An effective weight loss program is centered on modifications of diet, behavior and physical activity, with the goal of promoting the loss of excess body fat and maintaining the appropriate amount of lean body mass that is necessary for optimal health and work performance (American College of Sports Medicine 1983, American Medical Association 1988). Although restriction of dietary energy intake is an effective means for reducing body weight, the role of specific dietary factors in maximizing the proportion of fat loss and minimizing the loss of muscle mass is less clear. Certain aspects of eating behavior, such as the time of day that meals are ingested, may have important consequences for weight control (Halberg 1989). Studies of mice or rats on restricted feeding schedules indicate that time of meal presentation, in relation to the environmental lighting schedule, can affect body weight, body temperature and possibly energy expenditure (Nelson et al. 1975, Nelson and Halberg 1986, Philippens et al. 1977). In humans, very few controlled studies have been conducted to assess the importance of the manipulation of time of eating on weight or body composition.

In two studies with normal weight volunteers, weight loss occurred when a single daily meal was ingested in the morning, whereas weight loss was minimal or, in some cases, weight was gained when the single meal was ingested in the evening (Hirsh et al. 1975, Jacobs et al. 1975). A study of obese subjects found no difference in the quantity of weight loss when a very low energy diet (< 3 MJ/d) was consumed either as breakfast or as dinner (Sensi and Capani, 1987). Under this severe dietary restriction, consuming food in the evening consistently enhanced fat oxidation, measured by indirect calorimetry, but body composition changes were not reported.

These few studies that examined weight changes in response to altered eating patterns were of very short duration (1-3 wk), and it would be inappropriate to project long-term weight changes based solely on the outcome of these studies. Further, considering that eating behavior typically includes multiple episodes of eating on a given day (United States Department of Agriculture 1980 and 1986), the one-meal-per-day paradigm used in these studies would have limited applicability. Meanwhile, even though conclusive scientific evidence is lacking, reducing nocturnal eating is a weight loss strategy often recommended in the popular press.

The objective of this study was to determine whether altered meal timing patterns affected weight and body composition changes during a weight loss intervention that combined regular exercise with a controlled, moderately energy restricted diet. Additionally, we measured energy utilization under standardized conditions of rest, exercise, following meals and following exercise to determine whether altering meal patterns would affect the metabolic fate of energy-yielding nutrients.

METHODS

Table 1. Subject characteristics prior to weight loss intervention1 [View Table]

Subjects lived in the metabolic suite at the Western Human Nutrition Research Center, 24 h/d, 7d/wk for 105 d of the study. Times for meals, daily outdoor walk and exercise workouts were standardized to keep the order and type of activities similar throughout the study.

Four meals were served daily. Meal times were set and strictly adhered to with breakfast at 0800-0830 h, lunch at 1130-1200 h, dinner at 1630-1700 h, and evening snack at 2000-2030 h. During the stabilization period, energy intake was distributed among the meals: 15% at breakfast, 35% at lunch, 35% at dinner, and 15% at evening snack. During the experimental periods, the energy intake of the AM pattern was distributed: 35% at breakfast, 35% at lunch, 15% at dinner and 15% at evening snack; the PM pattern was distributed: 15% at breakfast, 15% at lunch, 35% at dinner and 35% at evening snack.

The experimental diets consisted of 4-d rotating menus, composed of conventional foods. The macronutrient composition of the daily diet remained constant throughout the study and consisted of (mean ± SD) 59.7 ± 1.4% carbohydrate, 18.1 ± 1.2% protein, and 22.3 ± 0.7% fat. The menus were based on the Dietary Guidelines for Americans (USDA 1990) with emphasis on including a variety of foods, and reducing total fat content. In addition, the foods served at breakfast, lunch, dinner and evening snack were considered typical foods for those meal times. The diets were designed to keep the day-to-day variance in energy content of specific meals < 1%. Also, the macronutrient intake was distributed so that 2/3 or more of the daily intake was taken in the two large meals, and 1/3 or less was taken in the two small meals (Table 2). Finally, daily intake met or exceeded the RDA for protein and the required vitamins and minerals so that nutritional supplementation was not necessary. Although the distribution of micronutrients among the meals was not a major consideration in planning the diets, in general, the micronutrient patterns followed the macronutrient patterns with the majority of intake occurring with the large meals in both diets (Table 2).

Table 2. Energy and selected nutrients in combined breakfast plus lunch meals (B + L) or dinner plus evening snack meals (D + S)1 [View Table]

Subjects' compliance with the diet was excellent, and actual energy intake, as a percentage of prescribed energy intake, was 99.4 ± 0.5% and 100.0 ± 1.3% for experimental periods 1 and 2, respectively. Actual protein intake, as a percentage of prescribed protein intake, was 99.5 ± 0.7% and 99.8 ± 2.0% for experimental periods 1 and 2, respectively and averaged 1.19 ± 0.06 g/kg for experimental period 1 and 1.22 ± 0.07 g/kg for experimental period 2. 

Body composition was determined twice weekly by total body electrical conductivity measured by the HA-2 body composition analyzer (EM-SCAN, Springfield, IL). All measurements were taken before the breakfast meal. A prediction equation specific for overweight women was used to estimate fat-free mass from body conductivity measurements (Van Loan et al. 1990). Body fat mass was determined by subtracting fat-free mass from body weight. Body energy content was calculated from the body composition measurements. Each kg of fat was assumed to be equivalent to 37.7 MJ (Acheson et al. 1980), and each kg of fat-free mass was assumed to be 19.4% protein with an energy equivalent of 16.7 MJ (Brozek et al. 1963).

Two different protocols were used to assess energy expenditure under various physiological conditions. In the first protocol, respiratory gas exchange was measured following an overnight fast, while resting in the morning for 20 min (AM-RMR), and intermittently for 10 min periods at 30, 60, 90, 120, 150, and 180 min following consumption of standard breakfast meals in the postprandial state (AM-PP). After an early morning void, a urine sample was collected at the end of a 1-h rest period that included the interval of the AM-RMR gas exchange measurement. A cumulative, postprandial urine sample was collected during the next 3-h, spanning the time of the AM-PP intermittent gas exchange measurements. The standard breakfast meals contained 35% and 15% of daily energy intake for the AM and PM patterns, respectively. The first protocol resumed in the afternoon at 1530, approximately 3.5 h after lunch and 1 h after the daily walk. Gas exchange measurements were taken again for 20 min while resting (PM-RMR), and intermittently for 3 h following consumption of standard dinner meals (PM-PP), following the same measurement intervals as the morning testing. The afternoon and evening urine sampling procedures were also the same as the morning sampling procedures. The standard dinner meals contained 15% and 35% of daily energy intake for the AM and PM patterns, respectively. This protocol was conducted twice during the last 2 wk of both experimental periods; each subject completed a protocol in one day and repeated the same protocol one week later.

In the second protocol, respiratory gas exchange was measured under three conditions: in the morning following an overnight fast, while resting (PRE-EX); during 30 min of moderate intensity aerobic exercise on a cycle ergometer at 50% O₂max (EX); and for 60 min following exercise, while resting (POST-EX). After an early morning void, a urine sample was collected at the end of a 1 h rest period, prior to the initiation of exercise. The combined exercise and post-exercise urine sample was collected at the end of 2-h, coinciding with the exercise and post-exercise periods. In this protocol, all subjects consumed the same light breakfast (1 MJ) 1 hour prior to the exercise bout. This protocol was conducted twice during each experimental period, spanning the second, third, and fourth weeks of each period. Each subject completed this protocol in one day and repeated the same protocol 10-12 days later.

The energy expenditure data reported for each experimental period represent the average of two tests conducted during that period.

RESULTS

Table 3. The effect of large morning meals (AM pattern) versus large evening meals (PM pattern) on changes in weight and body composition during weight reduction1 [View Table]

Loss of fat mass was affected by the order in which subjects received the meal patterns (P < 0.05). Consequently, the effect of meal pattern on change in fat mass was examined by period. There was a significantly greater loss of fat mass associated with the PM pattern in period 1, but not in period 2 (Table 4 and Fig. 1). Similarly, change in body energy stores was affected by order (P < 0.05); more energy was lost with the PM pattern during period 1, but not during period 2 (Table 4).

Table 4. The effect of large morning meals (AM pattern) versus large evening meals (PM pattern) on changes in fat mass and body energy stores during the first 6 wk of weight loss (period 1) and the second 6 wk of weight loss (period 2)1 [View Table]

Independent of the meal pattern effects, there were also significant period effects on loss of body weight, fat mass and fat-free mass (Fig. 1). Compared to period 2, during period 1, more weight was lost, 3.79 ± 0.15 vs. 3.38 ± 0.31 kg (P < 0.05), more fat mass was lost, 3.38 ± 0.16 vs. 2.35 ± 0.20 kg (P < 0.001), and less fat-free mass was lost, 0.46 ± 0.21 vs. 1.06 ± 0.20 kg (P < 0.05).

Substrate oxidation rates in response to meals and at PM rest were also affected by meal pattern. Higher rates of carbohydrate oxidation were measured with large AM meals during the AM postprandial condition, the PM resting condition, and with large PM meals during the PM postprandial condition (Fig. 2). Also with the PM pattern, a higher carbohydrate oxidation rate was measured during the post-exercise hour, even though the meal patterns produced similar rates of carbohydrate oxidation at pre-exercise rest and during exercise (Fig. 2). Fat oxidation rates were affected by meal pattern only during the pre-exercise condition when the AM pattern produced higher fat oxidation rates compared to the PM pattern (Fig. 2).

Fat oxidation rates associated with the AM and PM patterns were examined separately for period 1 and period 2 because there was a significant order effect on fat oxidation under the AM-RMR condition (P < 0.005). Also, fat mass loss was higher for the PM pattern only in the first period and we wanted to further explore the relationship between loss of body fat and fat oxidation rates by period. In response to the PM meal pattern in period 1, fat oxidation rate was higher during the PM resting condition (P < 0.05) and tended to be higher during the AM resting condition (P < 0.09) (Table 5). In period 2, there were no meal pattern differences in fat oxidation during the PM resting condition, but fat oxidation was higher during AM rest for the group ingesting the AM meal pattern (P < 0.01) (Table 5). The group ingesting the AM pattern in period 2 also had a tendency toward greater fat loss in period 2 (P = 0.18); (Table 4).

Table 5. The effect of large morning meals (AM pattern) versus large evening meals (PM pattern) on resting fat oxidation rate, during the first 6 wk of weight loss (period 1) and the second 6 wk of weight loss (period 2)1 [View Table]

DISCUSSION

In this study, the total body electrical conductivity method was used to measure body composition. This method is based on the principle that the electrical conductivity of lean tissue is much greater than that of fat, and it provides an accurate and reliable estimate of two body compartments, fat-free mass and fat mass (Van Loan et al. 1987). However, a major limitation of this method is that it cannot quantify the components of fat-free mass such as water compartments, bone, muscle mass, or other intracellular constituents, such as glycogen.

The attenuation of the loss of fat-free mass with the PM pattern may have been due, in part, to alterations in muscle glycogen content. Conlee et al. (1976) demonstrated that a diurnal pattern of skeletal muscle glycogen content exists in the rat, with peak concentrations preceded by periods of greater food consumption. In humans, muscle glycogen fluctuates in accordance with periods of muscular activity and subsequent carbohydrate consumption. In a review of factors affecting changes in muscle glycogen during and after exercise, Blom et al. (1986) concluded that there was an increasing rate of post-exercise muscle glycogen resynthesis with increasing carbohydrate intake up to a level of 0.7 g of carbohydrate per kg body weight per hour. These authors also reported that the rate of glycogen resynthesis was negligible when carbohydrate consumption was 0.1 g/(kg·h) but increased to 2.5 mmol glycosyl units/(kg muscle wet weight·h) when carbohydrate consumption was 0.27 g/(kg·h). In our study, two evening meals followed the morning and early afternoon periods of increased physical activity. For the group consuming the PM pattern, a greater intake of carbohydrate occurred in the evening. When averaged over an 8-h interval, 1630-0030 h, carbohydrate consumption rate was 0.29 g/(kg·h) with the PM pattern, whereas with the AM pattern, carbohydrate consumption was ~0.12 g/(kg·h). Assuming that muscle glycogen synthesis occurred at the rates reported by Blom et al. (1986) and that the skeletal muscle mass of our subjects was approximately 25% of body weight, as reported by Clarys et al. (1984), then in the course of the 8-h evening rest-sleep interval, an accumulation of an additional mass of 250 g (glycogen plus its associated water) could occur with the PM meal pattern. This value represents roughly 25% of the difference in fat-free mass that we measured. However, since a direct determination of muscle glycogen was not made, the net change in muscle glycogen levels over the 6-wk intervention period is unknown.

Certain endocrine influences might have contributed to the difference in fat-free mass change between the meal patterns. Growth hormone secretion displays an endogenous rhythm that is partially linked with the sleep cycle. At night pulsatile secretion increases after 1-2 hours of sleep, with maximal secretion occurring during stages 3 and 4 of sleep (Hartman 1993). Although the effect of prolonged changes in dietary intake or meal patterns on growth hormone release are not known, it is conceivable that a greater flux of dietary amino acids with the large evening meals, coupled with the known protein anabolic effect of growth hormone, might combine to favor deposition of lean tissue. Similarly, Goetz et al. (1976) found that following a large meal taken at breakfast, peak insulin and glucagon levels occurred approximately 2.5 h post-ingestion, in close proximity to each other. However, when the same meal was taken in the evening, insulin levels peaked on schedule, but the glucagon peak was delayed for an additional 5 h. Thus, with diminished glucagon activity to counteract insulin, the anabolic influence of insulin may span a longer period in the evening and contribute to lean tissue accretion and/or increased glycogen deposition, particularly when the exogenous nutrient supply is more plentiful, as with our PM pattern. However, the duration of this evening anabolic period is questionable since Van Cauter et al. (1992) reported that endogenous insulin clearance rate increased in response to meals consumed in the evening.

A good example of an effect of time of nutrient ingestion on energy utilization is the widely confirmed observation that glucose tolerance is best in the morning and decreases during the afternoon and night. Jarrett et al. (1972) found that when a glucose load was administered in the afternoon and evening, there was a delayed rise in insulin, overall insulin levels were lower and glycemia was higher compared to the response to a morning glucose load. Zimmet et al. (1974) reported that fasting levels of free fatty acids were higher in the afternoon than in the morning and, through their inhibitory effect on glucose utilization, free fatty acids may contribute to decreased insulin sensitivity occurring later in the day. Circulating level of cortisol, another insulin antagonist, also displays a prominent diurnal fluctuation, with levels peaking in the morning just before awakening and decreasing throughout the day into the evening (Van Cauter and Refetoff 1985). Theoretically, the PM drop in cortisol should enhance glucose tolerance, not decrease it. Clearly, more work must be done to document such modest shifts in the balance between insulin and the glucose counterregulatory hormones before we can predict with certainty the metabolic fate of macronutrients ingested at different times of day.

The second objective of the study was to investigate whether there was an effect of time pattern of meal ingestion on energy expenditure and utilization under conditions of rest, in response to standard meals, and during exercise and recovery. Other than the increases in energy expenditure and carbohydrate oxidation that were expected to occur in proportion to the energy and carbohydrate loads associated with the small and large meals, we found little evidence of consistent changes in energy utilization associated with meal pattern. It is unfortunate that we were unable to measure energy utilization throughout the 24-h cycle or during sleep in this study. Nevertheless, with the AM pattern of large morning meals and small evening meals, it is probable that subjects were near a postabsorptive state when sleep began, whereas with the PM pattern of small morning meals and large evening meals, subjects were still in the postprandial state when sleep began. It is tempting to speculate that with the onset of postprandial sleep, energy was conserved, and heat production fell. However, Zammit et al. (1992) demonstrated that despite a dampening of thermogenesis during postprandial sleep, repeated episodes of postprandial sleep did not result in important energy savings, even over extended periods of time. Our results support this notion, since we were unable to observe any consistent trends in change in body energy stores associated with either the AM or PM meal pattern.

Our estimates of energy utilization by meal pattern and period indicated that during the first 6-wk period, the rate of fat oxidation was higher at rest in the morning and the afternoon for the group receiving the PM meal pattern. The PM meal pattern group also lost more fat mass during this period. On the other hand, in the second 6-wk period, the rate of fat oxidation was higher in the morning, but not in the afternoon, for the group receiving the AM meal pattern. There was a tendency for this AM meal pattern group to lose more fat mass in period 2. Of the fat oxidation rates measured, the PM postabsorptive rate corresponded to the quantity of body fat mass lost, despite the lack of a consistent meal pattern effect on fat oxidation, fat mass loss, and total body energy loss.

Weight loss is more rapid at the initiation of a weight loss regimen. Indeed, we observed that both weight loss and particularly fat mass loss were greater in period 1 than in period 2. Because we did not have a perfectly balanced design (n = 4 in group A and n = 6 in group B), we cannot rule out that the period effects may have biased the meal pattern effects or vice versa. However, the order of presentation of the meal patterns only affected the loss of fat mass (order effect P < 0.05). The order effect on weight loss was not significant (P = 0.16). Further, greater weight loss occurred in period 1 when more subjects were consuming the PM pattern which was associated with less weight loss than the AM pattern. Thus, for weight loss, the period effect ran counter to that of the meal pattern effect. The order effect on loss of fat-free mass was not significant (P = 0.99). We also found that loss of fat-free mass was greater in period 2 than in period 1. In period 2, more subjects were consuming the AM pattern which was also associated with greater loss of fat-free mass. Consequently, it is possible that the period effect slightly biased the meal pattern effect. But considering the relative probabilities of the period effect (P < 0.05) and the meal pattern effect (P < 0.001) on loss of fat-free mass, it is more likely that the meal pattern effect biased the period effect. Furthermore, although there was a gradual progression in both the aerobic and weight training workouts from period 1 to period 2, we designed the study to keep the relative intensities of the workouts constant throughout the study. Although it is possible that the progressive nature of the exercise regimens may have exerted differential effects on fat-free mass in periods 1 and 2, it is also possible that the initiation of the exercise regimens at the beginning of the study may have been an even greater stimulus for fat-free mass preservation in period 1, when subjects made the transition from being mildly active during the stabilization period to being fully active during the experimental periods. More studies are needed to determine the impact of small differences in activity levels on fat-free mass.

To conclude, meal ingestion pattern affected body weight and composition changes during weight reduction. Most notably, the best preservation of fat-free mass occurred when large PM meals were consumed. Meal patterns had no consistent effect on loss of body fat mass since consumption of large PM meals in period 1 was associated with significantly more fat loss, but consumption of large PM meals in period 2 was not associated with greater fat loss. Energy expenditure and carbohydrate oxidation rates increased in a predictable manner in response to the large, carbohydrate-rich meals. The influence of meal pattern on fat oxidation rate varied but was significant under three conditions: i) the AM pattern was associated with greater fat oxidation during the morning pre-exercise condition, ii) the AM pattern was associated with greater fat oxidation during the morning resting condition in period 2, and iii) the PM pattern was associated with greater fat oxidation during the afternoon resting condition in period 1. It is interesting to note that the afternoon resting oxidation rate corresponded with the loss of body fat mass in period 1. A desired outcome of a weight loss intervention is to minimize loss of fat-free mass, and the consumption of larger meals in the evening may very well contribute to achieving this outcome, particularly if the retention of fat-free mass proves to be lean tissue and not merely water. Future studies should be directed toward quantifying the various components of the fat-free mass that change in response to altered meal patterns and identifying the mechanisms that regulate this response. In consideration of the inconsistent meal pattern effects on body fat mass, the nature of this observation should be explored further to determine if the greater loss of fat mass associated with large PM meals was a true, but transient meal pattern response or an artifact of our cross-over design.