Eur J Nutr DOI 10.1007/s00394-014-0658-3
ORIGINAL CONTRIBUTION
Plasma phospholipids indicate impaired fatty acid homeostasis in preterm infants Wolfgang Bernhard • Marco Raith • Vera Koch • Rebecca Kunze • Christoph Maas • Harald Abele Christian F. Poets • Axel R. Franz
•
Received: 18 July 2013 / Accepted: 13 January 2014 Ó Springer-Verlag Berlin Heidelberg 2014
Abstract Background During fetal development, docosahexaenoic (DHA) and arachidonic acid (ARA) are particularly enriched in brain phospholipids. After preterm delivery, fetal enrichment of DHA and ARA via placental transfer is replaced by enteral and parenteral nutrition, which is rich in linoleic acid (LA) instead. Specific DHA and ARA enrichment of lipoproteins is reflected by plasma phosphatidylcholine (PC) species, whereas plasma phosphatidylethanolamine (PE) composition reflects hepatic stores. Objective We profiled PC and PE species in preterm infant plasma, compared with cord and maternal blood, to assess whether current feeding practice meets fetal conditions in these patients. Design Preterm infant plasma (N = 171, 23–35 w postmenstrual age (PMA), postnatal day 1–103), cord plasma (N = 194) and maternal serum (N = 121) (both 24–41 w PMA) were collected. After lipid extraction, PC and PE
Electronic supplementary material The online version of this article (doi:10.1007/s00394-014-0658-3) contains supplementary material, which is available to authorized users. W. Bernhard (&) M. Raith V. Koch R. Kunze C. Maas C. F. Poets A. R. Franz Department of Neonatology, Faculty of Medicine, Eberhard-Karls-University, Calwer Straße 7, 72076 Tu¨bingen, Germany e-mail: [email protected] H. Abele Department of Gynecology, Faculty of Medicine, Eberhard-Karls-University, Tu¨bingen, Germany A. R. Franz Center for Pediatric Clinical Studies, Faculty of Medicine, Eberhard-Karls-University, Tu¨bingen, Germany
molecular species were analyzed using tandem mass spectrometry. Results Phospholipid concentrations were higher in preterm infant than in cord plasma after correction for PMA. This was mainly due to postnatal increases in LA-containing PC and PE, resulting in decreased fractions of their DHA- and ARA-containing counterparts. These changes in preterm infant plasma phospholipids occurred during the time of transition to full enteral feeds (day 0–10 after delivery). Thereafter, the fraction of ARA-containing phospholipids further decreased, whereas that of DHA slowly reincreased but remained at a level 50 % of that of PMA-matched cord blood. Conclusions The postnatal increase in LA–PC in preterm infant plasma results in decreased fractions of DHA–PC and ARA–PC. These changes are also reflected by PE molecular composition as an indicator of altered hepatic fatty acid homeostasis. They are presumably caused by inadequately high LA, and low ARA and DHA supply, at a stage of development when ARA–PC and DHA–PC should be high, probably reducing the availability of DHA and ARA to the developing brain and contributing to impaired neurodevelopment of preterm infants. Keywords Arachidonic acid Docosahexaenoic acid Linoleic acid Plasma phosphatidylcholine Tandem mass spectrometry Preterm infants Abbreviations ALA ARA DHA EDTA EPA
Alpha-linoleic acid Arachidonic acid Docosahexaenoic acid Ethylene diamine tetraacetate Eicosapentaenoic acid
123
Eur J Nutr
H-ESI HPLC LA LC–ESI–MS/MS
LC-PUFA NICU OA PC PE PEMT PMA SAMe SPH SRM TIC VLDL w
Heated electrospray ionization interface High-performance liquid chromatography Linoleic acid Liquid chromatography–electrospray ionization interface tandem mass spectrometry Long-chain poly-unsaturated fatty acid Neonatal intensive care unit Oleic acid Phosphatidylcholine Phosphatidylethanolamine PE-N-methyltransferase Postmenstrual age S-adenosylmethionine Sphingomyelin Specific reaction monitoring Total ion count Very low-density lipoproteins Week
Introduction Phospholipid homeostasis is essential to growth, differentiation and repair of all organisms. This also applies to glycerophospholipids due to their roles as membrane components, their fatty acids as precursors of messenger molecules (eicosanoids, resolvins) and their function as components in many secretions [1, 2]. Membrane phospholipids comprise 40–45 % phosphatidylcholine (PC) and 20–25 % phosphatidylethanolamine (PE), whereas PC predominates in secretions like lipoproteins [3, 4]. In liver tissue and brain gray matter, PC and PE are enriched in long-chain polyunsaturated fatty acids (LC-PUFA), such as docosahexaenoic acid (DHA), arachidonic acid (ARA) and—to a minor extent—eicosapentaenoic acid (EPA). In plasma, PC is the main carrier of these fatty acids [5–7]. From late second trimester onwards, maternal very low-density lipoproteins (VLDL) provide the placenta with high amounts of LCPUFA–PC, being predominantly synthesized by PE-Nmethyltransferase (PEMT) [5, 8–10]. The placenta then transfers free choline, DHA, ARA [11, 12] and lipoproteinscontaining LC-PUFA–PC to the fetus [13, 14]. Although the fetal intestine and liver contribute, the placenta is the major source of fetal plasma PC, choline and LC-PUFA [15, 16]. There is cumulating evidence that the supply with choline, DHA and ARA is critically important for normal human pre- and postnatal health and cognitive development. The placenta preferentially secretes DHA and ARA into the fetal circulation, but retains LA in the maternal
123
organism, presumably to ensure optimal development [17– 20]. LC-PUFA supply of the brain is mainly achieved by specific uptake of the PC moieties of high- and low-density lipoproteins (HDL, LDL), originating from VLDL and chylomicrons [21]. Plasma DHA–PC and ARA–PC are preferential substrates of endothelial lipase in brain capillaries to ensure DHA, ARA and choline uptake [22–24]. After delivery, the preterm infant is fed by intravenous or enteral route, including supply of fat high in LA. The brain’s DHA and ARA accretion is sensitive to external supply, as high LA uptake interferes with endogenous DHA formation from alpha-linoleic acid (ALA) and impairs cerebral DHA accretion [25]. Because—in contrast to the placenta—the intestine does not preferentially absorb ARA and DHA at the expense of LA, postnatal ARA and DHA availability for the brain depends on the composition of exogenous supply [26]. In preterm infants, this switch from placental provision high in LC-PUFA to supply-dependent fatty acid uptake (high in LA) occurs at a stage, when brain growth is maximal, and a continuous high ARA and DHA supply with low LA would be physiological [19, 27, 28]. Whereas plasma concentrations of DHA-, EPA- and ARA-containing PC species reflect hepatic lipoprotein secretion and availability of these components to peripheral tissues, plasma PE composition reflects the hepatic reservoir of LC-PUFA for specific LC-PUFA–PC synthesis via PEMT as the second most frequent liver phospholipid [5, 7, 29]. To date, it is unclear whether feeding regimens for neonatal intensive care unit (NICU) residents meet their LCPUFA requirements. Moreover, the importance of PC rather than neutral lipids as their plasma carriers and PE as representative for their liver tissue homeostasis has been underestimated [5, 30]. We therefore determined the concentration and molecular composition of PC and PE in preterm infant plasma during their NICU stay and compared these values with those found in cord and maternal blood samples matched for postmenstrual age (PMA). We investigated the kinetics of postnatal PC and PE changes in preterm infants, with particular attention to DHA, EPA and ARA compared with LAand oleic acid (OA)-containing PC and PE subgroups [5, 31].
Methods The study was approved by the local ethics committee, and written consent was obtained from patients or their legal representatives. This study was registered at www.clinicaltrials.gov. Identifier: NCT02027584. Study population Serum from the remainders of clinically indicated blood samples of 121 parturients and EDTA plasma of 194
Eur J Nutr Table 1 Characteristics of the study group
Parturients
Postnatal samples in preterm infants
Number of patients (samples)
121
194
56 (171)
Singleton pregnancy
101
154
42
Gender of child (m/f)
–
101/93
29/27
Samples per patient
1
1
2 (1–5) (1–14)*
Maternal age (years)
32.1 (28.3–36.0)
32.6 (28.7–36.2)
–
(17.9–43.8)*
(17.9–46.2)*
PMA at birth (weeks) Birth weight of child (g) Data are expressed as medians followed by 25th/75th percentiles in brackets. Additionally, min to max range is given where appropriate (*)
Cord blood samples
Postnatal age at time of blood sampling (days)
35.3 (28.3–36.0)
32.4 (28.4–36.7)
(24.0–41.8)*
(24.3–41.7)*
2,790 (1,603–3,321)
1,750 (1,013–2,880)
(505–4,200)*
(410–4,200)*
0 (-1 to ?1)*
0
umbilical cord blood samples of preterm and term infants (24–42 w PMA) was collected from 2008 to 2011 at the local University Women’s Hospital. Additionally, residual EDTA plasma from 171 venous punctures in 56 preterm infants of our neonatal intensive care unit (NICU) (23–35 w PMA; postnatal day 0–103; born at the local University Women’s Hospital) was collected. These punctures were mostly ([70 %) indicated by routine procedures during postnatal follow-up of patients. Data of the study group are shown in Table 1. Serum of parturients (±24 h relative to delivery) and plasma of preterm infants were harvested up to 12 h after venous puncture. Umbilical cord blood was harvested immediately after delivery. Samples were kept at 4 °C until plasma or serum isolation and centrifuged at 2,0009g for 10 min, and cell-free supernatant aspirated and stored in 50–100 lL aliquots at -80 °C. Unit guidelines for enteral and parenteral nutrition of preterm infants During the study period, enteral nutrition (breast milk or preterm infant formula (Beba preterm formula, Nestle Nutrition, Frankfurt, Germany)) was commenced on day 1 of life at 10–15 mL/kg/day and advanced by 15–20 mL/kg/day. Fortification of breast milk was initiated as soon as 150 mL/kg/day was tolerated adding 5–7.5 g multicomponent fortifier FM85Ò (Nestle Nutrition, Frankfurt, Germany) per 100 ml. Parenteral glucose (starting with 6–7 g/kg/day) and amino acid infusions (starting with 2–2.5 g/kg/day) were commenced on postnatal day 1 to complement enteral feeds. Parenteral lipid (20 % fat emulsion (ClinOleicÒ, Baxter, Unterschleißheim, Germany)) and fat soluble vitamins (4 ml/kg/
25.7 (24.6–28.1) (23.4–35.0)* 675 (480–910) (410–1,490)*
19.3 (6.9–36.9) (0.6–103.3)*
day Vitalipid infantÒ, Baxter, Unterschleißheim, Germany) were started on postnatal day 3, if enteral feeding advancements were slower than expected. According to these unit guidelines, supply with fatty acids was calculated for total enteral or total parenteral feeding at fully established nutrition (usually beyond postnatal day 10) based on the fatty acid composition of the respective feeds. This was performed for expressed maternal milk according to the literature [32–35], and for BebaÒ preterm formula and ClinoleicÒ according to the manufacturers’ information. Data of regular daily supply after completion of feeding advancement are indicated in absolute values, relative to the lower limits of recommended daily intake according to the ESPGHAN Committee on Nutrition for enteral [36] and parenteral nutrition [37], and relative to estimated transplacental uptake [38] (for details see: Online Resource 1). However, during periods of severe systemic inflammation or intestinal complications, enteral nutrition and parenteral lipid emulsions, respectively, were reduced or temporarily withheld. These practices were unchanged during the three-year observational period and have been described in detail elsewhere [18]. Chemicals Chloroform was of HPLC grade and from Baker (Deventer, The Netherlands). Methanol, water and ammonium hydroxide were of analytical grade and from Fluka Analytical/Sigma-Aldrich (Munich, Germany). Phospholipid standards were from Sigma-Aldrich or Avanti Polar Lipids Inc. (Alabaster, Alabama, USA). Other reagents were of analytical grade and from various commercial sources.
123
Eur J Nutr
Plasma and serum extraction Glass ware for extraction was cleaned with methanol prior to usage, to avoid sample contamination. Samples were extracted according to Bligh and Dyer [39]. Briefly, 50 lL plasma/serum was transferred into a 12-mL glass vial, spiked with 20 lmol 1,2-di-eicosanoyl-phosphatidylcholine (PC20:0/20:0) and 8 lmol 1,2-dimyristoyl-phosphatidylethanolamine (PE14:0/14:0) in 20 lL trifluoroethanol as internal standards, and 0,35 mL water, 1.2 mL methanol, and 0.4 mL chloroform added. The mixture was stirred and kept for 1 h at -20 °C. After adding 0.8 mL of water and 1.2 mL of chloroform, the samples were vigorously stirred, centrifuged at 3,0009g and 4 °C for 20 min to achieve an organic phase containing the phospholipids. The extract was adjusted to 4 mL and at -80 °C for further analysis with liquid chromatography–electrospray ionization tandem mass spectrometry (LC–ESI–MS/MS). LC–ESI–MS/MS analysis of phospholipids Molecular species of phosphatidylcholine (PC)- and other choline-containing phospholipids (sphingomyelin [SPH], lyso-PC) as well as of phosphatidylethanolamine (PE) were analyzed with liquid chromatography–electrospray ionization tandem mass spectrometry (LC–ESI–MS/MS) as described before [7]. The equipment comprised a Finnigan Surveyor Autosampler Plus equipped with a 100-lL sample loop and a heater, a Finnigan Surveyor MS Pump Plus and a TSQ Quantum Discovery MAX triple quadrupole mass spectrometer equipped with a heated electrospray ionization interface (H-ESI) (Thermo Fisher Scientific, Dreieich, Germany). Samples were dissolved in chloroform/methanol/water (20:78:2,v/v) at a concentration of 3 lmol/L. 25 lL samples were introduced into the mass spectrometer via loop injection from the autosampler maintained at 4 °C. The mobile phase comprised chloroform/methanol/water/25 % NH4OH (20:78:1.7:0.3,v/v). PC molecular species, 16:0-SPH and lyso-PC, were analyzed at positive ionization, using selective reaction monitoring (SRM) of individual components and phosphocholine (mass/charge [m/z] = ?184) as the diagnostic fragment [5, 7]. PE species were analyzed from the neutral loss fragment phosphoethanolamine (m/z = ?141) of the individual masses (for details see: Online Resource 2). Individual masses comprising less than 0.2 % of the total phospholipid class were ignored. These were mostly components with 2 fatty acyl residues containing 20 or 22 carbon units, comprising 0.45 ± 0.14 % (mean ± SD) of analyzed components in maternal and cord blood and 0.38 ± 0.07 % in preterm infants. To characterize the molecular identity of masses attributable to different components (e.g., mass by charge (m/z) = ? 808, which
123
can be oleoyl-arachidonoyl-PC (PC18:1/20:4) as well as stearoyl-eicosapentaenoyl-PC (PC10:8/20:5), PC and PE were isolated from total lipid extracts by solid-phase extraction [40], and daughter ion analysis of fatty acid fragments of individual components performed in negative ionization mode (for details see: Online Resource 2). Total ion counts were corrected for 13C effects of individual components differing in the number of carbon units, and differences in ionization rate according to chain length as described before [5, 7]. PC and PE compositions were expressed as mol% of the phospholipid class being analyzed, whereas plasma concentrations were calculated from the internal standards PC20:0/20:0 and PE14:0/14:0, respectively. For accurateness of repetitive measurement, see Online Resource 2. For comprehensive data presentation, PC and PE molecular species were grouped into those containing mono-unsaturated oleic acid (OA, C18:1), di-unsaturated linoleic acid (LA, C18:2) or an LC-PUFA, such as arachidonic acid (ARA, C20:4), eicosapentaenoic acid (EPA, C20:5) or docosahexaenoic acid (DHA, C22:6) residue. This resulted in a comprehensive structure of data, as all major individual molecular components could be attributed (for details see: Online Resource 3). Statistical analysis Statistical analysis was performed using Instat, version 3.10 (GraphPad, La Jolla, CA, USA). Data were controlled for normal distribution and—as the latter was frequently absent—are provided as medians and 25th/75th percentiles, if not indicated otherwise. Nonparametric Dunn test was used for group comparisons. Regression analysis of individual data points from stationary samples of preterm infants (Fig. 6) was performed, and deviation from linearity assessed. In case of nonlinearity, this was followed by breakpoint analysis according to Ryan et al. [41], using nonparametric smoothing, followed by estimation of a primary breakpoint according to the least squares method, and iteration for final breakpoint determination. P values below 5 % were regarded as being significant.
Results Demographic data of the study population are shown in Table 1. 83, 79 and 75 % of maternal, cord blood and preterm infant samples, respectively, were from singleton pregnancies. PMA at the time of sample harvesting was similar for samples from parturients and cord blood (24–42 w). NICU patients were all preterm, born at B35 w PMA and with \1,500 g birth weight. Postnatal age of preterm infants at the time of blood sampling ranged from
Total concentrations of PC and PE are indicated as the sum of all individual molecular species analyzed with ESI–MS/MS as indicated in ‘‘Methods’’. PE % is the fraction of PE/(PC ? PE). Data are medians (25th–75th percentiles). *, **, *** p \ 0.05, p \ 0.01 and p \ 0.001 versus parturient; p \ 0.01, p \ 0.001 versus fetus
(2.62–3.91) (6.26–8.60) (0.217–0.316) (1.91–2.79) (2.84–3.89)
(0.97–1.41)
(0.023–0.036)
(0.056–0.104)
(1.98–2.91)
3.20***,
(2.22–3.59)
7.4 0.266 2.35***, All
3.304
1.15***
0.029***
0.077***,
2.37***
2.82***
(6.87–9.87)
(2.30–3.30)
8.21
(0.045–0.085) (0.255–0.344)
(0.025–0.037)
0.291
(1.68–2.13) (0.90–1.27) (2.93–3.90)
2.01*, 3.4 37–42
1.04***
0.030***
0.059**
2.78***
3.11*** (2.41–3.93)
(2.64–3.47)
7.84 (7.20–8.55) 0.267 (0.236–0.309) 1.97*, (1.84–2.41) 3.208 (2.85–3.54) 34–36
1.04*** (0.96–1.15)
0.026*** (0.021–0.030)
0.064***, (0.048–0.075)
2.25*** (2.01–2.80)
3.06***
(6.81–8.19) (0.200–0.289) (1.91–2.84) (2.39–3.42)
(0.95–1.39)
(0.020–0.034)
(0.046–0.096)
(1.79–2.38)
(2.52–3.63)
7.39
2.09***
(4.98–7.87) (0.057–0.096)
0.269
0.026***
(0.160–0.282) (1.92–2.73)
2.36
(2.64–3.96)
(1.03–1.44)
32–33
3.145
1.15***
(0.023–0.036)
0.069*,
(1.93–2.82)
(2.87–4.11)
2.96*** 6.15
2.49***
(4.74–6.49) (0.073–0.122)
0.22 2.36
0.031***
(0.177–0.258) (2.12–2.91)
1.26*** 3.44 28–31
(3.03–4.27)
(1.11–1.61)
(0.025–0.037)
0.075**,
(1.99–3.05)
3.41 5.69 0.223 2.54 24–27
3.60
1.38***
0.032***
0.098
2.30***
NICU Fetus Parturient NICU Fetus Parturient NICU Fetus Parturient
Total plasma PE (mmol/L) Total plasma PC (mmol/L) PMA (w)
Table 2 Total PC and PE concentrations in plasma of parturients, fetuses and preterm infants during NICU treatment
PE fraction [(PE/(PC ? PE) 9 100 %]
Eur J Nutr
0.6 to 103 days after delivery, resulting in 24–42 w corrected PMA at sampling. Birth weights of the offspring from parturients and of neonates with cord blood samples ranged 505–4,200 and 410–4,200 g, respectively. Fatty acid supply of NICU patients in absolute values, relative to minima requirements according to recommendations [36, 37], and to placental transfer [38], is shown in Online Resource 1. Data show that with respect to official recommendations, full enteral nutrition with breast milk, which was supplemented with a fat-free multicomponent fortifier in this study, resulted in a high inter-individual variability of fatty acid supply. Nevertheless, for both breast milk and formula feeding, supply with ARA, DHA, LA and ALA was above or close to official recommendations for minimum supply. However, this was not the case for those patients requiring full parenteral nutrition. Remarkably, none of the lipid-feeding regimens hit the mark when taking estimated placental transfer [38] as the standard. Phospholipid concentrations in preterm infants compared with fetuses (cord blood) and parturients Plasma concentrations of total PC in preterm NICU patients were significantly higher than those in cord plasma throughout the PMA range (p \ 0.0001), but tended to be lower than maternal concentrations (Table 2). Total PE and the fraction of PE relative to PC were higher in postnatal plasma of NICU patients than in cord plasma, whereas both were lower than in maternal serum (Table 2). SPH was identical in all groups, with 5.0 (4.5–5.5) %, 5.3 (4.7–5.8) % and 5.1 (4.7–5.8) % of PC in postnatal plasma, cord plasma and maternal serum, respectively (p [ 0.05). Similarly, lyso-PC was 1.6 (1.3–1.9) %, 1.6 (1.3–2.6) % and 0.9 (0.7–1.4) %, respectively, in these groups (p [ 0.05). Alkyl-acyl-PC species were below 2 % of PC in all groups, too (1.16 [0.97–1.31] %, 0.89 [0.79–1.04] % and 0.92 [0.78–1.05] % in cord plasma, maternal serum and plasma from NICU residents, respectively). Alkenylacyl-PE species comprised 9.6 (7.7–11.9) % and 6.3 (5.2–7.3) %, in cord and maternal samples, respectively, but were increased in preterm infants (12.9 [9.9–16.9] %; p \ 0.001). Distribution of individual PC and PE molecular species The pattern of PC and PE molecular species in study samples is demonstrated in Fig. 1a, b. There are no principle differences in the distribution of individual molecular species between maternal, cord and neonatal blood samples, with abundant PC species ranging from m/z = 706 (myristoyl-palmitoyl-PC; PC14:0/16:0) to 808 (oleoyl-arachidonoyl-PC; PC18:1/20:4), and PE species
123
Eur J Nutr 35
A PC molecular species
Parturient Fetus
30
Preterm Infant
Mol% of PC
25 20 15 10 5
B PE molecular species
PC20:5/22:6
PC22:6/22:6
PC22:0/20:5
PC20:5/20:5
PC20:4/20:5
PC18:1/20:4
PC20:0/22:6
PC18:1/18:2
PC18:0/22:6
PC18:0/18:2*
PC18:0/22:5
PC18:0/22:4
aaPC18:0/20:5
PC18:0/18:1
PC18:0/20:4
PC16:0/22:6
PC18:0/18:0
aaPC16:0/22:6
PC16:0/20:4
PC16:0/20:5
PC16:0/18:3
PC16:0/18:1
PC16:0/18:2
PC16:0/18:0
PC16:0/16:0
PC16:0/16:1
PC14:0/16:0
30
PC14:0/22:6
0
Parturient Fetus Preterm Infant
25
20
Mol% of PE
Fig. 1 Pattern of molecular composition in maternal and fetal plasma PC (a) and PE (b) from 24 to 42 w postmenstrual age. PC and PE molecular species were analyzed with tandem mass spectrometry as described in ‘‘Methods’’. PE masses (b) below m/z = 704 (palmitoyl-linoleoyl-PE and beyond 806 (di-eicosanoyl PE) were below 0.1 % of total PE and are not indicated. PC phosphatidylcholine, PE phosphatidylethanolamine; aa in front of the PC or PE represents a component comprising one acyl and one alk(en)yl group rather than 2 acyl groups. The numbers, separated by a slash, define the two fatty acids in the sn-1 and sn-2 position of the glycerol backbone of PC and PE, respectively. The figure before the colon defines the number of carbon units, while the figure behind defines the number of double bonds of the respective fatty acid. Consequently, PC16:0/18:2 is a PC species comprising a hexadecanoic (palmitic) acid and a octadecadienoic (linoleic) acid residue, connected in ester bonds to the glycerol backbone. For details see Online Resource 2
15
10
5
PE16:0/18:3 PE16:0/18:2 PE16:0/18:1 PE16:0/18:0 aaPE16:0/20:5 aaPE16:0/20:4 aaPE18:1/18:2 aaPE18:0/18:2 aaPE18:0/18:1 aaPE18:0/18:0 PE14:0/22:6 PE16:0/20:5 PE16:0/20:4 PE18:1/18:2 PE18:0/18:2 PE18:0/18:1 aaPE16:0/22:6 aaPE18:1/20:4 aaPE18:0/20:4 aaPE18:0/20:3 PE16:0/22:6 PE18:1/20:4 PE18:0/20:4 PE18:0/20:3 aaPE18:0/22:6 aaPE18:0/22:5 aaPE18:0/22:4 PE20:5/20:5 PE20:4/20:5 PE20:4/20:4 PE18:0/22:6 PE18:0/22:5 PE18:0/22:4
0
ranging from m/z = 714 (palmitoyl-oleoyl-PE; PE16:0/ 18:2) to 796 (stearoyl-docosatetraenoyl-PC; PE18:0/ 22:4). PC was dominated by PC16:0/18:1, PC16:0/18:2, PC16:0/20:4, PC16:0/22:6, PC18:0/18:1, PC18:0/20:4, PC18:0/22:6, PC18:0/18:2, PC18:1/18:2 and PC18:1/20:4 (Fig. 1a). Alkyl-acyl-components were nearly absent in PC. PE comprised more alkenyl-acyl-components (aaPE), but was dominated by diacylated PE species as well. These were of the C16:0, C18:0 and C18:1 series, particularly comprising PE16:0/20:4, PE18:1/18:2, PE18:0/18:2, PE18:0/18:1, PE16:0/22:6, PE18:1/20:4, PE18:0/20:4 and PE18:0/22:6 (Fig. 1b).
123
In spite of these principal similarities, their molar fractions differed between parturient, cord and preterm infant serum/plasma. Notably, palmitoyl-linoleyl-PC (PC16:0/18:2) was the highest fraction in parturients, whereas in cord blood, palmitoyl-arachidonoyl-PC and stearoyl-arachidonoyl-PC (PC16:0/20:4, PC18:0/20:4) dominated. In preterm infant plasma, palmitoyl-docosahexaenoyl-PC (PC16:0/22:6) was low compared with both maternal and cord blood (p \ 0.001). Similarly for PE, stearoyl-linoleyl-PE (PE18:0/18:2) was low in cord blood, but in preterm infants, it was as high as in parturients, whereas palmitoyl-docosahexaenoyl-PE (PE16:0/
Eur J Nutr
∗∗∗ ∗∗
∗∗∗ ††
∗∗∗ †
†††
28-31
∗,†† ∗∗∗
∗∗∗ ∗,†††
†††
32-33
34-36
∗∗∗ ∗∗∗
24-27
∗∗∗ ∗∗∗
0.2
∗∗∗ ∗∗∗,†††
0.3
∗∗∗ ∗∗∗
F DHA-PC
∗∗∗ ∗∗∗
mmol/L
∗∗∗ ∗∗∗
0.4 ∗∗∗ ∗∗
∗∗∗ †
∗∗∗ †††
C LC-PUFA-PC
0.5 0
0.04
0.00
∗∗∗ †††
1.0
0.06
0.02
0.0 1.5
E EPA-PC
∗∗∗
mmol/L
∗∗∗
0.5
∗,†
∗,††† ∗∗∗
∗∗∗
†††
∗,††† ∗∗∗
∗,††† ∗∗∗
1.0
0.08
†††
0.0
∗∗∗
0.0
1.5
∗∗∗
0.4 0.2
B LA-PC
∗∗∗ †††
0.6
0.2
2.0
mmol/L
0.8
Parturient Fetus Preterm Infant
D ARA-PC
†††
mmol/L
∗,†† ∗∗∗
∗∗∗
∗∗∗
†††
††† ∗∗∗
0.6
2.5
mmol/L
†††
†††
0.8
0.4
1.0
∗∗∗
A OA-PC
∗∗∗
mmol/L
1.0
0.1 24-27
28-31
32-33
34-36
37-41
PMA (w)
0.0
37-41
PMA (w)
Fig. 2 Concentrations of PC subgroups in maternal, fetal and preterm infant plasma according to postmenstrual age (PMA). Maternal and cord blood samples are shown according to PMA at birth, whereas preterm infant samples, derived from clinically indicated blood samples of NICU patients, were shown according to corrected PMA at the time of blood sampling. Data are molar concentrations of PC subgroups and are depicted as medians and 25th/75th percentiles. For
individual PC species used for subgrouping see Online Resource 3. OA oleic acid, LA linoleic acid, LC-PUFA long-chain poly-unsaturated fatty acid, ARA arachidonic acid, EPA eicosapentaenoic acid, DHA docosahexaenoic acid, PC phosphatidylcholine. *p \ 0.05, **p \ 0.01, ***p \ 0.001 relative to maternal blood; p \ 0.05, p \ 0.01, p \ 0.001 relative to cord blood
22:6) was low in preterm infants compared with maternal and fetal samples.
plasma of preterm NICU patients compared with cord plasma and between maternal and fetal values (Fig. 2a, b). This applied to LC-PUFA–PC for NICU patients up to 33 w PMA. After 33 w PMA, LC-PUFA–PC concentrations were as low in postnatal as in fetal samples (Fig. 2c). Subdividing LC-PUFA–PC into ARA–PC, EPA–PC and DHA–PC (Fig. 2d–f), however, demonstrated that ARA–PC was higher in postnatal than in cord plasma until 36 w PMA, whereas EPA–PC was increased over fetal values throughout. In contrast, DHA–PC concentrations in postnatal preterm infant plasma were not increased, but identical to cord plasma throughout the PMA range investigated (Fig. 2f). As not only absolute concentration but also the relation between individual PC components may affect organ supply with LC-PUFA [25], fractions of PC subgroups relative to total PC are presented. Figure 3 demonstrates that the fraction of OA–PC is similar in all patient groups, except for slightly higher values in NICU patients at 24–27 w PMA (Fig. 3a). LA–PC, however, comprising only 26.2 (24.4–28.4) % of total PC in cord plasma throughout gestation, was increased in NICU patients throughout (24–42 w corrected PMA) (Fig. 3b). Consequently, the fraction of LC-
Concentration and composition of PC subgroups in preterm infants compared with fetuses (cord blood) and parturients For clarity, molecular species were summarized into PC subgroups containing an oleic (x9-C18:1, OA–PC), a linoleic (x6-C18:2, LA–PC) or a long-chain poly-unsaturated fatty acid (LC-PUFA–PC) residue, and concentrations depicted in Fig. 2a–c. LC-PUFA–PC were further subdivided into those containing an arachidonic (x6-C20:4; ARA–PC), eicosapentaenoic (x3-C20:5; EPA–PC) or docosahexaenoic acid (x3-C22:6; DHA–PC) residue (Fig. 2d–f). In Fig. 2, data from preterm infants were restricted to samples taken from 10-day postnatal age onwards, i.e., after establishing full enteral and/or parenteral nutrition, for comparison with their fetal and parturient counterparts. Samples were grouped according to postmenstrual age (PMA), as a surrogate parameter of developmental age. Concentrations of OA–PC and LA–PC were higher in
123
Eur J Nutr
∗∗∗ ††
∗∗∗ ∗,†††
∗∗∗ †††
∗∗∗ ∗,†††
∗ ∗
1.0 0.5 0.0
∗∗∗ †††
∗ ∗∗∗,†††
∗∗∗,†††
10
∗∗∗,†††
Mol %
20
15
F DHA-PC ∗∗∗,†††
∗∗∗
20 †††
∗∗∗ †††
∗∗∗ †††
∗∗∗ †††
∗∗∗
C LC-PUFA-PC †††
40
1.5
†
Mol %
∗∗∗
∗∗∗
∗∗∗
20
60
E EPA-PC ∗
†††
†††
††† ∗∗∗
∗,†††
40
2.0
∗∗
0 †††
10
0
B LA-PC
∗∗∗
20
10
0
Mol %
30
Mother Fetus Preterm Infant
D ARA-PC †††
Mol %
∗
40
20
60
Mol %
50
∗∗∗,†††
30
∗∗∗
Mol %
40
A OA-PC
5 0
0 24-27
28-31
32-33
34-36
37-41
PMA (w)
24-27
28-31
32-33
34-36
37-41
PMA (w)
Fig. 3 Fractions of PC subgroups in maternal, fetal and preterm infant plasma according to postmenstrual age (PMA). Maternal and cord blood samples are shown according to PMA at birth, whereas preterm infant samples, derived from clinically indicated blood samples of NICU patients, are depicted according to corrected PMA at the time of blood sampling. Data are molar percentages of PC subgroups relative to total PC and are depicted as medians and 25th/
75th percentiles. For individual PC species used for subgrouping see Online Resource 3. OA oleic acid, LA linoleic acid, LC-PUFA longchain poly-unsaturated fatty acid, ARA arachidonic acid, EPA eicosapentaenoic acid, DHA docosahexaenoic acid, PC phosphatidylcholine; *p \ 0.05, **p \ 0.01, ***p \ 0.001 relative to maternal blood; p \ 0.001 relative to cord blood
PUFA–PC was lower in NICU patients compared with cord plasma (overall median 29.2 [26.7–32.8] %; p \ 0.001) versus (47.0 [43.8–50.5] %), and as low as in maternal blood (Fig. 3c). In detail, ARA–PC was consistently lower in NICU patients compared with cord plasma, and similar to maternal values (Fig. 3d), whereas EPA–PC showed no group differences (Fig. 3e). In contrast, the fraction of DHA–PC continuously increased in cord plasma and surmounted maternal values from 34 w PMA onwards (Fig. 3f). In NICU patients, however, values were 50 % lower than in their fetal counterparts, and even lower than in parturients up to 36 w PMA (Fig. 3f). These differences between cord blood and NICU residents’ plasma were independent from gender throughout (data not shown).
throughout the PMA range studied (not shown), due to the general postnatal increase in plasma PE (Table 2). Analysis of PE subgroup fractions (Table 3b), however, demonstrates that PE composition in NICU patients resembled maternal rather than fetal values: LC-PUFA–PE was decreased from 68.9 (65.8–71.4) % in cord plasma to 50.3 (45.3–54.8) % in NICU patients (p \ 0.001). This applied to ARA– and EPA–PE (relative decrease by 18–20 %) as well as DHA–PE (relative decrease by 40 %) (Table 3b). As for PC, there were no gender differences in PE parameters (data not shown).
Concentrations and composition of PE subgroups in preterm infants compared with fetuses (cord blood) and parturients Concentrations of PE subgroups in plasma of NICU patients were between fetal and maternal values (Table 3a),
123
Postnatal changes in PC and PE in preterm infants We then assessed the kinetics of changes from the intrauterine PC and PE plasma profile to that of preterm infants during neonatal intensive care. The postnatal increase in total PC was solely due to an increase in LA–PC, indicated by its positive correlation with postnatal age (r = 0.5768; p \ 0.0001), whereas neither OA–PC nor LC-PUFA–PC concentrations correlated with postnatal age (Fig. 4a).
Eur J Nutr Table 3 Concentration (A) and composition (B) of mono-, di- and poly-unsaturated PE species in parturients, fetuses and NICU patients Molecular species
OA
A: PE concentration (lmol/L)
B: PE composition (%)
Parturient
Fetus
NICU
Parturient
Fetus
NICU
19.4
1.0***
6.4***,
7.9
4.2***
7.2
(13.6–25.1)
(0.5–1.5)
(3.9–8.7)
(6.8–8.9)
(3.2–5.4)
(5.7–10.2)
LA
62
1.8***
15.2***,
24.8
7.8***
18.5***,
(43.0–79.7)
(1.1–2.4)
(11.9–20.4)
(22.2–28.2)
(6.1–9.3)
(15.7–22.2)
LC–PUFA
150.06
18.7***
43.4***,
55
70.7***
50.6*,
(120.9–181.5)
(15.5–24.7)
(31.6–56.0)
(51.5–59.6)
(66.8–731.5)
(45.7–54.3)
ARA
71 (59.2–91.2)
9.8*** (7.3–13.4)
26.8***, (19.2–35.3)
26.7 (25.1–29.8)
39.5*** (36.5–43.8)
31.2***, (27.0–34.4)
EPA
10.4
1.1***
2.2***,
3.5
3.1
2.7*
(7.6–14.1)
(0.7–1.9)
(1.3–2.9)
(2.8–4.1)
(2.2–4.3)
(1.5–3.6)
64.5
7.4***
13.2***,
24.9
26.8
15.9***,
(44.7–80.8)
(6.0–9.7)
(9.7–17.2)
(21.0–27.8)
(23.1–30.6)
(12.8–19.2)
DHA
PE species were analyzed with ESI–MS/MS and grouped according to their content of oleic (OA), linoleic (LA), arachidonic (ARA), eicosapentaenoic (EPA) or docosahexaenoic acid (DHA). LC-PUFA is defined as the sum of ARA, EPA and DHA. Data are given as medians (25th– 75th percentile). For individual PE species used for subgrouping see Online Resource 3. ** p \ 0.01, *** p \ 0.001 versus parturient; p \ 0.01; p \ 0.001 versus fetus
Consequently, the fraction of LA–PC was directly (r = 0.7025), and those of OA–PC and LC-PUFA–PC inversely (r = -0.4543 and r = -0.4010, respectively) related to postnatal age (p \ 0.0001) (Fig. 4b). Similarly, LA–PE concentrations increased with postnatal age (r = 0.3398; p \ 0.0001), whereas those of OA–PE and LC-PUFA–PE did not (p [ 0.05) (Fig. 5a). The fraction of LA–PE continuously increased after delivery, whereas that of OA–PE remained unchanged (Fig. 5b). The fraction of LC-PUFA–PE decreased from fetal values within 1 w after delivery and then remained at a constant value (Fig. 5b). As these data indicate that plasma lipid profiles in NICU patients are determined by birth and/or postnatal care rather than a developmental program (depending on PMA), individual values for LC-PUFA–PC and –PE were plotted against postnatal age for regression analyses (Fig. 6). Only data for ARA–PE and EPA–PE fitted simple linear regression models (Fig. 6b), whereas all other data suggested models with two different first-order kinetics. Breakpoint analyses (Table 4) indicated that in NICU patients, a steep linear decrease for ARA–PC and DHA–PC (each p \ 0.0001) was followed by a continuing but slower decrease for ARA–PC and a slow increase for DHA–PC with breakpoints ranging from 3 to 8 days, and with ARA–PC decreasing the fastest. The fraction of EPA–PC initially increased until day 8, but did not change significantly thereafter. In contrast, ARA–PE continuously decreased with increasing postnatal age (p \ 0.0001), and EPA–PE did not seem to change with postnatal age (p [ 0.05), whereas the fraction of DHA–PE, just like DHA– PC, sharply decreased initially, followed by a slow increase beyond day 4 (Fig. 6b).
Discussion This study addresses the homeostasis and postnatal changes in essential and semi-essential LC-PUFA in preterm infants. It is well known that DHA and ARA are essential to brain and retina development, contribute to anti-inflammatory regulation together with EPA, and that a timely adjusted supply of these conditionally essential fatty acids to the fetus is required. This is particularly important as growth rate and lipid accretion, particularly by the brain and liver, is increasing from 24 w postmenstrual age (PMA) onwards, and much higher than in term infants after birth [42, 43]. Moreover, it is known that particularly PC is an important carrier and biomarker of DHA, EPA and ARA, in plasma [5, 7, 44]. In light of such knowledge, we analyzed the concentration and molecular composition of plasma PC and PE in preterm infants in comparison with cord blood and maternal samples matched for PMA, and in relation to postnatal age. We provide, for the first time, reference values derived from cord blood on plasma phospholipid homeostasis throughout the clinically relevant period of pregnancy (24–42 w PMA). We furthermore provide evidence that the homeostasis of DHA- and ARA-containing plasma lipids is impaired in preterm infants, probably due to current regimes in neonatal nutrition. We addressed this by level of individual phospholipid molecular species and subgroups of PC and PE, rather than by fatty acid composition of total phospholipids. This is important as plasma PC represents hepatic synthesis and secretion of lipoprotein phospholipids, whereas plasma PE, as an overflow from liver
123
Eur J Nutr
Concentration (mmol/L)
1.5
OA-PC LA-PC LC-PUFA-PC
r=0.5768 p
ORIGINAL CONTRIBUTION
Plasma phospholipids indicate impaired fatty acid homeostasis in preterm infants Wolfgang Bernhard • Marco Raith • Vera Koch • Rebecca Kunze • Christoph Maas • Harald Abele Christian F. Poets • Axel R. Franz
•
Received: 18 July 2013 / Accepted: 13 January 2014 Ó Springer-Verlag Berlin Heidelberg 2014
Abstract Background During fetal development, docosahexaenoic (DHA) and arachidonic acid (ARA) are particularly enriched in brain phospholipids. After preterm delivery, fetal enrichment of DHA and ARA via placental transfer is replaced by enteral and parenteral nutrition, which is rich in linoleic acid (LA) instead. Specific DHA and ARA enrichment of lipoproteins is reflected by plasma phosphatidylcholine (PC) species, whereas plasma phosphatidylethanolamine (PE) composition reflects hepatic stores. Objective We profiled PC and PE species in preterm infant plasma, compared with cord and maternal blood, to assess whether current feeding practice meets fetal conditions in these patients. Design Preterm infant plasma (N = 171, 23–35 w postmenstrual age (PMA), postnatal day 1–103), cord plasma (N = 194) and maternal serum (N = 121) (both 24–41 w PMA) were collected. After lipid extraction, PC and PE
Electronic supplementary material The online version of this article (doi:10.1007/s00394-014-0658-3) contains supplementary material, which is available to authorized users. W. Bernhard (&) M. Raith V. Koch R. Kunze C. Maas C. F. Poets A. R. Franz Department of Neonatology, Faculty of Medicine, Eberhard-Karls-University, Calwer Straße 7, 72076 Tu¨bingen, Germany e-mail: [email protected] H. Abele Department of Gynecology, Faculty of Medicine, Eberhard-Karls-University, Tu¨bingen, Germany A. R. Franz Center for Pediatric Clinical Studies, Faculty of Medicine, Eberhard-Karls-University, Tu¨bingen, Germany
molecular species were analyzed using tandem mass spectrometry. Results Phospholipid concentrations were higher in preterm infant than in cord plasma after correction for PMA. This was mainly due to postnatal increases in LA-containing PC and PE, resulting in decreased fractions of their DHA- and ARA-containing counterparts. These changes in preterm infant plasma phospholipids occurred during the time of transition to full enteral feeds (day 0–10 after delivery). Thereafter, the fraction of ARA-containing phospholipids further decreased, whereas that of DHA slowly reincreased but remained at a level 50 % of that of PMA-matched cord blood. Conclusions The postnatal increase in LA–PC in preterm infant plasma results in decreased fractions of DHA–PC and ARA–PC. These changes are also reflected by PE molecular composition as an indicator of altered hepatic fatty acid homeostasis. They are presumably caused by inadequately high LA, and low ARA and DHA supply, at a stage of development when ARA–PC and DHA–PC should be high, probably reducing the availability of DHA and ARA to the developing brain and contributing to impaired neurodevelopment of preterm infants. Keywords Arachidonic acid Docosahexaenoic acid Linoleic acid Plasma phosphatidylcholine Tandem mass spectrometry Preterm infants Abbreviations ALA ARA DHA EDTA EPA
Alpha-linoleic acid Arachidonic acid Docosahexaenoic acid Ethylene diamine tetraacetate Eicosapentaenoic acid
123
Eur J Nutr
H-ESI HPLC LA LC–ESI–MS/MS
LC-PUFA NICU OA PC PE PEMT PMA SAMe SPH SRM TIC VLDL w
Heated electrospray ionization interface High-performance liquid chromatography Linoleic acid Liquid chromatography–electrospray ionization interface tandem mass spectrometry Long-chain poly-unsaturated fatty acid Neonatal intensive care unit Oleic acid Phosphatidylcholine Phosphatidylethanolamine PE-N-methyltransferase Postmenstrual age S-adenosylmethionine Sphingomyelin Specific reaction monitoring Total ion count Very low-density lipoproteins Week
Introduction Phospholipid homeostasis is essential to growth, differentiation and repair of all organisms. This also applies to glycerophospholipids due to their roles as membrane components, their fatty acids as precursors of messenger molecules (eicosanoids, resolvins) and their function as components in many secretions [1, 2]. Membrane phospholipids comprise 40–45 % phosphatidylcholine (PC) and 20–25 % phosphatidylethanolamine (PE), whereas PC predominates in secretions like lipoproteins [3, 4]. In liver tissue and brain gray matter, PC and PE are enriched in long-chain polyunsaturated fatty acids (LC-PUFA), such as docosahexaenoic acid (DHA), arachidonic acid (ARA) and—to a minor extent—eicosapentaenoic acid (EPA). In plasma, PC is the main carrier of these fatty acids [5–7]. From late second trimester onwards, maternal very low-density lipoproteins (VLDL) provide the placenta with high amounts of LCPUFA–PC, being predominantly synthesized by PE-Nmethyltransferase (PEMT) [5, 8–10]. The placenta then transfers free choline, DHA, ARA [11, 12] and lipoproteinscontaining LC-PUFA–PC to the fetus [13, 14]. Although the fetal intestine and liver contribute, the placenta is the major source of fetal plasma PC, choline and LC-PUFA [15, 16]. There is cumulating evidence that the supply with choline, DHA and ARA is critically important for normal human pre- and postnatal health and cognitive development. The placenta preferentially secretes DHA and ARA into the fetal circulation, but retains LA in the maternal
123
organism, presumably to ensure optimal development [17– 20]. LC-PUFA supply of the brain is mainly achieved by specific uptake of the PC moieties of high- and low-density lipoproteins (HDL, LDL), originating from VLDL and chylomicrons [21]. Plasma DHA–PC and ARA–PC are preferential substrates of endothelial lipase in brain capillaries to ensure DHA, ARA and choline uptake [22–24]. After delivery, the preterm infant is fed by intravenous or enteral route, including supply of fat high in LA. The brain’s DHA and ARA accretion is sensitive to external supply, as high LA uptake interferes with endogenous DHA formation from alpha-linoleic acid (ALA) and impairs cerebral DHA accretion [25]. Because—in contrast to the placenta—the intestine does not preferentially absorb ARA and DHA at the expense of LA, postnatal ARA and DHA availability for the brain depends on the composition of exogenous supply [26]. In preterm infants, this switch from placental provision high in LC-PUFA to supply-dependent fatty acid uptake (high in LA) occurs at a stage, when brain growth is maximal, and a continuous high ARA and DHA supply with low LA would be physiological [19, 27, 28]. Whereas plasma concentrations of DHA-, EPA- and ARA-containing PC species reflect hepatic lipoprotein secretion and availability of these components to peripheral tissues, plasma PE composition reflects the hepatic reservoir of LC-PUFA for specific LC-PUFA–PC synthesis via PEMT as the second most frequent liver phospholipid [5, 7, 29]. To date, it is unclear whether feeding regimens for neonatal intensive care unit (NICU) residents meet their LCPUFA requirements. Moreover, the importance of PC rather than neutral lipids as their plasma carriers and PE as representative for their liver tissue homeostasis has been underestimated [5, 30]. We therefore determined the concentration and molecular composition of PC and PE in preterm infant plasma during their NICU stay and compared these values with those found in cord and maternal blood samples matched for postmenstrual age (PMA). We investigated the kinetics of postnatal PC and PE changes in preterm infants, with particular attention to DHA, EPA and ARA compared with LAand oleic acid (OA)-containing PC and PE subgroups [5, 31].
Methods The study was approved by the local ethics committee, and written consent was obtained from patients or their legal representatives. This study was registered at www.clinicaltrials.gov. Identifier: NCT02027584. Study population Serum from the remainders of clinically indicated blood samples of 121 parturients and EDTA plasma of 194
Eur J Nutr Table 1 Characteristics of the study group
Parturients
Postnatal samples in preterm infants
Number of patients (samples)
121
194
56 (171)
Singleton pregnancy
101
154
42
Gender of child (m/f)
–
101/93
29/27
Samples per patient
1
1
2 (1–5) (1–14)*
Maternal age (years)
32.1 (28.3–36.0)
32.6 (28.7–36.2)
–
(17.9–43.8)*
(17.9–46.2)*
PMA at birth (weeks) Birth weight of child (g) Data are expressed as medians followed by 25th/75th percentiles in brackets. Additionally, min to max range is given where appropriate (*)
Cord blood samples
Postnatal age at time of blood sampling (days)
35.3 (28.3–36.0)
32.4 (28.4–36.7)
(24.0–41.8)*
(24.3–41.7)*
2,790 (1,603–3,321)
1,750 (1,013–2,880)
(505–4,200)*
(410–4,200)*
0 (-1 to ?1)*
0
umbilical cord blood samples of preterm and term infants (24–42 w PMA) was collected from 2008 to 2011 at the local University Women’s Hospital. Additionally, residual EDTA plasma from 171 venous punctures in 56 preterm infants of our neonatal intensive care unit (NICU) (23–35 w PMA; postnatal day 0–103; born at the local University Women’s Hospital) was collected. These punctures were mostly ([70 %) indicated by routine procedures during postnatal follow-up of patients. Data of the study group are shown in Table 1. Serum of parturients (±24 h relative to delivery) and plasma of preterm infants were harvested up to 12 h after venous puncture. Umbilical cord blood was harvested immediately after delivery. Samples were kept at 4 °C until plasma or serum isolation and centrifuged at 2,0009g for 10 min, and cell-free supernatant aspirated and stored in 50–100 lL aliquots at -80 °C. Unit guidelines for enteral and parenteral nutrition of preterm infants During the study period, enteral nutrition (breast milk or preterm infant formula (Beba preterm formula, Nestle Nutrition, Frankfurt, Germany)) was commenced on day 1 of life at 10–15 mL/kg/day and advanced by 15–20 mL/kg/day. Fortification of breast milk was initiated as soon as 150 mL/kg/day was tolerated adding 5–7.5 g multicomponent fortifier FM85Ò (Nestle Nutrition, Frankfurt, Germany) per 100 ml. Parenteral glucose (starting with 6–7 g/kg/day) and amino acid infusions (starting with 2–2.5 g/kg/day) were commenced on postnatal day 1 to complement enteral feeds. Parenteral lipid (20 % fat emulsion (ClinOleicÒ, Baxter, Unterschleißheim, Germany)) and fat soluble vitamins (4 ml/kg/
25.7 (24.6–28.1) (23.4–35.0)* 675 (480–910) (410–1,490)*
19.3 (6.9–36.9) (0.6–103.3)*
day Vitalipid infantÒ, Baxter, Unterschleißheim, Germany) were started on postnatal day 3, if enteral feeding advancements were slower than expected. According to these unit guidelines, supply with fatty acids was calculated for total enteral or total parenteral feeding at fully established nutrition (usually beyond postnatal day 10) based on the fatty acid composition of the respective feeds. This was performed for expressed maternal milk according to the literature [32–35], and for BebaÒ preterm formula and ClinoleicÒ according to the manufacturers’ information. Data of regular daily supply after completion of feeding advancement are indicated in absolute values, relative to the lower limits of recommended daily intake according to the ESPGHAN Committee on Nutrition for enteral [36] and parenteral nutrition [37], and relative to estimated transplacental uptake [38] (for details see: Online Resource 1). However, during periods of severe systemic inflammation or intestinal complications, enteral nutrition and parenteral lipid emulsions, respectively, were reduced or temporarily withheld. These practices were unchanged during the three-year observational period and have been described in detail elsewhere [18]. Chemicals Chloroform was of HPLC grade and from Baker (Deventer, The Netherlands). Methanol, water and ammonium hydroxide were of analytical grade and from Fluka Analytical/Sigma-Aldrich (Munich, Germany). Phospholipid standards were from Sigma-Aldrich or Avanti Polar Lipids Inc. (Alabaster, Alabama, USA). Other reagents were of analytical grade and from various commercial sources.
123
Eur J Nutr
Plasma and serum extraction Glass ware for extraction was cleaned with methanol prior to usage, to avoid sample contamination. Samples were extracted according to Bligh and Dyer [39]. Briefly, 50 lL plasma/serum was transferred into a 12-mL glass vial, spiked with 20 lmol 1,2-di-eicosanoyl-phosphatidylcholine (PC20:0/20:0) and 8 lmol 1,2-dimyristoyl-phosphatidylethanolamine (PE14:0/14:0) in 20 lL trifluoroethanol as internal standards, and 0,35 mL water, 1.2 mL methanol, and 0.4 mL chloroform added. The mixture was stirred and kept for 1 h at -20 °C. After adding 0.8 mL of water and 1.2 mL of chloroform, the samples were vigorously stirred, centrifuged at 3,0009g and 4 °C for 20 min to achieve an organic phase containing the phospholipids. The extract was adjusted to 4 mL and at -80 °C for further analysis with liquid chromatography–electrospray ionization tandem mass spectrometry (LC–ESI–MS/MS). LC–ESI–MS/MS analysis of phospholipids Molecular species of phosphatidylcholine (PC)- and other choline-containing phospholipids (sphingomyelin [SPH], lyso-PC) as well as of phosphatidylethanolamine (PE) were analyzed with liquid chromatography–electrospray ionization tandem mass spectrometry (LC–ESI–MS/MS) as described before [7]. The equipment comprised a Finnigan Surveyor Autosampler Plus equipped with a 100-lL sample loop and a heater, a Finnigan Surveyor MS Pump Plus and a TSQ Quantum Discovery MAX triple quadrupole mass spectrometer equipped with a heated electrospray ionization interface (H-ESI) (Thermo Fisher Scientific, Dreieich, Germany). Samples were dissolved in chloroform/methanol/water (20:78:2,v/v) at a concentration of 3 lmol/L. 25 lL samples were introduced into the mass spectrometer via loop injection from the autosampler maintained at 4 °C. The mobile phase comprised chloroform/methanol/water/25 % NH4OH (20:78:1.7:0.3,v/v). PC molecular species, 16:0-SPH and lyso-PC, were analyzed at positive ionization, using selective reaction monitoring (SRM) of individual components and phosphocholine (mass/charge [m/z] = ?184) as the diagnostic fragment [5, 7]. PE species were analyzed from the neutral loss fragment phosphoethanolamine (m/z = ?141) of the individual masses (for details see: Online Resource 2). Individual masses comprising less than 0.2 % of the total phospholipid class were ignored. These were mostly components with 2 fatty acyl residues containing 20 or 22 carbon units, comprising 0.45 ± 0.14 % (mean ± SD) of analyzed components in maternal and cord blood and 0.38 ± 0.07 % in preterm infants. To characterize the molecular identity of masses attributable to different components (e.g., mass by charge (m/z) = ? 808, which
123
can be oleoyl-arachidonoyl-PC (PC18:1/20:4) as well as stearoyl-eicosapentaenoyl-PC (PC10:8/20:5), PC and PE were isolated from total lipid extracts by solid-phase extraction [40], and daughter ion analysis of fatty acid fragments of individual components performed in negative ionization mode (for details see: Online Resource 2). Total ion counts were corrected for 13C effects of individual components differing in the number of carbon units, and differences in ionization rate according to chain length as described before [5, 7]. PC and PE compositions were expressed as mol% of the phospholipid class being analyzed, whereas plasma concentrations were calculated from the internal standards PC20:0/20:0 and PE14:0/14:0, respectively. For accurateness of repetitive measurement, see Online Resource 2. For comprehensive data presentation, PC and PE molecular species were grouped into those containing mono-unsaturated oleic acid (OA, C18:1), di-unsaturated linoleic acid (LA, C18:2) or an LC-PUFA, such as arachidonic acid (ARA, C20:4), eicosapentaenoic acid (EPA, C20:5) or docosahexaenoic acid (DHA, C22:6) residue. This resulted in a comprehensive structure of data, as all major individual molecular components could be attributed (for details see: Online Resource 3). Statistical analysis Statistical analysis was performed using Instat, version 3.10 (GraphPad, La Jolla, CA, USA). Data were controlled for normal distribution and—as the latter was frequently absent—are provided as medians and 25th/75th percentiles, if not indicated otherwise. Nonparametric Dunn test was used for group comparisons. Regression analysis of individual data points from stationary samples of preterm infants (Fig. 6) was performed, and deviation from linearity assessed. In case of nonlinearity, this was followed by breakpoint analysis according to Ryan et al. [41], using nonparametric smoothing, followed by estimation of a primary breakpoint according to the least squares method, and iteration for final breakpoint determination. P values below 5 % were regarded as being significant.
Results Demographic data of the study population are shown in Table 1. 83, 79 and 75 % of maternal, cord blood and preterm infant samples, respectively, were from singleton pregnancies. PMA at the time of sample harvesting was similar for samples from parturients and cord blood (24–42 w). NICU patients were all preterm, born at B35 w PMA and with \1,500 g birth weight. Postnatal age of preterm infants at the time of blood sampling ranged from
Total concentrations of PC and PE are indicated as the sum of all individual molecular species analyzed with ESI–MS/MS as indicated in ‘‘Methods’’. PE % is the fraction of PE/(PC ? PE). Data are medians (25th–75th percentiles). *, **, *** p \ 0.05, p \ 0.01 and p \ 0.001 versus parturient; p \ 0.01, p \ 0.001 versus fetus
(2.62–3.91) (6.26–8.60) (0.217–0.316) (1.91–2.79) (2.84–3.89)
(0.97–1.41)
(0.023–0.036)
(0.056–0.104)
(1.98–2.91)
3.20***,
(2.22–3.59)
7.4 0.266 2.35***, All
3.304
1.15***
0.029***
0.077***,
2.37***
2.82***
(6.87–9.87)
(2.30–3.30)
8.21
(0.045–0.085) (0.255–0.344)
(0.025–0.037)
0.291
(1.68–2.13) (0.90–1.27) (2.93–3.90)
2.01*, 3.4 37–42
1.04***
0.030***
0.059**
2.78***
3.11*** (2.41–3.93)
(2.64–3.47)
7.84 (7.20–8.55) 0.267 (0.236–0.309) 1.97*, (1.84–2.41) 3.208 (2.85–3.54) 34–36
1.04*** (0.96–1.15)
0.026*** (0.021–0.030)
0.064***, (0.048–0.075)
2.25*** (2.01–2.80)
3.06***
(6.81–8.19) (0.200–0.289) (1.91–2.84) (2.39–3.42)
(0.95–1.39)
(0.020–0.034)
(0.046–0.096)
(1.79–2.38)
(2.52–3.63)
7.39
2.09***
(4.98–7.87) (0.057–0.096)
0.269
0.026***
(0.160–0.282) (1.92–2.73)
2.36
(2.64–3.96)
(1.03–1.44)
32–33
3.145
1.15***
(0.023–0.036)
0.069*,
(1.93–2.82)
(2.87–4.11)
2.96*** 6.15
2.49***
(4.74–6.49) (0.073–0.122)
0.22 2.36
0.031***
(0.177–0.258) (2.12–2.91)
1.26*** 3.44 28–31
(3.03–4.27)
(1.11–1.61)
(0.025–0.037)
0.075**,
(1.99–3.05)
3.41 5.69 0.223 2.54 24–27
3.60
1.38***
0.032***
0.098
2.30***
NICU Fetus Parturient NICU Fetus Parturient NICU Fetus Parturient
Total plasma PE (mmol/L) Total plasma PC (mmol/L) PMA (w)
Table 2 Total PC and PE concentrations in plasma of parturients, fetuses and preterm infants during NICU treatment
PE fraction [(PE/(PC ? PE) 9 100 %]
Eur J Nutr
0.6 to 103 days after delivery, resulting in 24–42 w corrected PMA at sampling. Birth weights of the offspring from parturients and of neonates with cord blood samples ranged 505–4,200 and 410–4,200 g, respectively. Fatty acid supply of NICU patients in absolute values, relative to minima requirements according to recommendations [36, 37], and to placental transfer [38], is shown in Online Resource 1. Data show that with respect to official recommendations, full enteral nutrition with breast milk, which was supplemented with a fat-free multicomponent fortifier in this study, resulted in a high inter-individual variability of fatty acid supply. Nevertheless, for both breast milk and formula feeding, supply with ARA, DHA, LA and ALA was above or close to official recommendations for minimum supply. However, this was not the case for those patients requiring full parenteral nutrition. Remarkably, none of the lipid-feeding regimens hit the mark when taking estimated placental transfer [38] as the standard. Phospholipid concentrations in preterm infants compared with fetuses (cord blood) and parturients Plasma concentrations of total PC in preterm NICU patients were significantly higher than those in cord plasma throughout the PMA range (p \ 0.0001), but tended to be lower than maternal concentrations (Table 2). Total PE and the fraction of PE relative to PC were higher in postnatal plasma of NICU patients than in cord plasma, whereas both were lower than in maternal serum (Table 2). SPH was identical in all groups, with 5.0 (4.5–5.5) %, 5.3 (4.7–5.8) % and 5.1 (4.7–5.8) % of PC in postnatal plasma, cord plasma and maternal serum, respectively (p [ 0.05). Similarly, lyso-PC was 1.6 (1.3–1.9) %, 1.6 (1.3–2.6) % and 0.9 (0.7–1.4) %, respectively, in these groups (p [ 0.05). Alkyl-acyl-PC species were below 2 % of PC in all groups, too (1.16 [0.97–1.31] %, 0.89 [0.79–1.04] % and 0.92 [0.78–1.05] % in cord plasma, maternal serum and plasma from NICU residents, respectively). Alkenylacyl-PE species comprised 9.6 (7.7–11.9) % and 6.3 (5.2–7.3) %, in cord and maternal samples, respectively, but were increased in preterm infants (12.9 [9.9–16.9] %; p \ 0.001). Distribution of individual PC and PE molecular species The pattern of PC and PE molecular species in study samples is demonstrated in Fig. 1a, b. There are no principle differences in the distribution of individual molecular species between maternal, cord and neonatal blood samples, with abundant PC species ranging from m/z = 706 (myristoyl-palmitoyl-PC; PC14:0/16:0) to 808 (oleoyl-arachidonoyl-PC; PC18:1/20:4), and PE species
123
Eur J Nutr 35
A PC molecular species
Parturient Fetus
30
Preterm Infant
Mol% of PC
25 20 15 10 5
B PE molecular species
PC20:5/22:6
PC22:6/22:6
PC22:0/20:5
PC20:5/20:5
PC20:4/20:5
PC18:1/20:4
PC20:0/22:6
PC18:1/18:2
PC18:0/22:6
PC18:0/18:2*
PC18:0/22:5
PC18:0/22:4
aaPC18:0/20:5
PC18:0/18:1
PC18:0/20:4
PC16:0/22:6
PC18:0/18:0
aaPC16:0/22:6
PC16:0/20:4
PC16:0/20:5
PC16:0/18:3
PC16:0/18:1
PC16:0/18:2
PC16:0/18:0
PC16:0/16:0
PC16:0/16:1
PC14:0/16:0
30
PC14:0/22:6
0
Parturient Fetus Preterm Infant
25
20
Mol% of PE
Fig. 1 Pattern of molecular composition in maternal and fetal plasma PC (a) and PE (b) from 24 to 42 w postmenstrual age. PC and PE molecular species were analyzed with tandem mass spectrometry as described in ‘‘Methods’’. PE masses (b) below m/z = 704 (palmitoyl-linoleoyl-PE and beyond 806 (di-eicosanoyl PE) were below 0.1 % of total PE and are not indicated. PC phosphatidylcholine, PE phosphatidylethanolamine; aa in front of the PC or PE represents a component comprising one acyl and one alk(en)yl group rather than 2 acyl groups. The numbers, separated by a slash, define the two fatty acids in the sn-1 and sn-2 position of the glycerol backbone of PC and PE, respectively. The figure before the colon defines the number of carbon units, while the figure behind defines the number of double bonds of the respective fatty acid. Consequently, PC16:0/18:2 is a PC species comprising a hexadecanoic (palmitic) acid and a octadecadienoic (linoleic) acid residue, connected in ester bonds to the glycerol backbone. For details see Online Resource 2
15
10
5
PE16:0/18:3 PE16:0/18:2 PE16:0/18:1 PE16:0/18:0 aaPE16:0/20:5 aaPE16:0/20:4 aaPE18:1/18:2 aaPE18:0/18:2 aaPE18:0/18:1 aaPE18:0/18:0 PE14:0/22:6 PE16:0/20:5 PE16:0/20:4 PE18:1/18:2 PE18:0/18:2 PE18:0/18:1 aaPE16:0/22:6 aaPE18:1/20:4 aaPE18:0/20:4 aaPE18:0/20:3 PE16:0/22:6 PE18:1/20:4 PE18:0/20:4 PE18:0/20:3 aaPE18:0/22:6 aaPE18:0/22:5 aaPE18:0/22:4 PE20:5/20:5 PE20:4/20:5 PE20:4/20:4 PE18:0/22:6 PE18:0/22:5 PE18:0/22:4
0
ranging from m/z = 714 (palmitoyl-oleoyl-PE; PE16:0/ 18:2) to 796 (stearoyl-docosatetraenoyl-PC; PE18:0/ 22:4). PC was dominated by PC16:0/18:1, PC16:0/18:2, PC16:0/20:4, PC16:0/22:6, PC18:0/18:1, PC18:0/20:4, PC18:0/22:6, PC18:0/18:2, PC18:1/18:2 and PC18:1/20:4 (Fig. 1a). Alkyl-acyl-components were nearly absent in PC. PE comprised more alkenyl-acyl-components (aaPE), but was dominated by diacylated PE species as well. These were of the C16:0, C18:0 and C18:1 series, particularly comprising PE16:0/20:4, PE18:1/18:2, PE18:0/18:2, PE18:0/18:1, PE16:0/22:6, PE18:1/20:4, PE18:0/20:4 and PE18:0/22:6 (Fig. 1b).
123
In spite of these principal similarities, their molar fractions differed between parturient, cord and preterm infant serum/plasma. Notably, palmitoyl-linoleyl-PC (PC16:0/18:2) was the highest fraction in parturients, whereas in cord blood, palmitoyl-arachidonoyl-PC and stearoyl-arachidonoyl-PC (PC16:0/20:4, PC18:0/20:4) dominated. In preterm infant plasma, palmitoyl-docosahexaenoyl-PC (PC16:0/22:6) was low compared with both maternal and cord blood (p \ 0.001). Similarly for PE, stearoyl-linoleyl-PE (PE18:0/18:2) was low in cord blood, but in preterm infants, it was as high as in parturients, whereas palmitoyl-docosahexaenoyl-PE (PE16:0/
Eur J Nutr
∗∗∗ ∗∗
∗∗∗ ††
∗∗∗ †
†††
28-31
∗,†† ∗∗∗
∗∗∗ ∗,†††
†††
32-33
34-36
∗∗∗ ∗∗∗
24-27
∗∗∗ ∗∗∗
0.2
∗∗∗ ∗∗∗,†††
0.3
∗∗∗ ∗∗∗
F DHA-PC
∗∗∗ ∗∗∗
mmol/L
∗∗∗ ∗∗∗
0.4 ∗∗∗ ∗∗
∗∗∗ †
∗∗∗ †††
C LC-PUFA-PC
0.5 0
0.04
0.00
∗∗∗ †††
1.0
0.06
0.02
0.0 1.5
E EPA-PC
∗∗∗
mmol/L
∗∗∗
0.5
∗,†
∗,††† ∗∗∗
∗∗∗
†††
∗,††† ∗∗∗
∗,††† ∗∗∗
1.0
0.08
†††
0.0
∗∗∗
0.0
1.5
∗∗∗
0.4 0.2
B LA-PC
∗∗∗ †††
0.6
0.2
2.0
mmol/L
0.8
Parturient Fetus Preterm Infant
D ARA-PC
†††
mmol/L
∗,†† ∗∗∗
∗∗∗
∗∗∗
†††
††† ∗∗∗
0.6
2.5
mmol/L
†††
†††
0.8
0.4
1.0
∗∗∗
A OA-PC
∗∗∗
mmol/L
1.0
0.1 24-27
28-31
32-33
34-36
37-41
PMA (w)
0.0
37-41
PMA (w)
Fig. 2 Concentrations of PC subgroups in maternal, fetal and preterm infant plasma according to postmenstrual age (PMA). Maternal and cord blood samples are shown according to PMA at birth, whereas preterm infant samples, derived from clinically indicated blood samples of NICU patients, were shown according to corrected PMA at the time of blood sampling. Data are molar concentrations of PC subgroups and are depicted as medians and 25th/75th percentiles. For
individual PC species used for subgrouping see Online Resource 3. OA oleic acid, LA linoleic acid, LC-PUFA long-chain poly-unsaturated fatty acid, ARA arachidonic acid, EPA eicosapentaenoic acid, DHA docosahexaenoic acid, PC phosphatidylcholine. *p \ 0.05, **p \ 0.01, ***p \ 0.001 relative to maternal blood; p \ 0.05, p \ 0.01, p \ 0.001 relative to cord blood
22:6) was low in preterm infants compared with maternal and fetal samples.
plasma of preterm NICU patients compared with cord plasma and between maternal and fetal values (Fig. 2a, b). This applied to LC-PUFA–PC for NICU patients up to 33 w PMA. After 33 w PMA, LC-PUFA–PC concentrations were as low in postnatal as in fetal samples (Fig. 2c). Subdividing LC-PUFA–PC into ARA–PC, EPA–PC and DHA–PC (Fig. 2d–f), however, demonstrated that ARA–PC was higher in postnatal than in cord plasma until 36 w PMA, whereas EPA–PC was increased over fetal values throughout. In contrast, DHA–PC concentrations in postnatal preterm infant plasma were not increased, but identical to cord plasma throughout the PMA range investigated (Fig. 2f). As not only absolute concentration but also the relation between individual PC components may affect organ supply with LC-PUFA [25], fractions of PC subgroups relative to total PC are presented. Figure 3 demonstrates that the fraction of OA–PC is similar in all patient groups, except for slightly higher values in NICU patients at 24–27 w PMA (Fig. 3a). LA–PC, however, comprising only 26.2 (24.4–28.4) % of total PC in cord plasma throughout gestation, was increased in NICU patients throughout (24–42 w corrected PMA) (Fig. 3b). Consequently, the fraction of LC-
Concentration and composition of PC subgroups in preterm infants compared with fetuses (cord blood) and parturients For clarity, molecular species were summarized into PC subgroups containing an oleic (x9-C18:1, OA–PC), a linoleic (x6-C18:2, LA–PC) or a long-chain poly-unsaturated fatty acid (LC-PUFA–PC) residue, and concentrations depicted in Fig. 2a–c. LC-PUFA–PC were further subdivided into those containing an arachidonic (x6-C20:4; ARA–PC), eicosapentaenoic (x3-C20:5; EPA–PC) or docosahexaenoic acid (x3-C22:6; DHA–PC) residue (Fig. 2d–f). In Fig. 2, data from preterm infants were restricted to samples taken from 10-day postnatal age onwards, i.e., after establishing full enteral and/or parenteral nutrition, for comparison with their fetal and parturient counterparts. Samples were grouped according to postmenstrual age (PMA), as a surrogate parameter of developmental age. Concentrations of OA–PC and LA–PC were higher in
123
Eur J Nutr
∗∗∗ ††
∗∗∗ ∗,†††
∗∗∗ †††
∗∗∗ ∗,†††
∗ ∗
1.0 0.5 0.0
∗∗∗ †††
∗ ∗∗∗,†††
∗∗∗,†††
10
∗∗∗,†††
Mol %
20
15
F DHA-PC ∗∗∗,†††
∗∗∗
20 †††
∗∗∗ †††
∗∗∗ †††
∗∗∗ †††
∗∗∗
C LC-PUFA-PC †††
40
1.5
†
Mol %
∗∗∗
∗∗∗
∗∗∗
20
60
E EPA-PC ∗
†††
†††
††† ∗∗∗
∗,†††
40
2.0
∗∗
0 †††
10
0
B LA-PC
∗∗∗
20
10
0
Mol %
30
Mother Fetus Preterm Infant
D ARA-PC †††
Mol %
∗
40
20
60
Mol %
50
∗∗∗,†††
30
∗∗∗
Mol %
40
A OA-PC
5 0
0 24-27
28-31
32-33
34-36
37-41
PMA (w)
24-27
28-31
32-33
34-36
37-41
PMA (w)
Fig. 3 Fractions of PC subgroups in maternal, fetal and preterm infant plasma according to postmenstrual age (PMA). Maternal and cord blood samples are shown according to PMA at birth, whereas preterm infant samples, derived from clinically indicated blood samples of NICU patients, are depicted according to corrected PMA at the time of blood sampling. Data are molar percentages of PC subgroups relative to total PC and are depicted as medians and 25th/
75th percentiles. For individual PC species used for subgrouping see Online Resource 3. OA oleic acid, LA linoleic acid, LC-PUFA longchain poly-unsaturated fatty acid, ARA arachidonic acid, EPA eicosapentaenoic acid, DHA docosahexaenoic acid, PC phosphatidylcholine; *p \ 0.05, **p \ 0.01, ***p \ 0.001 relative to maternal blood; p \ 0.001 relative to cord blood
PUFA–PC was lower in NICU patients compared with cord plasma (overall median 29.2 [26.7–32.8] %; p \ 0.001) versus (47.0 [43.8–50.5] %), and as low as in maternal blood (Fig. 3c). In detail, ARA–PC was consistently lower in NICU patients compared with cord plasma, and similar to maternal values (Fig. 3d), whereas EPA–PC showed no group differences (Fig. 3e). In contrast, the fraction of DHA–PC continuously increased in cord plasma and surmounted maternal values from 34 w PMA onwards (Fig. 3f). In NICU patients, however, values were 50 % lower than in their fetal counterparts, and even lower than in parturients up to 36 w PMA (Fig. 3f). These differences between cord blood and NICU residents’ plasma were independent from gender throughout (data not shown).
throughout the PMA range studied (not shown), due to the general postnatal increase in plasma PE (Table 2). Analysis of PE subgroup fractions (Table 3b), however, demonstrates that PE composition in NICU patients resembled maternal rather than fetal values: LC-PUFA–PE was decreased from 68.9 (65.8–71.4) % in cord plasma to 50.3 (45.3–54.8) % in NICU patients (p \ 0.001). This applied to ARA– and EPA–PE (relative decrease by 18–20 %) as well as DHA–PE (relative decrease by 40 %) (Table 3b). As for PC, there were no gender differences in PE parameters (data not shown).
Concentrations and composition of PE subgroups in preterm infants compared with fetuses (cord blood) and parturients Concentrations of PE subgroups in plasma of NICU patients were between fetal and maternal values (Table 3a),
123
Postnatal changes in PC and PE in preterm infants We then assessed the kinetics of changes from the intrauterine PC and PE plasma profile to that of preterm infants during neonatal intensive care. The postnatal increase in total PC was solely due to an increase in LA–PC, indicated by its positive correlation with postnatal age (r = 0.5768; p \ 0.0001), whereas neither OA–PC nor LC-PUFA–PC concentrations correlated with postnatal age (Fig. 4a).
Eur J Nutr Table 3 Concentration (A) and composition (B) of mono-, di- and poly-unsaturated PE species in parturients, fetuses and NICU patients Molecular species
OA
A: PE concentration (lmol/L)
B: PE composition (%)
Parturient
Fetus
NICU
Parturient
Fetus
NICU
19.4
1.0***
6.4***,
7.9
4.2***
7.2
(13.6–25.1)
(0.5–1.5)
(3.9–8.7)
(6.8–8.9)
(3.2–5.4)
(5.7–10.2)
LA
62
1.8***
15.2***,
24.8
7.8***
18.5***,
(43.0–79.7)
(1.1–2.4)
(11.9–20.4)
(22.2–28.2)
(6.1–9.3)
(15.7–22.2)
LC–PUFA
150.06
18.7***
43.4***,
55
70.7***
50.6*,
(120.9–181.5)
(15.5–24.7)
(31.6–56.0)
(51.5–59.6)
(66.8–731.5)
(45.7–54.3)
ARA
71 (59.2–91.2)
9.8*** (7.3–13.4)
26.8***, (19.2–35.3)
26.7 (25.1–29.8)
39.5*** (36.5–43.8)
31.2***, (27.0–34.4)
EPA
10.4
1.1***
2.2***,
3.5
3.1
2.7*
(7.6–14.1)
(0.7–1.9)
(1.3–2.9)
(2.8–4.1)
(2.2–4.3)
(1.5–3.6)
64.5
7.4***
13.2***,
24.9
26.8
15.9***,
(44.7–80.8)
(6.0–9.7)
(9.7–17.2)
(21.0–27.8)
(23.1–30.6)
(12.8–19.2)
DHA
PE species were analyzed with ESI–MS/MS and grouped according to their content of oleic (OA), linoleic (LA), arachidonic (ARA), eicosapentaenoic (EPA) or docosahexaenoic acid (DHA). LC-PUFA is defined as the sum of ARA, EPA and DHA. Data are given as medians (25th– 75th percentile). For individual PE species used for subgrouping see Online Resource 3. ** p \ 0.01, *** p \ 0.001 versus parturient; p \ 0.01; p \ 0.001 versus fetus
Consequently, the fraction of LA–PC was directly (r = 0.7025), and those of OA–PC and LC-PUFA–PC inversely (r = -0.4543 and r = -0.4010, respectively) related to postnatal age (p \ 0.0001) (Fig. 4b). Similarly, LA–PE concentrations increased with postnatal age (r = 0.3398; p \ 0.0001), whereas those of OA–PE and LC-PUFA–PE did not (p [ 0.05) (Fig. 5a). The fraction of LA–PE continuously increased after delivery, whereas that of OA–PE remained unchanged (Fig. 5b). The fraction of LC-PUFA–PE decreased from fetal values within 1 w after delivery and then remained at a constant value (Fig. 5b). As these data indicate that plasma lipid profiles in NICU patients are determined by birth and/or postnatal care rather than a developmental program (depending on PMA), individual values for LC-PUFA–PC and –PE were plotted against postnatal age for regression analyses (Fig. 6). Only data for ARA–PE and EPA–PE fitted simple linear regression models (Fig. 6b), whereas all other data suggested models with two different first-order kinetics. Breakpoint analyses (Table 4) indicated that in NICU patients, a steep linear decrease for ARA–PC and DHA–PC (each p \ 0.0001) was followed by a continuing but slower decrease for ARA–PC and a slow increase for DHA–PC with breakpoints ranging from 3 to 8 days, and with ARA–PC decreasing the fastest. The fraction of EPA–PC initially increased until day 8, but did not change significantly thereafter. In contrast, ARA–PE continuously decreased with increasing postnatal age (p \ 0.0001), and EPA–PE did not seem to change with postnatal age (p [ 0.05), whereas the fraction of DHA–PE, just like DHA– PC, sharply decreased initially, followed by a slow increase beyond day 4 (Fig. 6b).
Discussion This study addresses the homeostasis and postnatal changes in essential and semi-essential LC-PUFA in preterm infants. It is well known that DHA and ARA are essential to brain and retina development, contribute to anti-inflammatory regulation together with EPA, and that a timely adjusted supply of these conditionally essential fatty acids to the fetus is required. This is particularly important as growth rate and lipid accretion, particularly by the brain and liver, is increasing from 24 w postmenstrual age (PMA) onwards, and much higher than in term infants after birth [42, 43]. Moreover, it is known that particularly PC is an important carrier and biomarker of DHA, EPA and ARA, in plasma [5, 7, 44]. In light of such knowledge, we analyzed the concentration and molecular composition of plasma PC and PE in preterm infants in comparison with cord blood and maternal samples matched for PMA, and in relation to postnatal age. We provide, for the first time, reference values derived from cord blood on plasma phospholipid homeostasis throughout the clinically relevant period of pregnancy (24–42 w PMA). We furthermore provide evidence that the homeostasis of DHA- and ARA-containing plasma lipids is impaired in preterm infants, probably due to current regimes in neonatal nutrition. We addressed this by level of individual phospholipid molecular species and subgroups of PC and PE, rather than by fatty acid composition of total phospholipids. This is important as plasma PC represents hepatic synthesis and secretion of lipoprotein phospholipids, whereas plasma PE, as an overflow from liver
123
Eur J Nutr
Concentration (mmol/L)
1.5
OA-PC LA-PC LC-PUFA-PC
r=0.5768 p
