• Open Access

Resting Energy Expenditure per Lean Body Mass Determined by Indirect Calorimetry and Bioelectrical Impedance Analysis in Cats


  • S.A. Center,

    Corresponding author
    1. Pet Care Research and Development, Iams Co, Proctor & Gamble, Lewisburg, OH
    • Department of Clinical Sciences, College of Veterinary Medicine, Cornell University, Ithaca, NY
    Search for more papers by this author
  • K.L. Warner,

    1. Department of Clinical Sciences, College of Veterinary Medicine, Cornell University, Ithaca, NY
    2. Pet Care Research and Development, Iams Co, Proctor & Gamble, Lewisburg, OH
    Search for more papers by this author
  • J.F. Randolph,

    1. Department of Clinical Sciences, College of Veterinary Medicine, Cornell University, Ithaca, NY
    2. Pet Care Research and Development, Iams Co, Proctor & Gamble, Lewisburg, OH
    Search for more papers by this author
  • J.J. Wakshlag,

    1. Department of Clinical Sciences, College of Veterinary Medicine, Cornell University, Ithaca, NY
    2. Pet Care Research and Development, Iams Co, Proctor & Gamble, Lewisburg, OH
    Search for more papers by this author
  • G.D. Sunvold

    1. Department of Clinical Sciences, College of Veterinary Medicine, Cornell University, Ithaca, NY
    2. Pet Care Research and Development, Iams Co, Proctor & Gamble, Lewisburg, OH
    Search for more papers by this author

Corresponding author: Dr Sharon Center, Department of Clinical Sciences, College of Veterinary Medicine, Cornell University, Ithaca, NY 14853; e-mail: sac6@cornell.edu.



Resting energy expenditure (REE) approximates ≥60% of daily energy expenditure (DEE). Accurate REE determination could facilitate sequential comparisons among patients and diseases if normalized against lean body mass (LBM).


(1) Validate open-flow indirect calorimetry (IC) system and multifrequency bioelectrical impedance analysis (MF-BIA) to determine REE and LBM, respectively, in healthy nonsedated cats of varied body conditions; (2) normalize REE against LBM.


Fifty-seven adult neutered domestic short-haired cats with stable BW.


Continuous (45-min) IC-measurements determined least observed metabolism REE. Cage gas flow regulated with mass flow controllers was verified using nitrogen dilution; span gases calibrated gas measurements. Respiratory quotient accuracy was verified using alcohol combustion. IC-REE was compared to DEE, determined using doubly labeled water. MF-BIA LBM was validated against criterion references (deuterium, sodium bromide). Intra- and interassay variation was determined for IC and MF-BIA.


Mean IC-REE (175 ± 38.7 kcal; 1.5–14% intra- and interassay CV%) represented 61 ± 14.3% of DEE. Best MF-BIA measurements were collected in sternal recumbency and with electrodes in neck-tail configuration. MF-BIA LBM was not significantly different from criterion references and generated LBM interassay CV% of 6.6–10.1%. Over- and underconditioned cats had significantly (P ≤ .05) lower and higher IC-REE (kcal/kg) respectively, compared with normal-conditioned cats. However, differences resolved with REE/LBM (approximating 53 ± 10.3 kcal/LBM [kg]).

Conclusions and Clinical Importance

IC and MF-BIA validated herein reasonably estimate REE and LBM in cats. REE/LBM(kg) may permit comparison of energy utilization in sequential studies or among different cats.


body condition scoring


body weight




daily energy expenditure


extracellular body water


MF-BIA-determined ECW


NaBr-determined ECW




fat mass


MF-BIA-determined FM


food quotient


indirect calorimetry


intracellular body water


MF-BIA-determined ICW


lean body mass


LBM percentage of total BW


MF-BIA-determined LBM


D2O-determined LBM


LBMD percentage of total BW


multi-frequency bioelectrical impedance analysis


sodium bromide




resistivity of the ICW


resistivity of the ECW


respiratory quotient


resting energy expenditure


standard error of the estimate


total body water


MF-BIA-determined TBW


D2O-determined TBW

Total daily energy expenditure (DEE) ranges from 20 to 100 kcal/kg body weight (BW) in healthy adult research cats, reflecting resting energy expenditure (REE), physical activity, and postprandial thermogenesis.[1-4] After an overnight fast, REE usually represents 60–75% DEE.[1] Although the effects of overconditioning, gonadectomy, and age on energy requirements have been investigated in research cats, few studies have examined energy expenditure in pet cats.[3, 5-8]

Different techniques for estimating DEE include quantitative food intake and weight balance studies, doubly labeled water metabolism and dispersal, and whole body calorimetry.[5-12] Regrettably, each method has clinical disadvantages. Practically, open-flow indirect calorimetry (IC) is the gold standard for short-term energy assessments and can estimate REE at a fixed time point. Although a validated open-flow IC-system can estimate REE in <1 hour in an acclimatized patient, few investigations of energy utilization in cats using this methodology have been reported.[8]

Various formulas are used to calculate DEE in cats.[3, 5, 13] Yet, controversy surrounds application of each approach and the scaling factors designed to represent metabolically active tissue.[14-17] Differences in lean body mass (LBM), activity, and inherent metabolic adjustments (eg, disease, diet, feeding frequency) influence energy utilization, compromising accuracy of predictive equations. Studies in humans and experimental animals confirm that estimated DEE is most meaningful when normalized to LBM.[17] Still, normalization against LBM can be criticized because fat mass (FM) is not metabolically inert, contributing up to 15–20% equivalent LBM energy utilization.[17, 18] Nevertheless, practical rapid estimation of REE and LBM may have utility in nonsedated cats.

Estimation of LBM in cats has been accomplished using body condition scores (BCS), dual-energy X-ray absorptiometry, stable isotope (deuterium [D2O] or doubly labeled water [D2O with H218O]) dilution, or multifrequency bioelectrical impedance analysis (MF-BIA).[19-22] Only BIA has potential for practical objective clinical application.[21, 22] Considering that LBM importantly influences energy utilization, its accurate estimation might be used to normalize energy data allowing relevant comparisons among patients and disease states. To date, illness energy quotients used to modify DEE estimates are poorly substantiated.[9, 23-26] Merging techniques of open-flow IC-REE and MF-BIA LBM in cats could allow normalization of energy expenditures among cats with various body conditions and disease states.

The present study of cats had 3 aims: (1) develop and validate an open-flow IC-system for REE estimation; (2) validate an MF-BIA instrument in nonsedated cats to estimate total body water (TBW) and extracellular water (ECW) compartments against criterion reference standards (D2O determined TBW [TBWD] and sodium bromide [NaBr] determined ECW [ECWBr]) to estimate LBM; and (3) normalize and compare IC-REE/LBM (kg) in various body conditions.

Materials and Methods

The study was conducted in 3 phases over 4 years (Phase 1: years 1 and 2; Phase 2: year 3, Phase 3: year 4) involving 57 adult domestic short-haired cats. Before study, cats were assigned BCS by 2 experienced investigators (S.A.C. and K.L.W.) who independently scored each cat using a 5-point system (1/5 = thin; 2/5 = underweight; 3/5 = ideal weight; 4/5 = overweight; 5/5 = obese) with results averaged. All cats had REE estimated by IC and LBM determined either by D2O dilution or BIA.


Phase 1

A total of 30 cats housed in an experimental facility (14 spayed females, 16 castrated males; median age, 3 [range, 1.5–9] years; median weight, 5.30 [range, 3.84–8.53] kg) were used to validate an open-flow IC-system. Sixteen cats were normal-conditioned (BCS 3–3.5/5), whereas 14 cats were over-conditioned (BCS 4–5/5). A dry diet complete for maintenance of an adult cat was fed (metabolizable energy of 4.61 kcal/g; energy distribution of 29.2% protein, 36.0% fat, and 34.8% carbohydrate; calculated food quotient [FQ] of 0.87).[27]

Phase 2

An additional 18 cats housed in an experimental facility (13 spayed females, 5 castrated males; median age, 1 [range, 1–10] years; median weight, 3.22 [range, 2.51–6.83] kg) were used to validate an MF-BIA system and also underwent IC-measurements. Two cats were normal-conditioned (BCS 3/5), 14 under-conditioned (BCS ≤2.5/5), and 2 over-conditioned (BCS 4.5/5). A dry diet complete for maintenance of an adult cat was fed (metabolizable energy of 4.06 kcal/g; energy distribution of 29.3% protein, 42.9% fat, and 27.8% carbohydrates; calculated FQ of 0.84).[27]

Phase 3

An additional 9 cats housed in an experimental facility (all castrated males; median age, 4 [range, 4–8] years; median weight, 6.00 [range, 5.10–10.10] kg) had REE estimated by IC and LBM determined by MF-BIA (LBMBIA). These cats (5 with BCS 4–5/5; 4 with BCS 3/5) fed Phase 2 diet were added to diversify studied body conditions.

All cats were clinically healthy, individually housed, and fed dry food and water ad libitum. Daily group interaction and socialization were permitted for several hours. Food was withheld for 12 hours before open-flow IC, MF-BIA, and criterion reference standard studies. Body weight was determined to 0.01 kg using a calibrated scale. Morphometric measurements for MF-BIA validation followed a previously published scheme.[21] Calorimetry was done within 48 hours of MF-BIA. Animal care and study methods were compliant with Institutional Animal Care and Use Standards of Cornell University.

Open-Flow IC-Determined REE

Indirect calorimetry was done using custom-built plexiglass chambers with a perforated sliding door. Plastic-encased styrofoam inserts eliminated dead space, individualizing cage fit for each cat. Each cage was configured allowing space for the cat to sit or recline comfortably. Two cages with different dimensions (31.8 length × 33 height × 19.1 width cm [~20 L] and 62.2 length × 27.9 height × 26.7 width cm [~46 L]), designed and built by the authors, accommodated small- to -large-sized cats. The same assigned cage and custom inserts were consistently used for each cat's calorimetry studies. Vacuum-driven cross-cage gas flow allowed expiratory gas sampling as it exited the enclosure. Cage gas flow was adjusted based on CO2 concentrations in outflow, respiratory gases allowing <1% increase during the acclimation interval; initial gas flow was based on expected basal O2 consumption (BW [kg0.876] × 0.18) using a 10-fold greater flow that was subsequently down-titrated.[16] Mass flow controller1 (2–15 L/min) quantified cage gas flow was verified by nitrogen dilution.[28]

A water absorbent2 was positioned in-line before mass flow meters. Sampled gas was delivered to a CO2 sensor,3 scrubbed free of CO2,4 and then to an O2 analyzer.5 Gas analyzers were zeroed using pure nitrogen and then calibrated with a 99% guaranteed span gas (20.0% O2, 5.0% CO2). Fans kept ambient gases well mixed, and a vacuum system provided cross-cage gas flow evacuated through an environmental safety hood. Room air turnover averaged 10× per hour. After a 20-minute acclimation, expiratory gases were sampled every 5 seconds for 45 minutes and baselined automatically every 10 minutes (data acquisition system6). Oxygen consumption was calculated with gas concentrations corrected to STPD. Calorimetry measurements were done in a softly lit room closed to pedestrian traffic with stable room and cage temperatures (18–20°C and 22–24°C, respectively). REE was calculated using the abbreviated Weir Equation[29]: REE (kcal/day) = ([3.941 × Vo2 (L/min)] + [1.11 × Vco2 (L/min)]) × 1440 (min/day); 3.94 = heat generated from combusting 1 mole of O2; 1.11 = heat generated from producing 1 mole of CO2 based on the rate of O2 consumption (Vo2) and CO2 production (Vco2) from triplicate measurements using the lowest values from each recording interval representing “least-observed-metabolism” (lowest energy expenditure in the unfed resting state).

Cats were acclimated to IC first by introduction to the calorimetry room (free roaming access) and then were confined in IC cages for 10-minute sham recordings over several days. Thereafter, 45-minute recordings were undertaken (three 10-minute data collection sessions segregated by automated baseline ambient gas measurements). Three IC sessions (3 separate days) were recorded before collection of validation data (intra- and interassay repeatability). All cats were in stable BW (±7%) during validation studies, fed a consistent diet (Phase 1) with food restriction 12 hours before IC, with free choice water. Validation IC-REE studies were completed on 3 separate days in all cats (Day 1, Day 2 [60–90 days after Day 1], Day 3 [14 days after Day 2]). Data were recorded during 3 10-minute intervals during 1 REE session; cage flow was verified using nitrogen dilution after IC sessions. The IC-system was validated against alcohol combustion (methanol and ethanol) known to produce respiratory quotient (RQ) of 0.67 for complete combustion. Results compared favorably with mean RQ = 0.68 (accuracy ± 3.1% [n = 16]).

Doubly Labeled Water Determined DEE

For 12 cats in Phase I (6 normal-conditioned [BCS 3/5], 6 over-conditioned [BCS 4–5/5]), DEE was measured within 3 days of IC using a modification of a doubly labeled water method (D2O7 [0.15 g/kg] and oxygen-18 water [H218O; 0.15 g/kg]8 given SC).[6, 10] Cats were not manipulated for other tests during isotope studies, and food was withheld for 12 hours and water for 2 hours before studies, but food and water allowed 8 hours after dosing. Isotope enrichments were determined (in triplicate) by isotope ratio mass spectrometry in blood collected before, and 3 and 6 hours after D2O, and 6, 8, 24, 48, 72, and 96 hours after H218O administration.9,[30-34] DEE was estimated from CO2 production using IC-determined RQ and FQ.[27] TBW volume determined with each isotope was averaged to estimate LBM. Isotope-determined DEE was used to calculate the IC-REE percentage of DEE.

Criterion Reference Standards (TBWD and ECWBr) and MF-BIA to Determine LBM


Eighteen cats (Phase 2) received D2O (0.15 g/kg) SC with blood collected at 0, 6, and 8 hours. Serum D2O was measured using Fourier transform infrared spectroscopy,10,11 in water distillates.[20, 35] D2O was calculated by linear regression against four standards; triplicate analyses were averaged. Intra-assay repeatability (5 standards assayed 5 times) yielded 6.8% CV%; inter-assay repeatability (5 standards assayed in triplicate on 6 days) yielded 9.6% CV%. Specificity of the method has been demonstrated.[33, 36] TBWD was determined using a standard formula assuming the distribution pool represents TBW mass.[33] Body FM was calculated from BW minus LBM, where LBM was estimated by TBW/0.744 (fractional moisture of feline carcass).[37]


Eighteen cats (Phase 2) had ECWBr determined after pilot studies in 4 cats verified bromide equilibrium by 3 and 6 hours after SC dosing (NaBr12 30 mg/kg). Bromide was measured using HPLC in deproteinated serum (250 × 4.6 mm Partisil SAX-10 5 μ [anion exchange] column, 25°C, 30 mM KH2PO4 mobile phase in Milli-Q water, constant flow rate of 0.9 mL/min, UV detection at 200 nm, and chromatography software).13-15,[38] The assay was linear (R2 = 0.99) in feline serum (5–500 μg/mL) and recovered ≥98% of bromide added to pooled feline sera (7.5, 15, and 30 μg/mL NaBr). Interassay and intra-assay CV% were <3.1%. No interfering substances were identified. Limits for detection and quantification were set at signal-to-noise ratios ≥ 3.0 and ≥ 10, respectively.[39] Bromide peak area under the curve was integrated using HPLC software; linear regression calculated bromide concentration based on standards. Assay precision is within 1%.[38] NaBr dispersal served as the criterion reference for ECW, calculated using a distributional correction factor of 10% and 0.95 for the Donnan equilibrium.[38, 40] Determination of intracellular water (ICW) was derived from TBWD-ECWBR = ICW.

MF-BIA Determined LBM

A minimum of 10 simultaneous MF-BIA16 (50 frequencies spanning 5–1000 kHz) measurements were recorded from 2 tetrapolar platinum electrode17 configurations in all Phase 2 cats. In the neck-tail (NT) configuration, electrodes were situated 1-cm distal to the occipital protuberance and 1-cm proximal to the tail head. In the forelimb-hindlimb configuration, electrodes were situated at the lateral condyle of the humerus and lateral aspect of the proximal tibia at the level of the femorotibial joint. Cats were positioned in sternal and left lateral recumbency. Cats and analytic hardware were maintained on nonconductive surfaces, with cats minimally restrained with insulated latex industrial gloves. Pheromone spray18 was used to promote relaxation.

The MF-BIA recorded resistance and reactance with corresponding impedance and phase angle computed from resistance and reactance at 50 frequencies using Xitron utility software.19 The impedance and phase angle spectra data for each electrode configuration and body position were fitted to an enhanced version of the Cole-Cole model of current conduction through heterogeneous biologic tissues built into the sofware.[41-44] ECW and ICW were predicted from equations formulated from Hanai's mixture theory.[41-43] ECW volume was predicted from MF-BIA with each electrode configuration, body path, and position. Summation of estimated ECW and ICW was used to calculate TBW; LBM was inferred from TBW/0.744. Interassay repeatability (n = 14) was determined from data collected over 3 days using 2 cats. MF-BIA measurements in all cats were completed 24 hours before and 72 hours after criterion dilution studies for setting resistivity scalers.

Open-Flow IC-Determined REE Adjusted for LBM

In Phase 3, REE estimated by IC in 57 cats was normalized to gross BW (kg) and LBM (kg) assessed either by deuterium dilution or MF-BIA. BCS were compared with the LBM percentage of total BW (LBM%) where BCS ≤ 2.5 or LBM% ≥ 80% was considered underconditioned, and BCS ≥ 4.0 or LBM ≤ 60% was considered overconditioned.[45]


Data were inspected for Gaussian distribution (box and whisker plots, histograms, and Kolmogorov-Smirnov test). Ages and BWs were non-Gaussian and are provided as median and range. Isotope, MF-BIA, and calorimetry data were Gaussian and are provided as mean ± SD. REE was represented per cat, per gross BW (kg), per allometrically scaled BW (kg0.75), or per LBM (D2O [LBMD (kg)] or MF-BIA [LBMBIA (kg)]). Intra- and interassay REE and RQ repeatability were evaluated by calculation of CV% within a single day and between days, respectively, and by Bland-Altman analyses (bias and 95% limits of agreement [mean difference + 1.96 SD] between repeated measurements), and were examined for significant differences using analysis of variance (ANOVA).[46] Significance of differences between 6- and 8-hour D2O and between 3- and 6-hour NaBr concentrations were determined using a paired t-test. Differences in DEE (calculated using IC-RQ of calculated FQ) and in REE between under-conditioned (LBM% ≥ 80%), normal-conditioned (LBM% = 61–79%), and overconditioned (LBM% ≤ 60%) cats were examined using an analysis of covariance for significant model effects (dependent variable: DEE or REE, model: body condition, LBM [kg], covariates BW [kg, kg0.75]). No model effects influenced DEE probably because of the small number of animals. REE was not significantly affected by body condition or LBM (kg), but was significantly influenced by kg and kg0.75 (P = .005, .003, respectively). Subsequently, linear regression was used to predict REE for each cat using an average between group BW for each individual, with normalized data compared using a standard t-test (as described).[17] The percentage of DEE represented by REE and LBM% was calculated for each cat.

Differences between criterion reference standards and MF-BIA determined water distributional volumes (TBW, ECW, ICW), and body mass estimates of LBM and FM, were examined using paired t-tests. LBM determined by D2O and BIA were evaluated using linear regression and Pearson correlation test. Bland-Altman analyses were used to examine the agreement and bias of MF-BIA determined TBW, ECW, and ICW compared with criterion reference standards; correlation coefficients, standard error of the estimate (SEE), mean difference, and 95% limits of agreement were calculated. Interassay MF-BIA repeatability was examined by comparing MF-BIA determined volumes and LBM on 3 separate days in 2 cats (9 analyses in 1 cat, 5 analyses in 1 cat). An ANOVA for repeated measures examined data for significant differences and the CV% was determined. Regression analysis between the error of the criterion reference volumes and MF-BIA estimates and the mean of these 2 values was used to evaluate whether errors correlated with the magnitude (size) of the measured variable; R2  0.9 was deemed to indicate an association.[46]

All statistical evaluations were analyzed with commercial software packages,20,21 using an alpha = 0.05. Accuracy of BCS for correctly assigning cats to body condition categories relevant to LBM% was evaluated by data inspection.


Open-Flow IC-Determined REE

Mean REE (Phase 1, n = 30) was 175 ± 53.0 kcal, ranging from 120 to 215 kcal with a 30% CV among individuals (least-observed-metabolism).[47, 48] Mean REE/kg BW was 36 ± 10.3 kcal/kg with a 23% CV among individuals.

Mean intra-assay REE CV% was 5.3 ± 2.08, 5.7 ± 3.4, 5.7 ± 2.8% for validation Days 1, 2, and 3. Intra-assay REE CV% in individual cats was ≤14% with exception of 3 cats that had CV% of 15.5, 16.6, and 22.6%, respectively. There were no significant intra-assay differences in REE. The 95% limits of agreement of single day measurements ranged from −8.04 to 11.46 kcal with a bias ranging from −1.27 to 3.59 kcal. Interassay CV% for REE ranged from 5.3 to 12.8% (mean, 7.3 ± 2.08%). Bland-Altman plot of mean REE for each validation day in each cat (Fig 1) demonstrated a 95% limit of agreement ranging from −22.2 to +22.3, with a bias of 0 kcal.

Figure 1.

Bland-Altman plot illustrating the narrowest and widest difference between interassay resting energy expenditure (REE kcal) in 30 cats used to validate indirect calorimetry. Data represent the mean difference between each REE determination made on 3 days (Day 1 versus Day 2, Day 2 versus Day 3, Day 1 versus Day 3) with measurements spanning several months with cats in stable body weight (±7%).

Between weeks, RQ ranged from 0.74 to 0.91 (mean, 0.82 ± 0.04); intra-assay CV ranged from 4.7 to 5.7% for all cats. The 95% limits of agreement of RQ ranged from −0.07 to 0.07 with a bias ranging from −0.01 to 0 between measurements. There were no significant intra- or interassay differences in RQ measurements.

Doubly Labeled Water Determined DEE

A pilot study (4 cats) showed significantly (< .012) different D2O concentrations 6 and 8 hours after dosing (0.023 ± 0.001 and 0.025 ± 0.003, respectively) designating 8 hours for distributional volume. Mean isotope water pool dispersal sizes (D2O [ND] and H218O [NO]) were 173 ± 26.0 mols and 166 ± 25.3 mols, respectively; mean ND/NO was 1.04 ± 0.01. Mean elimination rate constants for D2O (KD) and H218O (KO) were 0.05 ± 0.01 (per day) and 0.08 ± 0.02 (per day), respectively. Mean DEE per cat, DEE/kg BW, and DEE/LBM (kg) based on IC-determined RQ were 325 ± 75.4 kcal, 49 ± 11.7 kcal/kg, and 82 ± 18.5 kcal/LBM (kg), respectively. Mean DEE per cat, DEE/kg of gross BW, and DEE/LBM (kg) based on the calculated FQ (0.87) were 307 ± 73.2 kcal, 47 ± 10.6 kcal/kg, and 78 ± 17.0 kcal/LBM (kg), respectively. There were no significant differences in DEE or REE in these cats when stratified according to their LBM% (with LBM% ≤ 60% classified as overconditioned [n = 7], LBM% ≥ 61% classified as normal conditioned or underconditioned [n = 5]). There also was no significant effect of body condition, LBM (kg) or BW kg0.75 on DEE. Based on IC-determined RQ, mean DEE was 284 ± 52.1 kcal and 74 ± 20.9 kcal/LBM (kg) in overconditioned cats and 332 ± 87.5 kcal and 81 ± 14.2 kcal/LBM (kg) in cats with LBM% ≥ 61%. Based on calculated FQ, mean DEE was 317 ± 61.9 kcal and 52 ± 18.4 kcal/LBM (kg) in overconditioned cats and 298 ± 87.7 kcal and 85 ± 14.2 kcal/LBM (kg) in cats with LBM% ≥ 61%. Mean REE was significantly (< .001) lower than mean DEE, representing 61 ± 14.3 and 60 ± 22.7% of DEE based on IC-RQ and calculated FQ, respectively. REE among these 12 cats had 29.3% CV%.

Criterion Reference Standards (TBWD and ECWBr) and MF-BIA Determined LBM

Mean 0-, 3-, and 6-hour bromide concentrations were 0, 101 ± 13.1, and 101 ± 12.6 mg/L, respectively. BW, BCS, water distributional spaces, and LBM (kg) determined by criterion reference standards and MF-BIA are summarized in Table 1. The LBMD and LBMBIA were not significantly different, and were significantly (< .001) and positively (R2 = 0.92) associated. Although mean difference in LBMD and LBMBIA was 0.02 ± 0.01 kg, differences ranged from 0.01 kg in an underconditioned cat to 0.49 kg in an overconditioned cat. Mean percent difference between LBM% determined by D2O (LBMD%) and MF-BIA (LBMBIA%) was 6.0 ± 3.8%. On inspection, BCS, LBMD%, and LBMBIA% were generally correlated. However, BCS assigned to 4 cats incorrectly classified them as underconditioned compared to LBMD and LBMBIA determined condition.

Table 1. Body condition score (BCS), body weight, total body water (TBW) distributional space, lean body mass (LBM), fat mass (FM), and extracellular water (ECW) and intracellular water (ICW) distributional spaces in 18 clinically healthy cats used to validate a multifrequency bioelectrical impedance analyzer (MF-BIA). Criterion reference fluid compartments determined by deuterium (TBWD) and bromide (ECWBr) were used to calculate ICWD, LBMD, and FMD to compare with MF-BIA estimated TBWBIA, ECWBIA, ICWBIA, LBMBIA, and FMBIA. LBMD%, deuterium-determined LBM (kg) percentage of total body weight; LBMBIA%, MF-BIA determined LBM (kg) percentage of total body weight.
BCS5 point scale2.51.8–4.2
Body Wt (kg)3.222.55–6.83
TBWD (L)1.841.56–3.11
TBWbia (L)1.831.61–3.21
LBMD (kg)2.472.09–4.18
LBMbia (kg)2.452.16–4.31
FMD (kg)0.740.14–2.96
FMBIA (kg)0.630.15–3.44
ECWBr (L)0.650.52–1.18
ECWBIA (L)0.690.53–1.24
ICWd (L)1.250.99–2.19
ICWBIA (L)1.180.94–1.97

Scaling constants for apparent extracellular (ρECW) and intracellular (ρICW) resistivity for MF-BIA arrays (body position, electrode configuration, and body-path lengths) are shown in Table 2. Although average Cole modeling characteristics for MF-BIA measurements (Table 3) were similar to values previously reported in sedated cats using the same model BIA equipment, scaling constants differed.[42] Best MF-BIA estimations and correlation with criterion references were achieved in sternal recumbency, the NT configuration, and nuchal crest to tail head body-path length (Table 4, Fig 2). There were no significant differences between TBW, ECW, ICW, LBM, or FM determined by criterion references and MF-BIA using this array. Subsequently, all MF-BIA measurements were made with this electrode configuration and optimal scaling constants (ρECW 10.59, ρICW 43.22) to set resistivity for ICW and ECW volume predictions. Interassay repeatability (n = 9 one cat, n = 5 another cat) established CV% for TBWBIA of 6.6 and 10.1%, ECWBIA of 2.8 and 6.9%, ICWBIA of 8.7 and 12.1%, LBMBIA of 6.6 and 10.1%, and FMBIA of 6.2 and 16.6%. Mean LBMD% of 80 ± 10.8 and LBMBIA% of 80 ± 13.3 had Pearson correlation coefficient of 0.86 and < .0001 (Fig 3). Regression analysis between the error of the criterion reference volume distributions and MF-BIA volume estimates and the mean of the 2 values indicated that the size of the error was neither significantly nor positively associated with the magnitude of the TBW, ECW, ICW, LBM, and FM (R2 = 0.03, 0.01, 0.07, 0.0, and 0.04, respectively).[46]

Figure 2.

Bland-Altman plot illustrating the narrowest and widest difference between total body water (TBW) estimated by deuterium (D2O) dilution and multifrequency bioelectrical impedance analysis (MF-BIA) plotted against the mean TBW of the two methods in 18 clinically healthy adult cats. MF-BIA was completed with cats in sternal recumbent body position with electrodes arranged in the neck tail (NT) configuration and using the nuchal crest to tail head body-path-length. The solid line represents the mean difference (bias), whereas the dashed lines represent the limits of agreement (mean difference ± 1.96 SD).

Figure 3.

Scatter plot demonstrating the lean body mass percent of total body weight (LBM%) based on deuterium dilution studies (LBMD%) or multifrequency bioelectrical impedance analysis estimations (LBMBIA%) in 18 clinically healthy cats. R2 represents the linear regression. Pearson correlation coefficient (CC) and P value provided.

Table 2. Modeled apparent extracellular and intracellular resistivity constants (ρECW and ρICW, respectively) determined from multifrequency bioelectrical impedance analysis using varied body positions (sternal or left lateral recumbency), electrode configurations (neck-tail [NT] or forelimb-hindlimb [FH]), and body-path-length measurements.
Position/Configuration/Body Length  
Table 3. Descriptive statistics for the Cole modeling characteristics: extracellular water resistance (RE), intracellular water resistance (RI), membrane capacitance (Cm), time delay (Td), alpha, and characteristic frequency (fc) for the four multi-frequency bioelectrical impedance analysis lead configurations in 18 clinically healthy adult cats (Ω, resistance in ohms; nF, nanofarads; ns, nanoseconds; kHz, kilohertz).
Position/ConfigurationRe(Ω)RI(Ω)Cm (nF)Td (ns)Alphafc (kHz)
Sternal/neck-tail215 ± 9.7196 ± 9.66 ± 0.47 ± 2.10.68 ± 0.0167 ± 1.6
Lateral/neck-tail244 ± 9.9225 ± 10.15 ± 0.37 ± 1.30.68 ± 0.0170 ± 2.5
Sternal/forelimb-hindlimb290 ± 15.5228 ± 10.25 ± 0.413 ± 0.90.71 ± 0.0172 ± 2.7
Lateral/forelimb-hindlimb239 ± 11.9213 ± 12.37 ± 0.512 ± 1.40.70 ± 0.0158 ± 1.6
Table 4. Standard error of the estimate (SEE), %SEE, correlation (R2), mean ± SE of the mean difference, 95% limits of agreement (mean difference ± 1.96 SD), bias (L or kg, as appropriate), bias %, and %difference of multifrequency bioelectrical impedance analysis (MF-BIA) determined water distribution and body composition compared with criterion reference standards. MF-BIA was completed with 18 clinically healthy adult cats in sternal posture with electrodes inserted in the neck-tail configuration and body-path-length measured from nuchal crest to tail head.
ParameterSEE%SEER2PMean Difference95% Limits of AgreementBias (L or kg)Bias (%)%Difference
  1. TBW, total body water; ECW, extracellular water; ICW, intracellular water; LBM, lean body mass; FM, fat mass.

TBW0.178.30.85<.0001−0.0008 ± 0.04−0.29–0.300.002−1.0−15.0–14.9
ECW0.056.30.90<.00010.008 ± 0.04−0.15–0.08−0.034−5.4−19.7–8.9
ICW0.1511.40.78<.0001−0.1 ± 0.05−0.27–0.300.0141.023.2–25.2
LBM0.197.00.89<.00010.01 ± 0.05−0.42–0.36−0.032−1.3−16.6–14.0
FM0.1922.40.95<.0001−0.01 ± 0.06−0.33–0.370.0191.7−73–77

Open-Flow IC-Determined REE Normalized with LBM or BW (kg0.75)

Of 57 cats, under-conditioned cats with significantly (P ≤ .0001) greater LBM% than normal-conditioned animals had significantly (< .0001) lower REE/cat compared with normal and overconditioned animals (Table 5). Underconditioned cats had significantly (< .05) higher REE/kg and REE/kg0.75, and over-conditioned cats had significantly (< .0001) lower REE/kg and REE/kg0.75 compared with normal-conditioned animals. Nevertheless, REE normalized by LBM (REE/LBM [kg]) resulted in cats of all body conditions having nearly identical results (Table 5). There were no significant differences in RQ based on group stratification according to BCS or LBM% (Table 5).

Table 5. Daily resting energy expenditure (REE) determined per cat, per total body weight, per lean body mass (LBM) and per kg0.75 in 57 clinically healthy adult cats classified as normal-conditioned, underconditioned, or overconditioned based on LBM percentage of BW (LBM%).
 All CatsNormal-ConditionedUnder-ConditionedOver-Conditioned
LBM% 61–79%LBM% > 80%LBM% ≤ 60%
  1. a

    < .0001; difference from normal-conditioned cats.

  2. b

    P < .05; difference from normal-conditioned cats.

  3. c

    P < .0001; difference between under-conditioned and over-conditioned cats.

Number of cats57261318
Body weight (kg)c5 ± 1.75 ± 1.23 ± 1.0a7 ± 1.4a
LBM%c72 ± 17.773 ± 6.286 ± 4.2a47 ± 11.0a
REE (kcal)/catc175 ± 38.7179 ± 42.6151 ± 14.7a187 ± 38.7
REE (kcal)/kgc36 ± 7.736 ± 0.247 ± 7.9b29 ± 0.2a
REE (kcal)/LBM (kg)53 ± 10.352 ± 4.555 ± 8.855 ± 16.0
REE (kcal)/kg0.75c53 ± 7.553 ± 3.3563 ± 6.7b46 ± 2.1a
RQ0.82 ± 0.040.81 ± 0.040.82 ± 0.040.81 ± 0.02

Comparing MF-BIA LBM% categorization of body condition against BCS demonstrated discordance in four cats where BCS was unreliable in discriminating LBM% ≥ 80% used to define an underconditioned status.


Our findings confirm that MF-BIA can reasonably estimate TBW and LBM in nonsedated cats of various body conditions. Results compare favorably with values previously established for this test system applied to sedated cats.[42] LBMBIA in nonsedated cats had ±10% reproducibility error and ±0.19 kg error, comparing favorably to criterion reference standards and suggesting potential clinical utility. Although MF-BIA is a reasonable method for estimating LBM in nonsedated cats, adoption of this method requires validation against criterion reference standards, as described herein, and in a former pioneering study.[42] Despite attempts to closely align study conditions to a previously described system, we were unable to directly adapt resistivity scalers for our equipment and procedure.[42] Unique differences in equipment circuitry, methods of criterion reference determinations, and environmental conditions (eg, conductivity of table surfaces, building circuitry), and our choice to avoid animal sedation probably contributed to this dilemma. However, in nonsedated cats we demonstrated best MF-BIA performance using a different electrode array then previously reported, probably reflecting the ease in replicating sternal recumbent posture and NT array in nonsedated animals.[21, 22, 42]

We found that MF-BIA estimation of LBM surpassed subjective BCS assignment in identifying under-conditioned cats. Because individuals assigning BCS were amply experienced, poor correlation with LBM% quantified by MF-BIA and D2O may reflect inability of subjective physical assessment to discriminate the 80% LBM cutoff used to assign underconditioned status. However, it also may suggest that BCS of 2.5 is not consistent with an underconditioned animal, supported by another study in which BCS was compared to dual-energy x-ray absorptiometry.[49] These findings suggest caution in using BCS assignment to validate MF-BIA equipment in lieu of criterion reference standards.22,[50]

Using methodologies described herein, and in a prior single report,[8] short-term open-flow IC can reasonably estimate REE in cats. However, measurement repeatability requires exacting control of equipment and variables and careful acclimatization to the procedure. Custom-built cages adapted to each subject thwarted formation of stagnant gas pockets and facilitated rapid flux of gases from the respiratory chamber in the present study. Care must be taken to acclimatize cats to IC to ensure that REE values reflect “least-observed-metabolism.” This study constitutes the largest data set of cats (n = 57) undergoing short-term IC and documents REE ranging between 120 and 215 kcal, 26 and 46 kcal/kg, and 38 and 68 kcal/LBM (kg). The high degree of REE repeatability allowed confident detection of ±15% change among serial REE assessments with 95% limit of agreement reconciling with ± 22 kcal/cat. Although short-term IC-REE and RQ reported herein were not compared with longer interval (ie, 12–24 hours) measurements, they do compare well with previously reported REE estimates (measured or estimated from DEE [ie, ≤ 60% DEE]) determined by a variety of methods.[8, 11, 12]

Diet-associated thermogenesis influences REE, with protein and fat digestion, and metabolism requiring energy and raising postabsorptive interval REE. Because all cats had food withheld for 12 hours before IC-REE, it is unlikely that meal-associated thermogenesis influenced findings. Although the 2 diets fed in this study had comparable protein, but different fat and carbohydrate contents, disparity in REE or RQ between groups was not realized. Measured REE herein also might reflect thermoregulatory metabolic activity considering that we maintained the ambient calorimetry temperature below the thermoneutral zone of cats (35–38°C) in an effort to avoid cage overheating during data collection.[3] Calculation of REE using a modified Weir formula estimates protein contribution (nitrogen) to energy utilization. Yet, REE estimation from respiratory gases (in humans) reportedly imparts an approximate 1.0% error for every 12.3% increment in protein contribution to total oxidative metabolism.[51] How much this error influences IC-estimated REE in cats remains unclear.

Isotope-estimated DEE (234–398 kcal/cat, 37–61 kcal/kg BW, 63–101 kcal/LBM [kg]) compared favorably with other reported estimates.[6, 7] Isotope distributional water space and elimination rate constants were consistent with previously published parameters in domestic cats.[6] We did not identify differences in DEE among cats with different body conditions, possibly because of the small number of cats investigated. Regardless of body condition, IC-estimated REE approximated 60% of DEE.

Expression of REE or REE/kg in overconditioned and underconditioned cats can misrepresent REE relative to normal-conditioned animals. Transforming BW with allometric scaling (BW0.75) did not alleviate this conundrum. Evaluating REE with cats stratified by body condition (LBM%) confirmed higher REE in over-conditioned and lower REE in underconditioned cats compared with normal-conditioned animals. Conversely, REE/kg BW was higher in underconditioned and lower in overconditioned cats compared with normal-conditioned animals. However, when REE was normalized for LBM (kg), cats with diverse body conditions had similar energy utilization (~52–55 kcal/LBM [kg]). This observation has important clinical and investigational relevance, and emphasizes the value of accurate LBM assessments when energy measurements are evaluated or sequentially compared.

Although FQ for the 2 diets fed were 0.87 and 0.84, a lower mean RQ was estimated by IC reconciling with the fasting status of studied cats (in neutral weight balance). Measured RQ reflects energy utilization during the postabsorptive interval as well as protein catabolism, consistent with the high rate of amino acid turnover associated with the pure carnivore status of cats.[52] The mean RQ determined herein (0.82 ± 0.04) is discordant with the mean RQ (0.86 ± 0.03) reported for a small group of moderately overconditioned experimental cats in stable BW.[12] However, in that study, IC was done using large cage enclosures (~10-fold larger volume than in the present study) with data collected over 12 hours with food and water available ad libitum. Thus, estimated RQ probably reflected ingested fuel sources.[12] Predictably, when those cats were energy restricted, mean RQ (0.81 ± 0.03) was similar to values reported herein.

In conclusion, this study validated an open-flow IC-system that is adaptable for short-term estimation of REE in domestic cats and an MF-BIA procedure that can be used for LBM estimation clinically in cats. These analytical tools may enable useful evaluations in cats with diverse health concerns. Indeed, serial MF-BIA hydration monitoring has been performed in animals undergoing hemodialysis (L.D. Cowgill, personal communication). The ability to rapidly determine quantitative LBM may critically modify dosing strategies for drugs and fluid treatment in over- and underconditioned cats, and might potentially assist in providing nutritional recommendations.


This work was supported by the Cornell Feline Health Center Research Grant Program and Iams Pet Food Company, subsidiary of Proctor & Gamble.

An abstract of this study was presented at the 2010 ACVIM Forum, Anaheim, CA.


  1. 1

    Sierra Series 840 mass flow controllers, Sierra Instruments Inc, Monterey, CA

  2. 2

    Drierite, anhydrous CaSO4, WA Hammond Drierite Co, Xenia, OH

  3. 3

    Sable Systems CA-1 Carbon Dioxide Analyzer, Sable Systems, Las Vegas, NV

  4. 4

    Ascarite II, Thomas Scientific, Swedesboro, NJ

  5. 5

    Sable Systems FC-1 oxygen analyzer, Sable Systems

  6. 6

    Datacam V, Analytical Software, Sable Systems

  7. 7

    Deuterium oxide, 99.98%, Cambridge Isotope Laboratories, Andover, MA

  8. 8

    18O Water, 10.05 APE, Europa Scientific, Europa House, Cheshire, UK

  9. 9

    Cornell University Isotope Laboratory, Thermo Finnegan Delta Plus Isotope Ratio Mass Spectrometer, Brennan, Germany

  10. 10

    Nexus 670 FT-IR ESP, Thermo Electron Corporation, Waltham, MA

  11. 11

    Omnic 5.2a software for FT-IR, Thermo Electron Corporation

  12. 12

    Sodium bromide ACS reagent assay >99%, Sigma-Aldrich, St. Louis, MO

  13. 13

    Nanosep Centrifugal Devices, 30K, PAL Life Sciences, Ann Arbor, MI

  14. 14

    SCL-10Avp System Controller including SIL-10ADvp Autoinjector Sample, LC-10ADvp Liquid Chromatograph Pumps, CTO-10ASvp Column Oven, and SPD-10Avp UV-Vis Detector, Shimadzu Class-VP Series HPLC system, Braintree, MA

  15. 15

    Shimadzu Class-VP software (Version 5.032)

  16. 16

    Hydra ECF/ICF Model 4200, Bioimpedance Spectrum Analyzer, Xitron Technologies Inc, San Diego, CA

  17. 17

    Grass platinum subdermal 27-gauge needle electrodes, Astro-Med, West Warwick, RI

  18. 18

    Feliway Pheromone Spray, Veterinary Product Laboratories, Phoenix, AZ

  19. 19

    Xitron utility software, Xitron Technologies Inc

  20. 20

    Statistix Analytical Software, 8.0, Tallahassee, FL

  21. 21

    Analyse-It-1997-2008, version 2.21 Excel 12+ , Analyse-It Software Ltd, Leeds, UK

  22. 22

    Wang W-L, Fan C-L, Yu Y-J, et al Comparison study of bioelectrical impedance analysis in body fat estimation in domestic cats. 32nd Annual WSAVA Congress, August 2007 (abstract)