Introduction

Reference ranges are needed for medical decision making. They are used for suspicion, confirmation or exclusion of disease, its prognosis, follow-up, and treatment, as well as for the evaluation of safety of nutritional or pharmaceutical products. Reference limits should be transposable to other laboratories; for the interpretation of data additional information should be given e.g. on pre-analytical factors, on the analytical methods and the statistics used. Recommendations have been made by expert panels of the International Federation of Clinical Chemistry (IFCC) on the theory, production and use of reference intervals (IFCC 1987a, b). Recent guidelines have been approved by the Clinical and Laboratory Standards Institute (CLSI 2013) as well as recommendations for pediatric reference limits (Jung and Khosrow 2009).

In pediatrics, reference limits should be established in healthy populations of newborns, infants, and children. Partition in age groups takes into account that major physiological changes occur already within the first few months of life (Dupont 2003), due to growth (velocity), organ development, changing nutrient composition, and protein requirements. In particular, plasma composition in young infants might be affected by nutritional factors. However, formula feeding has been modified in the last decades as reflected by the recommendations of the EFGHAN (Koletzko et al. 2005) in order to avoid an excessive supply or imbalance of protein composition and amino acids during early organ development. Reference limits need to adopt this and should generally allow distinction of breast and formula fed infants.

However the number of samples obtained from non-hospitalized healthy infants has often been too small to fulfill current international requirements for reference limits. Consequently, results of infants of 1 month up to 2 years of age have often been pooled (Lockitch et al. 1988; Applegarth et al. 1979; Lepage et al. 1997; Armstrong and Stave 1973a, b; Lepage et al. 1997; Ghoshal and Soldin 2003; Duran 2008) without further distinction of the post-neonatal period. Information on the reference population and pre-analytical factors is often missing and at least many older publications presented reference limits as mean values and standard deviation (SD), which is not adequate for parameters of which data are not normally distributed. To our knowledge age-related reference limits for plasma amino acids or common clinical chemistry parameters with sufficient numbers of infants below 1 year of age have not been published yet at all.

During three multicenter studies aiming the evaluation of safety of new infant formulas, we had the opportunity to measure plasma concentrations of amino acids and of common clinical chemistry parameters in healthy infants under highly-controlled feeding conditions. Collected data were sufficient to establish corresponding reference limits at one and four months of age and to test whether the results differ between these age groups.

Study populations and methods

Study populations

The target population in all three multicenter studies of Nestec Inc (Vevey, Switzerland) had to meet the same eligibility criteria including informed consent of the parents for the local or institutional ethical committees. The inclusion criteria were: healthy term infants, i.e. gestational age had to be between 37 and 42 weeks and birth weight between 2500 and 4200 g (>10th and <90th percentile for boys and girls). Infants were excluded if they had malformations, congenital coronary or vascular diseases, serious diseases of the gastrointestinal system, kidneys, liver, CNS and/or metabolic diseases; intensive care within the first 21 days of life or hospital admissions; evidently pathological biochemical markers; haemolytic plasma; gross outliers i.e. values below the 0.25 centile - 3 × interquartile range (IQR) or values above the 0.75 centile + (3 \( \overset{\cdot }{\times } \) IQR). Data from all three studies were pooled after verification that populations did not significantly differ in baseline criteria.

In all studies a control group was exclusively breast-fed at libitum since birth while the experimental groups received different formulas ad libitum. In the first study, formulas differed in protein quantity and quality (Räihä et al. 2002). The protein/energy content was either 2.2–2.3 g (control group) or 1.8 g protein/100 kcal (experimental groups). The second study used a whey-based formula of reduced protein content (1.8–1.9 g protein/100 kcal) along with pro- and/or prebiotics. In the third study, infants were fed a moderately hydrolyzed hypoallergenic formula (Beba Start HA®, Nestlé) with or without additional polyunsaturated fatty acids; the protein content was 2.3 g (control group) and 1.85 g/100 kcal (experimental group), respectively. The parents of formula-fed infants documented in diaries the consumption of formula during the three days preceding the control visits. The daily protein intake was averaged over these three days. In breast-fed infants the intake of milk was not assessed because the procedure of weighing before and after each feed would have been disruptive and stressful for mother and child.

Sample collection and handling

Blood samples of at least 2 ml were taken from a peripheral vein by butterfly needles and collected in heparinized tubes (Lithium heparinate monovettes, Sarstedt, Switzerland). Samples were centrifuged at 1500 g and 4 °C for 10 minutes within 30 minutes from draw. Plasma was removed and stored at -20 or -70 °C until dry ice shipment to the central laboratory every 2 or 6 weeks, respectively.

Analytical methods

All plasma analyses were done in the clinical chemistry laboratory at the CHUV (Lausanne Switzerland) which is accredited by the Swiss Federal Office of Metrology and Accreditation in concordance with standard ISO/IEC 17025.

Amino acids

Sample preparation: Lithium heparinate plasma which had been sent on dry ice was unfrozen and immediately deproteinized with a solution containing 5-sulfosalicylic acid (Merck; final concentration 32 g/L) and the internal standards. The plasma was vortexed for 60 seconds and left standing for 10 minutes before it was centrifuged twice (15 minutes 15600 × g at 4 °C; Beckman centrifuge J2-21).

Amino acids were analyzed using the automatic analyzer system 6300 (Beckman Instruments, Fullerton CA, USA) and corresponding standards for calibration. Peak surface integration was done with a TotalChrome, Nelson instrument v. 6.2.1 (Perkin Elmer AG, Schwerzenbach, Switzerland). Furthermore the retention time (CV% <0.4) and the ratio of the areas under the curve at 570 nm/440 nm were calculated automatically for each peak and compared with the ratio of the standards to exclude co-elution of other ninhydrin positive substances.

For the measurement of total tryptophan concentration, plasma was deproteinized with 10 % trichloroacetic acid (TCA) in order to liberate the albumin–bound tryptophan; the TCA-solution also contained 2-methyltryptophan as internal standard. Upon addition of the TCA solution the sample was vortexed, cooled in ice water for 1 h and centrifuged as for the other amino acids. Total tryptophan was separated by HPLC (Hewlett Packard model 1090) using a RP-18 precolumn and a Superspher 10-RP-18 column (Merck, Darmstadt, Germany). For detection, a spectrofluorimeter LC 240 (Perkin Elmer) was used with excitation wavelength at 278 nm and emission at 363 nm. The quantification of peaks was done with Borwin Software v. 1.21.60 (JMBS Developments SA; Varian medical systems, Baden Switzerland).

General clinical chemistry

All parameters were analyzed on Hitachi 917 automatic analyzers (Roche Diagnostics, Basel, Switzerland) equipped with Roche reagents and calibrators (Electronic supplemental material; Table S7).

Quality assurance

Internal quality control pools of amino acids were prepared, aliquoted and stored at -80 °C. Pools, blanks, and calibrators were run every 10th analysis (at least once per week) and after each calibration or change of reagents. Inter-series precision (CV%) was <2.5 %, 3.5 %, and <4 % at amino acid concentrations of >100, 50–100, and 20–50 μmol/L, respectively; except for asparagine which was 5.0 % at 55 μmol/L. Below concentrations of 20 μmol/L, CV% was 8.8 % for hydroxyproline and 12.4 % for aspartate (at 10 and 4 μmol/L, respectively). The laboratory participated in external quality assessment programs: schemes of the European Research Network for evaluation and improvement of Screening, diagnosis and treatment of Inherited Metabolic Disorders (ERNDIM) were used for amino acids (8 samples/year) and that of the Centre Suisse de Contrôle de Qualité (CSCQ) was applied for the clinical chemistry (12 samples/year; for details see electronic supplemental material; Table S8–10).

Statistics

Data of breast-and formula-fed infants were analyzed separately and after pooling. Data distribution was evaluated with Anderson Darling and Kolmogorov Smirnov tests. Differences between groups were tested by the Wilcoxon-Mann-Whitney (WMW) test. The 0.025, 0.05, 0.1, 0.5, 0.9, 0.95, 0.975 centiles and interquartile ranges (IQR) were computed. For 90 % confidence intervals (CI) of centiles a distribution-free method was used (IFCC 1987a). We corrected the fact that 195 infants were tested both at 1 and 4 months by correcting the P-values by the method of Hochberg and Benjamini (1990). P-values < 0.05 were considered significant.We used the Kolmogorov-Smirnov method for comparing the distribution of amino acids in infants aged 1 and 4 months fed formulas of low or high protein content. Statistical analyses were done using SAS 9.3 TS1M0 for Windows (SAS Institute Inc., Cary, NC) and Analyse-it 3 for Excel®, standard edition (Analyse-it Software Ltd., Leeds, UK).

Results

Plasma samples of 157 male and 135 female infants aged 1-month (23–43 days) were collected. Out of 292 of these infants, 62 were breast-fed, 197 formula-fed only, 28 were both breast- and formula-fed, and for 5 infants nutritional information was missing. At 4-months of age (113–133 days) plasma samples were collected from 121 male and 108 female infants. Of these 229 infants, 49 were breast-fed, 158 were formula-fed only, 18 were both breast- and formula-fed, and for 4 infants nutritional information was missing.

Tables 1 and 2 show the relevant upper and lower centiles of the analytes, median values and IQRs at both ages, for breast- and formula-fed infants combined (Tables 1 or 4) and separately for exclusively breast-fed (Tables 2, 5) and exclusively formula-fed infants (Tables 3, 6). The time elapsed from last feed until sampling was not significantly different for any two groups, neither age nor differences in diets or formulas. Reference limits and median values are given in bold. The distribution-free 90 % CI of the limits is given (in brackets) in Tables 1 and 4.

Table 1 Reference limits (centiles) for one month of age. based on all infants breast- and formula-fed combined
Full size table
Table 2 Reference limits (centiles) for one month of age. based on breast-fed infants
Full size table
Table 3 Reference limits (centiles) for one month of age based on formula-fed infants
Full size table
Table 4 Reference limits (centiles) for age 4 months. based on all infants breast- and formula-fed combined
Full size table

The data distribution of the amino acids and of most clinical chemistry parameters was not Gaussian (except for urea and total calcium); a log normal distribution fitted best the amino acid results with the exception of cystine (see below). Distribution of the reported protein intake (g/day) was Gaussian.

Differences between age groups

  • All infants (Tables 1 and 4)

    At 4 month of age most plasma amino acid concentrations were significantly lower or not different than at 1 month of age except for 2-aminobutyrate (Table 4).

    Regarding the general biochemistry, plasma levels of ferritin, alkaline phosphatase (AP) and potassium were significantly lower at the age of 4 months than at 1 month whereas those of glucose, cholesterol, total protein and albumin were significantly higher. No differences between both age groups were apparent for sodium, chloride, calcium, creatinine and urea (Table 4).

  • Breast-fed infants (Tables 2 and 5)

    Table 5 Reference limits (centiles) for age 4 months. based on breast-fed infants
    Full size table

    At 4 months of age no amino acid concentration was higher than at 1 month of age (p in Table 6). The difference between the two age groups was not significant for 2-aminobutyrate, alanine, asparagine, glutamine (and the sum of glutamine and glutamate), 1-methylhistidine, ornithine, taurine, tryptophan and tyrosine. All other amino acid concentrations were significantly lower at 4 months than at 1 month of age.

    Table 6 Reference limits (centiles) for age 4 months. based on formula-fed infants
    Full size table

    Total protein, albumin and ferritin were significantly higher and potassium significantly lower at 4 months than at 1 month of age. No significantly differences were found for the other biochemical markers.

  • Formula-fed infants (Tables 3 and 6)

    Most amino acids were significantly lower at 4 months than at 1 month of age except of 2-aminobutyrate which was higher at 4 months than at 1 month of age. Differences were not significant for alanine, isoleucine, 1-methylhistidine, phenylalanine, serine, total tryptophan, tyrosine, and valine.

    Regarding general biochemistry, beyond the significant differences revealed for the pooled data set, the formula-fed infants showed significantly lower potassium at 4 months than at 1 month of age. Urea, the urea/creatinine ratio, cholesterol, total protein and albumin were significantly higher at 4 months than at 1 month of age while ferritin and the alkaline phophatase were lower at 4 months as compared to 1 month of age (Table 6).

Differences between low and high protein formula

(Table S11 and 12 in electronic supplemental material)

Between infants fed formulas with high (2.2–2.3 g protein/100 kcal) and low protein content (1.8–1.9 g protein/100 kcal) the following significant differences were found for amino acids: infants of 1 month of age (Table S11) on high protein formulas showed higher threonine, 2-aminobutyrate and isoleucine but lower ornithine and arginine than infants fed low protein formulas. Infants aged 4 months (Table S12) on high protein formulas had higher threonine, 2-aminobutyrate, branched-chain amino acids and lysine, but lower aspartate, glycine and taurine than infants fed low protein formulas. As expected, infants fed with high-protein formulas had significantly higher urea concentrations than infants fed with low-protein formulas, regardless of age. The data distributions were tested with the Kolmogorov-Smirnov method by comparing the data at 1 and 4 months of age for infants fed formulas with a low or or a high protein content. There were no significant differences of data distribution except for 3-methylhistidine (p < 0.0001) and ferritin (p = 0.0214) for infants fed high protein and in those fed low protein formulas 1-methylhistidine (p = 0.0132) and 3-methylhistidine (p < 0.0001).

Differences between breast- and formula-fed infants

At 1 month of age most plasma amino acid levels were lower or about the same in breast-fed as compared to formula-fed infants (p in Table 2). The concentrations of proline, serine and tyrosine were significantly higher in breast fed than in formula-fed. At 4 months of age, only glutamine, proline and hydroxyproline were lower in formula-fed than in breast-fed infants; all other amino acids were as high or higher in formula fed infants (Table 6).

At 1 month of age, plasma levels of most of the clinical chemistry panel were higher in breast-fed than in formula-fed infants, most strikingly for cholesterol (median 3.95 versus 2.90 mmol/L). Only for potassium, chloride, and AP results were about the same in both groups.

At 4 months of age, higher levels in breast-fed infants persisted for glucose, cholesterol and ferritin whereas for urea and the urea/creatinine ratio they reversed to become significantly lower in breast-fed as compared to formula-fed infants. For the electrolytes, creatinine, total protein, and albumin, differences between both groups diminished or remained about the same.

Discussion

We propose pediatric reference limits of plasma amino acids and several clinical chemistry parameters for infants of one and four months of age. Centiles of Tables 1 and 4 can be used if the composition of the actual nutrition is unknown, e.g. in cases of emergency or if breast-milk is complemented with formula. In general, the adequate choice of centiles and tables depends on the clinical hypothesis, i.e. whether one aims at excluding, confirming or suspecting a disorder.

The reference population originated from three studies which tested the safety of new formulas on 3 cohorts of healthy infants. Amino acids were not assayed from whole blood or filter paper spots to avoid interference by changes in the aqueous volume of the sample due to the important variation of haematocrit (Bachmann 2000). We found that the concentrations of most amino acids were significantly lower at 4 months than at 1 month of age. Similarly, most of the usual clinical chemistry parameters showed significant differences between both age groups. This indicates that pooled reference of the two age groups should not be used for medical decisions. For both age groups, the proportion of infants receiving breast milk, a high or low protein formula only, and those fed on both were similar (at 1 month: 20 %, 19 %, 51 %, 10 % and at 4 months 21 %, 21 %, 47 % and 11 %, respectively). Thus the lower amino acid concentrations at 4 months and the differences in reference limits of usual clinical chemistry parameters cannot be explained by a food effect, neither through different food quality nor through different times elapsing from the end of the last feed until sampling.

The reference values of cystine extend to very low levels. The data distribution is biased by the low cystine values of one study center (Fig. 1), which are most likely due to a pre-analytical error, i.e. a delay between sampling and centrifugation and freezing. Interestingly, such low values have been reported by others in children and adults (Armstrong and Stave 1973a; Mayo Clinic 2014). Since the bias through results of one study center leads to a multimodal distribution we did not make inferential comparisons for cystine. We suggest that only the upper reference limit of cystine might be used, if at all.

Fig. 1
figure 1

Cystine: Predominance of very low values in one study center as compared to the other study centers. The distribution of cystine concentrations at one and four months is shown as percentage of the study center X and the other study centers

Full size image

Some high concentrations of potassium could be due to a mild hemolysis which could not be detected by inspection of the plasma.

The contribution of major organs and tissues which take up and release amino acids from and to the systemic circulation varies and has not been quantitated in infants. In addition to nutritional factors the postprandial time elapsed after the last feed affects some, though not all amino acid concentrations. Qualitative results concerning the postprandial absorption and elimination phase have been published (Bachmann and Haschke-Becher 2002; Bachmann 2003). When comparing nutritional effects between groups it is important to ensure that groups do not differ in times elapsing from last feed. One should note that the pattern of citrulline differs from other amino acids with trough plasma levels shortly after protein intake, followed by steady increase for up to 4–5 h (Fig. 2). This should be taken into account if low plasma citrulline is used as indicator of mitochondrial defects of urea cycle enzymes or transporters (Häberle et al. 2012) or control of enterocyte function after surgical resection. The pattern of the postprandial citrulline concentration in adults is not known.

Fig. 2
figure 2

Plasma citrulline concentration depends both on postprandial delay of sampling and on daily protein intake of age in formula fed infants aged one and four months. In contrast to other amino acids, citrulline concentrations in plasma not only increase with augmenting protein intake, but as well with increasing postprandial sampling delay

Full size image

Our findings of amino acid and urea concentrations in plasma at 1 and 4 months of age may shed some light on physiological mechanisms. Example (cp Tables S11 and 12 in the online supplemental material): formulas with low protein had a lower proline content (6.2–6.4 g aminoacyl/16 g N) than in breast milk (9 g aminoacyl/16 g N); Räihä et al. 2002). In infants aged 1 month fed low protein formula the median plasma urea concentration was the lowest compared to those of the other infant sub-groups; the median proline concentration in plasma did not differ between infants fed low or high protein formulas neither at 1 nor at 4 months of age. At 1 month but not at 4 months of age plasma ornithine was significantly lower in infants fed formula with low protein content compared to those fed high protein formula. This is compatible with a physiologically higher demand of proline and hydroxyproline (posttranslational modification) for procollagen synthesis at 1 month than at 4 months of age and with a flux direction of ornithine to proline. Findings in patients and animal experiments with some mitochondrial urea cycle disorders are in agreement with such an interpretation (Wang et al. 1995; Cleary et al. 2005; Ben-shalom et al. 2002; Baumgartner et al. 2005). In breast-fed infants however the median plasma urea and proline concentrations were lower at 4 months than at 1 month of age which is likely due to the higher proline content of breastmilk. This suggests that proline is a conditionally essential amino acid at 1–2 months of age. Increasing the proline content of low protein formulas to the level of breastmilk should perhaps be considered for that age group.

In conclusion, plasma concentrations of amino acids and clinical chemistry parameters in young infants are affected by age and diet. Both need to be taken into account through specific reference limits and standardization of pre-analytical procedures. Blood should be obtained from infants between 4.5 to 5.5 h after the last feed at trough levels to exclude deficiencies or imbalance of amino acids, except citrulline. After 6 h the plasma concentration of most amino acids is rising in infants probably due to endogenous protein catabolism.