The impact of higher protein dosing on outcomes in critically ill patients with acute kidney injury: a post hoc analysis of the EFFORT protein trial

Critical illness is frequently accompanied by acute kidney injury (AKI) [1‐4]. AKI impairs homeostatic functions, including the maintenance of acid–base balance and resultant metabolic acidosis, which further increases proteolysis, protein catabolism and impairs transcellular amino acid transportation [5]. This loss of renal metabolic function impairs amino acid conversions and utilization [6]. In severe AKI, the use of kidney replacement therapy (KRT) further exacerbates amino acid loss [7]. Moreover, critical illness defining conditions such as sepsis, respiratory failure, and trauma lead to proteolysis with a negative nitrogen balance and an acquired loss of muscle mass, which impairs physical functioning, recovery and ultimately quality of life in survivors [8, 9]. Therefore, in theory, critically ill patients with AKI may require a greater protein dose, compared to patients without AKI.

This is reflected in the current nutrition guidelines recommend targeting a protein dose up to 2.0 g/kg body weight (BW)/d for patients with AKI not on KRT and up to 2.5 g/kg BW/d for patients with AKI on KRT [10, 11]. The recently completed EFFORT Protein trial among mechanically ventilated critically ill patients demonstrated that higher (≥ 2.2 g/kg/BW/d), compared to usual (< 1.2 g/kg/BW/d) protein dose, did not improve time-to-discharge-alive from hospital (TTDA) nor 60-day mortality. However, in an a priori defined subgroup of patients with AKI, high-dose protein, compared to lower, was associated with worse outcomes [12]. Therefore, we performed this secondary analysis of the EFFORT protein trial to further explore the impact of protein dose among different subgroup of patients with AKI. Among these, the influence of protein in the different stages of AKI, the use and duration of KRT post randomization, and potential influence of chronic kidney disease (CKD) at baseline were of special interest. We hypothesize that compared to a lower dosing, higher protein dosing is associated with worse clinical outcomes in critically ill patients with AKI at any stage who were not treated with KRT.

In this secondary analysis of the largest RCT comparing protein dose in mechanically ventilated critically ill patients, higher protein dose, as compared to usual dose, in patients with AKI was associated with slower TTDA and increased 60-day mortality regardless of the stages of AKI. Interestingly, this harm did not persist in patients who received KRT, albeit the interaction was not statistically significant. High-dose protein was not associated with longer duration or higher incidence of KRT.

Critically ill patients who develop AKI are at risk of protein–energy malnutrition, which is a major negative prognostic factor. Severe malnutrition has been documented in up to 40% of critically ill patients, which is associated with a further increase in morbidity and mortality [19‐23]. Therefore, adequate energy and protein intake seems to be crucial component of ICU patient care to prevent deterioration of the nutritional status and its complications. A previous RCT demonstrated that patients with AKI and KRT who received up to 2.5 g/kg/d protein achieved positive nitrogen balance and had improved clinical outcomes, compared to those patients who received a lower dose [24]. Yet, this preliminary evidence was from a small single centre study only, which significantly limits the generalizability of the received findings. Therefore, the optimal nutritional support in AKI still remains an open issue and high-quality evidence in support of this hypothesis are still missing [5]. In this context, the present findings represent the largest analysis to-date, which highlights that higher protein dose, compared to usual dose, in critically patients with AKI is associated with slower TTDA and increased mortality. It is important to mention that the current findings are consistent with findings of previous analyses, which demonstrated that patients with acute kidney injury, were harmed by an additional administration of amino acids [25, 26]. A more careful approach of protein dosing in AKI patients seems warranted in the clinical practice.

There are potential explanations for our findings. First, experimental conditions and theoretical considerations differ from real-world clinical scenarios [27]. It has been previously shown that AKI in critically ill patients is not, per se, associated with increased protein catabolism. Even when present, the underlying mechanisms of protein catabolism are multifactorial and the catabolic state cannot be simply countered by the provision of a higher dose of proteins. The plasma and intracellular amino acid pool have been shown to be altered in critically ill patients with AKI and tissue amino acid utilization is impaired and transport into muscle is reduced, which may partially be attributed to metabolic acidosis [28‐31]. The higher urea levels observed in AKI patients, needs to be interpreted carefully within the limitations of an explorative analysis. Further mechanistic and translational studies are however encouraged to test, if the higher urea levels in the high protein group indicate a biological signal of harm, resulting from a reduced capacity to utilize the amino acids during critical illness. These impairments thus may limit the efficacy of delivered nutrition, particularly in preserving muscle mass in critically ill patients. Therefore, exogenously administered protein may increase metabolic stress.

Second, while the duration of dialysis and incidence of new dialysis were not significantly affected by the administration of higher proteins, KRT may have reduced a negative effect of higher proteins on outcomes in patients with AKI. KRT trigger a loss of amino acids, small proteins/peptides up to 5–22 g/day and other nutritional losses such as trace elements and vitamins [7, 32]. Due to its low molecular weight, amino acids such as cysteine, arginine, alanine, and glutamine can be readily filtered from the blood into effluent [7]. During continuous KRT, daily loss of almost 20 g of protein per day is reported [7, 32]. This removal of the excessive protein from the high protein group may explain why we found that higher protein had no effect on TTDA in patients with AKI who received dialysis. In this context, it must be acknowledged that rules for starting and discontinuing KRT are known to significantly vary between institutions and often not standardized across sites.

Third, based on expert consensus, current guidelines postulate that protein requirements are significantly increased in critically ill patients, including those with AKI [24, 32, 33]. Yet, our data could not confirm the benefits of delivering a high protein dose. In accordance with previous findings, it may be possible that the cohort of critically ill patients may have been too sick to benefit from high dose protein [34]. Despite relatively normal protein digestion and amino acid absorption in critically ill patients, the capacity to utilize the protein delivered may be blunted in critically ill patients [31], and high protein intake is unable to reverse this catabolic state [32, 33]. As multiple large RCTs repeatedly failed to demonstrate clinical benefits of nutritional interventions in a heterogenous population of critically ill patients, surrogate biologic markers may help to identify patients that may benefit from a nutritional intervention [35].

In hospitalized patients with AKI not receiving KRT, current nutrition guidelines suggest a range of 1.0–2.0 g/kg/BW/d of protein [10]. While for critically ill patients with AKI receiving KRT, guidelines recommend up to 2.5 g/kg/BW/d [11]. Our findings indicate that mechanically ventilated patients with AKI are harmed by higher protein dosing, particularly in patients who did not receive KRT, so that a careful re-evaluation of the current guideline recommendations for patients with AKI seems warranted.

Our secondary analysis has several strengths. First, we utilized data from the largest (to date) protein dose RCT in critically ill patients. Second, the data included a diverse sample from multiple practice settings worldwide, all of which enhance the generalizability of the received findings. Third, we identified a hypothesis-generating signal to inform practice and future trial design. We acknowledge several limitations of this exploratory secondary analysis. First, the modality of KRT could impact protein requirements, which was not evaluated in this study that included data from a pragmatic design. Second, post-hoc analysis of RCTs needs to be interpreted cautiously and should be considered hypothesis generating, precluding strong clinical recommendation. Third, as AKI was mainly diagnosed from randomization, we are unable to be certain whether the higher protein dose leads to AKI and worsened clinical outcome, or the continuous high protein dose after AKI is developed worsened clinical outcomes, but we note that the rate of AKI was similar in both protein groups so the later explanation seems more plausible. Nevertheless, we must interpret within subgroup treatment effects cautiously for AKI and all subgroups that were defined based on post-randomization data. Finally, there is a risk of type I errors since several subgroups were tested without adjustment for multiplicity, and conversely there is a risk of type II errors since the study may not be adequately powered for the interaction tests used to determine the significance of subgroup effects. Nevertheless, the overall signal of harm in critically ill patients with AKI cannot be ignored and should be considered in nutrition guidelines and future trial designs.

Higher protein dose, compared to usual dose, is associated with worse clinical outcomes in mechanically ventilated, critically ill patients with AKI. No effect of either higher or lower protein dosing was observed in patients with AKI who received KRT. Based on these findings, current guidelines for nutrition support in critically ill patients with AKI should be carefully re-evaluated and further research is warranted.