Effects of l -phenylalanine on energy intake in overweight and obese women: Interactions with dietary restraint status

Available online at www.sciencedirect.com

Appetite 51 (2008) 111–119 www.elsevier.com/locate/appet

Research report

Effects of L-phenylalanine on energy intake in overweight and obese women: Interactions with dietary restraint status Rachael J. Pohle-Krauza a, Juan L. Navia b, Elizabeth Y.M. Madore a, Jessica E. Nyrop a, Christine L. Pelkman a,* a

Department of Nutrition and Exercise Science, University at Buffalo, 15 Farber Hall, Buffalo, NY 14214-8001, United States b McNeil Nutritionals, Division of McNeil – PPC, Inc., 601 Office Center Drive, Fort Washington, PA 19034, United States Received 6 June 2007; received in revised form 21 December 2007; accepted 10 January 2008

Abstract L-Phenylalanine (Phe), is a potent releaser of the satiety hormone, cholecystokinin (CCK) and previous studies, conducted primarily in men, show that ingestion of Phe reduces energy intake. The objective of the current study was to test the effects of Phe on energy intake in overweight and obese women. Subjects (n = 32) received three treatments (high-dose (10 g Phe), low-dose (5 g Phe and 5 g glucose) or control (10 g glucose)) 20 min before an ad libitum lunch and dinner meal in a within-subjects’, counterbalanced, double-blind study. No effect of Phe was found, however, interactions with dietary restraint status were detected in post-hoc analyses. Energy intake over the day was 11% lower following high-dose Phe versus control for women classified in the lower tertile of rigid restraint, a subscale of the dietary restraint scale, whereas no effects were noted for women in the middle and upper tertiles. High-dose Phe increased ratings of nausea, however, reduced energy intake in the high-dose condition was noted only for subjects with low nausea ratings. These results suggest that the satiety response to Phe is modulated by rigid restraint status and that reductions in food intake occur independently of Phe’s effects on nausea. Published by Elsevier Ltd.

Keywords: Satiety; L-Phenylalanine; Dietary restraint; Energy intake; Nausea

Introduction The prevalence of obesity continues to rise in the United States and has increased by more than 60% since 1990 (Mokdad et al., 2001). Recent research has focused on the effects of dietary factors that enhance satiety and potentially influence energy intake and body weight. There is evidence that dietary macronutrients differentially influence satiety with previous studies showing that protein may be more satiating than carbohydrate or fat (Barkeling, Rossner, & Bjorvell, 1990; Butler, Davies, Gehling, & Grant, 1981; de Castro, 1987; Stubbs, van Wyk, Johnstone, & Harbron, 1996; see reviews by Westerterp-Plantenga & Lejeune, 2005 and Halton & Hu, 2004). Protein is thought to affect satiety through the release of biologically active peptides that engage a number of central and peripheral mechanisms. A key mechanism thought to mediate the effect of dietary protein on satiety is the release of gut

* Corresponding author. E-mail address: [email protected] (C.L. Pelkman). 0195-6663/$ – see front matter. Published by Elsevier Ltd. doi:10.1016/j.appet.2008.01.002

hormones, such as CCK. L-phenylalanine (Phe), an essential amino acid, has been shown to enhance satiety in humans, rodents and primates (Gibbs & Smith, 1977; Mathur & Manchanda, 1991; Muurahainen, Kissileff, & Pi-Sunyer, 1988) and to be a more potent releaser of CCK than other amino acids (Konturek, Radecki, Thor, & Dembinski, 1973). In human studies, 10 g of Phe has been shown to elicit a five-fold increase in plasma CCK within 20 min (Ballinger & Clark, 1994) and reduce energy intake by 16–30% 20 min, or one to two hours after consumption in a dose-dependent manner (Ballinger & Clark, 1994; Muurahainen et al., 1988; Rogers & Blundell, 1994; Ryan-Harshman, Leiter, & Anderson, 1987). In the four studies examining effects of Phe on energy intake, three used only male subjects and one included two women. Therefore, the purpose of the current study was to examine the effects of Phe on energy intake and satiety in women. We focused our investigation on overweight and obese women as this population potentially may benefit most from the development of dietary approaches that enhance satiety and reduce food intake. Recent results from our laboratory show that in women, dietary restraint, and in particular, classification based on scores

112

R.J. Pohle-Krauza et al. / Appetite 51 (2008) 111–119

for the rigid restraint subscale of the dietary restraint instrument, modulated the satiety effects of a fiber beverage in one study (Pelkman, Navia, Miller, & Pohle, 2007) and the effects of Phe in another (Pohle-Krauza, Carey, & Pelkman, in press). Other investigators also showed that behavioral characteristics related to eating, such as dietary restraint (Burton-Freeman, 2005; Ogden & Wardle, 1990; Rolls et al., 1994) and binge eating (Geliebter, Gluck, & Hashim, 2005) affected satiety responses. Thus, a secondary objective of our investigation was to determine if dietary restraint, as well as other baseline behavioral characteristics of the subjects such as binge eating, modulated the satiety effects of Phe. Methods Research design The study was of a double-blind, within-subjects’ design with 30 subjects to be given encapsulated treatments (low-dose Phe, high-dose Phe, and control). Order of treatment conditions was counterbalanced such that five subjects were randomly assigned to each of six possible sequences. Energy intake over one day was measured in each condition. Subjects consumed breakfast, lunch and dinner in the laboratory and were provided with a take-home evening snack. They consumed the capsules before lunch and dinner meals. Subjects were asked to complete visual analog (VAS) scales to rate subjective appetitive sensations over the day. Approval for the study was granted on March 2005 and subject testing was completed in March 2006. Subjects Premenopausal women of any racial/ethnic background were recruited through media advertisements, flyers and postcards. We recruited women between 20 and 50 years of age, who were non-smokers and overweight or obese (BMI 25.0–34.9 kg/m2). Subjects were screened to ensure they had no food restrictions, did not smoke, were not currently dieting and did not have any chronic diseases or take any medications known to affect food intake. Subjects were excluded if they scored >10 on the Beck Depression Inventory (Beck & Beamesderfer, 1974) or >30 on the Eating Attitudes Test (Garner, Olmsted, Bohr, & Garfinkel, 1982). Subjects also completed the Binge Eating Scale (Gormally, Black, Daston, & Rardin, 1982) and the Eating Inventory (Stunkard & Messick, 1985). Scales of the Eating Inventory were quantified and include disinhibition, hunger, and dietary restraint as well as the subscales of rigid and flexible restraint. Height was measured to the nearest tenth of a cm using standardized procedures (‘‘Plan and operation of the Third National Health and Nutrition Examination Survey, 1988–94. Series 1: programs and collection procedures,’’ 1994) and a portable stadiometer (Seca, Model #225). Subjects were asked to remove outer garments and shoes. Weight was then measured on an electronic scale (Seca, Model # 770) to the nearest tenth of a kg. A standard finger-prick screening test was performed to

ensure subjects did not have elevated serum levels of Lphenylalanine. Subjects were asked to abstain from alcohol for two days before each session, to stop use of any vitamins or other dietary supplements and to refrain from consumption of any foods after 10 p.m. in the evening before each session. The study was approved by the Health Sciences Institutional Review Board of the University at Buffalo and subjects gave their informed consent before participation. All procedures to be used in the study were fully disclosed, however, the true purpose of the experiment was not stated in the consent form. Subjects were told that the purpose was to examine the effects of dietary protein on energy levels and liking for foods. We routinely use such distractions as previous research demonstrates that demand characteristics of the laboratory setting can affect eating behavior in obese women (Faith, Wong, & Alison, 1998). Treatments Test capsules were prepared on site in three formulations – high-dose (10 g Phe), low-dose (5 g Phe and 5 g glucose) and control (10 g glucose). Doses were chosen based on findings in men showing a stepwise reduction in food intake from 20 to 30% for 5 and 10 g of Phe, respectively (Rogers & Blundell, 1994). Pharmaceutical-grade L-phenylalanine (Voight Global Distribution LLC, Kansas City, MO) and glucose (Cerelose1; Corn Products International, Westchester, IL) was used. Each dose contained 167.5 kJ (40 kcal) and was divided into 22 size 0 gelatin capsules. The capsules were an opaque orange color and had an orange flavor. Doses were prepared by laboratory staff using standard encapsulating equipment (CapsulineTM, Pompano Beach, Florida). Because Phe has a distinctly bitter taste, the capsules were lightly shaken before storage to ensure that the Phe powder did not adhere to the capsules. Doses were stored in plastic containers and labeled by a confederate with arbitrary letters to maintain the study blind. The composition of randomly selected, 22-capsule doses was verified by the study sponsor at the beginning, middle and end of the study. Test sessions Subjects were asked to report to the laboratory for meals (breakfast, lunch and dinner) on three occasions. Test days were scheduled on the same weekday (Monday – Thursday) for each subject within three consecutive weeks. Subjects were asked to consume a comparable meal in the evening before each session to reduce the effects of variations in pre-session meal composition on energy intake on test days. Sessions began between 7:00 and 9:00 a.m. Before breakfast, subjects were asked to report their evening meal and if they experienced any recent illness. They consumed breakfast in the laboratory and returned for lunch 4–5 h later and dinner 9–10 h after breakfast. Subjects were given a cooler containing snack items to be consumed in the evening after dinner and asked to return the cooler and its contents the following morning. They were instructed to consume only water between meals and to complete 100-mm VAS, anchored with the phrases ‘‘Not at all’’

R.J. Pohle-Krauza et al. / Appetite 51 (2008) 111–119

and ‘‘Extremely’’ for thirst, hunger, nausea and fullness and ‘‘Nothing at all’’ and ‘‘A large amount’’ for how much food they felt they could eat (prospective consumption). VAS measures of satiety have been shown to have good validity and reproducibility in experimental studies (Drapeau et al., 2007; Flint, Raben, Blundell, & Astrup, 2000; Parker et al., 2004; Stubbs et al., 2000) Subjects were also asked to use VAS to rate their energy level and liking of one of the foods to distract them from the true purpose of the study. VAS ratings were completed before and after ingestion of the capsules, before and after each meal, and hourly between meals. Subjects were asked to consume the capsules within a 5-min interval with 12 oz of water before the lunch and dinner meals. Timers were provided to aid pacing of consumption and an additional 8 oz of water was provided if needed. Lunch was served 15 min following ingestion of the capsules. Meals The study was designed to assess the effects of ingested Phe on energy intake at lunch, dinner and evening snack. Thus, food selection at breakfast was controlled to reduce effects on intake later in the day. Subjects were served the same breakfast each test session which included a bagel, juice, milk, coffee or tea and condiments. Lunch and dinner were served as individual, buffet-style meals allowing subjects ad libitum consumption from a variety of meal-appropriate foods, such as sliced meats, breads, cheeses and fruits for lunch, and hot entrees, including meats (roast beef or chicken) and vegetables for dinner. Subjects could choose from a variety of hot and cold beverages, including juice, soda, coffee and a variety of teas, and dessert items, including pudding, fruit-flavored yogurt or cake, at the lunch and dinner meals. They selected cold beverages and snack items, such as potato chips, cookies, chocolate bars, fruit and fresh vegetables, to consume after dinner. Cold bottled water was served with each meal and included in the evening snack pack. Foods were commercially available products and contained varying amounts of energy and macronutrients allowing subjects to vary intake of energy, fat, protein and carbohydrate. Foods were weighed on an electronic scale (Ohaus, Model N1B110) to the nearest 0.1 g before and after consumption to determine the amount consumed. Energy and macronutrient composition of the foods was obtained from the manufacturer’s food label or from a standard reference (Pennington, 1998) for unlabeled food, such as fresh produce. Data analyses Power analyses were conducted using NQuery (Version 4.0). We used estimates of mean intake (and standard deviations) at lunch, dinner and daily total reported by Rolls and colleagues (Rolls, Shide, Thorwart, & Ulbrecht, 1998) in a study of 12 obese women that consumed buffet-style meals in a laboratory setting using similar procedures for subject screening and testing to those we planned to employ. We conducted analyses using the paired t-test procedure for a one-sided t-test with

113

alpha adjusted to correct for multiple comparisons between three conditions (adjusted a = 0.0167). A sample size of 30 subjects yielded power estimates of 0.81, 0.86 and 0.94 to detect a 15% reduction in energy intake at lunch, dinner, and for total daily intake, respectively. Outcome data were analyzed using the Statistical Analysis System (SAS Version 9.0, Cary, NC). The mixed model procedure was used to test for differences between conditions (high-dose, low-dose and control). We entered session number (1–3) as a covariate to test for effects of repeated testing. It was not significant and was subsequently removed from the models. Endpoint measurements included the intake of energy over the day and at individual meals (breakfast, lunch, dinner and evening snack). Models were tested to determine the effects of condition on the VAS ratings of hunger, fullness, prospective consumption and nausea. Additionally, an appetite score was calculated to reflect the combined appetitive ratings, similar to the strategy employed by Anderson and colleagues (Anderson, Catherine, Woodend, & Wolever, 2002) using the following equation: appetite = (hunger + prospective consumption + (100 fullness))/3. Mixed model analysis of variance models were tested for each VAS rating and for appetite score with condition, time and the interaction term entered as factors. Separate models were run to test for effects on ratings taken immediately before meals. Ratings taken between lunch and dinner were examined separately to determine effects on postprandial satiety after lunch. Ratings during this interval were also used to calculate incremental area-under-the-curve (AUC) values (using the trapezoid rule). Tukey’s post-hoc tests were used to compare LSMEANS when the main effect of condition was significant. We tested interactions between condition and baseline characteristics on energy intake and subjective satiety, with covariates tested as continuous, binary (using the 50th percentile as a cut point) and tertiary variables (using the 33rd and 66th percentiles as cut points) in the mixed models. No interactions were found for age, BMI, binge score, or disinhibition. Significant interactions were found for daily energy intake between condition and type of restraint (flexible and rigid) when subjects were classified in tertiles and a trend was found for global restraint. We further examined these interactions using the SLICE command in SAS to test for an effect of condition within each tertile of flexible, rigid and global restraint. P < 0.05 was considered to be statistically significant for interaction terms and the effect of condition in the SLICE procedure. Where significant effects of condition were found within tertiles, adjusted P values were used to compare LSMEANS between the three conditions (adjusted P = 1 (1 P)3). Data given in the text, figures and tables are LSMEANS (S.E.M.) from the mixed models, unless stated otherwise. Results Thirty-two women were enrolled and completed the study (Table 1). Two additional subjects were enrolled to allow for attrition. These two subjects were randomly assigned to

114

R.J. Pohle-Krauza et al. / Appetite 51 (2008) 111–119

Table 1 Baseline characteristics of subjects (n = 32)

Table 2 Energy intake (kJ) at breakfast, lunch, dinner and evening snack by conditiona

Variable

Mean

S.D.

Range

Age (years) BMI (kg/m2) Beck Depression Inventory Eating Attitudes Test Binge Eating Scale

39.7 29.9 4.1 5.2 11.9

7.4 2.8 2.9 4.1 7.1

23–50 25.1–34.9 0–10 0–16 0–29

9.4 2.8 3.3 5.1 7.9

4.2 1.8 1.6 3.0 3.9

2–19 0–6 0–6 0–13 1–15

Breakfast Lunch Dinner Evening Snack Daily totalc a

Eating Inventory Global restraint Rigid restraint Flexible restraint Hunger Disinhibition

treatment sequences and were included in the final analysis as all subjects completed the protocol. Some exceptions to the study protocol were made due to subjects’ scheduling conflicts. One subject had two weeks off between test sessions, two subjects had one week off between test sessions, and two subjects required one shorter washout week (6 days). Energy intake As expected, there was no effect of condition on energy intake at breakfast (maximum difference of 35 kJ or 8 kcal between conditions) (Table 2). There was also no effect of condition on energy intake at lunch, dinner, evening snack or for total daily intake. Energy intake over the day varied less than 3% across conditions (Table 2). Interactions were found between condition and the tertile classifications of restraint. For total daily energy intake, the interaction with condition was significant for rigid (P = 0.009) and flexible restraint (P = 0.03) with a trend noted for global restraint score (P = 0.06) (Table 3). Using the SLICE procedure, a significant effect of condition was found in the low tertile of rigid and global restraint whereas no effect was

b c

High-dose

Low-dose

Control

P-value b

1846  69 2827  167 2856  172 1646  169 9511  403

1836  69 2758  167 3096  172 1758  169 9784  403

1811  69 2649  167 3111  172 1815  169 9721  403

0.82 0.30 0.12 0.54 0.50

n = 32. Data shown are LSMEANS  S.E.M. For main effect of condition in mixed model analysis of variance. Includes 335 kJ from L-Phe or dextrose.

found in the middle and high tertiles. Subjects in the low tertiles of global and rigid restraint consumed less in the high-dose versus low-dose conditions and tended to consume less in the high-dose condition versus control (Table 3). Despite the highly significant effects of condition on daily energy intake in the low tertiles of global and rigid restraint, no significant interactions were noted for individual meals. Trends were found at lunch with subjects in the low and middle tertiles of global and rigid restraint consuming approximately 10–15% less in the highdose verses low-dose conditions (data not shown). We compared subjects’ baseline characteristics between tertile groupings of flexible, rigid and global restraint using analysis of variance. No significant differences between tertiles were found for binge eating, disinhibition, hunger, age, BMI, Beck depression score or Eating Attitudes Test score for any of the restraint groupings (data not shown). Subjective satiety No effect of condition or interaction between condition and time was found for ratings of hunger, prospective consumption, fullness or appetite score over the morning or immediately before and after lunch and dinner. Similarly, no effects were found for ratings taken immediately before and after consumption of the capsules before lunch and dinner.

Table 3 Energy intake (kJ) by condition and dietary restraint statusa Tertile

Scores b

Low-dose

Interaction of flexible restraint and condition (P = 0.03) 1 10 <3 2 7 3 3 15 4–6

9119  738 10 022  882 9535  602

10 098  738 8987  882 9946  602

9462  738 9989  882 9768  602

0.06 0.06 0.46

Interaction of rigid restraint and condition (P = 0.009) 1 9 <2 2 13 2, 3 3 10 4–6

7658  680d,e 11 250  565 8919  645

9136  680d 10 749  565 9113  645

8560  680e 10 787  565 9379  645

0.003 0.28 0.51

Interaction of global restraint and condition (P = 0.06) 1 10 <8 2 9 8–10 3 13 11–19

8246  707d,e 10 874  745 9541  620

9519  707d 10 407  745 9556  620

9031  707e 10 557  745 9673  620

0.01 0.55 0.92

a

c d e

Control

P-valuec

High-dose

b

n

Data shown are LSMEANS  S.E.M. Includes energy intake at breakfast, lunch, dinner, evening snack and 335 kJ from L-Phe or dextrose. Scores on restraint subscales of the Eating Inventory. For results of SLICE procedure to test for effects of condition within tertiles. Significant difference between LSMEANS (adjusted P’s < 0.004). Trend for difference between LSMEANS (adjusted 0.05 < P’s < 0.10).

R.J. Pohle-Krauza et al. / Appetite 51 (2008) 111–119

115

Table 4 Ratings of subjective satiety and nausea following lunch (mm  4 h) by conditiona High-dose Hunger Fullness Prospective consumption Appetite scoree Nausea a b c d e

Low-dose

58.6  8.1c 75.0  11.1d 59.6  7.4c

95.5  8.0c 107.6  10.9d 94.4  7.4c

63.4  7.8c 53.8  9.1c,d

99.6  7.8c 23.8  9.0d

Control

P-value b

77.4  8.2 0.007 91.4  11.1 0.12 76.4  7.4 0.005 81.5  7.9 7.4  9.1 d

0.006 0.003

Data shown are LSMEANS  S.E.M. For effect of condition. Significant difference between LSMEANS (adjusted P’s < 0.05). Trend for difference between LSMEANS (adjusted 0.05 < P’s < 0.10). Appetite score = ((hunger + prospective consumption) + (100 fullness))/3.

Significant effects of condition were found for each of the ratings (all P’s < 0.0001) over the afternoon. When expressed as 4-h AUC values, significant effects of condition were found for hunger, prospective consumption and appetite score (Table 4). The effect of condition on subjective satiety over the afternoon differed as a function of time. A significant interaction of time and condition was found for appetite score (P = 0.047) and trends were found for prospective consumption and hunger (P’s < 0.10). The effect of condition was more pronounced later in the afternoon with significant effects of condition on appetite score at two, three and four hours after lunch (Fig. 1). No interactions with the restraint subscales (global, flexible, or rigid) were found for the effects of condition or condition  time over the afternoon, at any time point over the day or for the 4-h AUC measures on any of the ratings of subjective satiety. Significant effects of condition, time and their interaction were found for ratings of nausea. Subjects reported higher

Fig. 1. Appetite score over the day (LSMEANS). Significant interaction of time and condition was found in mixed model analysis of variance (P = 0.047). Using the SLICE command, significant effects of condition were found at 2 h (P = 0.01), 3 h (P < 0.0001) and 4 h (P = 0.004) after lunch. (@) Indicates significant difference between high-dose and low-dose conditions (adjusted P < 0.05). ($) Indicates significant difference between high-dose and control conditions. (#) Indicates a trend for a significant difference between high-dose and control conditions (adjusted P = 0.06).

Fig. 2. Ratings of nausea over the day (LSMEANS). Significant effects of condition (P < 0.0001), time (P < 0.0001) and condition  time (P < 0.0001) in mixed model analysis of variance. Using the SLICE command, significant effects of condition were found at 1, 2, 3 and 4 h after lunch. (@) Indicates significant difference between high-dose and low-dose conditions (adjusted P < 0.05). ($) Indicates significant difference between high-dose and control conditions.

nausea (11.3  1.5 mm) in the high-dose versus low-dose (6.4  1.5 mm; P < 0.0001) and control (5.5  1.5 mm; P < 0.0001) conditions whereas no difference between lowdose and control was found (P = 0.28). The effects of condition were significant at 1, 2, 3 and 4 h after lunch (Fig. 2). Relationships between variables We conducted exploratory data analyses to further examine relationships between restraint, nausea and energy intake at lunch, dinner and evening snack. No effects of global or flexible restraint were found for ratings of nausea. We found significant effects of condition (P < 0.0001), rigid restraint (P = 0.047) and the interaction of rigid restraint and condition (P = 0.007) on 4-h nausea ratings. Post-hoc analyses using the Slice procedure showed significant effects of condition for subjects classified in the low (P = 0.005) and middle (P = 0.02) tertiles of restraint but not in the highest tertile (P = 0.59). Ratings of nausea were consistently low in the control condition for all three rigid-restraint groups and ranged from 6.3 to 8.3 mm  4 h. In the low- and middle-rigid restraint groups there was a substantial increase (75 mm  4 h, on average) in nausea in the high-dose compared to control conditions (adjusted P’s < 0.05), whereas no significant increase (8 mm  4 h) was found in the highest-rigid restraint group. To further evaluate the possible confounding effects of nausea on energy intake, we classified subjects according to their nausea ratings and entered nausea as a factor in the mixed model analysis. We choose to split subjects into two groups based on the distribution of nausea ratings with approximately half of the subjects (n = 17) reporting low nausea (from 0 to 19 mm  4 h) and the remainder (n = 14) reporting higher ratings (from 26 to 249 mm  4 h) in the high-dose condition. Interestingly, significant three-way interactions of nausea group, condition and rigid restraint group were found for

116

R.J. Pohle-Krauza et al. / Appetite 51 (2008) 111–119

energy intake at dinner and evening snack (P’s < 0.05). Posthoc testing showed that reduced energy intake in the high-dose condition versus control was evident only in the low-nausea, low-rigid restraint group. However, due to the small sample sizes in some of the groups (e.g. n = 3 for the low-nausea, lowrigid group and n = 2 for the high-nausea, low-rigid group), the results of the three-way interaction will not be reported in detail here. Rather, we examined the effects of condition within nausea groups, with data collapsed across rigid-restraint groups. In the low-nausea group, energy intake was reduced by high-dose Phe (dinner 2445  239 and 3065  239 kJ; snack 1557  234 and 2052  234 kJ, for high-dose and control, respectively, adjusted P’s < 0.05). No effect was found in the low-dose condition (dinner 2801  239, snack 1892  234 kJ). In the high-nausea group, no differences in energy intake were found between conditions (dinner 3014  281, 3387  281, 3220  281 kJ; snack 1553  276, 1574  276, 1495  276 kJ for high-dose, low-dose and control conditions, respectively). Discussion The results show no effect of Phe on energy intake in overweight and obese women when it is ingested in capsule form 15 min before test meals. Intake at dinner was approximately 8% less in the high-dose versus low-dose and control conditions but this result was not statistically significant. The lack of effect may be attributed in part to the short time from dose ingestion to intake of test meals. In one report, comparing data from two studies involving 13 men, intake at a buffet-style lunch was suppressed by 31% following ingestion of encapsulated Phe (10.08 g) taken 1.75 h before lunch versus a low-dose (0.84 g) taken one hour before lunch (Rogers & Blundell, 1994; Ryan-Harshman et al., 1987). In another, involving four men and two women, a 20-min interval was used, however, doses were administered via a nasogastric tube allowing for more rapid absorption. Intake at a buffet-style lunch was suppressed by 31% by Phe versus control and 27% versus D-Phe (Ballinger & Clark, 1994). Significant results from another study, however, suggest that the shorter time interval does not account for the discrepant results. In a study of eight men (reported in abstract form only) ingestion of encapsulated Phe (10 g) versus a maltose control (10 g) with 500 g of tomato soup 20 min before a macaroni-and-beef lunch meal reduced energy intake by 17% (Muurahainen et al., 1988). Alternatively, the discrepancy may have been due to sex differences as our study included only women whereas previous studies included mostly men. It is possible that short-term effects of Phe on energy intake are more pronounced in men than women. It is also possible that the discrepancy may be due to differences in behavioral characteristics of the subjects. We found a trend for an interaction between response to Phe and global restraint scores, and significant interactions with the subscale scores (flexible and rigid) with significant effects noted only for women in the lowest tertile of rigid restraint. Dietary restraint, first introduced by Herman and colleagues (Herman & Mack, 1975), is defined as the tendency to restrict

energy intake in order to control body weight. Herman and Polivy postulated that restrained eaters develop anomalous eating patterns (Herman & Polivy, 1980). Recent experimental evidence suggests that restrained eating is associated with aberrant satiety responses. Individuals with high dietary restraint have been shown to report less hunger (Burley, Leeds, & Blundell, 1987; Smith et al., 1998), be less sensitive to the satiety value of dietary fat (Rolls et al., 1994), more responsive to external cues (Ogden & Wardle, 1990), and less responsive to food palatability (Yeomans, Tovey, Tinley, & Haynes, 2004). Three self-rating questionnaires have been developed to assess dietary restraint (Herman & Polivy, 1975; Stunkard & Messick, 1985; van Strien, Fritjers, Bergers, & Defares, 1986). We choose the scale developed by Stunkard and Messick (Stunkard & Messick, 1985) that delineates two subtypes of dietary restraint – rigid restraint, characterized by an all-ornothing approach to dieting and weight loss that is often maladaptive, or flexible restraint, characterized by a more graduated, adaptive approach. Results from two studies show that higher scores for rigid restraint were correlated with higher BMI and less success in weight reduction whereas higher scores for flexible restraint were associated with lower BMI and greater success in weight reduction (Westenhoefer, Stunkard, & Pudel, 1999). We found significant effects for women in the lower tertile of rigid restraint suggesting that subjects with higher scores were insensitive to the effects of Phe. In the lowest tertile group, energy intake over the day was reduced by 11% in the high-dose versus control condition. We detected similar effects in a subsequent study (Pohle-Krauza et al., in press). Treatments (10 g Phe and control) were administered over phases of the menstrual cycle with the expectation that the satiety effects of Phe would be evident in the follicular but not the luteal phase. Unexpectedly, three-way interactions with rigid restraint status were found. A 15% reduction in daily intake in the Phe versus control condition was noted in the follicular phase as expected, but only for women in the lower 50th percentile of rigid restraint (Pohle-Krauza et al., in press). Thus, in two studies we found that the effects of Phe were modulated by rigid restraint in women. These results, although consistent, should be interpreted with some caution due to the post-hoc nature of the findings. Further studies are needed to replicate these findings and to determine cutoff values that define high and low rigid restraint. Our strategy involved use of continuous covariates and groupings based on 50th percentile and tertile splits. We choose multiple strategies as normative values defining high and low values are not available and linearity cannot be assumed. For example, scores on the rigid subscale of the restraint scale were highly skewed (skewness = 0.45) in 53 women who participated in two of our studies (Pelkman et al., 2007; Pohle-Krauza et al., in press). Furthermore, in both studies dietary restraint and subscales of the restraint scale were significant covariates when tested as binary but not continuous variables. This suggests that some cut points exist that defines a group of women with aberrant satiety responses. We found trends for interactions between Phe treatment and scores on both the flexible and global restraint scales. This is not

R.J. Pohle-Krauza et al. / Appetite 51 (2008) 111–119

surprising as these scales can be highly correlated within individuals. In our sample, the correlations between subscales were 0.86 (global and flexible), 0.86 (global and rigid) and 0.64 (flexible and rigid). These high correlations suggest that distinguishing between types of restraint may not be possible in our sample. We used the short version of the Eating Inventory that includes seven items each for the rigid and flexible subscales (Stunkard & Messick, 1985). The longer version of the Eating Inventory [12 items for flexible restraint and 16 items for rigid restraint (Westenhoefer et al., 1999)], has been shown to have greater reliability to classify subjects into subcategories and should be used to further examine relationships between dietary restraint and the satiety response. The endogenous release of CCK is the proposed mechanism for the effects of Phe on energy intake. Ballinger and Clark demonstrated that suppression of energy intake coincided with the peak, five-fold increase in plasma CCK concentrations observed 20 min following nasogastric infusion of 10 g of Phe (Ballinger & Clark, 1994). Burton–Freeman demonstrated that men and women with high dietary restraint (defined using the global restraint scale of the Eating Inventory) have a blunted CCK response over a 45-min interval following ingestion of a fat preload (Burton-Freeman, 2005). This suggests that the lack of an effect that we found for women with high rigid restraint may be due to an aberrant physiological response to Phe. Further research is needed to determine if the CCK response to Phe is blunted in restrained women (or men) and if this response is more apparent using the rigid versus flexible subscales of the Eating Inventory. In contrast to previous findings demonstrating a step-wise suppression of energy intake with doses of 5 and 10 g of Phe (Rogers & Blundell, 1994; Ryan-Harshman et al., 1987), our results did not clearly support a dose–response effect. In the lowrigid restraint subjects the pattern of the results suggests that the effects of high-dose Phe (10 g) were more pronounced when compared to low-dose rather than control. For example, highdose Phe reduced energy intake by 16% compared to low-dose but only 11% compared to control. It is possible that use of glucose in our formulations affected energy intake. Ingestion of a small amount of glucose (9 g) reduced the glycemic response to an oral glucose tolerance test by 41% over a 40-min period in one study (Ezatagha et al., 2006). High glycemia within 1 h after consumption of a preload has been shown to reduce energy intake (Anderson et al., 2002) thus it is possible that the small amount of glucose that we used in our low-dose (5 g) and control conditions (10 g) reduced the glycemic response to lunch (and dinner) and hence, increased intake relative to the high-Phe condition. Nuttall and colleagues demonstrated that glucose and Phe have additive effects on the insulinemic and glycemic response (Nuttall, Schweim, & Gannon, 2006). In their study, the peak insulinemic response that occurred 30 min post-ingestion was 35% higher for glucose (25 g) plus Phe (9.3 g, on average) than for either substance ingested alone and resulted in a more rapid fall in blood glucose over the next hour. These effects suggest that relatively small amounts of Phe and glucose may affect postprandial metabolism in ways that affect energy intake. Further work is needed to explore these outcomes in multifactorial experiments.

117

We found that Phe affected subjective satiety ratings over several hours after lunch with no evidence of a dose response and ordering of the effects from greatest in the high-dose condition to least in the low-dose condition, with control in between. As cited above, a more precipitous fall in blood glucose was elicited by a combination of Phe and glucose relative to either ingested alone. Falling blood glucose is associated with an earlier return of hunger and meal initiation (Campfield & Smith, 2003) thus our findings are consistent with the hypothesis that post-meal satiety effects of Phe maybe mediated in part by effects on glycemia. Furthermore, our results suggest that this effect is not influenced by dietary restraint, in contrast to effects on food intake. This implies that dietary restraint affects some, but not all of the mechanisms engaged by ingestion of Phe. Further studies are needed to explore this hypothesis. High-dose Phe was associated with increases in ratings of nausea over the afternoon. Previous studies report no effects of Phe on nausea (Ballinger & Clark, 1994; Muurahainen et al., 1988; Rogers & Blundell, 1994; Ryan-Harshman et al., 1987). However, it is unclear if nausea was assessed in these studies as no data were reported. Furthermore, only short-term effects were assessed in these studies. Our results show a transient increase in nausea, peaking at two hours, and returning to baseline before dinner. The peak increase was moderate with a mean value of 28 on a 100-mm scale. Interestingly, the effect of high-dose Phe on nausea was evident only for women in the low and middle tertiles of rigid restraint. CCK has been associated with induction of malaise and aversion in animals (Ervin, Mosher, Birkemo, & Johnson, 1995; Mosher, Birkemo, Johnson, & Ervin, 1998; Swerdlow, van der Kooy, Koob, & Wenger, 1983) and in humans. In a study of 15 men, Greenough and colleagues showed that infusion of CCK induced gastrointestinal symptoms, such as nausea, indigestion and stomach ache, in eight subjects but that energy intake was equally suppressed 10 min following the start of CCK infusion for both symptomatic (57% reduction) and non-symptomatic (45% reduction) individuals (Greenough, Cole, Lewis, Lockton, & Blundell, 1998). We found suppression of energy intake by high-dose Phe in subjects who reported low nausea ratings whereas no effect was found in subjects who reported high nausea ratings. However, we also detected a three-way interaction involving rigid restraint, nausea and condition. Only unrestrained men were tested by Greenough and colleagues, thus direct comparisons between studies is not possible. Nonetheless, our findings are consistent in that the effects we observed on energy intake were not dependent on the induction of nausea. Small sample sizes hamper our ability to definitely conclude that rigid restraint status affects the nausea response to Phe. High dietary restraint has been associated with a blunted CCK response (Burton-Freeman, 2005), thus, our finding showing no effect of high-dose Phe on nausea in highrigid restraint women is consistent with the hypothesis that rigid restraint is associated with an aberrant CCK response that is manifest in a failure to suppress energy intake or experience nausea several hours after ingestion of Phe. In conclusion, our findings support the need for further study of the effects of rigid restraint status on satiety in experiments

118

R.J. Pohle-Krauza et al. / Appetite 51 (2008) 111–119

specifically designed to test for interactive effects. As Phe was ineffective in the total group, our findings suggest that demonstrating the effectiveness of foods or nutrients to enhance satiety may be hampered if heterogeneous groups of women are tested. Understanding how behavioral characteristics related to eating behavior, such as rigid restraint, affect the satiety response to dietary macronutrients is vitally important for the development of effective dietary strategies to enhance satiety and control body weight. References Anderson, G. H., Catherine, N. L., Woodend, D. M., & Wolever, T. M. (2002). Inverse association between the effect of carbohydrates on blood glucose and subsequent short-term food intake in young men. American Journal of Clinical Nutrition, 76(5), 1023–1030. Ballinger, A. B., & Clark, M. L. (1994). L-Phenylalanine releases cholecystokinin (CCK) and is associated with reduced food intake in humans: evidence for a physiological role of CCK in control of eating. Metabolism, 43(6), 735–738. Barkeling, B., Rossner, S., & Bjorvell, H. (1990). Effects of a high-protein meal (meat) and a high-carbohydrate meal (vegetarian) on satiety measured by automated computerized monitoring of subsequent food intake, motivation to eat and food preferences. International Journal of Obesity, 14(9), 743–751. Beck, A. T., & Beamesderfer, A. (1974). Assessment of depression: The depression inventory. Modern Problems in Pharmacopsychiatry, 7, 151– 169. Burley, V. J., Leeds, A. R., & Blundell, J. E. (1987). The effect of high and lowfibre breakfasts on hunger, satiety and food intake in a subsequent meal. International Journal of Obesity, 11(Suppl. 1), 87–93. Burton-Freeman, B. (2005). Sex and cognitive dietary restraint influence cholecystokinin release and satiety in response to preloads varying in fatty acid composition and content. Journal of Nutrition, 135(6), 1407–1414. Butler, R. N., Davies, M., Gehling, N. J., & Grant, A. K. (1981). The effect preloads of amino acid on short-term satiety. American Journal of Clinical Nutrition, 34(10), 2045–2047. Campfield, L. A., & Smith, F. J. (2003). Blood glucose dynamics and control of meal initiation: a pattern detection and recognition theory. Physiological Reviews, 83(1), 25–58. de Castro, J. M. (1987). Macronutrient relationships with meal patterns and mood in the spontaneous feeding behavior of humans. Physiology & Behavior, 39(5), 561–569. Drapeau, V., King, N., Hetherington, M., Doucet, E., Blundell, J., & Tremblay, A. (2007). Appetite sensations and satiety quotient: Predictors of energy intake and weight loss. Appetite, 48(2), 159–166. Ervin, G. N., Mosher, J. T., Birkemo, L. S., & Johnson, M. F. (1995). Multiple, small doses of cholecystokinin octapeptide are more efficacious at inducing taste aversion conditioning than single, large doses. Peptides, 16(3), 539– 545. Ezatagha, A., Dascalu, A., Sievenpiper, J. L., Jovanovski, E., Panahi, S., Jenkins, A., et al. (2006). Do all placebos fit the definition of a ‘‘placebo’’? The variation in glycemic response of different placebos in healthy individuals. FASEB Journal, 20(4), A580. Faith, M. S., Wong, F. Y., & Alison, D. B. (1998). Demand characteristics of the research setting can influence indexes of negative affect-induced eating in obese individuals. Obesity Research, 6(2), 134–136. Flint, A., Raben, A., Blundell, J. E., & Astrup, A. (2000). Reproducibility, power and validity of visual analogue scales in assessment of appetite sensations in single test meal studies. International Journal of Obesity, 24(1), 38–48. Garner, D. M., Olmsted, M. P., Bohr, Y., & Garfinkel, P. E. (1982). The eating attitudes test: Psychometric features and clinical correlates. Psychololgical Medicine, 12(4), 871–878. Geliebter, A., Gluck, M. E., & Hashim, S. A. (2005). Plasma ghrelin concentrations are lower in binge-eating disorder. Journal of Nutrition, 135(5), 1326–1330.

Gibbs, J., & Smith, G. P. (1977). Cholecystokinin and satiety in rats and rhesus monkeys. American Journal of Clinical Nutrition, 30(5), 758–761. Gormally, J., Black, S., Daston, S., & Rardin, D. (1982). The assessment of binge eating severity among obese persons. Addictive Behaviors, 7(1), 47–55. Greenough, A., Cole, G., Lewis, J., Lockton, A., & Blundell, J. (1998). Untangling the effects of hunger, anxiety, and nausea on energy intake during intravenous cholecystokinin octapeptide (CCK-8) infusion. Physiology & Behavior, 65(2), 303–310. Halton, T. L., & Hu, F. B. (2004). The effects of high protein diets on thermogenesis, satiety and weight loss: as critical review. Journal of the American College of Nutrition, 23(5), 373–385. Herman, C. P., & Mack, D. (1975). Restrained and unrestrained eating. Journal of Personality, 43(4), 647–660. Herman, C. P., & Polivy, J. (1975). Anxiety, restraint, and eating behavior. Journal of Abnormal Psychology, 84(6), 66–72. Herman, C. P., & Polivy, J. (1980). Restrained eating. In A. J. Stunkard (Ed.), Obesity. Philadelphia: Saunders. Konturek, S. J., Radecki, T., Thor, P., & Dembinski, A. (1973). Release of cholecystokinin by amino acids. Proceeding of the Society for Experimental and Biological Medicine, 143(2), 305–309. Mathur, R., & Manchanda, S. K. (1991). Ontogeny of phenylalanine (endogenous cholecystokinin) induced modulation of food intake in normal and undernourished rats. Progress in Neuro-Psychopharmacology & Biological Psychiatry, 15(3), 405–413. Mokdad, A. H., Bowman, B. A., Ford, E. S., Vinicor, F., Marks, J. S., & Koplan, J. P. (2001). The continuing epidemics of obesity and diabetes in the United States. JAMA, 286(10), 1195–1200. Mosher, J. T., Birkemo, L. S., Johnson, M. F., & Ervin, G. N. (1998). Sulfated cholecystokinin (26–33) induces mild taste aversion conditioning in rats when administered by three different routes. Peptides, 19(5), 849–857. Muurahainen, N. E., Kissileff, H. R., & Pi-Sunyer, F. X. (1988). L-Phenylalanine and L-Tryptophan reduce food intake when given with a food preload (abstract). American Journal of Clinical Nutrition, 47, 774. Nuttall, F. Q., Schweim, K. J., & Gannon, M. C. (2006). Effect of orally administered phenylalanine with and without glucose on insulin, glucagon and glucose concentrations. Hormones and Metabolism Research, 38(8), 518–523. Ogden, J., & Wardle, J. (1990). Cognitive restraint and sensitivity to cues for hunger and satiety. Physiology & Behavior, 47(3), 477–481. Parker, B. A., Sturm, K., MacIntosh, C. G., Feinle, C., Horowitz, M., & Chapman, I. M. (2004). Relation between food intake and visual analogue scale ratings of appetite and other sensations in healthy older and young subjects. European Journal of Clinical Nutrition, 58(2), 212–218. Pelkman, C. L., Navia, J. L., Miller, A. E., & Pohle, R. J. (2007). Novel calciumgelled, alginate-pectin beverage reduced energy intake in nondieting overweight and obese women: Interactions with dietary restraint status. American Journal of Clinical Nutrition, 86(6), 1595–1602. Pennington, J. A. T. (1998). Bowes and Church’s food values of portions commonly used (17 ed.). Philadelphia: Lippincott Williams & Wilkins. Plan and operation of the Third National Health and Nutrition Examination Survey, 1988–94. Series 1: programs and collection procedures. (1994). Vital Health Stat, 1(32), 1–407. Pohle-Krauza, R. J., Carey, K. H., & Pelkman, C. L. (in press). Dietary restraint and menstrual cycle phase modulated L-phenylalanine-induced satiety. Physiology & Behavior. Rogers, P. J., & Blundell, J. E. (1994). Reanalysis of the effects of phenylalanine, alanine, and aspartame on food intake in human subjects. Physiology & Behavior, 56(2), 247–250. Rolls, B. J., Kim-Harris, S., Fischman, M. W., Foltin, R. W., Moran, T. H., & Stoner, S. A. (1994). Satiety after preloads with different amounts of fat and carbohydrate: implications for obesity. American Journal of Clinical Nutrition, 60(4), 476–487. Rolls, B. J., Shide, D. J., Thorwart, M. L., & Ulbrecht, J. S. (1998). Sibutramine reduces food intake in non-dieting women with obesity. Obesity Research, 6(1), 1–11. Ryan-Harshman, M., Leiter, L. A., & Anderson, G. H. (1987). Phenylalanine and aspartame fail to alter feeding behavior, mood and arousal in men. Physiology & Behavior, 39(2), 247–253.

R.J. Pohle-Krauza et al. / Appetite 51 (2008) 111–119 Smith, C. F., Geiselman, P. J., Williamson, D. A., Champagne, C. M., Bray, G. A., & Ryan, D. H. (1998). Association of dietary restraint and disinhibition with eating behavior, body mass, and hunger. Eating and Weight Disorders, 3(1), 7–15. Stubbs, R. J., Hughes, D. A., Johnstone, A. M., Rowley, E., Reid, C., Elia, M., et al. (2000). The use of visual analogue scales to assess motivation to eat in human subjects: a review of their reliability and validity with an evaluation of new hand-held computerized systems for temporal tracking of appetite ratings. British Journal of Nutrition, 84(4), 405– 415. Stubbs, R. J., van Wyk, M. C., Johnstone, A. M., & Harbron, C. G. (1996). Breakfasts high in protein, fat or carbohydrate: effect on within-day appetite and energy balance. European Journal of Clinical Nutrition, 50(7), 409– 417. Stunkard, A. J., & Messick, S. (1985). The three-factor eating questionnaire to measure dietary restraint, disinhibition and hunger. Journal of Psychosomatic Research, 29(1), 71–83.

119

Swerdlow, N. R., van der Kooy, D., Koob, G. F., & Wenger, J. R. (1983). Cholecystokinin produces conditioned place-aversions, not place-preferences, in food-deprived rats: evidence against involvement in satiety. Life Sciences, 32(18), 2087–2093. van Strien, T., Fritjers, J. E. R., Bergers, G. P., & Defares, P. B. (1986). The Dutch Eating Behavior Questionnaire (DEBQ) for assessment of restrained, emotional and external eating behavior. International Journal of Eating Disorders, 5(2), 295–315. Westenhoefer, J., Stunkard, A. J., & Pudel, V. (1999). Validation of the flexible and rigid control dimensions of dietary restraint. International Journal of Eating Disorders, 26(1), 53–64. Westerterp-Plantenga, M. S., & Lejeune, M. P. (2005). Protein intake and bodyweight regulation. Appetite, 45(2), 187–190. Yeomans, M. R., Tovey, H. M., Tinley, E. M., & Haynes, C. J. (2004). Effects of manipulated palatability on appetite depend on restraint and disinhibition scores from the Three-Factor Eating Questionnaire. International Journal of Obesity, 28(1), 144–151.