International Journal of (1998) 22, 1084±1087 ß 1998 Stockton Press All rights reserved 0307±0565/98 $12.00 http://www.stockton-press.co.uk/ijo Association of and -satiety ratings in obese women

AF Heini1, C Lara-Castro2, KA Kirk3, RV Considine4, JF Caro4 and RL Weinsier*2

1Clinic of Internal Medicine, Inselspital, University of Berne, Switzerland; 2Department of Sciences and 3Biostatistics, University of Alabama at Birmingham, USA; and 4Department of Medicine, Jefferson Medical College of Thomas Jefferson University, Philadelphia, USA

OBJECTIVE: To measure leptin, and (CCK) concentrations in obese women on and to determine their correlation with hunger-satiety ratings. Although it has been proposed to play a role in regulation, the effects of physiological concentrations of these on hunger-satiety in humans have not yet been well established. DESIGN: Prospective metabolic study. A two week `wash-in period' followed by a three-week observation period, during which each subject underwent six measurements of satiety, blood parameters and body weight. SETTING: Energy Metabolism Research Unit, Department of Nutrition Sciences, University of Alabama at Birming- ham, Birmingham, Alabama, USA. SUBJECTS: 22 moderately to severely overweight women (mean age: 45Æ 8 y; body mass index (BMI): 33Æ 6kg=m2). INTERVENTION: Energy restriction, in the form of a 3.3 MJ (800 kcal) during ®ve weeks. MAIN OUTCOME MEASUREMENTS: blood levels of leptin, insulin, glucose and CCK, fasting hunger-satiety scores and body weight. RESULTS: The mean (Æ s.d.) fasting serum leptin concentration at the beginning of the observation period was 26.1Æ 15.9 ng=ml (range: 6.7±59.8 ng=ml). Leptin concentrations correlated positively with body weight (P < 0.0001). Furthermore, reductions in body weight were associated with decreases in fasting leptin levels (P ˆ 0.002). Leptin concentrations correlated with serum levels of insulin (P ˆ 0.0001) and CCK (P ˆ 0.06), but in multivariate analysis including insulin, CCK and glucose, only leptin had a signi®cant relationship with satiety (P ˆ 0.04). This relationship was linear. CONCLUSIONS: These results con®rm the association between leptin levels, body weight and serum insulin. We also showed that higher serum leptin levels correlated with greater feelings of fullness, a relationship which was not blunted in the more obese subjects. These ®ndings suggest that leptin is a satiety that reduces appetite, even in obese individuals, and that must be due to other factors, overriding this feed-back regulation.

Keywords: obesity; hormone; satiety; appetite regulation; leptin; cholecystokinin; insulin; body weight

Introduction hormone in humans and particularly in obese indivi- duals. For, contrary to speculations that subnormal leptin synthesis may cause obesity in humans, fasting Among hormones suggested to be involved in appetite serum leptin levels have been found to be higher with regulation (that is, cholecystokinin (CCK), insulin and increasing degrees of obesity.9,10 This ®nding has led Y), leptin, the gene product of the investigators to suggest that obesity is due to resis- recently discovered OB gene,1 represents a new can- tance to the action of leptin, rather than to mutations didate. Recent ®ndings which support the potential of the OB gene, which has not yet been shown in central action of leptin as a satiety hormone, include humans. Studying the relationship between leptin and the suppression of intake in mice given leptin satiety is complex, since leptin does not show an acute infusions,2±4 the presence of leptin in human cere- response to ,10,11 but seems to undergo changes brospinal ¯uid,5,6 the identi®cation of leptin receptors after prolonged changes of energy balance.11±13 Our in the hypothalamus7 and the suppressive effect of study focused on the relationship between fasting leptin on hypothalamic in mice.8 hormone levels and hunger-satiety ratings in obese Therefore, one of the remaining unanswered ques- women with steady-state negative energy balance tions is, whether indeed, leptin acts as a satiety conditions. Negative energy balance has been pro- posed to be necessary to activate leptin mechanisms, other than its direct relationship with body *Correspondence: Dr Roland L. Weinsier, CE Butterworth Jr. mass.11±13 To our knowledge, this study represents Professor and Chair, Department of Nutrition Sciences, University of Alabama at Birmingham, 1675 University the ®rst report showing a relationship between serum Boulevard, Birmingham, AL 35294, USA. leptin levels and feelings of hunger-satiety in humans. Leptin, a satiety hormone in obese women AF Heini et al 1085 Materials and methods among the studied parameters and their changes over time. RRCM was preferred for most analyses because, while both methods can accommodate miss- Study design ing data for some participants at random time points, The study was carried out at the Energy Metabolism the validity of the strict assumption of parallelism of Research Unit, Department of Nutrition Sciences, lines from subject to subject required to estimate University of Alabama at Birmingham, Birmingham, pooled within-subject slopes by ANCOVA methods, AL, USA. We studied 22 obese women (age range: was more problematic. This model provides an aver- 33±62 y; eight African American and 14 Caucasian) age regression based on the 22 individual regression with a body mass index (BMI) ranging from 28± lines (each comprising six separate measurements of a 51 kg=m2. None had diabetes (according to a two-hour metabolic parameter and of the hunger-satiety rating), glucose tolerance test) or , at the time which form the true basis for this analysis. Where of entry into the study. During ®ve weeks, they appropriate, the estimated ANCOVA pooled within- underwent a tightly controlled weight-loss program, subject correlation coef®cients, are also shown. For in which subjects were provided daily a low-fat, right-skewed distributions, data were log-transformed 3.3 MJ (800 kcal) formula diet (OptifastTM Novartis prior to analyses. The primary analyses were per- Nutrition Corporation, Minneapolis, MN). The for- formed using PROC MIXED of the Statistical Ana- mula provided 50% carbohydrate, 33% protein and lysis System (SAS). 17% fat (food quotient ˆ 0.89), such that each subject received at least 100 g=d carbohydrate, minimizing the likelihood of ketone body formation. Week 1 and Results week 2 were considered a `wash-in' period of adapta- tion to the 3.3 MJ diet. On the ®rst, third and last day of the third and ®fth weeks, the subjects underwent The study subjects' mean age was 45Æ 8 (s.d.) y. At assessments of fasting levels of serum glucose, insulin the beginning of the observation period (week 3) their and leptin, plasma CCK and hunger-satiety ratings body weights averaged 87.9Æ 12.2 kg and body mass between 06.00±08.00 h, after an overnight fast of at index (BMI) 33Æ 6 kg=m2. Fasting serum leptin con- least 10 h. Thus, each of the 22 subjects had a total of centrations averaged 26.1Æ 15.9 ng=ml (range 6.7± six assessments of each parameter in the course of the 59.8 ng=ml). There was a strong positive correlation study. On the last day of the third and ®fth weeks, between fasting serum leptin concentrations and body hunger-satiety ratings were also recorded at 0, 15, 30, weight throughout the observation period (by RRCM: 60 and 120 min after a test , consisting of 1331 kJ P < 0.0001; by ANCOVA: r ˆ 0.42, P < 0.0001). (333 kcal) of the formula diet. Decreases in body weight during weeks 3±5, which The study was approved by the institutional review averaged 2.8Æ 1.4 kg, correlated with decreases in board at the University of Alabama at Birmingham. serum leptin concentrations (RRCM: P ˆ 0.0015; All subjects gave informed consent. ANCOVA: r ˆ 0.43, P < 0.0001). In single-variable models, leptin correlated strongly with serum insulin (P ˆ 0.0001) and glucose (P ˆ 0.002) and weakly with Analyses of metabolic parameters and appetite Serum leptin was measured by a radioimmunoassay CCK (P ˆ 0.056), using the log scale for each para- (RIA), described by Considine et al.14 Plasma CCK meter. Hunger-satiety ratings demonstrated a positive samples were analyzed using a bioassay described by linear relationship with leptin (P ˆ 0.004) and with Liddle et al,15 which is based on the ability of CCK to log10 leptin (P ˆ 0.01); a weak relationship with insulin (P ˆ 0.06), glucose (P ˆ 0.10) and body stimulate amylase release from isolated rat pancreatic weight (P ˆ 0.16); and essentially no relationship acini. Serum glucose was analyzed by the glucose=HK with CCK (P ˆ 0.96). In multivariate analysis, includ- method (Boehringer Mannheim) and insulin by the ing leptin, glucose, insulin, CCK and body weight, AIA-PACK IRI (TOSOH) method. Ratings of hunger- hunger-satiety ratings only correlated signi®cantly satiety were assessed using a numeric category scale previously described by Haber et al16 and previously with leptin concentrations (P ˆ 0.04). Body weight used by us in studies of obese subjects.17 The scale, was included in this analysis with the possibility that which includes reference hedonic phrases at points the relationship of satiety to leptin might have been confounded by the contribution of weight changes to along a numerical spectrum, affords subjects the both these parameters. As shown in Table 1, leptin opportunity to describe their feelings of hunger-satiety contributed signi®cantly to hunger-satiety ratings anywhere in a numerical range from 0 (`painfully (P ˆ 0.04), whereas weight did not (P ˆ 0.86). The hungry') to 100 (`full to nausea'). regression line in Figure 1 describes the signi®cant, positive linear relationship between fasting leptin Statistical analysis levels and hunger-satiety ratings. The data depicted The random regression coef®cient model (RRCM) of in the ®gure, include the scatter in the original data Laird and Ware,18 and analysis of covariance points used as input to the random regression coef®- (ANCOVA), were used to estimate relationships cients analysis (that is, six data points for each of the Leptin, a satiety hormone in obese women AF Heini et al 1086 22 individuals). To simplify the graphical presenta- with weight reduction, are consistent with previously tion, we do not show the 22 individual regression lines reported studies.9,10 The women in our study were all (each comprised of six separate data points), which obese, based on the criterion used by NHANES19 are the true basis for this analysis and for the shown (BMI > 27.3); however, they represented a wide regression line. A large part of the apparent variation range of adiposity, favoring correlation analyses. in Figure 1 is due to different average satiety levels Since the observational period encompassed only among the 22 individuals, rather than to the differ- three weeks and a relatively small average weight ences in their individual regression lines. loss (2.8 kg), it is likely that the mechanism under- In contrast to the above ®ndings of a signi®cant lying the relationship between leptin and body weight correlation of serum leptin to ratings of hunger-satiety is a very sensitive one. Our data also con®rm the in the fasting state, no correlation could be shown previously described relationship between insulin and between fasting leptin levels and subjects' postpran- leptin concentrations in humans.20,21 dial satiety responses to the test meal, measured as New links have been established between leptin and area under the curve (AUC, P ˆ 0.314). its speci®c target genes,22 closing the gap between the mechanism of action of this weight-control hormone and behaviors.23 To our knowledge, however, Discussion no one has demonstrated a relationship between leptin levels and feelings of hunger-satiety in humans, thereby building on available evidence that leptin Our ®ndings of an association for circulating leptin injections reduce food intake in animals2±4 and that levels with body weight and of decreases in leptin leptin receptors are present in high levels in the .7 Among the different instruments Table 1 Five-variable model examining the relative available to assess hunger-satiety ratings, there is contribution of fasting levels of leptin and other study none which is clearly superior.24 The scale described parameters to hunger-satiety ratings among 22 obese women by Haber et al,16 which provides descriptive hedonic EstimateÆ s.e.m. P-value phrases along a numerical scale, was used in this Intercept 7 21.3Æ 50.6 0.68 study, because of our previous experience and success using the instrument to assess hunger-satiety ratings of Parameter Regression coefficient 17 Log10 (Leptin) 21.3Æ 9.6 0.04 lean and obese subjects. In the present study, sub- Log10 (Glucose) 20.9Æ 25.6 0.42 jects were provided with a formula diet of ®xed Log10 (Insulin) 3.8Æ 5.8 0.52 a energy content, throughout the three-week observa- Log10 (CCK ‡ 0.1) 7 2.1Æ 4.3 0.62 Body weight 7 0.05Æ 0.27 0.86 tion period, in order to minimize the potential con- founding effects of food choices and food variety on s.e.m. ˆ standard error of mean; CCK ˆ cholecystokinin. aThe value 0.1 was added because some CCK measurements their ratings of hunger-satiety. In addition, they were were below the limit of detection and resulted in a value of 0.00. not made aware that their ratings were being related to hormone levels or that their ratings might be expected to relate to hormone changes during weight loss. The ®nding of a highly signi®cant relationship between fasting levels of serum leptin and hunger-satiety, in a multivariate analysis including glucose, insulin and CCK, raises the possibility that leptin may play a direct role in weight regulation of humans. Although weight changes may in¯uence the of hunger-satiety, indirectly through an energy de®cit, our analysis suggests that the anorexic effect of leptin occurs independently of weight changes. Inclusion of lean control subjects might have strengthened the analysis of the relationship between leptin and satiety. This was not feasible, since it was our intent to study the effects of energy restriction on the relationship between satiety and leptin levels within individuals, during controlled weight loss. Figure 1 The regression line describes the signi®cant, positive We cannot rule out the possibility that leptin recep- linear relationship between fasting serum leptin levels and tors become more sensitive with the progression of hunger-satiety ratings of 22 obese women, using the random weight loss or that there may be distinct effects of regression coef®cient model. This model provides an average regression based on the 22 individual regression lines (each energy restriction and weight loss on leptin sensitivity. comprised six separate measurements of leptin and appetite), There was a broad distribution of values for each of which form the true basis for this analysis and for the shown these parameters in this group of mildly to severely regression line. For simplicity in the graphic presentation, the ®gure does not show the 22 individual regression lines (see obese women (see Figure 1). Based on this regression Results section for explanation). analysis, which appears to be linear, we found no Leptin, a satiety hormone in obese women AF Heini et al 1087 evidence in this relatively small study that the rela- 4 Pelleymounter MA, Cullen MJ, Baker MB, et al. Effects of the tionship between leptin and satiety was weaker or that obese gene product on body weight regulation in ob=ob mice. Science 1995; 269: 540±543. it reached a plateau at higher body weights. However, 5 Schwartz MW, Peskind E, Raskind M, Boyko EJ, Porte Jr D. to establish the presence or absence of resistance to Cerebrospinal ¯uid leptin levels: relationship to plasma levels leptin in terms of its effect on satiety, will entail larger and to adiposity in humans. Nature Med 1996; 2: 589±593. studies including normal-weight subjects. It is perhaps 6 Caro JF, Kolacznski JW, Nyce MR, et al. Decreased cere- not surprising that fasting leptin levels did not corre- brospinal-¯uid=serum leptin ratio in obesity: a possible mechanism for leptin resistance. Lancet 1996; 348: 159±161. late with postprandial satiety, since postprandial leptin 7 Lee GH, Proenca R, Montez JM, et al. Abnormal splicing of concentrations were found not to increase in previous the leptin in diabetic mice. Nature 1996; 379: studies.10,25 By contrast, we found a signi®cant rela- 632±635. tionship of serum leptin levels to ratings of hunger- 8 Stephens TW, Basinski M, Bristow PK, et al. The role of satiety, when examined over a more sustained period; neuropeptide Y in the antiobesity action of the obese gene product. Nature 1995; 377: 530±532. that is, as repeated measures taken in the postabsorp- 9 Maffai M, Halaas, Ravussin E, et al. Leptin levels in human tive (fasting) state. Since our study provided meals of and rodent: measurement of plasma leptin and ob RNA in ®xed energy content ± for weight loss and for mea- obese and weight-reduced subjects. Nature Med 1995; 1: surement of postprandial satiety ± we cannot rule out 1155±1161. the possibility that higher fasting leptin levels would 10 Considine RV, Sinha MK, Heiman ML, et al. Serum immu- noreactive-leptin concentrations in normal-weight and obese result in reduced voluntary food intake. This, however, humans. N Engl J Med 1996; 334: 292±295. failed to be shown in an experiment by Karhunen et 11 Karhunen L, Haffner S, Lappalainen R, Turpeinen A, Mietti- al,11 comparing fasting leptin levels and the amount of nen H, Uusitupa M. Serum leptin and short-term regulation of food eaten by obese women at energy balance. eating in obese women. Clin Sci 1997; 92: 573±578. 12 Scholz GH, Englaro P, Thiele I et al. Dissociation of serum leptin concentration and body fat content during long term dietary intervention in obese individuals. Horm Metab Res Conclusion 1996; 28: 718±723. 13 Weigle DS, Duell PB, Connor WE, Steiner RA, Soules MR, Kuijper JL. Effect of fasting, refeeding, and dietary fat We con®rmed the previously described relationship restriction on plasma leptin levels. J Clin Endocrinol Metab between leptin, body weight and insulin. In addition, 1997; 82: 561±565. 14 Considine RV, Considine EL, Williams CJ, et al. Evidence these preliminary data support the likelihood that against either a premature or the absence of obese leptin is a satiety hormone which is functioning gene mRNA in human obesity. J Clin Invest 1995; 95: normally to suppress appetite, even at higher body 2986±2988. weights. This ®nding is limited to sustained negative 15 Liddle RA, Gold®ne ID, Rosen MS, et al. Cholecystokinin energy balance and has to be con®rmed in energy- bioactivity in human plasma: molecular forms, responses to feeding and relationship to gallbladder contraction. J Clin balance conditions. If, as these data suggest, leptin Invest 1985; 75: 1144±1152. functions as a satiety hormone and greater degrees of 16 Haber GB, Heaton KW, Murphy D, et al. Depletion and adiposity result in greater leptin-induced satiety, then disruption of dietary ®bre: effects on satiety, plasma-glucose, this feed-back system must not be a primary factor in and serum-insulin. Lancet 1977; 2: 679±682. regulating body weight in humans. That is, other 17 Duncan KH, Bacon JA, Weinsier RL. The effects of high and low energy density diets on satiety, energy intake and eating internal or external factors (for example, food avail- time of obese and nonobese subjects. Am J Clin Nutr 1983; 37: ability, learned eating behaviors) may override leptin- 763±767. induced satiety and cause weight gain. 18 Laird NM, Ware JH. Random effects models for longitudinal data. Biometrics 1982; 38: 963±974. 19 Najjar MF, Rowland M. Anthropometric reference data and Acknowledgement prevalence of overweight, United States, 1976±80. Vital We thank Dr Rodger Liddle and his laboratory for Health Stat 11 1987; No. 238. CCK analysis and Ms Lisa Hughes for dietetic assis- 20 Boden G, Chen X, Mozzoli M, Ryan I. Effect of fasting on serum leptin in normal human subjects. J Clin Endocrinol tance. This study was supported by grants from the Metab 1996; 81: 3419±3423. Swiss National Fond and from Novartis Ltd. 21 Zimmer P, Hodge A, Nicolson M, et al. Serum leptin con- centration, obesity and in Western Samoans: cross sectional study. BMJ 1996; 313: 965±969. References 22 Ghilardi N, Ziegler S, Wiestner A, Stoffel R, Heim MH, 1 Zhang Y, Proenca R, Maffei M, Barone M, Friedman JM. Skoda RC. Defective STAT signaling by the Positional cloning of the mouse obese gene and its human in diabetic mice. Proc Natl Acad Sci USA 1996; 93: homologue. Nature 1994; 372: 425±432. [Erratum, Nature 6231±6235. 1995; 374: 479] 23 Darnell JE Jr. Re¯ections on STAT3, STAT5, and STAT6 as 2 Camp®eld LA, Smith FJ, Guisez Y, Devos R, Burn P. fat STATs. Proc Natl Acad Sci USA 1996; 93: 6221±6224. Recombinant mouse OB protein: evidence for a peripheral 24 Hill AJ, Rogers PJ, Blundell JE. Techniques for the experi- signal linking adiposity and central neural networks. Science mental measurement of human eating behaviour and food 1995; 269: 546±549. intake: a practical guide. Int J Obes 1995; 19: 361±375. 3 Halaas JL, Gajiwala KS, Maffei M, et al. Weight-reducing 25 Sinha MK, Channesian JP, Heiman ML, et al. Nocturnal rise effects of the plasma protein encoded by the obese gene. of leptin in lean, obese and non-insulin-dependent diabetes Science 1995; 269: 243±246. mellitus subjects. J Clin Invest 1996; 97: 1344±1347.