Early high protein intake and mortality in critically ill ICU patients with low skeletal muscle area and -density

(1)

Amsterdam University of Applied Sciences

Early high protein intake and mortality in critically ill ICU patients with low

skeletal muscle area and -density

Looijaard, Wilhelmus G.P.M.; Dekker, Ingeborg M.; Beishuizen, Albertus; Girbes, Armand

R.J.; Oudemans-van Straaten, Heleen M.; Weijs, Peter J.M.

DOI

10.1016/j.clnu.2019.09.007

Publication date

2020

Document Version

Final published version

Published in

Clinical nutrition

License

CC BY-NC-ND

Link to publication

Citation for published version (APA):

Looijaard, W. G. P. M., Dekker, I. M., Beishuizen, A., Girbes, A. R. J., Oudemans-van

Straaten, H. M., & Weijs, P. J. M. (2020). Early high protein intake and mortality in critically ill

ICU patients with low skeletal muscle area and -density. Clinical nutrition, 39(7), 2192-2201.

https://doi.org/10.1016/j.clnu.2019.09.007

General rights

It is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s)

and/or copyright holder(s), other than for strictly personal, individual use, unless the work is under an open

content license (like Creative Commons).

Disclaimer/Complaints regulations

If you believe that digital publication of certain material infringes any of your rights or (privacy) interests, please

let the Library know, stating your reasons. In case of a legitimate complaint, the Library will make the material

inaccessible and/or remove it from the website. Please contact the library:

https://www.amsterdamuas.com/library/contact/questions, or send a letter to: University Library (Library of the

University of Amsterdam and Amsterdam University of Applied Sciences), Secretariat, Singel 425, 1012 WP

Amsterdam, The Netherlands. You will be contacted as soon as possible.

(2)

Original article

Early high protein intake and mortality in critically ill ICU patients

with low skeletal muscle area and -density

Wilhelmus G.P.M. Looijaard

a,b,c,*

, Ingeborg M. Dekker

d

, Albertus Beishuizen

e

,

Armand R.J. Girbes

a,b,c

, Heleen M. Oudemans-van Straaten

a,b,c

, Peter J.M. Weijs

a,d,f

a_{Department of Adult Intensive Care Medicine, Amsterdam UMC, Vrije Universiteit Amsterdam, Amsterdam, the Netherlands} b_{Research VUmc Intensive Care (REVIVE), Amsterdam UMC, Vrije Universiteit Amsterdam, Amsterdam, the Netherlands} c_{Institute for Cardiovacular Research (ICaR), Amsterdam UMC, Vrije Universiteit Amsterdam, Amsterdam, the Netherlands} d_{Department of Nutrition and Dietetics, Amsterdam UMC, Vrije Universiteit Amsterdam, Amsterdam, the Netherlands} e_{Department of Intensive Care, Medisch Spectrum Twente, Enschede, the Netherlands}

f_{Department of Nutrition and Dietetics, Faculty of Sports and Nutrition, Amsterdam University of Applied Sciences, Amsterdam, the Netherlands}

a r t i c l e i n f o

Article history: Received 9 April 2019 Accepted 13 September 2019 Keywords: Protein

Skeletal muscle area Skeletal muscle density Computed tomography Critical care

Intensive care

s u m m a r y

Background & aims: Optimal nutritional support during the acute phase of critical illness remains controversial. We hypothesized that patients with low skeletal muscle area and -density may speciﬁcally beneﬁt from early high protein intake. Aim of the present study was to determine the association be-tween early protein intake (day 2e4) and mortality in critically ill intensive care unit (ICU) patients with normal skeletal muscle area, low skeletal muscle area, or combined low skeletal muscle area and -density.

Methods: Retrospective database study in mechanically ventilated, adult critically ill patients with an abdominal CT-scan suitable for skeletal muscle assessment around ICU admission, admitted from January 2004 to January 2016 (n¼ 739). Patients received protocolized nutrition with protein target 1.2e1.5 g/kg/ day. Skeletal muscle area and -density were assessed on abdominal CT-scans at the 3rd lumbar vertebra level using previously deﬁned cut-offs.

Results: Of 739 included patients (mean age 58 years, 483 male (65%), APACHE II score 23), 294 (40%) were admitted with normal skeletal muscle area and 445 (60%) with low skeletal muscle area. Two hundred (45% of the low skeletal muscle area group) had combined low skeletal muscle area and -density. In the normal skeletal muscle area group, no signiﬁcant associations were found. In the low skeletal muscle area group, higher early protein intake was associated with lower 60-day mortality (adjusted hazard ratio (HR) per 0.1 g/kg/day 0.82, 95%CI 0.73e0.94) and lower 6-month mortality (HR 0.88, 95%CI 0.79e0.98). Similar associations were found in the combined low skeletal muscle area and -density subgroup (HR 0.76, 95%CI 0.64e0.90 for 60-day mortality and HR 0.80, 95%CI 0.68e0.93 for 6-month mortality).

Conclusions: Early high protein intake is associated with lower mortality in critically ill patients with low skeletal muscle area and -density, but not in patients with normal skeletal muscle area on admission. Theseﬁndings may be a further step to personalized nutrition, although randomized studies are needed to assess causality.

1. Introduction

Optimal nutritional support for critically ill patients remains a topic of debate. Particularly the optimal dose of protein and energy during the early phase of critical illness is controversial[1]. Several observational studies found associations between (early or overall) high protein intake and improved outcomes[2e8], although these ﬁndings have not been conﬁrmed in RCTs[9e13], and even harmful

* Corresponding author. Department of Adult Intensive Care, Amsterdam UMC, Vrije Universiteit Amsterdam, De Boelelaan 1117, 1081 HV, Amsterdam, the Netherlands.

E-mail address:w.looijaard@amsterdamumc.nl(W.G.P.M. Looijaard).

Contents lists available atScienceDirect

Clinical Nutrition

j o u r n a l h o me p a g e :h t t p :/ /w w w .e l se v i e r. co m/ lo ca t e / cl n u

https://doi.org/10.1016/j.clnu.2019.09.007

(3)

associations have been found between very early[6,14]or overall

[15]protein intake and mortality. Possibly, optimal early nutrition may differ between individual patients and concomitant early en-ergy overfeeding may be harmful. Several studies found benefit of high protein intake only in specific groups of patients (e.g. non-septic, non-overfed patients [2], high nutrition risk in the criti-cally ill (NUTRIC) score patients[4], or patients with normal kidney function[16]), suggesting that identifying specific patients profiting from early high protein intake may be important.

Previous research shows that patients with low skeletal muscle area (SMA) or low skeletal muscle density (SMD) on intensive care unit (ICU) admission have an increased mortality, independent of severity of disease [17e19]. Since muscle comprises the largest protein pool in the body, these patients have lower protein reserves and may therefore beneﬁt from early high protein intake.

The objective of this retrospective database study was to determine whether the quantity of early protein intake is associ-ated with mortality and other clinical outcomes in critically ill pa-tients admitted with normal or low SMA and SMD. We hypothesized that patients with low SMA and low SMD on abdominal computed tomography (CT) scans may beneﬁt from early high protein intake.

2. Subjects and methods 2.1. Patients and data

This retrospective database study evaluated the association between early (day 2e4) protein intake and mortality in three groups of ventilated critically ill patients: normal SMA, low SMA, and a subgroup with combined low SMA and low SMD. Patients were admitted to a medical-surgical ICU of a university hospital (Amsterdam University Medical Centers, location VUmc) from January 2004 to January 2016. All patients admitted during this period were screened for eligibility. Inclusion criteria were age18 years, ICU stay4 days, mechanical ventilation, and an abdominal CT-scan made4 days before or after ICU admission. Patients were excluded if the CT-scan was not suitable for muscle analysis (Appendix 1), data on weight or height were missing, or oral intake was initiated within 4 days.

The study was approved by the VUmc institutional review board (IRB00002991, decision 2012/243). The need for informed consent was waived because of the retrospective nature of the study using coded data obtained from standard care. The study has been registered atClinicalTrials.gov(NCT02817646).

Patient data including age, sex, weight, height, admission diagnosis, Acute Physiology and Chronic Health Evaluation (APACHE) II score, daily protein and energy intake (including non-nutritional sources), length of ventilation, ICU- and hospital length of stay, discharge destination, and mortality were obtained from the ICU patient data management system (Metavision, IMDsoft, Tel-Aviv, Israel), hospital information system (Mirador, iSOFT Neder-land BV, Leiden, The NetherNeder-lands), civil registry, or general practitioner.

Primary endpoints were short-term mortality 60 days after ICU admission and long-term mortality 6 months after ICU admission. Secondary endpoints were the odds of being discharged to home, length of ventilation, and ICU- and hospital length of stay. 2.2. CT-scan analysis

Patients were categorized into three groups; admitted with normal SMA, with low SMA, and a subgroup of the low SMA group admitted with combined low SMA and low SMD[18,19].

All abdominal CT-scans made during the study period for diagnostic or interventional purposes4 days before or after ICU admission were reviewed for suitability for muscle analysis. CT-scans were analyzed using Slice-O-matic versions 4.3 and 5.0 (TomoVision, Montreal, QC, Canada) by two certiﬁed investigators (WGPML and IMD, trained by the Cross Cancer Institute, Canada). CT-scans were analyzed at the level of the third lumbar vertebra (L3). The precision of single L3 slice CT-scan analysis is high (inter-and intra-observer variability<2%) [20]and L3 SMA is strongly related to whole-body skeletal muscle volume (r ¼ 0.83e0.99, p< .01)[21,22].

Muscle tissue was identified using boundaries in Hounsfield Units (HU) set to29 to þ150[23]. All muscles present at the L3 level were analyzed. Low SMA was de_{fined using previously found} ICU-specific cut-off points: males <170 cm2_{and females}_{<110 cm}2_, which were associated with hospital mortality. Patients with a low SMA according to these cut-off points had an odds ratio for hospital mortality of 4.3 (95% confidence interval (CI) 2.0e9.0, p < .001) compared to those with normal SMA[18]. The software automati-cally calculates SMD from the mean radiological attenuation of all L3 muscle. Low SMD was defined using cut-off points the for the 5th percentile from a healthy population of kidney donors (95% of this healthy population had a SMD above these values and 5% below these values): males<29.3 HU and females <22.0 HU[24]. 2.3. Protein intake and nutritional protocol

Average daily protein intake over ICU admission day 2e4 was used as early protein intake. Day 1 was excluded to include only full nutrition days. Protein intake was analyzed both as continuous variable and dichotomized using mean day 2e4 intake 1.2 g/kg/ d or<1.2 g/kg/d.

Enteral nutrition (EN) was initiated within 24 h from ICU admission or after hemodynamic stabilisation. The preferable route was enteral, parenteral nutrition (PN) was provided only when the gut failed, not as supplemental nutrition in theﬁrst week.

The protein intake target was 1.2e1.5 g/kg pre-admission body weight per day. Weight was adjusted to weight at body-mass index (BMI) 20 kg/m2for patients with BMI<20 kg/m2_{and to weight at} BMI 27.5 kg/m2for BMI_{>30 kg/m}2[25]. Protein provision was not adjusted in case of renal failure or renal replacement therapy.

Energy target was estimated resting energy expenditure (REE) using the Harris and Benedict 1984 equation[26]_{þ30% for stress}

and activity.

We previously developed an algorithm to select the best nutri-tional formula and feeding rate to meet both energy- and protein targets, using several nutritional formulae with a range of energy-to-protein ratios[27].

2.4. Statistical analysis

Independent samples T-tests and ManneWhitney U-tests were used to compare continuous variables, and Fisher Exact tests and Chi2-tests with post-hoc z-test with Bonferroni correction for cat-egorical variables. Kaplan Meier survival curves were made for the normal SMA, low SMA, and combined low SMA and low SMD groups; and for protein intake (1.2 g/kg/d vs. <1.2 g/kg/d) within these groups, with Log-rank tests to compare survival curves.

Cox regression analysis was used to evaluate the association between protein intake (as continuous variable and dichotomous 1.2 g/kg/d vs. <1.2 g/kg/d) and 60-day- and 6-month mortality, with adjustments for APACHE II score and energy intake as pro-portion of calculated needs.

To evaluate secondary outcome measures in ICU- or hospital survivors, logistic regression analysis was used for discharge to

(4)

home and linear multiple regression analyses for length of venti-lation, and ICU- and hospital length of stay. Finally, we performed sensitivity analyses including only patients who were adequately fed (80e120% of energy target); pre-ICU hospital stay of <1 week; excluding trauma patients; with additional adjustments for sex, age, and pre-ICU hospital stay; with cut-off points for skeletal muscle index[28]; and with protein intake expressed in g/kg ideal body weight[29]. Additional information on these statistical ana-lyses can be found inAppendix 2.

IBM SPSS Statistics 22 (IBM Corp, Armonk, NY, USA); R 3.5.0 (R Foundation for Statistical Computing, Vienna, Austria) with sur-vival-, tidyverse-, and ggfortify packages; and GraphPad Prism 7 (GraphPad Software, La Jolla, CA, USA) were used for statistical analysis. Values are reported as mean (_{±standard deviation, SD) or} median (25e75% interquartile range, IQR). All statistical tests were conducted two-sided. A p < .05 was considered statistically signiﬁcant.

3. Results

A total of 3.851 patients were admitted to the ICU during the study period for at least 4 days and mechanically ventilated, with a mean APACHE II score of 24.8 (Supplemental Fig. 1). Nine hundred eighty-two patients fulﬁlled inclusion criteria. After excluding pa-tients with CT-scans not suitable for muscle analysis (n¼ 210) or missing data (n¼ 33), 739 patients were available for ﬁnal analysis. The CT-scan was made within one day of ICU admission in 595 patients (81%). Of 739 patients, 445 patients (60%) were admitted to the ICU with low SMA, and among these 200 patients (45% of the low SMA group) with combined low SMA and low SMD. Nutritional intake is presented inSupplemental Table 1.

3.1. Patient characteristics of muscle groups

Patients admitted with low SMA were signiﬁcantly older, more often male, had a lower weight and BMI, higher APACHE II score, longer pre-ICU hospital stay, and were less often trauma patients when compared to the normal SMA group (Table 1). Early (day 2e4) protein intake was not signiﬁcantly different between the low- and normal SMA groups (0.70± 0.39 vs. 0.75 ± 0.42 g/kg/d, respectively, p¼ .13). The low SMA group received more energy (90 ± 38% vs. 84± 36% of REE, p ¼ .03).

Similar differences were seen between the combined low SMA and low SMD subgroup and the normal SMA group, except that BMI and energy intake were not signiﬁcantly different (Table 1). 3.2. Mortality in muscle groups

Sixty-day and six-month mortality were 14.6% and 22.1% in the normal SMA group, 32.7% and 42.7% in the low SMA group, and 38.3% and 50.0% in the combined low SMA and low SMD subgroup (all p_{< .001 vs. normal SMA). Kaplan Meier survival curves of the} latter two groups were signiﬁcantly lower than the normal SMA group (Fig. 1A).

3.3. Patient characteristics and mortality of protein intake groups within muscle groups

In the normal- and low SMA groups, patients with an early protein intake1.2 g/kg/d had a lower body weight and BMI, and longer pre-ICU hospital stay than patients who received<1.2 g/kg/ d (Table 1). In the combined low SMA and low SMD subgroup, patients with an early protein intake 1.2 g/kg/d had a lower APACHE II score and longer pre-ICU hospital stay.

Mortality was not signi_{ﬁcantly different between protein intake} 1.2 g/kg/d vs. <1.2 g/kg/d in the normal SMA and low SMA groups (Table 2). However, in the combined low SMA and low SMD sub-group, both 60-day and 6-month mortality were signiﬁcantly lower in patients with an early protein intake1.2 g/kg/d (11% vs. 43%, p¼ .001 and 29% vs. 54%, p ¼ .02, respectively).

3.4. Protein intake as continuous variable

In adjusted Cox regression analysis with protein intake expressed as continuous variable, no signiﬁcant association be-tween protein intake and mortality was found in the normal SMA group (Table 3). In the low SMA group, higher early protein intake was associated with lower 60-day mortality (adjusted hazard ratio (HR) per 0.1 g/kg/d 0.82, 95% CI 0.73e0.94) and lower 6-month mortality (HR 0.88, 95%CI 0.80e0.98). Similar associations were found in the combined low SMA and low SMD subgroup (HR 0.76, 95%CI 0.64e0.90 for 60-day mortality and HR 0.80, 95%CI 0.68e0.93 for 6-month mortality). Higher early energy intake was associated with higher 60-day and 6-month mortality in the low SMA group and in the combined low SMA and low SMD subgroup, but not in the normal SMA group. The hazard ratios associated with different levels of protein- and energy intake are visualised inFig. 2

(60-day mortality) andSupplemental Fig. 2(6-month mortality).

3.5. Protein intake as dichotomized variable

Kaplan Meier survival curves of patients with an early protein intake 1.2 g/kg/d were signiﬁcantly higher than those who received<1.2 g/kg/d in the combined low SMA and low SMD sub-group (Fig. 1D), but not in the normal- and low SMA groups (Fig. 1B,C).

In the normal SMA group, no signiﬁcant association was found between protein intake and mortality (Table 3). In the low SMA group, early protein intake1.2 g/kg/d was associated with lower 60-day mortality (HR 0.53, 95%CI 0.29e0.98), and in the combined low SMA and low SMD subgroup with lower 60-day mortality (HR 0.16, 95%CI 0.05e0.55) and lower 6-month mortality (HR 0.32, 95% CI 0.14e0.74).

3.6. Secondary outcomes

Protein intake was not associated with the odds of discharge to home (Supplemental Table 2). However, higher protein intake as continuous variable was associated with a shorter ICU stay, and both as continuous and dichotomized variable with shorter me-chanical ventilation in patients with normal- and low SMA, but not in patients with combined low SMA and low SMD.

3.7. Sensitivity analyses

In sensitivity analyses the results remained robust (Supplemental Table 3).

4. Discussion

This study in mechanically ventilated patients admitted to the ICU for at least four days and having an abdominal CT scan made around admission demonstrates that an early higher protein intake is associated with lower mortality in patients admitted with low skeletal muscle area and -density but not in patients admitted with normal skeletal muscle area when adjusted for confounders energy intake and severity of disease. Theseﬁndings are relevant because low skeletal muscle area and the combination with low skeletal muscle density on admission are associated with high mortality

W.G.P.M. Looijaard et al. / Clinical Nutrition 39 (2020) 2192e2201 2194

(5)

Table 1

Characteristics of patients with mean day 2e4 protein intake <1.2 g/kg/day or 1.2 g/kg/day in patients admitted with low skeletal muscle area or combined low skeletal muscle area and -density.

Muscle groups Normal skeletal muscle areaa _{Low skeletal muscle area}a _{Combined low skeletal muscle area and}

-density subgroupa Normal SMA n¼ 294 Low SMA n¼ 445 P-value vs. normal SMA

Low SMA&

low SMD n¼ 200 P-value vs. normal SMA Protein< 1.2 g/ kg/dayb_n_{¼ 260} Protein 1.2 g/ kg/dayb_n_{¼ 34} P-value protein groups Protein< 1.2 g/ kg/dayb n¼ 372 Protein 1.2 g/ kg/dayb_n_{¼ 73} P-value protein groups Protein< 1.2 g/ kg/dayb n¼ 171 Protein 1.2 g/ kg/dayb_n_{¼ 29} P-value protein groups Age, median (IQR),

y

52 (36e65) 66 (54e75) <.001 71 (62e77) <.001 52 (37e65) 50 (30e63) .78 67 (54e75) 62 (55e72) .12 72 (62e77) 68 (57e75) .10

Sex male, No. (%) 179 (61) 304 (68) .04 144 (72) .01 163 (63) 16 (47) .09 259 (70) 45 (62) .22 125 (73) 19 (66) .50

Weight, median (IQR), kg

80 (71e90) 75 (65e82) <.001 79 (70e85) .02 80 (73e90) 75 (65e80) .006 75 (65e83) 70 (60e80) .04 77 (70e85) 80 (70e87) .39

BMI, median (IQR),

kg/m2 25.6 (23.5 e27.8) 24.4 (22.5 e26.4) <.001 25.1 (23.2 e27.9) .41 25.7 (23.5 e27.8) 24.5 (23.0 e25.8) .02 24.5 (22.8 e26.8) 23.5 (21.5 e24.9) .02 25.3 (23.1 e27.8) 24.8 (23.3 e29.4) .65 Under-weight, No. (%)c 2 (1) 25 (6) 8 (4) 2 (1)a 0 (0)a 18 (5)a 7 (9)a 6 (3)a 2 (7)a Normal weight, No. (%)c 135 (46) 252 (56) 92 (46) 113 (43)a 22 (65)b 203 (55)a 49 (67)b 77 (45)a 15 (52)a Over-weight, No. (%)c 122 (41) 129 (29) 73 (37) 111 (43)a 11 (32)a 119 (32)a 10 (15)b 68 (40)a 5 (17)b Obese, No. (%)c _{35 (12)} _{39 (9)} _{27 (13)} _{34 (13)} a 1 (3)a 32 (8)a 7 (9)a 20 (12)a 7 (24)a APACHE II score, mean (SD) 22 (8) 25 (8) <.001 26 (8) <.001 21 (7) 23 (8) .34 25 (8) 23 (8) .09 26 (8) 22 (8) .01 Admission category, No. (%) <.001 <.001 .70 .25 .42 Medical 95 (32) 224 (50) 117 (58) 83 (32) 12 (35) 192 (52) 32 (43) 102 (60) 15 (52) Surgical 199 (68) 221 (50) 83 (42) 177 (68) 22 (65) 180 (48) 41 (57) 69 (40) 14 (48) Admission diagnosis, No. (%) <.001 <.001 .10 .18 .67 Cardio-vascular 6 (2) 29 (6) 13 (7) 6 (2)a 0 (0)a 25 (7)a 4 (5)a 11 (6)a 2 (7)a Metabolic/Renal 8 (3) 11 (3) 10 (5) 8 (3)a 0 (0)a 11 (3)a 0 (0)a 10 (6)a 0 (0)a Neurologic 16 (5) 17 (4) 4 (2) 16 (6)a 0 (0)a 14 (4)a 3 (4)a 3 (2)a 1 (3)a Post-resuscitation 13 (4) 29 (6) 14 (7) 13 (5)a 0 (0)a 29 (8)a 0 (0)b 14 (8)a 0 (0)a Post-surgery 60 (21) 146 (33) 74 (37) 53 (20)a 7 (21)a 115 (31)a 31 (43)a 61 (36)a 13 (45)a Respiratory insufﬁciency 23 (8) 66 (15) 36 (18) 17 (7)a 6 (17)b 54 (14)a 12 (16)a 30 (18)a 6 (21)a Sepsis 22 (8) 48 (11) 28 (14) 18 (7)a 4 (12)a 40 (11)a 8 (11)a 23 (13)a 5 (17)a Trauma 139 (47) 75 (17) 9 (4) 124 (48)a 15 (44)a 65 (17)a 10 (14)a 8 (5)a 1 (3)a Other 7 (2) 24 (5) 12 (6) 5 (2)a 2 (6)a 19 (5)a 5 (7)a 11 (6)a 1 (3)a Length of pre-ICU hospital stay, median (IQR), d

0 (0e1) 1 (0e4) <.001 2 (0e6) <.001 0 (0e1) 2 (0e13) <.001 0 (0e3) 4 (0e10) <.001 1 (0e5) 4 (1e14) .009

Time from ICU admission to CT-scan, mean (SD), d

0 (1) 0 (1) .78 0 (2) .14 0 (1) 0 (2) .05 0 (1) 0 (2) .20 1 (2) 0 (2) .17

Skeletal muscle area, median (IQR), cm2

Female 123 (116

e138)

96 (84 e102)

<.001 89 (76e99) <.001 122 (116e136) 128 (118e141) .53 95 (84e101) 97 (90e104) .21 88 (74e98) 96 (87e101) .12

Male 194 (181 e209) 142 (124 e158) <.001 135 (116 e150)

<.001 194 (182e209) 185 (176e200) .22 144 (125e158) 133 (121e149) .01 136 (117e152) 133 (113e149) .40

Skeletal muscle index, median (IQR), cm2_/m2

Female 44.5 (40.9 e48.7) 34.7 (31.2 e37.6) <.001 32.7 (28.9 e35.5) <.001 44.4 (40.7 e48.2) 46.6 (41.1 e50.1) .50 34.5 (30.7 e37.7) 35.8 (33.1 e37.4) .27 32.0 (27.8 e35.0) 34.0 (32.8 e36.9) .13 (continued on next page)

W .G.P .M. Looijaard et al. / Clin ical Nutrition 39 (2020) 2 192 e 220 1 21 9 5

(6)

and appear to be common among critically ill patients (60% and 27% respectively). This study suggests that these patients may benefit from an early high protein intake of1.2 g/kg/d. Although our findings have a physiological rationale and were robust in sensi-tivity analyses, no inferences about causality can be made in this retrospective study and randomized studies are needed to exclude residual confounding and assess causality. Nevertheless, this is the largest study op to now combining both muscle- and nutritional data and ourfindings may be a first step to personalized nutritional support during critical illness.

4.1. High protein intake

We previously found an association between day 4 protein intake of1.2 g/kg/d and lower hospital mortality in non-septic, non-overfed critically ill patients [2]. In the current study, we identify a subgroup which may speciﬁcally beneﬁt from early high protein intake.

An association between higher protein intake and lower mor-tality was demonstrated in several observational studies in a het-erogeneous ICU population[3,5,7]. Few studies specifically report early protein intake. In a retrospective cohort study, Bendavid et al. found an association between a protein intake of>0.7 g/kg/d dur-ing thefirst 3 days of ICU admission and lower 60-day all-cause mortality[8]. Additionally, in another large retrospective cohort study, Zusman et al. also found a signi_{ficant association between} day 3 protein intake of 1 g/kg/d and lower mortality[30]. However, higher protein delivery in thefirst week was found to be associated with greater muscle wasting in a small selected cohort of patients with prolonged critical illness[15]. Additionally, Koekkoek et al. found lower mortality when protein intake was gradually increased during 7 days[6]. Similarly, a post-hoc analysis of the EPaNIC trial suggested that the day 3 protein/amino acid dose, rather than the glucose dose, explained the delayed recovery in the early PN group

[14]. However, on day 5 and day 7 this association was not found. Thus, especially the optimal protein dose early during critical illness is still controversial.

These seemingly contrasting findings suggest that optimal nutritional strategies may differ between patients. Certain sub-groups may benefit while others may not. Indeed, in a post-hoc subgroup analysis of the Nephro-protective randomized trial, reduced mortality was observed only in patients with normal kidney function allocated to receive high protein[16]. Additionally, an association between greater protein adequacy and lower mor-tality was found only in patients with a high NUTRIC score[4]. The present study suggests that specifically patients with low SMA and low SMD may benefit from early high protein intake.

The contrastingfindings regarding early protein intake may also be attributed to concomitant energy overfeeding. Because of inflammation-induced endogenous energy production, hypocaloric nutrition is recommended during early critical illness[2,3,31]. In the well-designed EAT-ICU trial, early goal directed nutrition was not associated with improved outcomes [12]. Furthermore, the INTACT trial was stopped prematurely because of higher mortality in the intensive medical nutrition therapy group[13]. In both trials, the high-protein groups received full energy from day 1 with inherent risk of early energy overfeeding. In the current study, mean energy intake was frequently at or above target as well and higher early energy intake was associated with higher mortality. We therefore adjusted for energy intake and validated our results in a sensitivity analysis including only patients who were not under-nor overfed, and found similar results. Hence, our results are robust but observational. The benefit of early high protein in patients admitted with low SMA and low SMD should now be assessed in a

Table 1 (continued ) Muscle groups Normal skeletal muscle area a Low skeletal muscle area a Combined low skeletal muscle area and -density subgroup a Normal SMA n¼ 294 Low SMA n ¼ 445 P -value vs. normal SMA Low SMA & low SMD n ¼ 200 P -value vs. normal SMA Protein < 1.2 g/ kg/day bn ¼ 260 Protein 1.2 g/ kg/day bn ¼ 34 P

-value _protein _groups

Protein < 1.2 g/ kg/day b n ¼ 372 Protein 1.2 g/ kg/day bn ¼ 73 P

Protein < 1.2 g/ kg/day b n ¼ 171 Protein 1.2 g/ kg/day bn ¼ 29 P

Male 59.3 (55.2 e 64.1) 45.5 (40.1 e 49.1) < .001 42.8 (37.0 e 48.2) < .001 59.3 (55.2 e 64.2) 57.7 (55.3 e 62.7) .71 45.9 (40.4 e 49 .5) 42.3 (37.4 e 47.1) .03 43.0 (37.7 e 48.2) 38.3 (36.1 e 46.9) .19 Skeletal muscle density, mean (SD), HU Female 31.6 (11.6) 25.7 (10.3) < .001 15.7 (4.1) < .001 32.0 (11.4) 30.0 (12.9) .52 25.4 (10.4) 26.9 (10.0) .48 15.6 (4.2) 16.1 (3.8) .70 Male 42.0 (11.3) 30.9 (11.3) < .001 21.5 (5.5) < .001 42.0 (11.1) 41.9 (13.4) .95 31.0 (11.6) 30.1 (9.6) .61 21.5 (5.5) 21.7 (5.2) .89 Protein intake day 2 e 4, mean (SD), g/ kg/d 0.70 (0.39) 0.75 (0.42) .13 0.73 (0.39) .41 0.61 (0.31) 1.38 (0.14) < .001 0.62 (0.33) 1.39 (0.17) < .001 0.62 (0.32) 1.35 (0.12) < .001 Energy intake day 2 e 4, mean (SD), kcal/kg/d 17.5 (7.8) 18.8 (8.5) .04 17.7 (7.6) .76 16.0 (6.8) 28.9 (4.0) < .001 16.6 (6.9) 30.3 (6.5) < .001 16.0 (6.5) 28.1 (5.5) < .001 Energy intake day 2 e 4, mean (SD), % of REE/d 84 (36) 90 (38) .03 89 (37) .17 77 (32) 138 (14) < .001 80 (32) 143 (23) < .001 80 (32) 140 (19) < .001 P -values in bold indicate a signi ﬁ cant test result. APACHE: acute physiologic, age, and chronic health eval uation, BMI: body mass index, CT: computed tomography, REE: resting ener gy expenditure, ICU : intensive care unit. a, b Values within a m uscle group on the same row not shar ing the same subscript were signi ﬁ cantly different on post-hoc z-test with Bonferroni correction. a Skeletal muscle area cut-offs: 170 cm 2for males and 110 cm 2for females, [15] skeletal muscle density cut-offs: 29.3 HU for males and 22.0 HU for females [21] . b Mean protein intake on day 2e 4. c WHO categories; underweight: BMI < 18.5 kg/m 2, normal weight: BMI 18.5 e 24.9 kg/m 2, overweight: BMI 2 5e 29.9 kg/m 2, obesity: BMI 30 kg/m 2.

(7)

randomized study avoiding energy overfeeding by the use of pro-tein supplements or a high propro-tein-to-energy ratio nutrition. 4.2. Low skeletal muscle density and protein intake

In the present study, the association between high protein intake and lower mortality was more pronounced in patients admitted with combined low SMA and SMD. While a low SMA is an indication of low muscle mass and therefore low protein reserves, low SMD is associated with qualitative changes in muscle such as fatty inﬁltration or myosteatosis[32]. Myosteatosis may create an environment with low-grade inﬂammation and insulin resistance

[33e35], which contribute to anabolic resistance [36]. A higher protein intake may be needed to overcome this anabolic resistance. This may additionally explain why the most apparent bene_{ﬁt of} protein intake was seen in the patients with combined low SMA (low reserves) and low SMD (anabolic resistance). However, these explanations remain speculation.

4.3. Low skeletal muscle area and -density

The high prevalence of both low SMA and low SMD found in this study, as well as the increased mortality in these groups, are in line with other studies[17e19,37,38]. Identifying these patients may improve risk-stratiﬁcation and help guide treatments. However, accurately doing so remains a challenge. BMI or other simple anthropometric measurements are not accurate, because they do not detect sarcopenic obesity [39]. Although CT-scanning may provide accurate measurements of muscle area and density, routine CT-scanning is not feasible in critically ill patients due to costs, time, risks associated with transport, and exposure to radiation. How-ever, some of these limitations may be offset and automatic CT-scans analysis for determining SMA and SMD may become clini-cally available when novel artiﬁcial intelligence-based methods are integrated into routine image analysis[40,41]. Additionally, alter-native bedside methods to measure body composition in the ICU are available, although each has their own limitations [42]. Musculoskeletal ultrasound provides both muscle mass and -quality, although standardized protocols and cut-off points are lacking. For bio-electrical impedance analysis these are available, however, concerns exist about the applicability of algorithms to calculate muscle mass in critically ill patients.

4.4. Strengths and limitations

The high accuracy of CT-scan analysis to measure SMA and SMD adds to the validity of ourﬁndings. Furthermore, the use of an al-gorithm to select the optimal nutrition from several nutritional formulae with a range of energy-to-protein ratios, rather than using one nutritional formula with aﬁxed energy-to-protein ratio, pro-vided enough statistical variation to analyze protein intake and energy intake separately.

We also acknowledge several limitations to this study. It is a retrospective study and therefore no inferences about causality can be made: our results are hypothesis-generating only. Possibly, the association between high protein intake and lower mortality is confounded by less severely ill patients reaching higher protein intakes. However, we corrected for severity of illness and the re-sults were robust in a sensitivity analysis including only patients who were adequately fed. Additionally, baseline differences

Fig. 1. Kaplan Meier survival curves. Kaplan Meier survival curves comparing pa-tients admitted with normal SMA, low SMA, and combined low SMA and low SMD (A), and Kaplan Meier survival curves in patients admitted with normal SMA (B), low SMA (C), and combined low SMA and low SMD (D) comparing mean day 2e4 protein intake <1.2 g/kg/d vs. 1.2 g/kg/d. Patients admitted with combined low SMA and low SMD had the lowest 6-month survival. Within this group, those with an early protein intake

1.2 g/kg/d had a better 6-month survival than those with <1.2 g/kg/d. Log-rank tests were used to compare survival curves. Light-coloured areas denote the 95% conﬁdence interval. SMA: skeletal muscle area, SMD: skeletal muscle density.

(8)

Table 2

Outcomes of patients with mean day 2e4 protein intake <1.2 g/kg/day or 1.2 g/kg/day within muscle groups.

Normal skeletal muscle areaa _{Low skeletal muscle area}a _{Combined low skeletal muscle area and -density subgroup}a Protein < 1.2 g/kg/dayb n¼ 260 Protein 1.2 g/kg/dayb n¼ 34 P-value Protein < 1.2 g/kg/dayb n¼ 372 Protein 1.2 g/kg/dayb n¼ 73 P-value Protein < 1.2 g/kg/dayb n¼ 171 Protein 1.2 g/kg/dayb n¼ 29 P-value

60-day mortality, No. (%)c

37 (15) 3 (9) .44 122 (34) 17 (25) .16 72 (43) 3 (11) .001

6-month mortality, No.

(%)c 46 (20) 5 (15) .64 154 (44) 26 (38) .50 90 (54) 8 (29) .02

Length of ventilation,

median (IQR), dd 10 (5e18) 10 (5e17)

.72 9 (6e17) 9 (5e18) .37 9 (5e17) 10 (4e28) .85

Length of ICU stay, median (IQR), dd

13 (7e22) 14 (6e23) .86 13 (8e22) 12 (7e21) .31 12 (8e21) 13 (6e28) .86

Length of hospital stay,

median (IQR), de 33 (20e50) 41 (25e57) .19 34 (22e57) 45 (30e69) .01 35 (23e61) 63 (37e77) .01 Destination after discharge, No. (%)e .85 .10 .49 Home 87 (38) 13 (43) 80 (33) 25 (45) 33 (35) 11 (44) Other hospital 49 (22) 4 (13) 68 (28) 11 (19) 25 (28) 4 (16) Nursing home 42 (18) 6 (20) 53 (21) 14 (25) 27 (29) 7 (28) Rehabilitation unit 31 (14) 5 (17) 25 (10) 6 (11) 5 (5) 3 (12) Other 19 (8) 2 (7) 20 (8) 0 (0) 3 (3) 0 (0)

ICU: intensive care unit.

P-values in bold indicate a signiﬁcant test result.

a_{Skeletal muscle area cut-offs: 170 cm}2_{for males and 110 cm}2_{for females,}_[15]_{skeletal muscle density cut-offs: 29.3 HU for males and 22.0 HU for females}_[21]_. b_{Mean protein intake on day 2e4.}

c _{Due to losses to follow-up, standardized mortality was known for a subset of patients. Sixty-day mortality n}_{¼ 699. Six-month mortality n ¼ 690.} d _{Length of ventilation and length of ICU stay in ICU survivors only, n}_{¼ 619.}

e_{Length of hospital stay and destination after discharge in hospital survivors only, n}_{¼ 561.}

Table 3

Cox regression analyses on the association between mean day 2e4 protein intake and 60-day- and 6-month mortality.

Protein intake - continuousa _{Protein intake - dichotomized}b

Unadjusted model Adjusted model Unadjusted model Adjusted model

HR 95%CI HR 95%CI HR 95%CI HR 95%CI

60-day mortality Normal SMAc

Protein intake 0.91d _0.79e1.05 _0.92d _0.72e1.18 _0.57 _0.18e1.86 _0.28 _0.07e1.08

APACHE II score 1.07 1.03e1.12 1.08 1.04e1.13

Energy intake 0.92 0.09e9.33 2.33 0.78e7.01

Low SMAc

Energy intake 3.93 1.20e12.85 1.64 0.94e2.87

Low SMA& low SMDc

Protein intake 0.96 0.90e1.01 0.76 0.64e0.90 0.20 0.06e0.63 0.16 0.05e0.55

Energy intake 13.65 2.39e77.95 1.98 0.90e4.37

6-month mortality Normal SMAc

Protein intake 1.02 0.96e1.09 0.99 0.83e1.19 0.75 0.30e1.90 0.40 0.13e1.19

Energy intake 1.37 0.18e10.42 2.05 0.78e5.41

Low SMAc

Energy intake 2.72 0.95e7.79 1.64 1.01e2.68

Low SMA& low SMDc

Energy intake 4.96 1.10e22.34 1.82 0.92e3.60

APACHE: acute physiologic, age, and chronic health evaluation, HR: hazard ratio, SMA: skeletal muscle area, SMD: skeletal muscle density. Values in bold indicate a signiﬁcant result, p < .05.

a_{Average day 2e4 protein intake per 0.1 g/kg/day increase.} b_{Average day 2e4 protein intake 1.2 g/kg/day vs. <1.2 g/kg/day.}

c _{Skeletal muscle area cut-offs: 170 cm}2_{for males and 110 cm}2_{for females,}_[15]_{skeletal muscle density cut-offs: 29.3 HU for males and 22.0 HU for females}_[21]_{. Normal}

SMA n¼ 294. Low SMA n ¼ 444. Low SMA & low SMD n ¼ 200.

d _{Proportional hazards assumption not met, therefore the time-dependent covariate was added to the Cox regression model.}

(9)

between the protein intake groups exist. For example, patients with an early high protein intake had a significantly lower body weight. To account for these baseline differences we performed sensitivity analyses including all significantly different baseline variables, and recalculated protein intake into g/kg ideal body weight. Results were robust in these analyses. Additionally, while higher protein intake was associated with lower mortality, higher energy intake was associated with higher mortality, supporting a specific role of protein and not volume of nutrition. Nevertheless, residual con-founding is possible. Furthermore, we included only patients in whom an early abdominal CT-scan was available. This selection bias limits the generalizability of ourfindings. Finally, we used ICU- and sex-specific cut-off points for SMA which have not yet been vali-dated elsewhere and are not normalized to height[18]. However, in sensitivity analysis using commonly used cut-off points for oncology patients by Martin et al. which are normalized to height

[28], we found similar results.

In this retrospective database study in mechanically ventilated critically ill patients, an early high protein intake, particularly of more than 1.2 g/kg/d, was associated with lower mortality in pa-tients admitted with low skeletal muscle area and -density, but not in those with normal muscle area. Further studies are needed to evaluate theseﬁndings in a prospective randomized design.

Author contributions

WL, AB, and PW designed research; WL, ID, HO, and PW con-ducted research; AG provided essential materials; WL and PW

analyzed data and performed statistical analysis; WL, HO, and PW wrote the paper; WL had primary responsibility forﬁnal content. All authors read and approved theﬁnal manuscript.

Con_{ﬂicts of interest}

WL has received congress support and speaker's honorary from Baxter and Fresenius-Kabi.

AG holds stock options as commissioner of a start-up company for development of new antibiotics.

HO has received congress support and speaker's honorary from Abbott, Baxter/Gambro, Fresenius-Kabi, Nestle and Nutricia.

PW has received funds from Baxter, Fresenius-Kabi, Nestle, and Nutricia.

ID and AB have no con_{ﬂicts of interest.}

Sources of support

A research grant provided by Baxter Healthcare was used for acquisition of CT scan analysis software and for a part of CT scan analysis. Fresenius Kabi Deutschland GmbH provided an unre-stricted grant for writing and publication costs. The funders had no role in the design and conduct of the study; collection, manage-ment, analysis, and interpretation of the data; review or approval of the manuscript; and decision to submit the manuscript for publication.

Fig. 2. Adjusted hazard ratios for 60-day mortality of day 2e4 protein intake (top row) and day 2e4 energy intake (bottom row). Hazard ratios in patients admitted with normal SMA (left), low SMA (middle), and combined low SMA and low SMD (right). Data werefit using a cox regression model. The graphs were truncated at a protein intake of 1.5 g/kg and an energy intake of 150% of REE because of the limited number of patients with higher intakes. Protein intake was adjusted for APACHE II score and energy intake, while energy intake was adjusted for APACHE II score and protein intake. In the low SMA group and in the combined low SMA and low SMD group, higher protein intake was associated with lower 60-day mortality and higher energy intake was associated with higher 60-day mortality. The blue area shows the 95% confidence interval. REE: resting energy expenditure, SMA: skeletal muscle area, SMD: skeletal muscle density. (For interpretation of the references to colour in thisfigure legend, the reader is referred to the Web version of this article.)

(10)

Acknowledgements

We thank Ronald Driessen from the Amsterdam UMC, Vrije Universiteit, Department of Intensive Care Medicine for his contribution in the collection of data.

Appendix A. Supplementary data

Supplementary data to this article can be found online at

https://doi.org/10.1016/j.clnu.2019.09.007.

References

[1] Patel JJ, Martindale RG, McClave SA. Controversies surrounding critical care nutrition: an appraisal of permissive underfeeding, protein, and outcomes. J Parenter Enteral Nutr 2017;42(3):508e15. https://doi.org/10.1177/ 0148607117721908.

[2] Weijs PJM, Looijaard WGPM, Beishuizen A, Girbes ARJ, Oudemans-van Straaten HM. Early high protein intake is associated with low mortality and energy overfeeding with high mortality in non-septic mechanically ventilated critically ill patients. Crit Care 2014;18(6):701. https://doi.org/10.1186/ s13054-014-0701-z.

[3] Zusman O, Theilla M, Cohen J, Kagan I, Bendavid I, Singer P. Resting energy expenditure, calorie and protein consumption in critically ill patients: a retrospective cohort study. Crit Care 2016;20(1):367.https://doi.org/10.1186/ s13054-016-1538-4.

[4] Compher C, Chittams J, Sammarco T, Nicolo M, Heyland DK. Greater protein and energy intake may be associated with improved mortality in higher risk critically ill patients: a multicenter, multinational observational study. Crit Care Med 2017;45(2):156e63. https://doi.org/10.1097/ ccm.0000000000002083.

[5] Allingstrup MJ, Esmailzadeh N, Wilkens Knudsen A, Espersen K, Hartvig Jensen T, Wiis J, et al. Provision of protein and energy in relation to measured requirements in intensive care patients. Clin Nutr 2012;31(4):462e8. [6] Koekkoek W, van Setten CHC, Olthof LE, Kars J, van Zanten ARH. Timing of

PROTein INtake and clinical outcomes of adult critically ill patients on pro-longed mechanical VENTilation: the PROTINVENT retrospective study. Clin Nutr 2018. https://doi.org/10.1016/j.clnu.2018.02.012. S0261-5614(18): 30075-X.

[7] Nicolo M, Heyland DK, Chittams J, Sammarco T, Compher C. Clinical outcomes related to protein delivery in a critically ill population: a multicenter, multi-national observation study. J Parenter Enteral Nutr 2016;40(1):45e51.https:// doi.org/10.1177/0148607115583675.

[8] Bendavid I, Zusman O, Kagan I, Theilla M, Cohen J, Singer P. Early adminis-tration of protein in critically ill patients: a retrospective cohort study. Nu-trients 2019;11(1).https://doi.org/10.3390/nu11010106.

[9] Ferrie S, Allman-Farinelli M, Daley M, Smith K. Protein requirements in the critically ill: a randomized controlled trial using parenteral nutrition. J Parenter Enteral Nutr 2016;40(6):795e805. https://doi.org/10.1177/ 0148607115618449.

[10] Rugeles SJ, Rueda JD, Diaz CE, Rosselli D. Hyperproteic hypocaloric enteral nutrition in the critically ill patient: a randomized controlled clinical trial. Indian J Crit Care Med: Peer-rev Off Publ Indian Soc Crit Care Med 2013;17(6): 343e9.https://doi.org/10.4103/0972-5229.123438.

[11] Doig GS, Simpson F, Bellomo R, Heighes PT, Sweetman EA, Chesher D, et al. Intravenous amino acid therapy for kidney function in critically ill patients: a randomized controlled trial. Intensive Care Med 2015;41(7):1197e208.

https://doi.org/10.1007/s00134-015-3827-9.

[12] Allingstrup MJ, Kondrup J, Wiis J, Claudius C, Pedersen UG, Hein-Rasmussen R, et al. Early goal-directed nutrition versus standard of care in adult intensive care patients: the single-centre, randomised, outcome assessor-blinded EAT-ICU trial. Intensive Care Med 2017;43(11):1637e47.https://doi.org/10.1007/ s00134-017-4880-3.

[13] Braunschweig CA, Sheean PM, Peterson SJ, Gomez Perez S, Freels S, Lateef O, et al. Intensive nutrition in acute lung injury: a clinical trial (INTACT). J Parenter Enteral Nutr 2015;39(1):13e20. https://doi.org/10.1177/ 0148607114528541.

[14] Casaer MP, Wilmer A, Hermans G, Wouters PJ, Mesotten D, Van den Berghe G. Role of disease and macronutrient dose in the randomized controlled EPaNIC trial: a post hoc analysis. Am J Respir Crit Care Med 2013;187(3):247e55. [15] Puthucheary ZA, Rawal J, McPhail M, Connolly B, Ratnayake G, Chan P, et al.

Acute skeletal muscle wasting in critical illness. JAMA 2013;310(15): 1591e600.

[16] Zhu R, Allingstrup MJ, Perner A, Doig GS. The effect of IV amino acid supple-mentation on mortality in ICU patients may be dependent on kidney function: post hoc subgroup analyses of a multicenter randomized trial. Crit Care Med 2018;46(8):1293e301.https://doi.org/10.1097/ccm.0000000000003221. [17] Moisey LL, Mourtzakis M, Cotton BA, Premji T, Heyland DK, Wade CE, et al.

Skeletal muscle predicts ventilator-free days, ICU-free days, and mortality in

elderly ICU patients. Crit Care 2013;17(5):R206. https://doi.org/10.1186/ cc12901.

[18] Weijs PJM, Looijaard WGPM, Dekker IM, Stapel SN, Girbes ARJ, Oudemans-van Straaten HM, et al. Low skeletal muscle area is a risk factor for mortality in mechanically ventilated critically ill patients. Crit Care 2014;18(1):R12.

https://doi.org/10.1186/cc13189.

[19] Looijaard WGPM, Dekker IM, Stapel SN, Girbes ARJ, Twisk JW, Oudemans-van Straaten HM, et al. Skeletal muscle quality as assessed by CT-derived skeletal muscle density is associated with 6-month mortality in mechanically venti-lated critically ill patients. Crit Care 2016;20(1):386.https://doi.org/10.1186/ s13054-016-1563-3.

[20] MacDonald AJ, Greig CA, Baracos V. The advantages and limitations of cross-sectional body composition analysis. Curr Opin Support Palliat Care 2011;5(4):342e9.https://doi.org/10.1097/SPC.0b013e32834c49eb. [21] Shen W, Punyanitya M, Wang Z, Gallagher D, St-Onge MP, Albu J, et al. Total

body skeletal muscle and adipose tissue volumes: estimation from a single abdominal cross-sectional image. J Appl Physiol (Bethesda, Md: 1985) 2004;97(6):2333e8.https://doi.org/10.1152/japplphysiol.00744.2004. [22] Schweitzer L, Geisler C, Pourhassan M, Braun W, Gluer CC, Bosy-Westphal A,

et al. What is the best reference site for a single MRI slice to assess whole-body skeletal muscle and adipose tissue volumes in healthy adults? Am J Clin Nutr 2015;102(1):58e65.https://doi.org/10.3945/ajcn.115.111203. [23] Mourtzakis M, Prado CM, Lieffers JR, Reiman T, McCargar LJ, Baracos VE.

A practical and precise approach to quantiﬁcation of body composition in cancer patients using computed tomography images acquired during routine care. Appl Physiol Nutr Metabol Physiol Appl Nutr et Metabol 2008;33(5): 997e1006.https://doi.org/10.1139/h08-075.

[24] van der Werf A, Langius JAE, de van der Schueren MAE, Nurmohamed SA, van der Pant K, Blauwhoff-Buskermolen S, et al. Percentiles for skeletal muscle index, area and radiation attenuation based on computed tomography im-aging in a healthy Caucasian population. Eur J Clin Nutr 2018;72(2):288e96.

https://doi.org/10.1038/s41430-017-0034-5.

[25] Weijs PJM, Sauerwein HP, Kondrup J. Protein recommendations in the ICU: g protein/kg body weight - which body weight for underweight and obese patients? Clin Nutr 2012;31(5):774e5.

[26] Roza AM, Shizgal HM. The Harris Benedict equation reevaluated: resting energy requirements and the body cell mass. Am J Clin Nutr 1984;40(1): 168e82.

[27] Strack van Schijndel RJM, Weijs PJM, Sauerwein HP, de Groot SDW, Beishuizen A, Girbes ARJ. An algorithm for balanced protein/energy provision in critically ill mechanically ventilated patients. e-SPEN Eur e-J Clin Nutr Metab 2007;2(4):69e74.

[28] Martin L, Birdsell L, Macdonald N, Reiman T, Clandinin MT, McCargar LJ, et al. Cancer cachexia in the age of obesity: skeletal muscle depletion is a powerful prognostic factor, independent of body mass index. J Clin Oncol: Off J Am Soc Clin Oncol 2013;31(12):1539e47.https://doi.org/10.1200/jco.2012.45.2722. [29] Hamwi GJ. Therapy: changing dietary concepts. Edition ed. In: Danowski T,

editor. Diabetes mellitus: diagnosis and treatment. New York: American Diabetes Association; 1964. p. 73e8.

[30] Bendavid I, Zusman O, Kagan I, Theilla M, Cohen J, Singer P. Early adminis-tration of protein in critically Ill patients: A retrospective cohort study. Nu-trients 2019;11(1).https://doi.org/10.3390/nu11010106.

[31] Singer P, Blaser AR, Berger MM, Alhazzani W, Calder PC, Casaer MP, et al. ESPEN guideline on clinical nutrition in the intensive care unit. Clin Nutr 2018.

https://doi.org/10.1016/j.clnu.2018.08.037.

[32] Goodpaster BH, Kelley DE, Thaete FL, He J, Ross R. Skeletal muscle attenuation determined by computed tomography is associated with skeletal muscle lipid content. J Appl Physiol (Bethesda, Md: 1985) 2000;89(1):104e10.

[33] Manini TM, Clark BC, Nalls MA, Goodpaster BH, Ploutz-Snyder LL, Harris TB. Reduced physical activity increases intermuscular adipose tissue in healthy young adults. Am J Clin Nutr 2007;85(2):377e84.

[34] Delmonico MJ, Harris TB, Visser M, Park SW, Conroy MB, Velasquez-Mieyer P, et al. Longitudinal study of muscle strength, quality, and adipose tissue inﬁltration. Am J Clin Nutr 2009;90(6):1579e85. https://doi.org/10.3945/ ajcn.2009.28047.

[35] Khan IM, Perrard XY, Brunner G, Lui H, Sparks LM, Smith SR, et al. Inter-muscular and periInter-muscular fat expansion in obesity correlates with skeletal muscle T cell and macrophage inﬁltration and insulin resistance. Int J Obes 2015;39(11):1607e18.https://doi.org/10.1038/ijo.2015.104.

[36] Morton RW, Traylor DA, Weijs PJM, Phillips SM. Deﬁning anabolic resistance: implications for delivery of clinical care nutrition. Curr Opin Crit Care 2018;24(2):124e30.https://doi.org/10.1097/mcc.0000000000000488. [37] Shibahashi K, Sugiyama K, Kashiura M, Hamabe Y. Decreasing skeletal muscle

as a risk factor for mortality in elderly patients with sepsis: a retrospective cohort study. J Intensive Care 2017;5:8. https://doi.org/10.1186/s40560-016-0205-9.

[38] Fuchs G, Thevathasan T, Chretien YR, Mario J, Piriyapatsom A, Schmidt U, et al. Lumbar skeletal muscle index derived from routine computed tomography exams predict adverse post-extubation outcomes in critically ill patients. J Crit Care 2018;44:117e23.https://doi.org/10.1016/j.jcrc.2017.10.033.

[39] Trouwborst I, Verreijen A, Memelink R, Massanet P, Boirie Y, Weijs P, et al. Exercise and nutrition strategies to counteract sarcopenic obesity. Nutrients 2018;10(5).https://doi.org/10.3390/nu10050605.

(11)

[40] Popuri K, Cobzas D, Esfandiari N, Baracos V, Jagersand M. Body composition assessment in axial CT images using FEM-based automatic segmentation of skeletal muscle. IEEE Trans Med Imaging 2016;35(2):512e20.https://doi.org/ 10.1109/tmi.2015.2479252.

[41] Lee H, Troschel FM, Tajmir S, Fuchs G, Mario J, Fintelmann FJ, et al. Pixel-level deep segmentation: artiﬁcial intelligence quantiﬁes muscle on computed

tomography for body morphometric analysis. J Digit Imaging 2017;30(4): 487e98.https://doi.org/10.1007/s10278-017-9988-z.

[42] Looijaard WGPM, Molinger J, Weijs PJM. Measuring and monitoring lean body mass in critical illness. Curr Opin Crit Care 2018;24(4):241e7.https://doi.org/ 10.1097/mcc.0000000000000511.