EATing in the ICU: TARGETs for Measuring Energy, Nutrition Dose, and Clinical Outcomes
Reviewer: Jayshil (Jay) Patel, MD
Associate Professor of Medicine, Division of Pulmonary & Critical Care Medicine, Medical College of Wisconsin, Milwaukee, Wisconsin
Reference: Allingstrup MJ, Kondrup J, Wiis J et al. Early goal-directed nutrition versus standard of care in adult intensive care unit patients: the single-center, randomized, outcome assessor-blinded EAT-ICU trial. Intensive Care Med. 2017;43:1637-1647
Why is this paper important? The 2016 ASPEN and 2019 ESPEN critical care nutrition support guidelines agree that indirect calorimetry (IC) be used to assess energy requirements but disagree in nutrition dose during the first week of critical illness. The EAT-ICU trial was the first randomized controlled trial that compared IC-derived early-goal directed nutrition (EGDN) to meet full energy requirement versus standard of care nutrition delivery in mechanically ventilated critically ill adults with an expected length of ICU stay of 3 or more days.
Summary: Citing observational studies showing benefits of achieving IC-derived energy and protein goals in critically ill adults, the EAT-ICU study sought to answer the question: In mechanically ventilated critically ill adults expected to remain in the ICU for at least 3 days, does intervention with EGDN, compared to standard of care, improve the primary outcome of physical quality of life score6-months after randomization? There were numerous secondary outcomes, including 28-day, 90-day and 6-month mortality and length of ICU stay. Under concealed allocation, patients were randomized 1:1 to EGDN or standard nutrition therapy.
The EGDN arm aimed to deliver 100% of IC-derived energy requirements within 24-hours using enteral and supplemental parenteral nutrition. EGDN patients underwent daily 24-hour urea excretion and were prescribed protein at 1.5 g/kg/day regardless of urea excretion. The standard nutrition therapy arm had energy requirements calculated using 25 kcal/kg/day and had EN started within 24-hours and received supplemental PN if the calculated energy requirement goal was not met by day 7. Both arms had the same blood glucose targets, received trace elements as needed, had gastric residual volumes measured, and received prokinetic agents at the clinicians’ discretion.
Over approximately three years, 586 patients were eligible and 203 were randomized: 102 to EGDN and 101 to standard of care. The mixed medical-surgical ICU patients had comparable baseline characteristics: EGDN and standard nutrition therapy groups were 63 and 68 years old, respectively; and both groups had a median BMI of 22 kg/m2. Of all patients, 47% had severe sepsis and had a SOFA score of 8. Over the first 7-days of ICU admission, IC-derived energy requirement was 2069 kcal in the EGDN group, compared to 1887 kcal in standard of care. Calculated EGDN group energy requirement was 1950 kcal, compared to 1875 kcal in standard of care. Energy balance was -66 kcal/day in EGDN, compared to -787 kcal/day in standard of care. Protein intake was 1.47 g/kg/day in EGDN, compared to 0.50 g/kg/day in standard of care. There was no difference in the primary outcome of 6-month physical quality life score between EGDN and standard of care (22.9 vs 23, p=0.99). There were no differences in 28-day, 90-day, or 6-month mortality rate between groups. EGDN had better cumulative energy and protein balance compared to standard of care. More EGDN patients had at least one episode of hyperglycemia compared to standard of care (52 vs 25%, p=0.0001) and received higher doses of insulin (86 vs 0 international units, p=0.008).
The authors concluded EGDN, compared to standard of care, led to more energy and protein delivery and lower nutritional deficits in mechanically ventilated ICU adults but no differences in the primary outcome of physical quality of life at 6 months or secondary outcomes of mortality, organ failure, complications or length of ICU stay.
Commentary: The EAT-ICU trial results shed light on four important critical care nutrition questions:  Is IC needed to estimate energy requirements in critically ill patients?  What is the optimal nutrition dose during the first week of critical illness?  What is the role for supplemental parenteral nutrition during the first week of critical illness? and  What outcome measures should critical care nutrition trials evaluate?
First, the 2016 ASPEN and 2019 ESPEN critical care nutrition support guidelines recommend using IC to define energy requirement. Despite these recommendations, lack of resources and personnel and the cumbersome nature of conducting IC may limit widespread use, prompting bedside clinicians to utilize predictive equations to estimate energy requirements. EAT-ICU results challenge guidelines’ call for IC use to estimate energy requirements in critically ill patients by finding a difference of 119 kcal between IC-derived and calculated (25 kcal/kg/day) energy requirement in the EGDN group (2069 vs 1950 kcal), suggesting there’s no clinical difference in deriving energy estimation between the two methods. However, EAT-ICU was a single-center study where medical-surgical patients had a mean BMI of 22 kg/m2 and a SOFA score of 8. Therefore, it is important to consider the patient population before concluding IC and equation-derived energy requirements are equal. In fact, sicker patients with greater catabolic burden or those with lower or higher BMI may have drastically different IC- and equation-derived energy requirements. The EAT-ICU investigators should be congratulated for developing transparent and reproducible protocols to identify energy requirements, which paves the path for future EGDN studies.
Next, ASPEN and ESPEN guidelines provide disparate recommendations for nutrition dose in critically ill adults. When IC is used, the ESPEN guideline recommends isocaloric nutrition be progressively implemented after the early phase of acute illness, but when a predictive equation is used to estimate energy requirements, hypocaloric nutrition is preferred (over isocaloric) during the first week of ICU admission. The ASPEN guideline does not distinguish between IC- or predictive equation-based energy requirements and nutrition dose. Rather, the 2016 ASPEN guideline calls for determining nutritional risk using NUTRIC or NRS-2002 scores and recommends optimizing nutrition dose (>80% estimated or calculated energy goal) within 48-72 hours in patients deemed high-nutritional risk. EAT-ICU did not find any difference in primary or secondary outcomes despite EGDN patients receiving 90% of both energy and protein prescriptions, compared to standard of care receiving 56 and 43% of energy and protein prescriptions, respectively. A “one size fits all” nutrition dose may not be appropriate since no two ICU patients are alike. EAT-ICU randomized medical-surgical ICU patients to EGDN or standard of care and a signal for benefit may have been lost in patient heterogeneity. Rather, sicker patients at nutritional risk may derive greater benefit from energy and protein optimization. Observational studies using NUTRIC to stratify patients to high and low nutritional found an association between more energy and protein and reduced mortality. Studies comparing EGDN to standard of care in patients most likely to benefit from nutrition optimization (i.e., high nutritional risk) are warranted.
Third, both ASPEN and ESPEN critical care guidelines recommend commencing enteral nutrition within 24-48 hours in critically ill adults unable to maintain volitional intake. ASPEN recommends waiting 7-10 days and ESPEN recommends starting supplemental PN on a case-by-case basis. As with other supplemental PN trials (EPaNIC, SPN, TOP-UP), EAT-ICU weaves an important thread into the fabric of critical care nutrition trials that have demonstrated feasibility of using supplemental PN to meet energy requirements during the first week of critical illness. The EGDN nutrition protocol were started 16 hours after ICU admission (table S2) and 95% of EGDN patients adhered to the protocol to achieve >90% IC-measured energy requirement during the first week of critical illness. However, findings from the EAT-ICU trial challenge guideline recommendations for supplemental PN use during the first week of critical illness. EAT-ICU showed no differences in primary or secondary outcomes between EGDN and standard of care arms despite greater nutrition delivery using supplemental PN in the EGDN group during the first week of critical illness. Even though there were no significant differences in nosocomial infections between groups, there were more hyperglycemia episodes and greater insulin use in the EGDN group, which are likely due to PN use. The contrast in blood glucose level was not associated with differences in clinical outcomes but does suggest pushing nutrition dose closer to full energy requirement during the first week of critical illness may worsen intolerance.
Finally, the EAT-ICU investigators evaluated the effect of greater nutrition during the first week of critical illness on physical functioning at 6-months. With enhanced vigilance and better support systems, ICU mortality has decreased over time. In EAT-ICU, the observed 28-day mortality was approximately 20%. In general, a lower baseline event rate makes it challenging to design critical care nutrition trials that are powered for mortality benefit. Stated differently, to test the hypothesis that nutrition therapy reduces mortality a few percentage points (from a baseline 20% mortality) would require thousands of patients. EAT-ICU assessed a patient-centered outcome like physical function, which holds biologic plausibility as a primary outcome for a nutrition intervention. Greater protein and energy delivery during critical illness could, in theory, maintain, if not enhance, lean muscle function; which may augment physical function. However, the lack of benefit observed in EAT-ICU may lie in the timing of the primary outcome. In other words, does a difference of 498 kcal at ICU day 7 translate into a difference in physical function at 6-months? Future studies should evaluate the impact of nutrition dose on short and long-term non-mortality outcomes, including muscle content and bioenergetics and physical function.
Reviewer: Stephen A. McClave, MD; University of Louisville, Louisville, KY
Reference: Chapman M, Peake S, and the Australian and New Zealand Intensive Care Society (ANZICS) Clinical Trials Group. Energy-Dense versus Routine Enteral Nutrition in the Critically Ill. The TARGET Trial. NEJM 2018;379:1823-34.
Why this paper is important: While study results indicate that an energy-dense formula is a successful strategy for meeting caloric requirements in critical illness, no outcome benefits were realized. The study suggests that feeding less calories is safe and better tolerated than full feeds and leads to a strategy that may pose less risk to the patient in the early phases of critical illness.
Summary: After completing a pilot trial which showed that a calorically-dense formula delivered calories closer to goal than a routine formula resulting in reduced mortality, the ANZICS group set up this TARGET Trial, a 4000 patient DBPRCT in 46 ICUs in Australia and New Zealand involving critically ill patients receiving invasive mechanical ventilation on EN, randomized to Augmented EN 1.5 kcal/ml (Fresubin® Energy Fibre) vs Standard EN 1.0 kcal/ml (Fresubin® 1000 Complete) with content of protein similar at 56 g/L vs 55 g/L, respectively. Patients on average were 75% medical, 67 years of age, had a BMI 29 with APACHE II score 22, and 62% were on vasopressor support. EN was started within 15 hours of admission to ICU and given for average of 6 days. Study patients got 29.1 kcal/kg IBW/d while controls got 19.6, a difference of 46.7% more calories, both groups getting the same protein at 1.1 gm/kg IBW/d. The primary endpoint, 90-day all-cause mortality, was no different at 26.8% for study patients vs 25.7% for controls (with no differences found on 7 different subgroup analyses for age, disease severity, BMI, admission diagnosis, etc.). No other outcomes were different between groups. Study patients on the energy-dense formula showed evidence of greater intolerance with increased gastric residual volumes, regurgitation, need for promotility agents, elevated glucose levels, and higher dose of insulin required than controls. Researchers concluded that use of the energy-dense formula was a successful strategy for delivering EN closer to goal than routine formulas, but the practice provided no mortality benefit.
Commentary: In their power analysis prior to the study which indicated that an enrollment of 3774 patients was needed, the authors accurately estimated the event rate for their primary endpoint of 90-day mortality between 20-30 % (actual study results 25.7%) but overestimated the treatment effect as being a reduction of 17% (actual was only 1.1%). This means that despite the 3957 patients that ultimately were entered, the study was still underpowered to show a difference between groups. As patients in the control group received 47% less calories than study patients, the study represents yet another trial where the permissive underfeeding of calories yields the same outcome as patients receiving full feeds. Such findings suggest that the amount of feeding is less important than timing, which may be explained by the possibility that underfeeding still provides the non-nutritional benefits of enteral feeding at a low dose (to maintain gut defenses and microbiome) and that the nutritional benefits at full dose (to maintain lean body mass and maximize protein synthesis) are less important in the early phases of critical illness. Slower ramp-ups and less feeding may be a more optimal strategy to avoid overfeeding, risk of ischemia or refeeding syndrome, to support autophagy, and to monitor tolerance as feedings are advanced. Reaching the protein goal may be more important early in the ICU than reaching caloric goal, and keeping the caloric goal at less than 70-80% of energy requirements may have better outcome than delivering full caloric regimen. Mitochondrial failure in the early phases of critical illness may mean that patients are not ready to assimilate full nutritional therapy. An emphasis on getting EN started shortly after admission to the ICU should quickly shift to adding supplemental therapy to maintain the gut barrier, prevent disordered immune responses, and promote commensalism (such as soluble prebiotic fiber, probiotics, pegylated phosphate, intact food formulas, and even fecal microbial transplantation).
Reviewers: D. Clark Files, MD; Wake Forest School of Medicine, Winston-Salem NC 27157
Sadeq Quraishi, MD, MHA, MMSc; Harvard Medical School, Boston MA 02114
Authors: Arabi YM, Al-Balwi M, Hajeer AH, Jawdat D, Sadat M, Al-Dorzi HM, Tamim H, Afesh L, Almashaqbeh W, Alkadi H, Alwadaani D, UdayaRaja GK, Abdulkareem IBA, Al-Dawood A.
Title: Differential Gene Expression in Peripheral White Blood Cells with Permissive Underfeeding and Standard Feeding in Critically Ill Patients: A Descriptive Sub-study of the PermiT Randomized Controlled Trial. Sci Rep. 2018;8(1):17984.
Why is the paper important: This secondary mechanistic analysis of gene expression from patients enrolled in the PermiT trial suggests broad biological differences that occur in response to feeding strategies, supporting the central importance of enteric nutrition on regulation of critical biological pathways.
Summary and Commentary:
In this study1, Arabi et al., performed a secondary analysis of peripheral blood mononuclear cell (PBMC) gene expression from participants included in the PermiT trial.2 PermiT was a 7 center, international, randomized, controlled trial of mechanically ventilated patients (n=894) comparing a permissive underfeeding strategy (40-60% of caloric targets) to standard feeding strategy (70-100% of caloric targets). The PermiT investigators reported that 90-day mortality, as well as a number of secondary outcomes, did not differ between the treatment arms.
For the current manuscript, Arabi et al., analyzed over 1,700 peripheral blood mononuclear cell (PBMC) gene transcripts from 25 permissively underfed and 25 standard feeding participants in the PermiT trial. The most notable downregulated genes in the “underfed” group included those involved in phospholipase pathways (membrane associated phospholipase A1), autophagy inhibition (GADD45B), glycolysis (muscle pyruvate kinase), and inflammation (sPLA2-IIE), as compared to the standard feeding group. On the other hand, the most notable upregulated genes with “underfeeding” included Pan2, Cytochrome C, and VEGF-C, suggesting increased protein catabolism, mitochondrial activity, and endothelial cell proliferation, respectively, as compared to the standard feeding group.
Although the study suggests a potentially beneficial metabolic profile and autophagy promotion in the “underfed” group, several limitations are worth noting. First, the study cohort was limited to only 50 patients. Second, there was only a small difference between groups in terms of actual calories delivered, i.e. 57: IQR 51-59 % of calculated requirements in the “underfed” group vs. 68: IQR 57-90 % in the standard feeding group (p <0.001). Third, while protein delivery was not statistically significant between groups (82: IQR 69-89 % vs. 66: IQR 50-84 %; p=0.07, respectively), the “underfed” group received significantly higher amounts of protein as additional enteral supplements (31: IQR 22-40 g/day vs. 0: IQR 0-6 g/day; p<0.0001, respectively). And fourth, though not statistically significant, the “underfed” group had a numerically lower median age compared to the standard feeding group (28: IQR 23-58 years vs. 45: IQR 38-65 years; p=0.21, respectively). This may be important since impaired autophagy is implicated as an underlying driver of aging in multiple studies.3
Taken all together, these findings suggest that how protein is delivered, and potentially age, in addition to differences in caloric delivery may have played a role in the observed results. Of note, the authors suggest that a downregulation of GADD45B in the “underfed” group may signal enhanced autophagy. While GADD45B is an important regulator at the later parts of the autophagy pathway,4autophagy is a complex pathway involving many proteins. Therefore, it may be premature to conclude that this study demonstrates that “underfeeding” promotes autophagy through inhibition of this single gene. Nonetheless, these data support the concept that nutritional interventions delivered to critically ill patients likely have meaningful and broad impacts on important biological pathways. As such, more studies of this type are needed to better understand the biology of nutrition during acute illness.
1Arabi YM, Al-Balwi M, Hajeer AH, Jawdat D, Sadat M, Al-Dorzi HM, Tamim H, Afesh L, Almashaqbeh W, Alkadi H, Alwadaani D, UdayaRaja GK, Abdulkareem IBA, Al-Dawood A. Differential Gene Expression in Peripheral White Blood Cells with Permissive Underfeeding and Standard Feeding in Critically Ill Patients: A Descriptive Sub-study of the PermiT Randomized Controlled Trial. Sci Rep. 2018;8(1):17984.
2Arabi YM, Aldawood AS, Haddad SH, Al-Dorzi HM, Tamim HM, Jones G, Mehta S, McIntyre L, Solaiman O, Sakkijha MH, Sadat M, Afesh L. Permissive Underfeeding or Standard Enteral Feeding in Critically Ill Adults. New England Journal of Medicine. 2015.
3Levine B, Kroemer G. Biological Functions of Autophagy Genes: A Disease Perspective. Cell. 2019;176(1-2):11-42.
4Keil E, Hocker R, Schuster M, Essmann F, Ueffing N, Hoffman B, Liebermann DA, Pfeffer K, Schulze-Osthoff K, Schmitz I. Phosphorylation of Atg5 by the Gadd45beta-MEKK4-p38 pathway inhibits autophagy. Cell Death Differ. 2013;20(2):321-32.