| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Original Studies |
SOS Secretariat (J.S.T., K.S., L.S.), Clinical Metabolic Laboratory (J.S.T., B.C., K.S., E.B., L.S.), and Research Center for Endocrinology and Metabolism (B.C., L.M.S.C.), Department of Medicine, Sahlgrenska University Hospital, 413 45 Göteborg, Sweden
Address all correspondence and requests for reprints to: Lars Sjöström, SOS Secretariat, Vita Str
ket 15, Sahlgrenska University Hospital, 413 45 Göteborg, Sweden. E-mail:
lars.sjostrom{at}medfak.gu.se
 |
Abstract |
|---|
 Top Abstract IntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
|
Introduction |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
It is a notorious problem in the treatment of obesity that, almost regardless of treatment strategy, it is difficult to maintain weight losses long term (4, 5). One reason for this could be that severe caloric restriction may trigger a starvation response, including a neuroendocrine adaptation, that counteracts the therapeutic efforts (6, 7). It could be hypothesized that the obese patients that show the most marked reductions in leptin levels during treatment would be at a greater risk for weight relapse (1, 8).
We conducted a 1-yr dietary intervention in obese subjects, including 16 initial weeks of a very low calorie diet (VLCD). We examined the predictive value of the baseline level and short term changes in serum leptin for the 1-yr weight reduction. Furthermore, the change in serum leptin levels during the VLCD phase was examined in relation to changes in body weight and serum levels of insulin, cortisol, and thyroid hormones.
|
Subjects and Methods |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Patients examined at the Clinical Metabolic Laboratory at
Sahlgrenska University Hospital from January 1996 were invited to
participate in a randomized, 2-yr obesity treatment program if aged
18–60 yr and with a body mass index (BMI) 
30.0
kg/m2. The study was approved by the ethics
committee of the Faculty of Medicine, University of Göteborg, and
all patients gave written informed consent.
The patients were randomized to two treatment groups. However, all subjects started with a 16-week VLCD, using Modifast (NOVARTIS Nutrition, Bern, Switzerland) and were recommended an intake of about 450 kcal/day. After the VLCD period, ordinary food was gradually introduced during a 3-week refeeding phase. All patients were then advised to consume an individualized hypocaloric diet (approximately -500 kcal/day) for up to 2 yr. According to the initial randomization, patients in group I were scheduled to repeat VLCD periods 2 weeks every third month. Patients in group II were to use VLCD whenever the body weight was above an individualized cut-off level, corresponding to the body weight after 16 weeks of VLCD treatment plus 3 kg. The cut-off was lowered after any further weight loss. In case of weight gain, the cut-off level remained unchanged. The results of the 2-yr trial will be reported separately, once finalized.
Twenty-four men and 45 women, free from drug treatment for diabetes or hypothyreosis, who all completed the initial treatment year, were available for the present 1-yr substudy. Among these 69 subjects, there were no differences between the 2 treatment groups in age, sex distribution, baseline levels of BMI and serum leptin, or the weight change during the first year. Thus, for the purpose of this report the treatment groups were not separated.
Measurements
Weight was measured to the nearest 0.1 kg using electronic
scales calibrated monthly. Height was determined to the nearest
0.01 m. The total adipose tissue mass (kilograms) was estimated
using previously published and validated anthropometric equations (9, 10). At baseline and after 8 and 18 weeks the following serum levels
were analyzed: leptin; insulin; cortisol; TSH; free and total
T4 (fT4 and
TT4); and free, total and
rT3 (fT3,
TT3, and rT3). The serum
concentrations of leptin were determined in duplicate using a human
leptin RIA (Linco Research, Inc., St. Charles, MO) at
Research Center for Endocrinology and Metabolism; all other analyses
were performed at the Department of Clinical Chemistry, Sahlgrenska
University Hospital. The laboratory is accredited according to European
norm, EN 45 001. All patients had fasted overnight before blood
sampling. Baseline characteristics of the study patients are shown in
Table 1
.
|
The change in a given variable (
) was calculated by
subtracting the baseline value from a subsequent value (a decline thus
becoming negative). Male sex was coded as 1, and female sex as 2. A
paired t test was used to analyze changes in normally
distributed variables, and for not normally distributed variables the
Wilcoxon signed rank test was used. To compensate for multiple analyses
performed on each variable, the 
level was decreased, so that
was divided by the number of comparisons. Pearson product moment
correlation coefficients (normally distributed variables) or Spearman
rank correlation coefficients (not normally distributed variables) were
calculated. Multiple linear regression analysis and two-way ANOVA
(general linear model) were performed. The Minitab statistical package
was used (Minitab, Inc., State College, PA).
|
Results |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Figure 1 shows the relative change
in body weight in response to caloric restriction. After the VLCD
period, at week 16, the reduction in body weight was 20.3 ± 6.7%
(mean ± SD) in men (P 
0.001) and
15.6 ± 6.9% in women (P 0.001). Between weeks
16–18, body weight remained stable, with an average increase of
0.2 ± 1.8 kg (P = NS). The maintained weight
losses after 48 weeks were 14.4 ± 7.5% (P
0.001) and 11.8 ± 8.4% (P 0.001) in men and
women, respectively.
|
.
Compared to baseline there were highly significant reductions in the
serum levels of leptin and insulin after both 8 and 18 weeks
(P 0.001). The cortisol levels were elevated on the same
occasions, but to a significant extent only in women (P
0.001). The changes in thyroid hormone levels showed similar patterns
in both genders, with significant reductions in the
TT3 levels after both 8 (P 0.001)
and 18 weeks (P 0.001 for women, P
0.05 for men).
|
The baseline leptin levels were not significantly correlated to
the changes in body weight after 48 weeks in either men (r =
-0.09) or women (r = -0.06). However, in a multivariate
regression also taking sex, leptin18,
weight, and height into account (Table 2,
Eq. I, all subjects), weight48 was
positively related to baseline leptin as well as to
leptin18, but was negatively associated with
baseline weight. This model indicated that a large maintained weight
loss after 48 weeks (
weight48) was associated with a
low initial leptin level, a large decline in serum leptin after 18
weeks ( leptin18), and a large body weight at baseline.
Similar results were obtained when the model was applied to men and
women separately (Table 2, Eq. I, men, and Eq. I, women).
|
weight48 was most likely real and only to a
limited extent explained by a regression to the mean error. This is
supported by the fact that when a median-split analysis was performed,
we found patterns similar to those in Eq I (all subjects) among heavier
as well as lighter subjects (not shown).
When baseline levels and changes after 18 weeks in insulin, cortisol,
or the thyroid hormones were added to the model, which was not feasible
to do in men and women separately due to the groups being too small, no
further significant associations were found (not shown). Using the
baseline leptin/BMI ratio in Eq I instead of the separate values for
height, weight, and leptin resulted in similar patterns in both
genders. There were positive and highly significant relations between
weight48 and the leptin/BMI ratio as well as
leptin18 in both men and women (not
shown).
When weight18 was added as a predictor (Eq
II), the explanatory power of baseline leptin, body weight, and
leptin18 disappeared. This was mainly related to
the correlations between weight18, on the one
hand, and leptin18 (r = 0.48 in men and
r = 0.60 in women), on the other. When
weight48 was instead regressed by the 8-week
changes, patterns similar to those in Table 2
were found. However, the
explained variance of the model (Eq I) was reduced to about 28% in men
and 9% in women (not shown).
Baseline body weight and leptin in relation to weight loss
In Fig. 3a,
weight48 is shown as a function of the baseline
leptin/BMI ratio and leptin18. The leptin/BMI
ratio was divided at the median in a low and a high group, and
leptin18 was similarly divided into one group
with a small and one with a large decline, resulting in four groups of
subjects (A–D). Subjects in group A had a
leptin18 larger than the median and a leptin/BMI
ratio below the median, whereas subjects in group B also had a
leptin18 larger than the median but a leptin/BMI
above the median. The subjects in group C had a
leptin18 smaller than the median and a leptin/BMI
ratio below the median, whereas subjects in group D also had a
leptin18 smaller than the median but a leptin/BMI
ratio above the median.
|
leptin18 was related to a larger
weight48 than a small decline in leptin after 18
weeks. A two-way ANOVA showed that for
weight48 there was a main effect of the baseline
leptin/BMI ratio (F3,65 = 7.06; P
0.01) and of leptin18
(F3,65 = 17.09; P 0.001), but
no significant interaction was found between the two factors
(F3,65 = 0.95; P = 0.33). Similar
results were obtained when the analyses were based on the baseline
leptin/total adipose tissue mass ratio or on the unadjusted baseline
leptin level (not shown).
The baseline leptin levels were 32.6 ± 6.2 ng/mL in group A,
57.8 ± 22.2 ng/mL in group B, 24.2 ± 8.1 ng/mL in group C,
and 50.5 ± 15.2 ng/mL in group D. The relative decline in serum
leptin after 18 weeks were 75 ± 10%, 64 ± 16%, 41 ±
25%, and 21 ± 12%, respectively. Thus, among subjects with
similar leptin18 [large (A and B) or small
(C and D)], the largest relative decline in serum leptin were found
among subjects with the lowest baseline leptin levels (A and C), also
resulting in the lowest absolute serum leptin levels after 18 weeks in
these two groups.
In Fig. 3b, weight48 is shown as a function
of the baseline weight and the change in serum leptin after 18 weeks.
Both body weight and leptin18 were divided at
the median, resulting in four groups of subjects (E–H). In analogy
with Eq I in Table 2, a baseline body weight above the median was
related to a larger weight48 than a low
initial weight. As in Fig. 3a, a large
leptin18 was related to a large maintained weight
reduction after 48 weeks. A two-way ANOVA indicated that for
weight48 there was a main effect of
leptin18 (F3,65 = 12.84;
P 0.001) and a strong tendency for a main effect of
baseline weight (F3,65 = 3.87; P
= 0.054). There was no significant interaction between the two factors
(F3,65 = 0.12; P = 0.73).
A post-hoc analysis of the weight changes in groups A, B, C,
and D in Fig. 3a showed that groups A and B had the largest weight
reductions after both 18 and 48 weeks (Fig. 4). Group A had the highest and group B
the lowest baseline body weight. Groups C and D, with the smallest
maintained weight reductions after 48 weeks, also experienced smaller
18-week weight losses and had intermediate baseline weights. Figure 5 illustrates that subjects in group A
had larger BMI and lower leptin levels at baseline compared to group B,
with an overlap in BMI but not in leptin. Groups C and D were
distributed over the whole observed BMI range, with low baseline leptin
levels in C and high levels in D, almost without any overlap.
|
|
The change in serum leptin after 18 weeks was significantly
correlated to insulin18 in women (r =
0.30; P 0.05) but not in men (r = 0.30;
P = NS). There were no significant correlations between
leptin18 and the 18-week changes in cortisol
(r = 0.16 and r = -0.02) or TT3
(r = -0.25 and r = 0.26) in men and women, respectively.
To examine whether leptin18 was independently
related to the observed endocrine alterations (Fig. 2), multivariate
regression analysis was performed (Table 3). leptin18
was not related to changes in TT3 or cortisol.
The insignificant contribution of leptin18
increased the explained variance only by 0.5–1.6%. However,
leptin18 tended to be independently
associated to insulin18 (P =
0.06). The explained variance increased by 4.7% when adding
leptin18.
|
in men and women separately was not feasible, as
the groups were too small. However, when
TT3,18, cortisol18, or
insulin18 was regressed by baseline leptin,
leptin18, baseline BMI, and
weight18 in each gender separately, no
significant associations were found between the dependant variables and
baseline leptin or leptin18 (not shown). |
Discussion |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Our observations suggest that reduced leptin levels in response to an energy deficit do not appear to regulate the hormonal starvation response in obese men and women and do not predict weight relapse but, rather, a maintained weight reduction. This is in accordance with the report by Wing et al., who found no evidence that obese women with large initial reductions in leptin had difficulties in maintaining weight losses during a 6-month follow-up (12). The fact that the maintained weight reduction after 1 yr was more strongly predicted by the early weight loss than by the early drop in serum leptin might suggest that the predictive power of the initial leptin decrease was mediated by the weight reduction. However, this suggestion must be confirmed with other types of experiments because the strong explanatory power of the early weight loss can be explained by a part-whole correlation. It should also be pointed out that a larger energy deficit in the heaviest subjects was not the sole explanation for large early weight reductions, as subjects in group B, with the second largest weight reduction at 18 weeks, had the lowest baseline body weight.
Our second finding was that on prediction of the maintained weight
reduction, there was a difference between baseline body weight and
baseline leptin. Thus, a large maintained weight reduction after 1 yr
was associated with a large initial body weight, as expected. Although
there was no significant simple correlation between
weight48 and baseline leptin, low serum leptin
levels at baseline were associated with a large
weight48 in multivariate regressions. As we see
it, the latter finding was not expected. Leptin is believed to be
transported from the blood to the brain by a saturable transport
system. Thus, the relationship between leptin in the cerebrospinal
fluid (CSF) and in serum is curvilinear, with a small increase in CSF
leptin concentration for any given increase in serum leptin at high
serum leptin levels. The saturable leptin transport is reflected by low
CSF/serum leptin ratios in the obese (13, 14) and by an increased ratio
after weight loss with a concomitant reduction in serum leptin levels
(15). A given serum leptin reduction in response to weight loss would
consequently result in the smallest CSF leptin reductions in subjects
with the highest serum leptin levels. Thus, a hypothalamic starvation
response with accompanying weight relapse would be expected to be
greatest in subjects with low serum leptin levels. In contrast, the
largest maintained weight reductions were found in subjects with low
serum leptin levels. In fact, many of these patients had serum leptin
levels below the estimated level where the leptin transport becomes
impaired (16). Observational, longitudinal studies have found that
baseline leptin levels correlate positively (17), negatively (18, 19),
or not at all (20) with future weight change. In dietary intervention
studies, Nicklas et al. found that the baseline leptin level
was negatively correlated to the loss of fat after 6 months in older
women (21), but other clinical trials have not shown any relation
between the initial leptin level and the subsequent weight change (22, 12). Most of these studies have not adjusted for factors that could
influence the weight reduction response.
Our third unexpected finding was that the hormonal starvation response was not or was only weakly related to the early changes in serum leptin. The negative energy balance resulted in the expected decreases in insulin (11), TT3 (23), and leptin (3) and, at least in women, in the expected increase in cortisol (11). However, the decrease in leptin levels was not related to the changes in cortisol or TT3 and was only weakly related to the changes in insulin. These observations together with the fact that large early declines in leptin were associated with the largest and not the smallest long term weight reductions suggest that leptin by itself is of minor or no importance for the response to energy restriction in obese humans. These results are in contrast to those obtained in rodents, as discussed above. Besides the present study, there are additional indications that there may be species differences in the role of leptin as a starvation signal. Rodents without leptin display changes in the neuroendocrine system resembling those present during starvation. In contrast, humans with inactivating mutations in the genes encoding leptin and the leptin receptor have less severe endocrine disturbances (24, 25, 26, 27). Furthermore, rodents without leptin have a reduced basal metabolic rate, in contrast to humans (27). The regulation of the GH system in response to starvation further indicates the difference between humans and rodents. In man, GH secretion increases in response to starvation (28), whereas it is blunted in rodents (29). The decreased GH secretion in rodents is reversed by the administration of leptin (29).
There are several limitations in this study. It was performed in 69 subjects only, and it may be that a larger number of subjects would have revealed more convincing relationships between the changes in insulin and those in serum leptin. On the other hand, the study was large enough to detect some highly significant relationships in both men and women, speaking against the importance of leptin as a regulator of the starvation response in obese subjects. Although confirming studies are needed, it is not likely that larger trials would arrive at opposite results. Another limitation could be that we may have chosen a nonoptimal timing for endocrine measurements. We cannot exclude this possibility. However, the fact that conclusions based on measurements at week 18, after 2 weeks of unchanged body weight, were similar to those based on examinations after 8 weeks of ongoing weight reduction, indicates that timing was not critical for our results. A third limitation is that our observations are based on serum leptin concentrations without taking possible changes in leptin sensitivity into consideration. Conclusions based on measurements of changes in serum leptin levels assume that leptin sensitivity is unaltered or is altered in an identical manner in all individuals included in the study. Experiments in rodents suggest that leptin sensitivity may be under hormonal and nutritional control (30, 31). Thus, it might be that the relative lack of relations between changes in the serum levels of leptin and other endocrine variables could depend on alterations in leptin sensitivity in response to caloric restriction. As discussed above, this is not likely, because humans without leptin do not display a starvation-like endocrine pattern. However, our findings call for further analyses of basic aspects on the leptin system in humans, including studies of leptin sensitivity.
In conclusion, this study suggests that leptin by itself is not an important regulator of the starvation response in obese subjects. Taken together with previous reports on the relation between the leptin system and the starvation response in man, these results suggest that additional signals, yet to be identified, may act in concert with leptin in humans.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
Received April 15, 1999.
Revised June 24, 1999.
Accepted July 12, 1999.
|
References |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
rd U, Tylén U,
Sjöström L. 1988 Total and visceral adipose tissue
volumes derived from measurements with computed tomography in adult men
and women: predictive equations. Am J Clin Nutr. 48:1351–1361.This article has been cited by other articles:
 |
|
 |
|
  K. K. Koh, S. M. Park, and M. J. Quon Leptin and Cardiovascular Disease: Response to Therapeutic Interventions Circulation, June 24, 2008; 117(25): 3238 - 3249. [Full Text] [PDF]
|
|
|
|
 |
|
 C. Langenberg, J. Bergstrom, G. A. Laughlin, and E. Barrett-Connor Ghrelin, Adiponectin, and Leptin Do Not Predict Long-term Changes in Weight and Body Mass Index in Older Adults: Longitudinal Analysis of the Rancho Bernardo Cohort Am. J. Epidemiol., December 15, 2005; 162(12): 1189 - 1197. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. K. Das, S. B Roberts, M. A McCrory, L. G. Hsu, S. A Shikora, J. J Kehayias, G. E Dallal, and E. Saltzman Long-term changes in energy expenditure and body composition after massive weight loss induced by gastric bypass surgery Am. J. Clinical Nutrition, July 1, 2003; 78(1): 22 - 30. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Endocrinology | Endocrine Reviews | J. Clin. End. & Metab. |
| Molecular Endocrinology | Recent Prog. Horm. Res. | All Endocrine Journals |
