A supplemental intravenous amino acid infusion sustains a positive protein balance for 24 hours in critically ill patients

Several observational studies have demonstrated correlation between a low protein intake in the ICU and poor outcomes, making protein supplementation an appealing therapy [1‐3]. Current guidelines recommend between 1.2 and 2.5 g/kg/day for critically ill patients [4, 5]. There are no high-quality randomized controlled trials (RCTs) demonstrating improved patient outcomes in support of these recommendations. A number of studies with surrogate endpoints have failed to demonstrate any advantage of protein deliveries above 1.5 g/kg/day [6‐9]. Studies employing short-term nitrogen balance also report improved nitrogen balance at intakes above 1.5 g/kg/day [10].

There is a paucity of evidence to inform clinical practice. Clinical trials are urgently needed, but these in turn require physiological studies to elucidate the biological effects of protein supplementation during critical illness. The design of these RCTs should rest upon knowledge of physiology and safety of the suggested feeding regimens in subjects with compromised organ function.

Stable isotope tracer techniques provide a method of quantifying protein synthesis, breakdown and oxidation in vivo [11]. Our research group has previously demonstrated that an intravenous infusion of amino acids in addition to enteral feeding improves net whole-body protein balance for up to 3 h during the initial week of ICU stay, and that a similar response is seen when this therapy is repeated 2–4 days later [12]. In addition, safety in terms of unaltered amino acid oxidation and urea levels was demonstrated. We therefore wanted to investigate if a prolonged intravenous amino acid infusion could sustain improved protein balance for up to 24 h (primary outcome). Changes in amino acid oxidation rate, serum urea and plasma free amino acid concentrations were monitored as secondary outcomes.

A total of 12 patients in the ICU of Karolinska University Hospital Huddinge were recruited between January and December of 2016. Baseline protein balance was assessed in all patients, with eight completing the entire protocol. In four cases protein balance could not be assessed at 3 h (n = 1) or 24 h (n = 3) due to protocol violations or clinical circumstances (Fig. 3). One patient died during the study period. The proportion of males to females was 3:1, median age was 65 (47–74) years, and median body mass index (BMI) 27.9 (22–38.9) kg/m². Other patient characteristics and nutritional therapy are summarized in Table 1 and 2.

In the eight patients in whom protein kinetics were analyzed at all three time points, the amino acid infusion resulted in improved net protein balance over time (p = 0.0016). Protein balance increased between 0 and 3 h (p = 0.01) and remained unaltered between 3 and 24 h (p = 1.00). A similar pattern was seen in the four patients in whom change in protein balance could only be measured at a single time point. The temporal changes in individual patients are illustrated in Fig. 4. The relative contributions from changes in synthesis and breakdown towards the improvement in net protein balance were not statistically significant (p = 0.500 and p = 0.292, respectively). Protein kinetics are summarized in Table 3.

The amino acid supplementation did not result in altered amino acid oxidation (p = 0.147) or plasma urea concentrations (p = 0.053). Plasma total free amino acid concentrations increased over time (p = 0.038, Fig. 5).

In this study we investigated the effects on whole-body protein kinetics of adding a supplemental intravenous amino acid infusion to enteral feeding for 24 h in critically ill patients. We found that intravenous amino acid supplementation resulted in improved net protein balance that was sustained unaltered from 3 h up to 24 h. This response was not associated with increases in protein oxidation or plasma urea concentrations, but plasma total free amino acid concentrations increased over time. The results after 3 h confirmed our earlier reported results [12].

Our results reinforce the physiological rationale for protein supplementation during critical illness. However, several gaps in the knowledge remain to be clarified. The optimal dosing of protein delivery is uncertain. Our study employed a pragmatic protocol, meaning that nutrition was not protocolized beyond the nutritional guidelines of the unit and the supplementation studied. This resulted in median amino acid delivery of 1.11 and 2.07 g/kg/day prior to and during therapy, respectively. It is uncertain if higher or lower delivery within the recommended intervals of different guidelines (1.2–2.5 g protein/kg/day) would result in a similar or improved protein balance, although pooled results from tracer studies indicate linear dose-response correlation [14]. The potential patient-centered benefits of protein supplementation and the optimal dose and timing of the intervention require further exploration in the context of RCTs.

The finding that the additional intravenous dose in our study resulted in improved net protein balance without increasing amino acid oxidation supports its adequacy. Our results are only partially in agreement with those of Shaw et al., who examined protein kinetics in patients with sepsis during total parenteral nutrition with varying protein content [15]. They observed an optimal protein-sparing effect at 1.5 g/kg/day, with no improvement at higher doses. The mean amino acid dose in our study was similar to that delivered in the largest current RCT of supplemental intravenous amino acids [8]. Although this trial was designed to study the possible protective effect of extra protein supplementation on kidney function, there was no difference in 90-day mortality between groups. This may be indicative of safety, but it must be emphasized that inclusion criteria make the study population selective, and the limited statistical power in this secondary outcome parameter calls for caution in interpretations. Furthermore, there was a significant increase in serum urea concentration in the intervention group, but this did not translate to increased RRT requirements. The numeric increase in mean serum urea at 24 h in our study was not statistically significant, but as a safety parameter it needs to be explored further in future research. In another recent study, 1.5 g/kg/day of protein as compared to 0.5 g/kg/day resulted in marginally improved protein balance but no difference in outcomes. Both serum urea concentration and urinary urea excretion were increased [9]. This study may also be regarded as an indication of safety, but again a specific selected patient group was eligible for inclusion.

Beyond assessing the effects on protein balance, physiological and kinetic studies are essential to investigating safety aspects of therapeutic regimens. Although the prevalent opinion in critical care nutrition is leaning towards high protein delivery early in the course of illness, there are limited data supporting this position [16]. The increased load of non-volatile acids and azotemia associated with protein feeding are potentially harmful in patients in the ICU, who have altered metabolism and compromised organ function. The REDOXS trial by Heyland et al. is the largest RCT of amino acid (glutamine + alanine + glycine) supplementation to date, indicating a trend towards greater mortality in the intervention group [17]. It should be noted that 13% of patients in the treatment group had urea levels >50 mmol/L, compared to 4% in the control arm. This result warrants caution when administering high doses of amino acids, and especially as an unbalanced amino acid composition, to vulnerable patients [18]. We propose that indices of amino acid oxidation, plasma urea and free amino acid concentrations should be routinely monitored as an adjunct to other measures of outcomes in future clinical trials or observational studies on protein delivery. This would provide important information regarding safety and a physiological framework to explain treatment effects. In our study we did not observe any significant increases in amino acid oxidation or ureagenesis. This may be due to the small sample size, but our data provide some basis for power calculations to determine the appropriate recruitment targets for monitoring these parameters in future studies.

It is important to stress that the single-pool model we used provides no information on where the changes in synthesis and breakdown take place and which proteins are synthesized and broken down. Studying the protein balance of individual organs or the immune system in the critically ill is technically demanding. In a study of 15 patients in the ICU, Essen et al. found that protein synthesis rate was upregulated in peripheral lymphocytes and liver (albumin synthesis), but on average was unaffected in muscle, as compared to healthy subjects [19]. Several investigators have also used the arteriovenous balance technique to demonstrate negative protein balance in the leg muscle in patients with sepsis or burns [20‐22]. These findings support the theory that skeletal muscle acts as a protein reservoir during critical illness, providing indispensable host functions with necessary substrates [23]. The effects of protein supplementation on regional synthesis and breakdown are yet to be characterized. The mean phenylalanine balance observed at 3 h and 24 h in our study corresponds to net synthesis of approximately 45–55 g protein/day (containing 4% phenylalanine), in a patient weighing 75 kg. Assuming a protein content of 20%, this would equate to 250 g of lean body mass. Given the biological implausibility of patients gaining muscle mass during critical illness, any increases in net synthesis from protein feeding must occur mainly elsewhere. On the other hand, the loss of muscle protein during critical illness is mainly the result of increased breakdown, and inhibition of breakdown due to protein feeding might prevent larger losses. Simultaneous quantification of regional protein balance in skeletal muscle and whole-body kinetics in response to amino acid supplementation would be a logical next step towards elucidating the response to protein feeding in critically ill patients.

Our study has several strengths. Measuring protein kinetics during enteral feeding in adult patients in the ICU has, to our knowledge, so far only been performed by our research group. As we used both enteral and parenteral tracers to determine protein kinetics, no changes were made to any ongoing nutritional therapy. Patients were primarily fed by the enteral route in accordance with international guidelines. The repeated quantification of protein balance in patients in the ICU is also novel. Although the stable isotope tracer method only provides a snapshot of whole-body protein metabolism, the repeatability of measurements between 3 h and 24 h demonstrates that this technique can be used to assess the physiological response to therapeutic interventions by repeated measures. Related work exploring the effects of increased protein intake as part of parenteral nutrition in an ICU setting with stable isotope tracer methodology has been performed in preterm neonates, infants undergoing cardiac surgery, infants with bronchiolitis and adolescents with sepsis [24‐27]. The fundamental differences in the metabolism of children preclude any extrapolations of these results to an adult population.

Limitations of our study include the small sample size and heterogeneity in age, time course of the ICU stay and anthropometric characteristics of patients. Despite these differences the trend in the change in protein balance over time was similar in all patients, regardless of baseline conditions (Fig. 4). The non-randomized design, non-standardized nutrition and the fact that not all patients were eligible for analysis at three time points due to protocol violations also limit generalizability of the results. All 12 patients were analyzed on a minimum of two occasions, and the results of partial analyses were in agreement with those of patients who completed the full protocol. There are also inherent limitations in the stable isotope tracer technique and single-pool model used to determine whole-body protein kinetics. It is possible that synthesis and oxidation rates are underestimated because the technique only allows measurements from plasma precursors of protein synthesis, neglecting intracellular amino acid recycling or accumulation. The choice of tracer may influence the measured rates of synthesis, breakdown and oxidation depending on the metabolism of different amino acids. As demonstrated in earlier work by our group [11, 28], simultaneous quantification of protein balance with leucine and phenylalanine tracers in critically ill patients results in different rates of synthesis and breakdown. Absolute values in the magnitude of change in protein balance should therefore be interpreted with caution. However, as net changes with different tracers are clearly correlated, we believe the trend towards improved protein balance seen in this study is independent of the choice of tracer. Also, the fact that we measure whole-body protein kinetics does not allow us to determine which proteins/tissues represent the more positive net balance. Despite these limitations we consider the methodology used to be the best available technique to quantify protein kinetics under relevant clinical circumstances. Another limitation is that our study only concerns the effects of intravenous amino acid supplementation. As the route of protein delivery may have importance for the physiological and therapeutic effects during critical illness, future studies should consider investigating protein kinetics during exclusively enteral nutrition.

We demonstrated that a supplemental intravenous infusion of amino acids in addition to routine nutrition, mainly by the enteral route, sustained improved protein balance from 3 h up to 24 h in critically ill patients. These findings reinforce the potential importance of protein supplementation during critical illness. Given the unique properties of protein kinetic studies to investigate physiological effects and safety endpoints of protein delivery during critical illness, we advocate this methodology to guide the design of future research.