BACKGROUND: A relatively high percentage of energy intake as protein has been shown to increase satiety and decrease energy efficiency during overfeeding.
AIM: To investigate whether addition of protein may improve weight maintenance by preventing or limiting weight regain after weight loss of 5–10% in moderately obese subjects.
DESIGN OF THE STUDY: In a randomized parallel design, 148 male and female subjects (age 44.2±10.1 y; body mass index (BMI) 29.5±2.5 kg/m2; body fat 37.2±5.0%) followed a very low-energy diet (2.1 MJ/day) during 4 weeks. For subsequent 3 months weight-maintenance assessment, they were stratified according to age, BMI, body weight, restrained eating, and resting energy expenditure (REE), and randomized over two groups. Both groups visited the University with the same frequency, receiving the same counseling on demand by the dietitian. One group (n=73) received 48.2 g/day additional protein to their diet. Measurements at baseline, after weight loss, and after 3 months weight maintenance were body weight, body composition, metabolic measurements, appetite profile, eating attitude, and relevant blood parameters.
RESULTS: Changes in body mass, waist circumference, REE, respiratory quotient (RQ), total energy expenditure (TEE), dietary restraint, fasting blood-glucose, insulin, triacylglycerol, leptin, β-hydroxybutyrate, glycerol, and free fatty acids were significant during weight loss and did not differ between groups. During weight maintenance, the ‘additional-protein group’ showed in comparison to the nonadditional-protein group 18 vs 15 en% protein intake, a 50% lower body weight regain only consisting of fat-free mass, a 50% decreased energy efficiency, increased satiety while energy intake did not differ, and a lower increase in triacylglycerol and in leptin; REE, RQ, TEE, and increases in other blood parameters measured did not differ.
CONCLUSION: A 20% higher protein intake, that is, 18% of energy vs 15% of energy during weight maintenance after weight loss, resulted in a 50% lower body weight regain, only consisting of fat-free mass, and related to increased satiety and decreased energy efficiency.
The increasing incidence of obesity is a recognized medical problem in developed countries.1 For instance, obesity is a major factor for a number of diseases, including coronary heart diseases, hypertension, noninsulin-dependent diabetes mellitus, pulmonary dysfunction, osteoarthritis, and certain types of cancer.2,3,4
Factors related to the development of obesity are decreased physical activity (PA) and increased energy intake (EI); thus weight loss and loss of body fat can be achieved by reducing EI and/or increasing energy expenditure.
Treatment of obesity is beneficial. Weight loss reduces the risk for mortality and morbidity in obese subjects. Even modest weight loss, 5–10% of the initial body weight, already leads to beneficial health effects.5,6,7 Modest weight loss is a realistic goal for most subjects.5,7 However, long-term maintenance of the body weight lost can be described as unsuccessful.8
Most studies on weight maintenance show that weight regain is usually the case,9,10,11,12,13 indicating that subjects are not able to change their eating and activity behavior adequately.8 Interventions to improve long-term weight maintenance are therefore needed in order to treat obesity effectively. The limited long-term effectiveness of conventional weight management (dietary intervention, PA, and behavioral therapy) leads to the development of alternative weight reduction strategies. In this context, we suggest that aiming for a more favorable body composition during weight regain, that is, consisting of a larger fat-free mass (FFM), may slow down weight regain. This is based on our previous observation that weight regain was slower, when body composition of the weight regained consisted of a greater FFM,13 due to PA intervention. Therefore, we hypothesize that if we would be able to reduce increase in fat mass (FM) and promote increase in FFM during the inevitable weight regain, by an appropriate ingredient, we might limit weight regain. For this, we suggest to elevate protein intake, because of its contribution to storage of FFM,14 its low energy efficiency during overfeeding,15,16 and its increased satiety effect despite similar EI.17 The ‘Stock hypothesis’ states that during overfeeding, a relatively high percentage of energy as protein might have a limiting effect on body weight gain in humans through an energy efficiency effect.15,16 We suggest that this may also be applicable during weight regain. Thus, energy efficiency expressed as kg body weight regained over EI in kJ may be relatively lower on a low- and on a high-protein diet, and relatively higher on a 12–15 energy% (en%) protein diet. Until now, this has not been applied in a weight-maintenance situation. Thus, if on a relatively high-protein diet during weight regain, energy efficiency is low, this may result in a slower weight regain at a given EI. The low energy efficiency may be partly due to the body composition of the weight regained if this consists mainly of FFM (see above), which makes the cost of energy storage (ES) high.18 The third part of the hypothesis is based on our observation of a sustained higher satiety over 24 h during a high-protein diet despite the same EI,17 which also may imply that a relatively low EI during weight maintenance is bearable.
Therefore, the aim of the present study was to investigate whether addition of protein to the diet may improve weight maintenance by preventing or limiting weight regain after a weight loss of 5–10% in moderately obese subjects.
A total of 200 male and female subjects, aged between 18 and 60 y, were recruited for this study. They underwent a medical screening. Selection resulted in 180 subjects who were in good health, nonsmokers, not using medication, and at most moderate alcohol users. These were overweight and moderately obese subjects, with a body mass index (BMI) between 25 and 35 kg/m2. They all gave their written informed consent. The Medical Ethics Committee of the Academic Hospital in Maastricht approved the study. Subjects were randomized to an extensive group (n=60) and to an intensive group (n=120). The intensive group underwent all measurements; the extensive group underwent the same protocol and measurements, except the ventilated hood measurements determining resting energy expenditure (REE) and accelerometer measurements determining PA. During the first week, 30 subjects did not start the study due to moving house, changing jobs, or not being able to fulfill the schedule with visits to the University. Intention to treat applied for 150 subjects. Then two subjects dropped out during the first week of the very low-energy diet. They were aimed to participate in the ‘additional-protein’ group. Their baseline characteristics did not affect the averages of the base-line measurements of the treatment group. Finally, 148 subjects fulfilled the study, that is, 103 in the intensive group and 45 in the extensive group.
To determine the subject characteristics, the following measurements were executed at baseline, after weight loss, and after 3 months weight maintenance (Table 1):
Body weight and body mass index. Body weight was measured with a digital balance (Seca, model 707, Hamburg, Germany; weighing accuracy 0.1 kg) with subjects in underwear, in a fasted state and after voiding their bladder. Height was measured using a wall-mounted stadiometer (Seca, model 220, Hamburg, Germany). BMI was calculated as body weight/height2 (kg/m2).
Waist. The distribution of fat was investigated by measuring the waist circumference at the site of the smallest circumference between the rib cage and the ileac crest, with the subjects in standing position.
Body composition. Total body water (TBW) was measured using the deuterium (2H2O) dilution technique.19,20 In the evening, the subjects ingested a dose of deuterium-enriched water (2H2O) after collecting a background urine sample. After consumption of the 2H2O, no more fluid and food was consumed. The following morning, a urine sample from the second voiding was collected between 800 and 1000 h. Deuterium concentration in the urine samples was measured using an isotope ratio mass spectrometer (Micromass Optima, Manchester, UK). TBW was obtained by dividing the measured deuterium dilution space by 1.04.19 FFM was calculated by dividing the TBW by the hydration factor 0.73. By subtracting FFM from body weight, FM was obtained. FM expressed as a percentage of body weight is body fat percentage.
Attitude toward eating. To determine whether attitude towards food intake changed during the experiment, the Three Factor Eating Questionnaire (TFEQ) was used.21 To determine the frequency of dieting, the Herman Polivy Questionnaire (HP)22 was used.
Postabsorptive appetite profile. To determine the postabsorptive appetite profile, hunger and satiety were rated on anchored 100 mm visual analogue scales in the morning before breakfast.
Blood parameters. A fasted blood sample of 10 ml was taken and mixed with EDTA to prevent clotting. Plasma was obtained by centrifugation, frozen in liquid nitrogen, and stored at −80°C until further analysis. Plasma glucose concentrations were determined using the hexokinase method (Glucose HK 125 kit, ABX Diagnostics, Montpellier, France). The Wako NEFA C-kit (Wako Chemicals, Neuss, Germany) was used to determine free fatty acid concentrations. Insulin concentrations were measured using the RIA-kit (Insulin RIA-100, Kabi-Pharmacia). The glycerolkinase method was used to determine glycerol concentrations (Boehringer Mannheim GmbH, Mannheim, Germany). Triacylglycerol was measured using the GPO-trinder kit (Sigma Diagnostics Inc., St Louis, MO, USA). Concentrations of triacylglycerol were corrected for glycerol.
The β-hydroxybutyrate dehydrogenase method (Sigma Diagnostics Inc., St Louis, MO, USA) was used to determine β-hydroxybutyrate concentrations. Leptin concentrations were measured using the human leptin RIA-kit (Linco Research Inc., St Charles, USA).
Adverse events. Adverse events during treatment were recorded and the severity and outcome specified.
Resting energy expenditure and substrate oxidation. REE and substrate oxidation were measured by means of an open-circuit ventilated hood system. The measurements were executed in the morning, while subjects were in a fasted state and lying supine for 30 min. Gas analyses were performed by a paramagnetic oxygen analyzer (Servomex type 500A, Crowborough, Sussex, UK) and an infrared carbon dioxide analyzer (Servomex type 500A), similar to the analysis system described by Schoffelen et al.23 Calculation of REE was based on Weir's formula.24 Respiratory quotient (RQ) was calculated as CO2 produced/O2 consumed.
Physical activity. PA was partly determined using a CSA,25 partly with a triaxial accelerometer for movement registration (Tracmor) during 1 week. The Tracmor is a small device (7 × 2 × 0.8 cm, 30 g), which measures accelerations in the anteroposterior, mediolateral, and vertical directions of the trunk.26 Subjects were wearing the same type of accelerometer during waking hours in a belt at the back of the waist, during three different phases of the study.
Physical activity level (PAL) was calculated using the following equations: in which TEE is the total energy expenditure (MJ/day) and REE is the resting energy expenditure (MJ/day).
Energy expenditure. TEE was calculated by multiplying REE by PAL.
Energy intake. EI was calculated as EI=TEE+ES (energy storage). ES was calculated from the composition of the energy stored, that is, 30 MJ/kg for a usual ES as 60% fat and 40% FFM, and 52 MJ/kg for protein storage.18
Protein intake. Compliance to additional protein intake was checked by taking 24 h urine samples, and analyzing these for nitrogen. EI from protein was calculated from the 24 h nitrogen output according to the formula of Isaksson:27 protein intake (g)=(nitrogen output in 24 h urine (g/day)+2 g) × 6.25.
Body temperature. Core body temperature was measured in a subset of subjects (n=10 in the additional protein group and n=10 in the control group) during 24 h using a wireless core body temperature monitoring system (CorTemp TM 2000), consisting of calibrated disposable temperature sensors (COR-100) with a miniaturized ambulatory recorder (CT2000) (HTI Technologies, Inc., Palmetto, FL, USA). This way, every minute a core body temperature measurement was stored. Moving averages were calculated over 30 min. The 24 h average core body temperature, and the average 24 h minimum, 24 h maximum, and minimum core body temperature during the day, over 30 min was calculated. The hypothesis behind measuring core body temperature is that it decreases due to body weight loss, and increases again during weight regain. However, due to the expected increased thermogenesis due to a high-protein diet during weight regain, increase in core temperature may be greater in the group on the additional protein diet.
Weight loss period. After determining the subject's baseline measurements, a very low-energy diet (2.1 MJ/day) intervention followed for 4 weeks, in order to let the subjects lose weight. The diet (Modifast®) was supplied in three sachets daily, to be dissolved in water in order to obtain a milkshake, pudding, soup, or muesli. Vegetables and fruit were allowed in addition to Modifast. The aim was a body weight loss of at least 4 kg over 4 weeks.
After this weight loss period, the measurements described above were repeated (Table 1).
Weight maintenance. During the weight-maintenance phase, the subjects were divided into two identical groups, stratified for gender, BMI, age, eating behavior (TFEQ, factor 1), and REE. Subjects from both groups visited the University with the same frequency, and received the same attention from the researchers with respect to questions, as well as the same counseling on demand by the dietician.
To one of the groups, 48.2 g additional protein/day was provided as one sachet of a meal replacer (Modifast)/day (17 g protein; 0.7 MJ/day) plus two sachets of protein (31.2 g; protein source Ca-caseinate; 0.5 MJ/day), to be dissolved in water resulting in two vanilla-drinks. This amounted together to 1.2 MJ/day. Subjects were required to consume the meal replacer and one protein drink as part of their ad libitum lunch, and one protein drink in the afternoon. This way we aimed at an EI of 18–20 percentage energy from protein/day. Although the treatment with respect to number of visits, measurements, and attention was identical in both groups of subjects, there was no placebo used for the additional protein, similar to previous meal-replacement studies.28 We included the measurements of dietary restraint to check whether additional protein to the diet would affect attitude toward eating.
Measurements as described under baseline measurements were executed again 3 months (ie 13 weeks) later. In addition, body weight was determined 1 and 2 months after the start of the weight-maintenance phase. Also, at 2 months weight maintenance, nitrogen in the urine was determined in order to be able to check for compliance with respect to additional protein intake.
Body weight maintenance was expressed as rate of regain over the first 3 months after weight loss. The 3 months period to measure weight maintenance was based on the stable rate of regain (0.67 kg/month) that we showed before over the first 3 months, and on the strong relationship of rates of regain between 3, and 8 or 14 months (r=0.9; P<0.0001).13
Effects of the protocol including additional protein intake over time on the parameters measured were compared with effects of the protocol followed without additional protein on these parameters, using ANOVA repeated measures (Statview SE+Graphics, Abacus Concepts, Berkeley, CA, 1988). When appropriate, differences between groups were analyzed using a factorial ANOVA, or a Student's t-test (Statview SE+Graphics). Relationships of changes in parameters over time with changes in leptin were analyzed using regression analysis (Statview SE+Graphics). Statistical significance was set at P<0.05. Energy efficiency was calculated as kg body mass regained over kJ EI.
No different effects of additional protein consumption for men or women were observed; therefore, these data have been taken together. No adverse events occurred.
Body weight loss
The following variables changed over time, but were not significantly different between groups (Tables 1, Tables 2 and 3). Body weight decreased, that is, 6.4 kg±1.8 (s.d.) or 7.5%±2.0 (s.d.) of their original body weight (P<0.001) (Figure 1). This consisted of 3.9±3.2 kg FM (61%) and 2.5±2.2 kg FFM (39%) (Figure 2). Waist circumference decreased. Attitude toward eating showed an increase in cognitive dietary restraint (Table 1). REE, RQ, and TEE reduced, while PAL remained the same (Table 2). The blood parameters from the fasting blood-samples (glucose, insulin, triacylglycerol, and leptin) significantly decreased, while β-hydroxybutyrate, glycerol, and free fatty acids increased (Table 3).
During the weight-maintenance phase, the ‘additional-protein group’ showed compliance by a significantly higher amount of nitrogen in 24 h urine collection, representing a significantly higher protein intake. Percentage EI from protein appeared to be 18% in the ‘additional-protein group’, being significantly higher than the 15% in the control group. Dietary restraint, that is, cognitive restraint remained at the level reached during weight loss, with no differences between the groups, indicating no different role of dietary restraint.
The following variables changed differently between the groups during the weight-maintenance period (Tables 1, 2 and 3). Satiety ratings in the fasted state before breakfast were increased significantly more in the additional-protein group, despite lack of difference in 24 h EI. Percentage regain of the body mass lost (Figure 2), and related to this the rate of body mass regain, kg body mass regained, the resulting BMI, FM regained (Figure 3), the resulting % body fat, and regained waist circumference were significantly lower in the additional-protein group. Energy efficiency, calculated as kg body mass regain/EI, was significantly lower in the ‘additional-protein group’. Changes in REE expressed as a function of FFM, PAL, TEE, and core body temperature did not differ significantly between the groups. Increase in triacylglycerol and in leptin was significantly lower in the ‘additional-protein’ group. Increases in glucose, insulin, and decreases in β-hydroxybutyrate, glycerol, and free fatty acids did not reach the original values again. They did not differ between the groups.
Core body temperature measurements did not show significant differences over time or between groups (Table 4).
In both groups, the decrease in leptin during weight loss was related to the body weight lost (r2=0.3; P<0.0001) and to the decrease in REE (r2=0.5; P<0.0001). The increase in leptin during body weight regain was related to body weight regain (r2=0.3; P<0.0001), the increase in REE (r2=0.5; P<0.0001), and to the increase of maximum core body temperature of 30 min moving average per 24 h (r2=0.4; P<0.0001). The increase in leptin was only related to the increase in FM in the nonadditional-protein group (r2=0.2; P<0.0001); there was no relationship with the increase in FFM.
Overweight to moderately obese men and women who consumed 18% of EI as protein regained less weight, that is, 1 kg, during 3 months after 7.5±2.0% body weight loss over 4 weeks, compared to the 2 kg that their counterparts who consumed 15% of EI as protein regained. Related to this percentage regain of the body mass lost, regain of waist circumference, rate of body mass regain, regain of FM and of percentage body fat, and the resulting BMI were lower in the additional-protein group, without a difference in dietary restraint or PA between the groups.
Body composition of the body mass regained was more favorably in the additional-protein group, that is, no regain of FM, but only of FFM, resulting in a lower percentage body fat. This was independent of PA, since PAL did not change over time or between groups. Leptin concentrations from fasting blood-samples during weight regain increased significantly slower in the additional-protein group. This may be related to the observation that only in the control group the increase of leptin was related to the increase of FM. Moreover, metabolic risk characteristics were reduced in the additional protein group, such as regain of waist circumference and triacylglycerol. In addition to dietary restraint and PA, also REE, TEE, or core body temperature played no significant role in the difference in body weight regain, since these did not differ between the groups. The reason for this may be that in the additional protein group only FFM was increased, causing increases in REE, but that in the control group body mass increased more, causing a similar increase in REE.
Thus, the slower body weight regain with the additional protein diet related to a difference in body composition of the body mass regained (without differences in PA or dietary restraint) was in line with a part of our hypothesis.
Energy efficiency (kg body mass regain/EI) was significantly lower in the additional-protein group, which contributes to some evidence for the second part of our hypothesis. The observation with respect to energy efficiency during weight regain is similar to the ‘Stock hypothesis’, as described for weight gain.15 The background of the difference in energy efficiency was entirely due to the difference in body composition of the weight regained, as described above. It is likely that the diet-induced thermogenesis was higher on the high-protein diet, as we have shown before,17 but this did not appear in a possible difference in TEE.
The third part of our hypothesis was supported in that satiety was higher on the high-protein diet, under similar EI conditions, which is also in line with our previous observations.17
Taken together, we showed evidence for the combination hypothesis we suggested for weight maintenance. Increased protein intake sustained weight maintenance by (i) favoring regain of FFM at the cost of FM at a similar PAL, (ii) reducing the energy efficiency with respect to the body mass regained, and (iii) increasing satiety independent of EI or dietary restraint.
Although the treatment with respect to number of visits, measurements, and attention was identical in both groups of subjects, there was no placebo used for the additional protein, similar to previous meal-replacement studies.28 Therefore, the results in the control group could be strongly related to increased dietary restraint. However, cognitive restraint (factor 1 of the TFEQ) was similarly increased during weight reduction and weight maintenance in both groups. So both groups increased their dietary restraint; this increase early during treatment is an important predictor of weight-maintenance success, as we have shown before.8
The strong drop in leptin during weight loss, and the increase during weight regain in our overweight subjects, is in line with short-term observations in normal weight subjects.29,30,31,32,33,34 The decrease in leptin shows the ability of the adipocytes to downregulate leptin expression in response to a proceeding negative energy balance. This is in line with Flier's suggestion that leptin responds perhaps relatively strongly to a negative energy balance because ‘leptin may act as an indicator of neuroendocrine factors signaling energy deficiency rather than fulfilling the role of an antiobesity hormone’.35 However, we also found that the increase in leptin during weight regain was positively related to the percentage body weight regain, regain of FM, the increase in REE, and body temperature during weight regain. This indicates that the effect of leptin during ‘refeeding’ may be rather a metabolic than a satiety effect. Only in the control group, leptin was correlated to regain of FM, which shows a stronger relationship of leptin with body fat than with body mass. In the ‘additional protein group’, no correlation between changes in leptin and in FM was shown; FM still decreased while FFM increased and represented the body weight regained.
In conclusion, additional protein consumption during weight maintenance after weight loss resulting in 18 vs 15 en% protein, resulted in a 50% lower body weight regain, only consisting of FFM, and related to increased satiety and decreased energy efficiency.