Insulin Sensitivity as a Predictor of Weight Regain

Transcription

1 Insulin Sensitivity as a Predictor of Weight Regain Rena R. Wing Abstract WING RENA R. Insulin sensitivity as a predictor of weight regain. Obes Res. 1997;5:2429. A recent study found that increases in insulin sensitivity following weight loss and stabilization were strongly related to subsequent weight regain. The present paper analyzed this relationship in two behavioral weightloss programs. In the first study, 125 nondiabetic subjects were followed over 30 months; weight losses averaged 10 kg at six months, and subjects had regained 8 kg of their weight loss by their 30month followup. Neither fasting insulin levels at six months nor changes in fasting insulin from zero to six months were related to subsequent weight regain. Similarly, insulin levels measured two hours after a 75 g glucose load were unrelated to subsequent weight regain. The second study followed 33 individuals with Type I1 diabetes, treated with behavior modification, and either a low calorie diet or a very low calorie diet. Weight losses averaged 18 kg at six months, and subjects had regained 10 kg by their 24month followup. The Bergman minimal model was used to assess insulin sensitivity at 6month intervals. Initial analyses suggested that changes in insulin sensitivity from zero to six months were related to subsequent weight regain, but this effect was strongly influenced by an outlier. After removing this individual, there were no significant relationships between the changes in insulin sensitivity that accompanied weight loss and future weight regain. Likewise, insulin sensitivity at 12 months did not predict weight regain from 12 to 24 months. These data do not support the hypothesis that increases in insulin sensitivity with weight loss are associated with subsequent weight regain. Submitted for publication March Accepted for publication July 5, From the Department of Psychiatry, University of Pittsburgh School of Medicine, Pittsburgh, PA. Reprint requests to Dr. Wing. Western Psychiatric Institute and Clinic, 381 I O Hara Street. Pittsburgh, PA Copyright NAASO. Key words: weight regain, insulin sensitivity, NIDDM Introduction Participants in weight loss programs typically lose 10% of their body weight (715 kg) during the initial phase of treatment (1 1). However, approximately 40% of this weight is regained within one year, and by five years participants have, on average, returned to their starting weight (1 1,12). Efforts to enhance longterm outcome in weightloss programs have been largely unsuccessful. A better understanding of the variables associated with weight regain may help in the development of longterm weightcontrol programs. To date, the best predictors of weight regain have been variables associated with adherence to weightcontrol strategies (13). For example, those individuals who continue to selfmonitor their behavior, to exercise, or to avoid high caloriehigh fat foods show better maintenance of weight loss (1,4,7). Initial weight loss is also a predictor of weight regain, with greater weight loss associated with greater weight regain (15). A recent study by Yost et al. (16) suggests that increases in insulin sensitivity which accompany initial weight loss may also predict weight regain. They studied 10 moderately obese women, who lost 11 kg during a 3month weightloss program with a 900 kcal/day liquid diet and 3month weight stabilization. When reexamined at 12 and 18month followups, body weight had returned to baseline. This study found that the increases in insulin sensitivity which occurred after initial weight loss and stabilization were strongly correlated with subsequent weight regain (Spearman correlation of for the association between improvements in insulin sensitivity from zero to six months and weight regain from 6 to 12 or 6 to 18 months, respectively). The association between changes in insulin sensitivity and weight regain remained highly significant after adjusting for the effect of initial weight loss on subsequent weight regain. The goal of the present study was to further examine the question of whether changes in insulin sensitivity in response to weight loss are associated with weight regain. This question was examined in two separate weight loss 24 OBESITY RESEARCH Vol. 5 No. 1 Jan. 1997

2 studies that have previously been published, one involving nondiabetic overweight individuals on a balanced low calorie diet (5) and the other with overweight patients with Type I1 diabetes on a low calorie diet (LCD) or a very low calorie diet (VLCD) (5). Methods and Procedures Study I Subjects. Subjects for this analysis were 125 overweight individuals (65 men; 60 women), between the ages of 25 and 45 (38.7 f 5.0 yrs; Mean f SD) and kg (3070 lbs) over ideal body weight (weight = 90.5 f 9.8 kg; BMI = 30.9 f 2.2). All subjects were healthy nonsmokers with no evidence of hypertension (DBP < 89 mm Hg), not taking any prescription medications or oral contraceptives, and consuming less than three alcoholic beverages per day. Method. All subjects were participants in the TRIM Study, a randomized controlled trial investigating behavioral strategies to enhance weight loss (5). Subjects who were randomly assigned to one of the four treatment groups and completed assessments at baseline.6, 12, 18, and 30 months have been included in the present analysis. Subjects who were randomly assigned to the notreatment control condition in that study have been excluded. The four treatment conditions all participated in a behavioral weight loss program involving weekly treatment meetings for six months and monthly meetings for the following year (between months 6 and 18). Subjects were then restudied at 30 months, after a year with no treatment contact. The treatment program included a calorie restricted diet (100015OO kcal/day) and a recommendation to gradually increase exercise (brisk walking primarily) to 1000 kcall week. The present analyses collapsed across the four treatment conditions, because the outcome at 30 months was comparable across conditions. Assessments were completed at 6month intervals. Subjects were weighed in hospital gowns, without shoes, on a balance beam scale. Insulin levels were assessed after a 12hour fast, and again two hours after a 75 g glucose load. Insulin levels were analyzed with a modification of the immunoassay techniques of Herbert et al. (2). Low, medium, and high control samples were run in each insulin assay with appropriate crossovers. The interassay coefficient of variation is 17% at 4 ku/ml, 9.4% at 36 ku/ml, and 9.6% at 74 ku/ml. The intraassay coefficient of variation is 5.9% at 7 ku/ml and 18% at 17 ku/ml. Fasting insulin levels have been shown to be correlated with glucose clamp measures of insulin sensitivity in nondiabetic subjects (6). Insulin levels were log transformed for analysis. Results. Subjects lost 9.8 f 5.5 kg during the initial 24 weeks of the program. After that time, they gradually regained weight (Table 1). However, even at 30 months subjects still remained slightly, but significantly, below their baseline weight. Fasting insulin also decreased significantly during treatment, with a decrease of 4.4 f 12.0 pu/ml between baseline and 6 months. The lowest levels of fasting insulin, as with weight, were seen at six months. Fasting insulin gradually increased, returning to baseline levels by 30 months. Twohour insulin levels likewise decreased significantly during treatment, and remained significantly below baseline throughout the remainder of the study. Neither fasting insulin levels, changes in fasting insulin from baseline to six months, or changes in 2hour insulin levels were significantly related to weight regain (Table 2). In contrast to the Yost et al. study (16), changes in fasting Table 1. Weight, Fasting, and 2Hour Insulin LevelsStudy 1 (n = 125)* 0 6 Months Weight (kg) Mean ** 82.9** 84.3** 88.1** SD Fasting Insulin (ku/ml) Mean ** 19.8** 19.1** 21.8 SD Hour Insulin (ku/ml) Mean ** 72.1 ** 80.8** SD *n for 2hour insulin = 121. **Differs significantly Q~0.01) from baseline. OBESITY RESEARCH Vol. 5 No. 1 Jan

3 Table 2. Correlation Coefficients for Relationships between Weight Changes and Fasting Insulin LevelsStudy 1 Fasting Insulin 2 Hour Insulin Insulin Levels Insulin Levels Change in Insulin Insulin Levels Insulin Levels Change in Insulin at at from at at from Baseline 6 Months 0 to 6 Months Baseline 6 Months 06 Months Weight Loss 06 Months * * Weight Regain 6 12 Months Months Months *p 0<.05. insulin from zero to six months were unrelated to weight regain from 612, 618, or 630 months with correlations of I' = 0.03 (p = 0.77), r = 0.02 (p = 0.82) and I' = (p = 0.96), respectively. Changes in 2hour insulin did not predict subsequent weight regain. Similar analyses conducted using BMI or using Spearman correlations all remained nonsignificant. In addition, neither changes in insulin from baseline to 12 or baseline to 18 months nor insulin levels at 12 or 18 months predicted later weight regain (data not shown). Analyses done separately for men and women and for those subjects who were above the median in weight loss (>9.6 kg) also showed no significant correlations. Figure 1 shows the lack of association between initial changes in insulin sensitivity and subsequent weight regain. Removing the outliers from this correlation did not affect the results; in all cases the correlations remained nonsignificant. The strongest predictor of weight regain was initial weight loss. Subjects who lost the most weight from zero to six months regained the most from six to 30 months (r = 0.42, p < ). Study 2 The second study assessed the same relationship in a sample of 33 individuals with Type I1 diabetes who were participating in a behavioral weight loss program. The Bergman minimal model was used to assess insulin sensitivity in this cohort of diabetic subjects (14) because fasting insulin levels are not as highly related to insulin sensitivity in diabetic subjects as they are in nondiabetics (6). Subjects. Thirtythree subjects (1 1 men; 22 women) with NIDDM were studied. These subjects averaged 55.8 f 9.2 years of age. Their BMI at baseline was 36.5 f 4.8 (weight f 14.8 kg). Sixtyfour percent were treated with oral medication, 21% with insulin, and 15% with diet only. These 33 subjects were selected for analysis because they had insulin sensitivity assessments at 0, 6, and 12 months and weight data at 0, 6, 12, and 24 months. Methods. All subjects participated in a yearlong behavioral weightloss program (15). The program involved weekly meetings for a full year with instruction in behavior modification, diet, and physical activity. Subjects were randomly assigned to follow a VLCD intermittently ( kcal/day of liquid formula and/or lean meat, fish, and fowl used during weeks 112 and 2436) or to use a balanced low calorie diet (loocr1200 kcal) throughout the year. The 30 c 25 e 5: 20 s f 15 In s E s E 5 6 O I I I I I I I Change In Fasting Insulin ( Bawl1110 to 6 month8 ) Figure 1 : Association between changes in fasting insulin (06 months) and weight regain (630 months)study OBESITY RESEARCH Vol. 5 No. 1 Jan. 1997

4 ~~ Insulin Sensitivity and Weight Regain, Wing data have been analyzed for the two treatment conditions separately and collapsed across conditions. All subjects were removed from their oral medication for one week before assessments of insulin sensitivity and subjects on insulin stopped their insulin for 48 hours. Insulin sensitivity was measured in the General Clinical Research Centre (GCRC), using the Bergman minimal model analysis of frequently sampled intravenous glucose tolerance tests (FSIGTs) modified for use in diabetic patients (14). Baseline samples for insulin and glucose were obtained at 15 and 0. Glucose (300 m ag as dextrose 50 g/dl) was administered within 2 minutes at t = 0. Samples were then obtained at 3,6, 10, 14, 19,22, 25, 30,40, 50,70, 100, 140, and 180 minutes (8). To allow use of this model with obese diabetic subjects, a bolus of insulin (0.05 U/kg) was administered during the FSIGT (at 20 minutes). Insulin levels were assessed by radioimmunoassay and glucose levels were analyzed with a Beckman glucose analyzer. Insulin sensitivity was determined using the MINMOD computer program (copyright RNB) at the University of Southern California s Department of Physiology and Biophysics. Results. Subjects reduced their weight from f 14.8 kg at baseline (Mean f SD) to 85.2 f 12.8,85.7 f 12.2, and 95.1 f 15.3 kg at 6, 12, and 24 months respectively. All later time points were significantly different from baseline. Insulin sensitivity increased significantly from 0.62 f 0.56 at baseline to I.69 f 1.40 at six months and 1.29 f 0.83 at one year. The apparent decrease in sensitivity from 6 to 12 months was not statistically significant. Changes in weight and insulin sensitivity were comparable for the VLCD vs. LCD conditions. As seen in Table 3, weight loss from 012 months was correlated with increases in insulin sensitivity over this time period and with insulin sensitivity levels at 12 months. This effect was not observed with weight loss from zero to six months. When reexamined at two years, subjects had regained 9.7 f 7.6 kg of their initial weight loss (54%). Weight regain was significantly associated with initial weight loss (r = 0.52; p <.001); that is, subjects who lost the most weight during the first year of treatment regained the most during the second year. Initial analyses suggested that changes in insulin sensitivity from zero to six months and insulin sensitivity at six months were related to weight regain from 624 months (r = 0.38, p < 0.02 and r = 0.35, p < 0.05, respectively). However, as shown in Figure 2, these associations resulted from an outlier in the data, who had a change in insulin sensitivity of 6.7, and a 6month insulin sensitivity level of 7.2. After removing this subject from the data, the correlations of changes in insulin sensitivity and 6month sensitivity level with weight regain were reduced to r = 0.20, and r = 0.17, respectively, both insignificant (p values >.25). Removing other extreme values revealed no significant correlations. Similarly there were no significant correlations for the VLCD or LCD condition considered separately after removing the outlier from the data. Insulin sensitivity at 12 months and changes in insulin sensitivity from 0 to 12 months were unrelated to weight regain from 12 to 24 months. Discussion. These studies fail to support the Yost et al. (16) finding that changes in insulin sensitivity that occur with initial weight loss predict subsequent weight regain. In the first study, a large cohort of nondiabetic subjects were followed over 30 months, during which time they experienced almost a 10 kg weight loss (1 1% of initial body weight) and gradual weight regain. Initial weight loss was associated with subsequent weight regain, such that those participants who initially lost the most weight regained the most. However, there was no relation between changes in insulin sensitivity and weight regain. In Study 2, an association was seen between changes in insulin sensitivity from zero to six months (and insulin sensitivity at six months) and subsequent weight regain. How Table 3. Correlation Coefficients for Relationships between Insulin Sensitivity and Weight ChangesStudy 2 (n = 33) Insulin Insulin Change Insulin Insulin Change Insulin Sensitivity at Sensitivity at Sensitivity Sensitivity at Sensitivity 0 Months 6 Months 04 Months 12 Months 012 Months Weight Loss 06 Months Weight Loss 012 Months ** 0.38* Weight Regain 624 Months * 0.38* After Removing Outlier Weight Regain 1224 Months OBESITY RESEARCH Vol. 5 No. 1 Jan

5 20. 0' *. '0. *. I I I I I Change in S, ( B &h to 6 month.) Figure 2: Association between changes in insulin sensitivity (06 months) and weight regain (624 months)study 2. ever, this effect resulted entirely from one outlier and became nonsignificant after removing this subject. Despite maintenance of weight loss from 6 to 12 months, insulin sensitivity at one year was not related to subsequent weight regain. It is difficult to explain why the present data differ from those of Yost et al. (16). In the Yost et al. study insulin sensitivity was measured with a glucose clamp, which is considered to be the gold standard for this measurement. The use of fasting and 2hour insulin levels as indices of insulin sensitivity may have limited the results of Study 1. However, Study 2 utilized the Bergman minimal model; several studies have shown good correlations between this model and glucose clamp measures of insulin sensitivity (8,14). Restricting the current analyses to men or women or to large weight losers did not affect the findings. Because the original paper by Yost is based on only 10 women, replication of their findings is clearly needed to determine whether the original observation is supported, and if so, under what conditions. The finding reported by Yost et a1.that increases in insulin sensitivity following weight loss predict subsequent weight regainmay reflect the fact that with increases in insulin sensitivity, there is a shunting of lipid fuels away from oxidation in muscle toward storage in adipose tissue. However, increased insulin sensitivity will result in weight gain only if accompanied by positive energy balance. Consequently, the amount of weight regain experienced by a participant during the year after weight loss will depend not only on changes in insulin sensitivity, but also on their degree of continued adherence to the low fat, low calorie goals of the program and their longterm adherence to exercise. Numerous studies have shown that longterm changes in these behaviors are related to weight maintenance (1,4,7). In a previous paper from subjects in the TRIM Study (1), we found that weight changes were associated with changes in overall dietary fat intake (and specifically with changes in the number of servings/week of vegetables, beef, hot dogs, and sweets) and with changes in selfreported level of physical activity. These variables accounted for 10% of the variance in BMI change, whereas insulin sensitivity showed no effect. Likewise in Study 2 physical activity at two years was associated with BMI (r = 0.35, p = 0.05) and change in activity from baseline to 2 years was associated with change in BMI from baseline to two years (r = 0.32, p = 0.08). (Data not reported; no measures of dietary intake were obtained at two years in Study 2). Thus behavioral variables may be as important or more important than changes in insulin sensitivity in predicting weight regain. Three previous studies have shown that within a given population, individuals who are the most insulin sensitive at baseline are those who will gain the most weight over the next three to eight years (3,9,10). However, there are several important differences between these three prior studies and the current study and that by Yost et al. (16) of weight regain after successful weight loss. First, in the studies of spontaneous weight gain, the measure of insulin sensitivity is assumed to represent an enduring characteristic of the individual (not an acute response to weight loss). Second, the average weight gain was 0.2 kg1.5 kdyear in the studies of spontaneous weight gain, and a substantial number of individuals within the population lost weight spontaneously. In contrast, in the weight regain studies, almost all subjects gained weight and the rate of regain was much faster (410 kdyear). Finally, the more dramatic rate of weight regain in the latter studies would suggest that there were greater changes in diet and exercise behaviors; these marked changes in behavior may overshadow the effect of differences in insulin sensitivity. Additional studies are needed to determine whether changes in insulin sensitivity with weight loss are related to weight regain, and if so, to determine whether this is because increases in insulin sensitivity make it more difficult to continue to adhere to the low calorie/low fat diet and activity regimen, and/or whether increases in insulin sensitivity and positive energy balance are two independent variables that when combined are most likely to result in weight gain. Acknowledgment This research was supported by funds from NIH grants HL4 1330, DK29757, and NIHNCRRGCRC grant 5MOIRR OBESITY RESEARCH Vol. 5 No. 1 Jan. 1997

6 References 1. Harris JK, French SA, Jeffery RW, McGovern PG, Wing RR. Dietary and physical activity correlates of longterm weight loss. Obes Res. 1994;2(4): Herbert V, Lau K, Gottlieb C, Bleicher S. Coated charcoal immunoassay of insulin. J Clin Endocrinol Metab. 1965;25: Hoag S, Marshall JA, Jones RH, Hamman RF. High fasting insulin levels associated with lower rates of weight gain in persons with normal glucose tolerance: The San Luis Valley Diabetes Study. Int J Obes. 1995;19: Holden JH, Darga LL, Olson SM, Stettner DC, Ardito EA, Lucas CP. Longterm followup of patients attending a combination verylow calorie diet and behaviour therapy weight loss programme. Int J Ohes. 1992;16: Jeffery RW, Wing RR, Thorson C, et al. Strengthening behavioral interventions for weight loss: A randomized trial of food provision and monetary incentives. J Consult Clin Psychol. 1993;61: Laakso M. How good a marker is insulin level for insulin resistance? Am J Epidemiol. 1993; 137: Pronk NP, Wing RR. Physical activity and longterm maintenance of weight loss. Obes Res. 1994;2: Steil GM, Volund A, Kahn SE, Bergman RN. Reduced sample number for calculation of insulin sensitivity and glucose effectiveness from the minimal model. Suitability for use in population studies. Diabetes. 1993;42: Swinburn BA, Nyomba BL, Saad MF, et al. Insulin resis tance associated with lower rates of weight gain in Pima Indians. J Clin Invest. 1991;88: Valdez R, Mitchell BD, Haffner SM, et al. Predictors of weight change in biethnic population. The San Antonio Heart Study. Int J Obes. 1994;18: Wadden TA. The treatment of obesity: an overview. In: Stunkard AJ, Wadden TA, eds. Obesity: Theory and Therapy. New York: Raven Press, Ltd.; 1993: Wadden TA, Sternberg JA, Letizia KA, Stunkard AJ, Foster GD. Treatment of obesity by very low calorie diet, behaviour therapy, and their combination: A fiveyear perspective. Int J Ohes. 1989; 13: Wadden TA, Letizia KA. Predictors of attrition and weight loss in patients treated by moderate and severe caloric restriction. In: Wadden TA, VanItallie TB, eds. Treatment of the seriously obese patient. New York: Guilford Press; 1992: Welch S, Gebhart SSP, Bergman RN, Phillips LS. Minimal model analysis of intravenous glucose tolerance testderived insulin sensitivity in diabetic subjects. J Clin Endocrinol Metab. l990;7 1 : Wing RR, Blair E, Marcus M, Epstein LH, Harvey J. Yearlong weight loss treatment for obese patients with Type I1 diabetes: Does inclusion of an intermittent very low calorie diet improve outcome? Am J Med. 1994;97: Yost TJ, Jensen DR, Eckel RH. Weight regain following sustained weight reduction is predicted by relative insulin sensitivity. Ohes Res. 1995;3: OBESITY RESEARCH Vol. 5 No. 1 Jan