We planned to measure REE by indirect calorimetry if a right heart catheterization was scheduled in a hemodynamically stable condition in 293 patients who were hospitalized for close investigation or treatment for advanced HF, which was considered by the attending physician according to the HF guidelines (14), at Osaka University Hospital from June 2015 to November 2020. As a result, REE data were successfully obtained in 72 patients. Serum and urinary laboratory samples were collected from 17 of the 72 patients, excluding patients with proteinuria. To analyze the relationship between REE and BMI, we classified the patients into three groups according to BMI (kg/m²) as follows: <18.5 in the underweight group, ≥18.5 and <25 in the normal weight group, and ≥25 in the overweight group (15).

This study was approved by the Institutional Review Board (IRB) of Osaka University (IRB number: 16209–2). Informed consent was obtained in the form of an opt-out option on the website.

The oxygen consumption (VO₂) and carbon dioxide production (VCO₂) were measured using a face-covering mask connected to an indirect calorimetry (AE-100i; Minato Medical Science Co., Osaka, Japan). During the measurement, the patients were in a lying position at rest for at least 20 min after a 4-h fast in a hemodynamically stable phase, defined as no requirement for up-titration of intravenous inotropes or mechanical circulatory supports. REE was calculated from the average of VO₂ and VCO₂ during the last 5 min of indirect calorimetry measurements, using the abbreviated Weir equation (16). Other indices related to energy expenditure, including the TEE, were calculated according to previous reports (17–20) (Table 1). REE per body weight (REEBW) was evaluated because it is commonly used to consider the required energy in routine clinical practice. In this study, the activity index was set at 1.3 because the hemodynamically stable inpatients with HF were physically active away from the bed (20). We modified the European Society for Clinical Nutrition and Metabolism (ESPEN) formula by adjusting the coefficients with obtained REEBW data and proposed it as a new predictive formula for TEE, indicating the energy required per day (Table 1).

We performed right heart catheterization based on clinical indication in a hemodynamically stable condition to evaluate cardiac function implying pathophysiological information. Hemodynamic parameters, right atrial pressure (RAP), which mainly reflects right heart function (21), pulmonary arterial wedge pressure (PAWP), which mainly reflects left heart function, and venous oxygen saturation (SvO2) and cardiac index by Fick’s formula (22), which reflect both heart functions, were measured.

Blood samples were collected before breakfast after fasting for at least 8 h, and centrifuged to obtain serum or plasma separately. Urine samples were collected from urine stored for 24 h overnight before blood sample collection. Plasma and urinary amino acid (AA) analyses were performed using liquid chromatography/mass spectrometry and high-performance liquid chromatography (23).

The amounts of macronutrients were determined for each type of hospital meal. Energy and protein intakes were calculated from the records of the type and intake of each meal recorded by the nurses in the electronic medical record. The average of the consecutive 3 days prior to the day of blood sample collection was used for the analysis (24).

Statistical analyses were performed using International Business Machines Statistical Package for the Social Sciences Statistics (version 26). The model assumption of normality was assessed and confirmed using the Shapiro–Wilk test and Q–Q plots of residuals. Continuous variables are expressed as mean ± standard deviation or median (interquartile range), as appropriate. The paired or unpaired t test was used for comparisons between the two groups. The 2-way ANOVA without replication was used to investigate the effects of BMI category and sex on REEBW and the differences between TEE and pTEE by each equation.

Correlations between two variables were analyzed by Pearson’s or Spearman’s rank correlation coefficient test, as appropriate, and each correlation coefficient is expressed as r. Multiple regression analysis was performed using the forced-entry model to estimate associations between REE and clinical parameters related to HF severity, with adjustment for body weight.

Patient characteristics are summarized in Table 2. This study primarily included patients with advanced HF, as evident in 43 (60%) patients on the transplant waiting list and 61 (85%) patients with stage D HF. REE and REEBW were significantly higher than the predicted values of REE (pREE) and REEBW (pREEBW) by the Harris-Benedict (HB) equation (p < 0.05 and p < 0.05, respectively, Figures 1A,B). TEE was also significantly higher than the predicted values of TEE (pTEE) based on the ESPEN formula (p < 0.01, Figure 1C) or the Mifflin-St Jeor equation (p < 0.01, Figure 1D). In the analysis of each sex, the difference between REE/REEBW and pREE/REEBW by HB equation was significant in men (p < 0.05, p < 0.01, respectively), but not in women (p = 0.457, p = 0.488, respectively). In contrast, the difference between TEE and pTEE using the ESPEN or Mifflin-St Jeor equations remained significant in both sexes (all p < 0.05). The average individual differences between REE and pREE by the HB equation was 49.0 ± 169.4 kcal/day (male: 59.0 ± 177.3 kcal/day, female: 24.8 ± 149.6 kcal/day). Such individual differences were also observed between REEBW and pREEBW by the HB equation (1.0 ± 0.4 kcal/kg/day) or between TEE and pTEE by the ESPEN formula (373.9 ± 243.5 kcal/day) or Mifflin-St Jeor equations (106.7 ± 216.3 kcal/day).

The REEBW was not associated with clinical parameters related to hemodynamics and HF severity or inotropes use (mean blood pressure, p = 0.818; heart rate, p = 0.376; pulmonary artery wedge pressure, p = 0.400; right atrial pressure, p = 0.843; cardiac index, p = 0.599; mixed SvO₂, p = 0.752; log brain natriuretic peptide [BNP], p = 0.352; inotropes use, p = 0.758). These associations were confirmed by analyses using multiple regression models, where REE was not associated with the above parameters after adjusting for BW. REEBW was not associated with age and sex (Figures 2A,B) but was significantly associated with BMI (r = −0.691, p < 0.001, Figure 2C). The significant association of BMI with REEBW remained after adjusting for sex. Comparisons of REEBW among each group classified by BMI showed a higher REEBW in the underweight group than in the normal- and overweight groups and a higher REEBW in the normal-weight group than in the overweight group (Figure 3A). The mean REEBW was approximately 28 kcal/kg/day, 24 kcal/kg/day, and 21 kcal/kg/day for underweight, normal-, and overweight groups, respectively, and the coefficients for pTEE (proposed formula) in each BMI group were 36 kcal/kg/day for underweight patients, 32 kcal/kg/day for normal weight patients, and 27 kcal/kg/day for overweight patients. The difference between TEE and pTEE by the ESPEN formula was large and was the greatest in the underweight group (502.4 ± 206.9 kcal/day) and greater in the normal-weight group (370.6 ± 192.5 kcal/day) than in the overweight group (148.8 ± 296.8 kcal/day) (Figure 3B). In contrast, there was no significant difference between TEE and pTEE by the HB equation, the Mifflin-St Jeor equation, and the proposed formula among each BMI group (Figures 3C–E). These associations remained after adjusting for sex.

Table 3 shows the relationships between the candidate biomarkers and the ratio of daily energy intake to REE, protein intake, representative variables of HF features, and renal function. The serum albumin and plasma AA levels, including fractional concentrations, such as essential AA, branched-chain AA, and each essential AA concentration, were maintained in the normal range and had no significant correlation with the ratio of daily energy intake to REE (Figures 4A–F). In contrast, the urinary essential AA excretion was mostly below the normal range. Notably, urinary excretions of histidine and 3-methylhistidine were significantly correlated with the ratio of daily energy intake to REE (Figures 4G,H). Urinary threonine and lysine excretion tended to be positively correlated with the ratio of daily energy intake to REE (Figures 4I,J). These significant correlations remained after adjusting for sex.

We performed REE measurements 4 h after meals because of practical reasons. For REE measurements, a minimum fast of 5 h after meals and 4 h after a small meal is recommended if further fasting is clinically inappropriate. The peak of the increase in metabolic rate associated with the diet occurs between 60 and 180 min. Therefore, there may be a diet-induced metabolic rate increase in measured REE. However, measurement 4 h after small meals is considered acceptable (42). REE is generally reported to be negatively correlated with age, decreasing in individuals aged over 60 years (26). Our observation that age was not associated with REE may be attributed to the relatively young age of our study population, with an average age of 50 years, because heart transplantation candidates accounted for 60% of the patients (Table 2). Furthermore, in our advanced HF cohort, there were few patients (5 of 72) with HF and preserved ejection fraction, which is common in older patients with HF. The single-center design and small study sample limited the statistical power, resulting in possible type I and II errors. The mechanism of AA profile change and its effectiveness for nutritional management need further investigation.