55 mg/kg/day for preterm infants, representing the AI and UL values of adults [19]. However, only the UL of this latter recommendation, i.e., enteral intakes of 50–60 mg/kg/day choline equivalent in postnatal preterm infants, resulted in plasma choline concentrations similar to those observed in umbilical cord plasma (which may represent fetal concentrations). [20].

Total choline intake through breastmilk and current formulas is lower than prenatal supply. Choline content of breast milk fortifiers varies greatly, resulting in infants’ plasma-free choline concentrations being only half those of the fetus, and even decreasing postnatally [21]. Estimated AI of choline is neither achieved by most formulas, frequently comprising free choline or PC, nor by breast milk predominantly containing phosphoryl-choline and GPC, together with PC and SPH [21].

In this study, we aimed to investigate potential differences in the plasma kinetics of choline, betaine and PC following supplementation of ‘free’ choline as a salt compared to its major physiologic water-soluble and lipophilic organic esters. We used their stable-isotope-labeled derivatives, namely D9-choline chloride, D9-phosphoryl-choline, D9-GPC and D9-POPC. This was to inform future enteral choline supplementation in preterm infants’ nutrition.

This is a randomized parallel-group comparison in 32 preterm infants, carried out at Tübingen University Hospital, Germany. Patients were recruited from January 2021 to December 2021. The Institutional Review Board (project number 322/2019BO1) approved the protocol, and written informed consent was obtained prior to enrollment.

Patient flow and demographics are shown in Fig. 1 and Table 1. There were no significant differences in weight (p = 0.65) or gestational age (p = 0.83) at birth between groups, and all patients completed the study.

In this study, we investigated the metabolism of four different choline compounds, which all might be suitable to optimize choline supplementation of very preterm infants. To better follow the differences in the kinetics of individual compounds, we applied stable isotope labeling of the tri-methyl-ammonium group of the choline unit with Deuterium (D9-choline) of supplements, as previously used for choline chloride only [3, 20]. This study was limited by ethical restrictions in infants, namely with respect to the number of venous punctures in this highly vulnerable population. Very sparse blood sampling was carried out (only 2 samples per patient), to avoid venipunctures in addition to those being clinically indicated. Consequently, exact kinetics and time points of maximum concentrations could not be determined. Hence, results need to be considered in the context of previous and future reports on (D9-labeled) choline compounds enterally administered to adults, where more frequent blood sampling is possible [3, 15, 25]. In this study on very preterm infants, enteral administration of any water-soluble D9-labeled choline compound used resulted in the rapid appearance of D9-choline and D9-betaine in plasma, suggesting that they all are rapidly absorbed, and that a significant portion is metabolized to D9-betaine. D9-choline was absent from plasma at 12 h and beyond, suggesting complete cellular uptake and metabolism. However, the maximum time point and half-life (t1/2) of free plasma D9-choline, that is 1–2 h in adults [3, 15], could not be assessed in these preterm because the number of blood samples was limited to two for ethical considerations. Although choline is released into plasma from degraded PC in neonatal rats, this was not measurable here, likely due to the lower D9-choline label in human studies compared to animal experiments (2.7 vs. 37.5 mg/kg) [2]. Rapid increase of D9-betaine at 1 h further accentuates the rapid assimilation and metabolism of water-soluble D9-choline compounds. It is important to note, however, that D9-betaine maximum values were at 12 h rather than 1 h as determined for D9-choline, and that values had not returned to zero at 60 h. This indicates that a single choline administration, although not having a sustainable effect on plasma choline, increases the pool of betaine as a methyl donor for a longer period.

Following D9-POPC administration, however, D9-choline was not detected at any time point. D9-betaine was detected only at 12–60 h, but not at 1 h, and D9-betaine maximum plasma concentrations were lower compared to water-soluble compounds. This is important for three reasons: first, betaine is synthesized from free choline by choline dehydrogenase and betaine-aldehyde dehydrogenase [26]. Hence, part of D9-POPC must have been hydrolyzed to free D9-choline, but did not enter the circulation at all, or did so between 1 and 12 h following D9-POPC administration. Second, as D9-betaine increased later, with zero values at 1 h, D9-choline release from D9-POPC was delayed. This is consistent with previous data in adults, where water-soluble compounds resulted in choline maxima at 1–2 h, while choline concentrations following egg-PC levels peaked at 3 h [25]. Third, these data support that intestinal choline metabolism is different for PC compared to its water-soluble counterparts in preterm infants as previously described in adults: free choline from choline chloride is absorbed and released to the portal vein, whereas the precise mechanisms of phosphoryl-choline and GPC absorption and enterocytic metabolism are still unclear, but its choline moiety apparently enters the portal vein as well. (PO)PC, however, is initially hydrolyzed by secretory pancreatic phospholipase A2IB (pPLaseA2IB), followed by the absorption of lyso-PC and fatty acid. About 50% of lyso-PC is then resynthesized to PC for chylomicron assembly and lymphatic transport bypassing the liver, using a fatty acid (acyl-coenzyme A) within the enterocyte, whereas only the remaining 50% are transported to the liver via the portal vein [26, 27]. This re-acylation occurs by enterocytic lyso-PC acyl transferase 3 (LPCAT3), being important for plasma lipid transport and favoring incorporation of poly-unsaturated fatty acids [28]. In line with these mechanisms of choline assimilation from PC, and the involvement of pPLaseA2IB and enterocytic LPCAT3, preserving ingested D9-POPC from complete degradation and D9-choline release, D9-POPC resulted in the highest D9-PC/D9-betaine ratio in these preterm infants. Consequently, (PO)PC administration favors PC formation at the expense of choline oxidation to betaine. As betaine increased to values above fetal and neonatal levels upon choline supplementation in preterm infants [20], it may be adequate to use PC as a choline supplement in combination with water-soluble compounds. However, it must be considered that some preterm infants appear to have exocrine pancreatic insufficiency [29, 30], which may impair cleavage of PC and reduce its uptake as lyso-PC; hence, the effectiveness of choline supplementation using PC. This could be especially important in very preterm infants, which would be the population with the greatest need for choline supplementation.

The concept of PC supplementation in preterm infants may additionally be useful with respect to the de-acylation/re-acylation process that PC undergoes during assimilation. We have previously shown that following preterm birth, linoleic acid-containing PC (C18:2-PC) increases at the expense of compounds comprising arachidonic (C20:4-PC) or docosahexaenoic acid (C22:6-PC). This is due to the low supply with these fatty acids and high supply with linoleic acid from breast milk and formula, which is in contrast to placental retention of linoleic acid and preferential arachidonic and docosahexaenoic acid delivery to the fetus [31]. In line with this, D9-POPC was rapidly converted to linoleic acid-containing D9-C18:2-PC, with low yields of D9-C20:4-PC and D9-C22:6-PC. It is reasonable to expect, but must be tested in future experiments, that higher dietary administration of arachidonic and docosahexaenoic acid in combination with PC as a choline supplement, at the expense of linoleic acid supply, will result in PUFA-PC composition that resembles fetal plasma PC profile [32].

Finally, experiments in adults have shown that TMAO is formed from water-soluble choline supplements but not from PC [33]. In these preterm infants, D9-TMAO was not detected in plasma in response to any (D9-) choline compounds, which confirms previous data [20]. Unlabeled TMAO was present in plasma of all infants, but concentrations were about 1/10 of those observed in fasting plasma samples of healthy adults [25]. This may be due to the intestinal microbiome of predominantly breast milk-fed infants or due to the immaturity of their liver, being not yet able to oxidize trimethylamine to TMAO [8]. Nevertheless, administered free choline may undergo intestinal bacterial degradation without being represented by increased TMAO, as a result of small intestinal bacterial overgrowth (SIBO) in intestinally compromised patients [27].

Our findings are important to optimize future preterm infant nutrition. Assuming that the prenatal choline plasma concentrations are optimal for rapid fetal growth and development, we previously showed that an enteral supply of 50–60 mg/kg/d choline is needed to achieve these prenatal choline plasma concentrations in enterally fed preterm infants [20]. Breast milk or many preterm infant formulae do not meet these requirements, resulting in the pertinence of low plasma choline [1, 10, 21]. Since 2010, the ESPGHAN recommends an enteral choline supply of 8–55 mg/kg/day to very preterm infants [19]. Our data have shown that the upper limit of this broad recommendation should be aimed for [20].

We detected similarly high D9-choline concentrations in plasma after supplementation with D9-choline chloride, D9-phosphoryl-choline and D9-GPC. Due to sparse sampling, it remains unclear whether the minor differences in maximal plasma concentrations at 1 h are clinically relevant and due to substance-specific differences in the absorption kinetics. Previous data from adults may point in this direction [25]: phosphoryl-choline and GPC, which are the major natural choline components in breast milk [21, 34, 35], showed a lower formation of TMAO in adults [25] and tended to result in a lower increase in plasma D9-betaine compared to choline chloride (see Fig. 2B). Since phosphoryl-choline is currently not commercially available as a supplement, we consider GPC the most suitable water-soluble organic choline ester to increase choline plasma concentrations in preterm infants. Although GPC has been used in infant formulas, without apparent adverse effects, its use is still an exception [21].

The observation that native choline concentrations weakly correlated with D9 choline plasma concentrations may indicate that there may be inter-individual differences in absorption and availability of enterally administered choline, which must be considered in future studies. The observation that the GPC group had lower native choline concentrations may indicate that we underestimate the effect of supplementation with GPC on choline plasma concentrations. It may have been of additional interest to determine exact daily baseline choline intake in the study subjects, to see whether native choline concentrations correlate with this intake.

Optimizing fatty acid composition of plasma PC is key to plasma transport and organ delivery of PUFA, such as arachidonic and docosahexaenoic acid [32]. This may be of particular importance for the developing brain and eye, where combined arachidonic and docosahexaenoic acid supplementation improved retinopathy outcome [36]. Docosahexaenoic acid is needed for synaptogenesis, membrane fluidity and retinal receptor activity [37], but arachidonic acid and its elongation product adrenic acid surmount docosahexaenoic acid in brain lipids [12]. Arachidonic acid is a precursor of eicosanoids and also needed for growth and cell signal systems [38‐40]. Decreased levels are associated with neonatal morbidity [41]. It is of special interest that, while D9-POPC (18:1) was supplemented, predominantly D9-PC with linoleic acid (18:2) was measured in plasma, presumably because linoleic acid is the predominant poly-unsaturated fatty acid in western breast milk and infant formulas [31, 42]. In a previous study, we showed that in adipose tissue of term infants, studied immediately after birth, the proportion of arachidonic and docosahexaenoic acid was high in triglycerides and other lipids [43], because these are actively enriched in the fetus prenatally, whereas linoleic acid is retained in the mother [44]. However, in preterm infants at term (after about 10 weeks of feeding in the NICU), the adipose tissue content of arachidonic and docosahexaenoic acid was significantly lower while linoleic acid was greatly increased [43]. According to the processing of D9-POPC during its assimilation described here, co-supplementation of docosahexaenoic and arachidonic acid together with POPC in preterm infants could improve the non-physiologic molecular composition of plasma PC, and potentially help to improve long-term outcome in preterm infants by improving PUFA supply to the brain and other organs. It could also be an option to supply purified DHA-PC and ARA-PC, which can be found in milk fat globular membranes but, at present, these complex mixtures, with many additional compounds and variable content by batch, require further investigation.

All water-soluble D9-choline supplements increased plasma levels of D9-choline and D9-betaine. Highest levels were detected after the intake of D9-choline chloride and D9-GPC, with D9-GPC showing lower D9-betaine formation. Since GPC is a natural component of breast milk, we suggest this as the most suitable water-soluble choline supplement for preterm infants. Absence of plasma D9-choline after D9-POPC may be due to delayed absorption and different metabolism, as it increased plasma D9-PC at the expense of D9-betaine compared to water-soluble components. De- and re-acylation of administered PC as a choline source has the potential to shuttle arachidonic and docosahexaenoic acid into the PC-fraction of chylomicrons, with PC as their major transport form in plasma. (PO)PC also leads to lowest levels of betaine, which is strongly increased in response to water-soluble choline supplements. We therefore conclude that a combination of GPC and PC may be the most suitable choline supplement for preterm infants.