Keywords

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

Introduction

When infants cannot be breast-fed formulas with low protein content (1.8–1.9 g/100 kcal) have been recommended in order to avoid an excessive protein supply or amino acid imbalance during organ development in infants (Koletzko et al. 2005). Furthermore, infant formulas with higher protein content are risk factors for later obesity (Kirchberg et al. 2015; Ong et al. 2009; Toschke et al. 2004). Formulas containing less than 1.8 g protein might be adequate but not safe as shown by Fomon et al. (1999). Frequent controls of amino acid concentrations in plasma of infants are thus needed when low protein formulas are exclusively fed in order to exclude that very low amino acid concentrations limit protein synthesis of infants during this phase of change of growth rate. For the interpretation of amino acid results the effect of pre-analytical factors must be taken into account; however, most often pre-analytical factors are not documented and/or communicated.

Relevant pre-analytical factors which modify the plasma levels are (1) the time elapsed between the end of the last feed and the sampling of blood and (2) the actual daily intake of protein and its composition. For following the course of a patient or for excluding protein malnutrition blood sampling should be done at trough levels, i.e. between 4.5 and 5.5 h postprandial (p.p.) except of citrulline (Haschke-Becher et al. 2016; Windmueller and Spaeth 1981). However this is not always done or feasible. If blood is sampled during food absorption (i.e. until 2 h p.p.) the results are useless since no reference data of healthy infants exist. In practice samples are most often taken during the exponential decrease of essential and of some non-essential amino acids before trough level is reached, e.g. if the infants are fed on demand. Such results might, however, lead to pitfalls of interpretation: Most essential amino acids and some non-essential ones decrease exponentially after 2.25 postprandial (p.p.) hours until the trough level (>4.5 h) is reached; when blood is sampled during the exponential drop, the actual amino acid concentration determined will be higher than at trough level. If this is not taken into account in infants fed low protein formulas there is a risk of missing amino acid deficiencies and to interpret the amino acid concentration as safe level.

To our knowledge the effect of time of blood sampling and of the actual daily intake of formulas with low protein content on plasma amino acids has not been quantitatively estimated in healthy infants of 1 and 4 months of age. In this period of rapid growth the concentrations of indispensible amino acids in plasma should be safe, adapted to the needs and neither rate limiting for protein synthesis nor in excess.

We show quantitative data of plasma amino acid concentrations of healthy infants aged 1 and 4 months fed low protein formula on demand. We focus on the combined effect of the p.p. sampling delay from 2.25 to 4.5 h and of the individual daily protein intake on the amino acid concentrations in plasma. Furthermore we present low limits (the 10th percentile) of amino acid concentrations in plasma of healthy infants aged 1 and 4 months after extrapolating the time of sampling to trough levels (5 h p.p.). By comparison of patient results with these data the risk of amino acid deficiency might be evaluated in patients fed amino acid mixtures of low protein content.

Study Population and Methods

In order to quantify and estimate the influence of postprandial sampling delay after the end of last feed and of daily protein intake on the plasma amino acids we used a population of healthy infants of 1 and 4 months of age fed exclusively formulas with low protein content (1.8–1.9 g protein/100 kcal). The population and the analytical methods used for the amino acid analysis including total plasma tryptophan have been published elsewhere (Haschke-Becher et al. 2016). For this study we included only those infants who had information on actual protein intake per day, on postprandial sampling delay (independent variables) and plasma amino acid concentrations (dependent variables). 102 samples were of infants aged 1 month and 79 of infants aged 4 months. We do not report data of the following amino acids: taurine (added to the protein mixture of the formulas), hydroxyproline (posttranslational product of proline), cystine (pre-analytical bias) and 1- and 3-methylhistidine (low precision of low values); as shown in Haschke-Becher et al. (2016) on Fig 1 and in its electronic supplemental material. We focussed on the kinetics of plasma amino acids in blood samples obtained between 2.25 and 4.5 h postprandial. We used the natural logarithm of amino acid concentrations and determined intercept and parameter estimates of both independent variables. Furthermore we used standardized Beta coefficients of the multiple regressions, i.e. the coefficients of the dependent variables transformed to a standard deviation of 1 for weighting the impact of the two independent variables.

Statistics

Sample distribution of amino acids was analysed with Shapiro-Wilk, Kolmogorov Smirnov and Cramer von Mises tests and quantile regression with SAS 9.4 TS1M2 software for Windows (SAS Institute Inc., Cary, NC). The fit of combined effects of both independent factors was calculated by multiple regression of the natural logarithm (ln) of amino acid concentrations in dependence of p.p. sampling delay and daily protein intake. Analyse-it 3 for Excel software was used for these calculations (Analyse-it Software Ltd., Leeds, UK). In addition, the fit of each dependant variable by linear regression was controlled separately. R software (Koenker and Bassett 1978) was used for computing the fit at the 10th percentile.

Values of p < 0.05 were considered significant. Half-life was calculated by: T1/2 = − ln(2)/estimate of the regression coefficient of the sampling delay.

Results

Post-absorptive Kinetics of Amino Acid Concentrations in Plasma of Healthy Infants: Effect of p.p. Sampling Delay and Amount of Daily Protein Intake

The kinetic parameters of samples taken between 2.25 and 4.5 h p.p. differed among amino acids. The results of multivariate regressions are shown in Tables 1 and 2 for blood samples taken at 1 and 4 months of age, respectively. Biological half-life is also shown. The equations with the factors of multiple regressions are presented in Table 3. The sample distribution of amino acids was logarithmic; thus the natural logarithms (ln) of the amino acid concentrations were used for calculations. For infants of 1 month of age all the regression parameters of sampling delay of the ln of essential amino acids had a significant negative slope except for total tryptophan (Table 1). This was also found for the following non-essential amino acids: alanine, arginine, asparagine, ornithine, proline, serine and the sum of aspartate and asparagine. The regression coefficients of daily protein intake were significant and positive only for lysine and threonine at 1 month of age.

Table 1 Kinetic parameters of plasma amino acids sampled at 1 month of age between 2.25 and 4.5 h p.p. fed formula of low protein content
Full size table
Table 2 Kinetic parameters of plasma amino acids sampled at 4 months of age between 2.25 and 4.5 h p.p. fed formula of low protein content
Full size table
Table 3 Equations of multiple regression of low protein formula-fed healthy infants
Full size table

At 4 month of age the significance of regression coefficients of sampling delay differed from those at 1 month insofar as the regression coefficient of ornithine was not significant anymore (p = 0.098); the coefficients of histidine, tryptophan and tyrosine were significant as well as the sum of glutamine and glutamate (Table 2). At 4 months of age the sole significant coefficient of protein intake/d was that of glutamate (p = 0.006). Citrulline concentration increased during the post-absorptive phase; the lowest level was reached during the first hour after the end of last feed (Haschke-Becher et al. 2016).

There were not enough data for calculating the kinetics during the initial absorption of the gut (0–2.0 h) and after 5.5 postprandial hours when concentrations of many amino acids tended to increase presumably by incipient proteolysis.

Kinetic parameters of both independent variables calculated separately by univariate linear regression are presented in the electronic supplemental material (Tables 5–8).

Effect of the Independent Variables on the Mean of Amino Acid Concentrations

The standardized squared Beta values were higher for the p.p. sampling delay than for the protein intake/day. At 1 month of age the percentage of the total variance which exceeded 10% and can be attributed to the sampling delay was found for: alanine (21%), isoleucine (20%), leucine (14%), threonine, asparagine, lysine (13% each), ornithine and proline (each 12%). At 4 months the effect of sampling delay was 21% for leucine, 20% for isoleucine and lysine, 19% for methionine, 17% for asparagine, 14% for valine and 11% for arginine and alanine. The effect of protein intake per day on total variance was lower than 5% for all the amino acids, both at 1 and 4 months of age.

Univariate regression was done for each independent variable on the same set of data as used for the multivariate regression. As expected, most estimates differed between the two methods. The differences were, however, minor.

Procedure to Avoid the Pitfall of Overestimated Amino Acid Concentrations

The goal was to avoid the pitfall of overestimating amino acid concentrations sampled during the post-absorptive phase (2.25–4.5 h). We estimated by extrapolation to trough level (5 h) the ln of those amino acids which showed a significant p.p. drop in healthy infants fed formula with low protein content. We defined lower limits at the 10th percentile for such amino acids and back-transformed the logarithmic results into μmol/L in dependence of sampling delay during the exponential drop and at trough level (Tables 4 and 5). These data of healthy infants aged 1 or 4 months may be used for evaluating amino acid results (μmol/L) of controls made on patients aged 1 or 4 months fed low protein formula to make sure that there is no risk of protein malnutrition, e.g. if albumin is below the reference limits.

Table 4 Tenth percentile of significant amino acid concentrations (μmol/L) depending on p.p. sampling delays with 95 % confidence intervals (CI) in infants aged 1 month
Full size table
Table 5 Tenth percentile of significant amino acid concentrations (μmol/L) depending on p.p. sampling delay in infants aged 4 months
Full size table

Discussion

The quantitative effects of p.p. sampling delay indicate that during the post-absorptive phase of 2.25–4.5 h the effect of sampling delay on the ln of essential amino acids (except total tryptophan) and of some non-essential amino acids leads to a significant negative slope, while several non-essential amino acids in plasma are not significantly affected by the p.p. sampling delay. Ornithine concentration is affected by the sampling delay at 1 month of age (p < 0.001), but not at 4 months of age (p = 0.098). This supports quantitatively the hypothesis that the flux direction at 1 month of ornithine to proline synthesis prevails (Haschke-Becher et al. 2016) due to the high demand of proline for collagen I synthesis. Since the proline content in the formula is lower than in breast-milk additional proline should be provided. Collagen I synthesis further requires glycine which is not affected by sampling delay in contrast to its precursor serine. Hydroxyproline, as a posttranslational product of proline is needed as well; the ln of proline and of hydroxyproline correlate positively (Spearman ρ = 0.422 and 0.454 at 1 and 4 months, respectively; p < 0.001 for both).

The effect of the actual daily intake of low protein formula is significant for lysine and threonine in infants of 1 month of age and at 4 months solely for glutamate (precursor of Δ1-pyrroline 5-carboxylate and proline or ornithine). It is not clear if the significant effects of intake of lysine and threonine at 1 month of age are due to a relatively high concentration of these amino acids in formulas as compared to breast-milk and/or to reduced utilization for protein synthesis.

As shown in the equations of multiple regression (Table 3) the coefficients of sampling delay are more than 10 times higher than the coefficients of actual daily protein intake. Despite our demand to the supervisors of the original studies to obtain blood samples later than 3.5 h after the last feed, this recommendation was ignored. In fact 82% of the samples were taken before 3.75 h p.p. at 1 month and 80% at 4 months of age. This is due to feeding the infants on-demand. We wonder if after 3.5 h satiety was not reached any more with the low protein formula.

Conclusion

For the interpretation of plasma amino acid concentrations data on the postprandial delay of sampling must be obtained and taken into account; otherwise low amino acid concentrations could be missed in patients.