Prolonged starvation has been reported to result in reduction of both energy expenditure and protein turnover rates in healthy and obese subjects.1-4 With acute starvation, this hypometabolism has been shown to be in excess of that anticipated by a reduction of lean body mass and may be considered as an adaptive mechanism allowing for survival during periods of food shortage.5 This metabolic adaptation, or increased metabolic efficiency,6 would have the effect of maintaining fat mass, and reducing the need for essential amino acids.
Although it is generally assumed that in chronic undernutrition the physiologic and metabolic responses occur in much the same way as in the studies of experimental or therapeutic semistarvation,4 several investigators have failed to demonstrate the phenomenon of metabolic adaptation in chronically undernourished patients or in those with coexistent disease.5,7-11
The aim of this study was to determine the functional consequences of undernutrition and subsequent refeeding on respiratory quotient (RQ), resting energy expenditure (REE), and whole-body protein kinetics in human subjects.
MATERIALS AND METHODS
Subjects
Twenty-five severely undernourished hospitalized patients, with a body mass index (BMI) <17 kg/m^sup 2^, were studied during routine clinical management. Patients were recruited simply according to severe undernutrition and the need for nutrition support. Although a number of different parameters have been used in the assessment of nutrition status, the BMI has been shown to be a reliable and easy-to-determine index.12 The patients were stratified into a disease patient group and an anorexia nervosa patient group.
Disease patient group. This group consisted of 17 patients with undernutrition associated with a number of disease states. These included patients with Crohn's disease (n = 8), tuberculosis (n = 3), immunoproliferative small intestinal disease (n = 2), carcinoma of the lung (n = 1), metastatic melanoma (n = 1), disseminated amyloid (n = 1), and short bowel after previous trauma (n = 1).
Anorexia patient group. This group of 8 patients with anorexia nervosa was studied to assess the effects of undernutrition consequent solely to an inadequate intake.
None of the subjects recruited were overtly septic at time of study, and none had evidence of significant cardiac, renal, or hepatic disease. After admission to hospital, studies were commenced once dehydration and electrolyte imbalances, if present, had been corrected.
The undernourished patients were evaluated in comparison to 17 normal healthy volunteers.
Nutrition Support and Repeat Studies
After the initial studies, the undernourished patients received intensive nutrition support as inpatients, aimed at providing 30 kcal/kg ideal body weight/d energy and 1.5 g/kg ideal body weight/d protein. The diet initially consisted of a semielemental, hydrolyzed protein formulation (Vital or Alitraq, Ross Laboratories, Columbus, OH) requiring a minimum of digestion before absorption, administered via a finebore nasogastric tube. After 1-2 weeks as tolerated, a polymeric feeding formula (Ensure, Ross Laboratories), also administered via a fine-bore nasogastric tube, was introduced, and subsequently a normal ward diet, in addition to the tube-feeding, permitted. Four patients initially required parenteral nutrition (PN) (30 kcal/kg/d energy and protein/amino acids 1.5 g/kg/d). One patient was profoundly malnourished due to anorexia nervosa, 1 patient with Crohn's disease and 1 with immunoproliferative small intestinal disease experienced exacerbation of their diarrhea on initiation of enteral feeding, and the fourth had a distal small bowel fistula due to enteric tuberculosis. Enterai feeding was subsequently successfully reintroduced after 2 weeks in the anorexic patient, after 2 weeks in the patient with Crohn's disease, after 3 weeks in the patient with immunoproliferative bowel disease, and after 2 months in the patient with tuberculosis, following closure of the fistula.
Nineteen patients (disease patients, n = 13; anorexia patients, n = 6) consented to further study of digestive function and protein synthesis after a period of nutrition support. Studies were performed once the patients were receiving a normal ward diet and demonstrating sustained weight gain.
Ethical Considerations
Approval for the project was granted by the Research and Ethics Committee, University of Cape Town. Informed, written consent was obtained from all participants before study. The study was carried out in accordance with the declaration of Helsinki.
METHODS
RQ and REE
Total carbon dioxide production and oxygen consumption were measured in the rested, fasted state by means of indirect calorimetry, using a metabolic monitor (MedGraphics CPX/D, St Paul, MN). This allowed determination of the RQ and assessment of the REE.
Isotope Incorporation Studies
In order to assess protein kinetics, a 4-hour primed, continuous infusion of L-1-^sup 14^C leucine (0.30 μCi/kg + 0.30 μCi/kg/h) was administered. Venous blood samples were taken from the contralateral arm at 0, 1, 2.5, and 4 hours for subsequent measure of serum leucine specific activity (SA), and expired breath samples were collected at 1-hour intervals for determination of amino acid oxidation rates. To determine plasma ^sup 14^C leucine SA, 400 μmol of homocysteic acid was added to 2-mL samples of plasma, and the plasma-free amino acids were separated after precipitation of the proteins with 50 mg 5-sulfosalicylic acid by centrifugation at 2500 rpm for 10 minutes.7 Plasma leucine concentrations were measured by high-performance liquid chromatography (HPLC),13 (Chromjet Integrator, SP8800 Ternary HPLC pump, SpectroSeries UV150 detector; Spectra-Physics, Mountain View, CA). Radioactivity due to ^sup 14^C was measured by dual-window liquid scintigraphy counting (Tricarb 1500; Packard Instrument, Downers Grove, IL). The SA of ^sup 14^C leucine in plasma was then calculated. Carbon dioxide excretion in the breath (mmol/h) was measured by indirect calorimetry, and 1-mmol samples of expired breath carbon dioxide were trapped in hyamine and counted to calculate the SA of ^sup 14^C carbon dioxide (dmp/mmol). Total isotope excretion in the breath (dpm/h) was then calculated.
Calculations
A stochastic model for whole-body amino acid metabolism was used to calculate rates.14 The leucine flux (Q) in plasma was calculated from the relationship between the rate of isotope infusion and the plateau SA of ^sup 14^C leucine in the blood (SA during infusion - basal SA).
Leucine excretion (oxidation) rate (E) was measured from the relationship between the rate of excretion of isotope in the breath carbon dioxide and the SA of ^sup 14^C leucine in the blood.
Intake (I) = 0, and therefore flux (Q) = breakdown (B).
The leucine flux, oxidation, and synthesis rates are converted to "whole-body protein" rates by assuming that human body mixed protein consists of 8% leucine (by weight).18-20
Statistics
Results are given as the means ± SEM. Statistical analysis was by the Statistica statistical package using analysis of variance (ANOVA with post hoc planned comparison), Student's t test, and paired t test for parametric data, and Mann-Whitney U test and Wilcoxon matched paired test for nonparametric data, whichever was appropriate. A p value of < .05 was considered significant.
RESULTS
On entry to the study, the mean BMI of the anorexia patients was 12.46 ± 0.53 kg/m^sup 2^, and the disease patients 13.81 ± 0.40 kg/m^sup 2^ (p = not significant), compared with 23.71 ± 0.72 kg/m^sup 2^ in the control subjects (p < .001). After the period of refeeding (anorexia patients 6.83 ± 0.75 weeks, disease patients 6.08 ± 1.03 weeks; p = not significant), the mean BMI of the anorexia patients increased to 15.26 ± 0.62 kg/m^sup 2^; p < .001, and the disease patients increased to 16.32 ± 0.49 kg/m^sup 2^; p < .001.
The results of the studies of RQ, REE, and wholebody protein kinetics in the control subjects, and anorexia patients and disease patients before and after nutrition support are illustrated in Table I. There were no significant differences in RQ, REE or protein kinetics between those patients who subsequently underwent studies after nutrition support and those who declined further study, or those who received PN and those who received only enterai nutrition.
RQ
The mean RQ of the anorexia patient group was similar to the control values, (0.85 ± 0.05 vs 0.90 ± 0.05), whereas that of the disease patient group was lower (0.76 ± 0.03 vs 0.90 ± 0.05; p < .05). After feeding the RQ's of the disease patients increased significantly (0.84 ± 0.03 vs 0.76 ± 0.03; p < .05), with the mean RQs of both the anorexia patients (1.00 ± 0.07) and disease patients (0.84 ± 0.03) similar to the controls (0.90 ± 0.05).
REE
Mean REE was significantly reduced in both the anorexic patient group and the disease patient group compared with controls (anorexic patient group 1058 ± 134.0 kcal/d us 1828 ± 89.76 kcal/d; p < .001, disease patient group 1189 ± 101.4 kcal/d vs 1828 ± 89.76 kcal/d; p < .001). There was no significant difference between the anorexic patients group and the disease patient group. There was a significant correlation between REE and body weight (r = .78; p < .001; Fig. 1).
Although mean REE of the undernourished patients, when expressed in absolute terms (kcal/d), was lower than the control value, when expressed per kg body weight, it was significantly higher (anorexic patient group 32.17 ± 4.25 kcal/kg/d vs 25.07 ± 1.00 kcal/kg/d; p < .05, disease patient group 31.30 ± 2.14 kcal/kg/d vs 25.07 ± 1.00 kcal/kg/d; p < .01).
After refeeding, there was no change in the REE of the anorexic patients and the disease patients, which remained significantly lower than the control value (anorexia patients 1133 ± 94.83 kcal/d vs 1828 ± 89.76 kcal/d; p < .001, disease patients 1308 ± 102.2 kcal/d us 1828 ± 89.76 kcal/d; p < .001). Expressed per kg body weight, there was now, however, no significant difference between the undernourished patients and the controls (anorexia patients 27.65 ± 3.05 kcal/kg/d us 25.07 ± 1.00 kcal/kg/d, disease patients 28.90 ± 1.85 kcal/kg/d us 25.07 ± 1.00 kcal/kg/d).
Whole-Body Protein Kinetics
Mean whole-body protein flux/breakdown, oxidation and synthesis were all significantly lower in the undernourished patients before nutrition support, compared with the control group (anorexic patient group: flux/breakdown 160.0 ± 12.24 g/d us 349.1 ± 22.11 g/d; p < .001, oxidation 19.11 ± 4.28 g/d vs 44.09 ± 4.18 g/d; p < .01, synthesis 140.9 ± 10.54 g/d us 305.0 ± 21.64 g/d; p < .001: disease patient group: flux/breakdown 139.5 ± 10.31 g/d vs 349.1 ± 22.11 g/d; p < .001, oxidation 19.84 ± 3.24 g/d vs 44.09 ± 4.18 g/d; p < .001, synthesis 119.8 ± 8.57 g/d us 305.0 ± 21.64 g/d; p < .001.) Body weight correlated significantly with wholebody protein flux/breakdown rates (r = .74; p < .001), protein oxidation rates (r = .58; p = .001), and protein synthesis rates (r = .72; p < .001; Fig. 2).
When expressed per kg body weight, mean wholebody protein flux/breakdown and synthesis of the anorexic patient group was similar to controls (flux 4.94 ± 0.60 g/kg/d us 4.88 ± 0.33 g/kg/d, synthesis 4.36 ± 0.55 g/kg/d us 4.27 ± 0.32 g/kg/d), whereas the disease patient group had significantly lower values than either controls (flux/breakdown 3.61 ± 0.27 g/kg/d us 4.88 ± 0.33 g/kg/d; p = .01, synthesis 3.11 ± 0.24 g/kg/d us 4.27 ± 0.32 g/kg/d; p < .05) or the anorexia patient group (flux/breakdown 3.61 ± 0.27 g/kg/d us 4.94 ± 0.60 g/kg/d; p < .05, synthesis 3.11 ± 0.24 g/kg/d us 4.36 ± 0.55; p = .02; Fig. 3). Mean whole-body protein oxidation rates, expressed per kg body weight were similar in all groups (controls 0.61 ± 0.06 g/kg/d, anorexia patient group 0.58 ± 0.13 g/kg/d, disease patient group 0.53 ± 0.10 g/kg/d).
After the period of nutrition support, the disease patients demonstrated a significant increase in wholebody protein flux/breakdown (200.2 ± 19.88 g/d us 135.9 ± 12.18 g/d; p = .01) and synthesis (173.6 ± 16.38 g/d us 116.5 ± 10.15 g/d; p < .01), whereas the values of the anorexia patients remained similar to the prefeeding levels (flux/breakdown 192.3 ± 28.17 g/d us 150.1 ± 12.92 g/d, synthesis 158.2 ± 26.20 us 134.0 ± 12.75 g/d). The rates remained significantly lower than controls (flux/breakdown 349.1 ± 22.11 g/d; p < .001, synthesis 305.0 ± 21.64 g/d; p < .001). Expressed per kg body weight, mean total body flux/breakdown, oxidation, and synthesis in the anorexia patients and disease patients after refeeding were similar to control levels (anorexia patients: flux/breakdown 4.56 ± 0.57 g/kg/d vs 4.88 ± 0.33 g/kg/d, oxidation 0.80 ± 0.08 g/kg/d vs 0.61 ± 0.06 g/kg/d, synthesis 3.75 ± 0.56 g/kg/d vs 4.27 ± 0.32 g/kg/d, disease patients: flux/breakdown 4.34 ± 0.38 g/kg/d vs 4.88 ± 0.33 g/kg/d, oxidation 0.57 ± 0.09 g/kg/d vs 0.61 ± 0.06, synthesis 3.77 ± 0.31 g/kg/d vs 4.27 ± 0.32 g/kg/d).
DISCUSSION
Although the fasting RQ of the anorexic patients was similar to the controls, the disease patient group had a lower value. This suggests reduced carbohydrate and increased lipid metabolism in malnourished patients with coexistent disease. This is likely to be the consequence of depleted glycogen stores, which appears to be more pronounced in the patients with disease. Lauvin et al have previously reported that anorexia nervosa patients tended to catabolize carbohydrate, possibly reflecting the nature of their diets, whereas patients with both malignant and nontumoral disease catabolized lipid.
After refeeding, the patients demonstrated a significant increase in RQ. This indicates replenishment of carbohydrate (glycogen) stores, and a reduction in fat metabolism in these patients.
The mean REE, considered in absolute terms (kcal/d), was significantly lower in the malnourished patients compared with controls; however, when assessed in relation to body mass (kcal/kg/d), it was significantly higher. Mean REE of the anorexia patient and disease patient groups was similar both before and after nutrition support.
These findings contrast with previous studies of semistarvation in normal individuals. The classic studies of Keys et al1 demonstrated a 14% reduction of REE per unit fat-free mass (FFM). Shetty22 demonstrated a similar reduction of REE related to body mass and to body surface area in undernourished but otherwise healthy Indian laborers, and studies of therapeutic semistarvation in obese patients have also produced similar results.23 Results of these studies suggest metabolic adaptation, or increased metabolic efficiency, associated with the undernourished state.6
Other studies have, however, failed to demonstrate evidence of metabolic adaptation in chronically undernourished subjects or in patients with coexistent disease. Ashworth,7 although reporting a 12% reduction of REE in malnourished Jamaican subjects, was unable to demonstrate metabolic adaptation. Likewise, Lindmark et al24 did not observe any reduction of REE per unit body cell mass. Other reports by McNeill et al,10 Carbonnel et al,5 and a recent study by FerroLuzzi et al11 have also failed to demonstrate metabolic adaptation in stable undernourished patients. Similar to our results, a study by Soares et al demonstrated a tendency to increased REE, expressed per unit active body mass, in undernourished Indian men.9
Our finding of an increased REE, expressed per unit body weight, does not necessarily imply reduced metabolic efficiency in the malnourished patients. Shetty4 has pointed out that metabolic rate, expressed per unit body weight, may not reflect true variations in metabolic efficiency and is likely to be due largely to variations in body composition. Skeletal muscle, although comprising 40%-50% of body weight, contributes only 18%-22% to the basal metabolic rate.25'26 Fat likewise contributes little to metabolic rate. On the other hand, the brain and liver, although comprising only 3%-5% of total body weight, use 40% of energy.27 Chronic undernutrition is associated with a disproportionate loss of relatively metabolically inactive fat and muscle tissue, with sparing of the more metabolically active visceral tissue.4 Our studies related REE to body weight, and as the undernourished patients would be depleted, particularly of fat stores, this would result in an apparent increase in metabolic rate expressed per unit body mass.
There appears, therefore, to be a variable metabolic adaptive response to undernutrition. It has been suggested that the reduction in basal metabolic rate occurs in 2 phases.4 After acute energy restriction, the decrease in metabolic rate is greater than can be attributed to loss of active tissue, providing evidence for increased metabolic efficiency, or metabolic adaptation. After 2-3 weeks, the degree of metabolic adaptation remains constant, and further decrease in metabolic rate is a consequence of loss of active tissue.
Studies of metabolism during recovery from undernutrition have produced conflicting results. High-energy feeding of semistarved, otherwise healthy volunteers has been reported as resulting in either a small (± 5%), or no increase in REE per unit surface area, relative to baseline.1'28 In patients with anorexia nervosa, after 4 weeks of refeeding, Melchior et al29 did not observe any change in REE per unit FFM. On the other-hand, Obarzanek et al,30 also studying anorexia nervosa patients, reported no significant change in REE per unit FFM after 2 weeks of feeding, whereas after 10 weeks there was a dramatic 37% increase. Piers et al31 documented a 7% increase of REE per unit FFM in malnourished Indian patients who received energy supplementation for 3 weeks, which increased to 20% after 12 weeks of feeding. REE subsequently decreased to presupplementation levels 12 weeks after cessation of supplementation.32 Conversely, Carbonnel et al33 recently reported no significant change in REE per unit FFM after a mean gain of 6.5 kg body weight in malnourished patients receiving PN, although absolute values of REE were increased.
In our patients after the period of nutrition support, although there was no significant change in mean REE expressed in absolute terms, which remained significantly lower than the control value, when expressed per unit body weight, there was now no significant difference compared with controls. This suggests an increase in the relatively metabolically inactive fat and protein stores, and a trend to restoration of normal body composition.
Our studies demonstrated significantly lower wholebody protein flux, oxidation, and synthesis in the undernourished patients, compared with the controls when considered in absolute terms (g/d). In comparing the anorexia patient group with the disease patient group, although there was no significant difference in mean absolute rates of whole-body flux, oxidation, and synthesis between the 2 groups, when expressed per kg body weight the anorexia patient group had rates similar to the control subjects, whereas the disease patient group had significantly lower protein flux and synthesis rates than either controls or the anorexia patient group. After nutrition support, the disease patient group demonstrated improvement in total body flux and synthesis, with flux and synthesis, expressed per kg body weight, now similar to both the anorexia patient group and controls.
Previous studies of the effects of nutrition status on total body protein kinetics have produced apparently conflicting results. A study by Norton et al demonstrated 23% reduction of whole-body protein synthesis and turnover (g/kg/d) induced by short-term fasting in healthy controls. Carbonnel et al5 also studied a group of undernourished patients with nonneoplastic disease, and, although in comparison to healthy controls they demonstrated reduced rates of whole-protein synthesis and turnover when expressed in absolute terms (g/d), when expressed in relation to body weight (g/kg/d), the rates were similar. The authors also reported no significant change in whole-body protein kinetics, expressed per kg body weight, after nutrition recovery with PN, although there was a slight increase when expressed per kg FFM. Holt et al35 reported similar whole-body protein synthesis rates in undernourished cystic fibrosis patients with stable pulmonary disease, compared with normal healthy children. However, patients with active infectious disease had markedly reduced protein synthesis. These results indicated that increased metabolic demands during acute disease, in the absence of sufficient energy, resulted in reduced protein synthesis.
Our study demonstrated significant reduction in whole-body protein turnover and synthesis (g/kg/d) only in patients with coexistent disease (disease patient group), whereas chronically undernourished patients without overt disease (anorexia patient group) had values similar to healthy controls. Thus, although we did not specifically study the nature of the antecedent diet in our malnourished patients, which may have been of a higher quality (although inadequate in quantity) in the anorexia patients compared with those with disease, and this may have influenced the results, the presence of disease in association with the undernutrition appears to have a significant adverse affect on whole-body flux and synthesis. Furthermore, our patients were heterogenous, with a number of different diseases, and it is likely that this would result in variable effects on metabolism. Previous studies have indicated that protein synthesis is increased in patients with active Crohn's disease,36 with reduction occurring after therapy with either steroids or elemental diets.3 However, the patients admitted to our study had relatively inactive disease and had been admitted primarily for nutrition support. The effect of specific disease entities, such as cancer, infection, inflammation, and malabsorption, on whole-body metabolism and protein synthesis requires further investigation.
CONCLUSIONS
Undernutrition is associated with an increase in REE (kcal/kg/d). Reduction in RQ and protein synthesis (g/kg/d) was evident only in those patients with coexistent disease. Refeeding resulted in normalization of RQ, REE (kcal/kg/d), and protein synthesis (g/kg/d).
