|
Are Leptin and Cytokines Involved in Body Weight Gain During Treatment
With Antipsychotic Drugs?
Causality in Chronic Diseases
In obesity, as in many chronic diseases, the causal relations between the
postulated factors and the pathological consequences are seldom linear
and simple. Hill summarized 9 postulates in the complex field of causation
that may assist researchers with the essential process of distinguishing
between “association” and “causation” in human health and disease (33).
The 9 postulates are strength, consistency, specificity, biological gradient,
plausibility, coherence, experimental support, analogy, and temporality.
In more recent reviews, the criterion of temporality (that is, the cause
precedes the effect) often includes complex relations between the events
such as simultaneity and circularity (34). Let us examine, as an example,
how insulin (one of the most studied factors in obesity) fulfills the criteria
of causality.
Most people with primary or secondary obesity display hyperinsulinemia
(which satisfies the criteria of strength and consistency). Chronic hyperinsulinemia
generally leads to obesity (which meets the criterion of specificity).
The extent of hyperinsulinemia (and other components of the metabolic syndrome,
such as dyslipidemia, hypertension, and insulin resistance) increases as
the BMI progresses, reaching a plateau around BMI 34 (35). In addition,
BWG in insulin-treated rats is dosage-dependent (meeting the criteria of
biological gradient and experimental support) (36). The metabolic effects
of insulin promote BWG since it enhances lipogenesis, glucogenesis, and
protein synthesis (37). Hence, it may be deduced that hyperinsulinemia
induces obesity (meeting the criteria of plausibility, coherence, and analogy).
In relation to the criterion of temporality, there are clear examples where
hyperinsulinemia precedes BWG. These include insulinomas in humans, administration
of exogenous insulin (36), and the high serum insulin levels quickly observed
in rats after lesions of the ventromedial hypothalamus (38).
Unfortunately, this simple scheme of causality for insulin in obesity is
further complicated by at least 2 important facts. First, insulin decreases
appetite by acting in the brain, and this action is opposite to its peripheral
effects on metabolism (12). Second, other factors are involved in primary
obesity, such as abnormalities in appetite, in resting metabolic rate,
in energy cost of activity, in fat oxidation, and in leptin and insulin
sensitivity. Because of this, hyperinsulinemia develops simultaneously
with BWG (39).
Let us briefly review the evidence for the role of leptin and TNF-a as
factors that induce obesity. Both hormones are elevated in obese people,
and their serum levels correlate with BMI and insulin levels (fulfilling
the criteria of strength, consistency, and biological gradient). Yet, TNF-a
levels are also increased in cachexia (21), and high leptin levels may
be observed in normal-weight subjects (7). Hence, the criterion of specificity
is not fulfilled. Most experimental evidence described above suggests that
leptin and TNF-a decrease appetite and BW. No published study has reported
BWG or obesity directly induced by any of the hormones (displaying a lack
of experimental support). Perhaps researchers are prone to make an analogy
between leptin, TNF-a , and insulin, thus meeting the criteria for plausibility
and coherence, but more empirical support is needed. We speculate that
studies showing leptin and TNF-a to impair insulin sensitivity may have
led researchers to suspect that these hormones may also impair the mechanisms
of BW regulation (and hence cause or aggravate obesity). This is plausible;
however, such deleterious effects generally seem to be induced after BWG
has already occurred. Thus, more empirical data are needed to sustain the
contention.
This leads us to the important criteria of temporality. Few experimental
studies have shown high leptin levels before actual BWG. A remarkable example
but one difficult to extrapolate in the daily clinical arena is the hyperleptinemia
detected in rats immediately after an experimentally induced lesion of
the ventromedial hypothalamus, followed by a progressive BWG and subsequent
obesity (38). In fact, Ravusin and others showed that low serum leptin
levels predicted subsequent BWG in Pima Indians, a population at high risk
for obesity and diabetes (39).
Regarding TNF-a, Hinze-Selch and others showed that, in clozapine-treated
subjects, the effects on sTNFRp75 (a soluble receptor of the hormone) preceded
the increase in the BMI (40). Hence, they conclude: "Our data suggest that
weight gain may be a consequence of an induction of the TNF-a system" (40,
p 17). We discuss this study in detail later, but to conclude this section,
we emphasize that most studies have shown that leptin and TNF-a increase
simultaneously with BW. It seems then, that both BWG and high hormone levels
are triggered by a suprafactor (for example, increased appetite, low levels
of resting metabolic rate, energy cost of activity, fat oxidation, or high
insulin sensitivity) (41).
Do leptin or TNF-" induce BWG in humans or animals treated with APs?
We conducted a Medline search using the words leptin, cytokines, antipsychotic
drugs, neuroleptics, psychotropic drugs, weight gain, and obesity. We found
11 articles dealing directly with the topic of AP-induced BWG. In this
section, we briefly review the reports in which leptin or TNF-a were assessed
during treatment with APs. As previously discussed, some authors suggested
a causal relation between leptin or TNF-a and AP-induced BWG, based on
anomalies in the expected correlation between these hormones and some physiological
variables. Our main arguments are that these studies were not primarily
designed to establish causality and that the postulated anomalies warrant
replication after improvement in some relevant methodological issues. For
the benefit of interested readers, we present a detailed description of
some papers, their main results, and the proposed anomalies observed in
some studies (Table 1).
What do these results tell us about causality between leptin or cytokines
and AP-induced BWG?
Before discussing the above-indicated articles, we will summarize the type
of results (and the anomalies) that might be expected to support a causal
link between leptin and cytokines and excessive BWG. Experimental studies
should demonstrate that exogenous administration of these hormones induces
obesity in animals. High serum levels of the hormones should be observed
before obesity has developed, and blocking the effects of leptin and TNF-a
might prevent BWG. An important variation could be that the hormones might
further impair the mechanisms of BW regulation once obesity is established,
which would make it difficult to return to a normal BW. The anomalous findings
could be the lack of correlation between the basal hormone levels and insulin,
BMI, BWG, and appetite. The expected sex-related difference for leptin
(that is, higher levels in women) might also be lost.
In the following paragraphs we critically discuss the above-mentioned reports,
with regard to causality. It will be apparent that criteria for a causal
relation between leptin, TNF-a , and AP-induced BWG have not yet been fulfilled.
All the human studies reported a significant increase in leptin and BW
during AP treatment (7,8,40,42–48). A serum leptin increase (by fat accumulation)
is expected during BWG in most people. Hence, the criterion of specificity
is not fulfilled. No leptin increase was observed in studies wherein BW
was not significantly affected during administration of haloperidol (43),
sulpiride (44), or lithium (49). In an early experimental study, we reported
that rats with sulpiride-induced obesity had leptin levels similar to those
of vehicle-treated animals (50). We recently demonstrated the same lack
of leptin elevation in risperidone-treated rats, in spite of significant
BWG (unpublished research). In both cases, the difference in BWG between
AP-treated rats and control animals was small, though statistically significant.
However, a different experimental protocol that induced robust BWG in sulpiride-treated
rats indeed showed a significant increase in serum leptin levels. It thus
appears that, in contrast to humans (44), the magnitude (threshold) of
BWG necessary to observe significant changes in leptin levels is higher
in genetically intact rats.
Melkerson reported that the expected correlation between leptin levels,
BMI, and insulin was absent and that olanzapine-treated women did not display
significantly higher leptin levels than did olanzapine-treated men (7,8).
These interesting findings seem to show an abnormal or anomalous leptin
regulation during olanzapine administration and are implicitly used by
the authors to support a causal relation between leptin and AP-induced
BWG. However, in our prior review of the Melkerson and others study (51),
we commented that 2 methodological limitations precluded any firm conclusion:
first, the lack of pretreatment values for leptin and insulin and, second,
the presence of 2 subjects with very abnormal leptin and insulin levels,
despite normal BMI (see Table 1). In fact, when the data were reanalyzed
excluding these patients, the expected correlation between leptin, BMI,
and insulin was actually observed. Nonetheless, in a response to our letter,
Melkerson and Hulting argued that there were no reasons to exclude these
subjects.
Regarding the lack of sex differences in leptin levels (8), we also suggested
that, ideally, this study should have been conducted by strictly pairing
men and women by BMI, age, and ethnic group. We think this should be considered
in future comparisons between olanzapine and other APs, since the positive
correlation between BMI and leptin is one of the strongest and most consistent
findings in the field. In fact, Baptista and others showed that women with
obesity associated with typical AP-administration had leptin levels similar
to healthy control subjects matched by age, BMI, and day of the menstrual
cycle (45). This result seems to prove that observed elevated leptin levels
are not specific to subjects who gain BW during AP administration.
Similarly, we commented elsewhere (52) that the anomalies observed by Herràn
and others (47) are inconclusive because the sample was too small (n =
5), and the subject’s sex was not specified in the report. Herran and others
reported that clozapine-treated patients displayed lower leptin levels
than expected for their BMI; however, an overrepresentation of men in the
clozapine group (or women in the olanzapine group) may explain their results,
since women display higher leptin levels than do men after controlling
for BMI. Herràn and others’ claim that "olanzapine appeared to induce a
greater increase in leptin independently of weight gain" is very important,
since again, it points to an olanzapine-related deleterious metabolic profile.
Unfortunately again, these results are inconclusive because of the small
sample size (n = 7) and the absence of data directly comparing the olanzapine-treated
subjects and matched control subjects.
|
Table 1 Human studies assessing leptin or TNF-a during AP treatment
|
|
Authors
|
Treatment
|
Results
|
Anomalies
|
|
Melkerson and others (7) and Melkerson and Hulting (8)
|
Conventional APs (n = 19) clozapine (n = 14) olanzapine (n = 14)
|
Leptin levels increased after olanzapine administration
|
In olanzapine-treated patients, the expected correlation between leptin,
insulin, and BMI was absent.
The normally observed sex difference in leptin levels (higher in women)
was found in the conventional AP group, but not in the clozapine or olanzapine
groups.
|
|
Hinze-Selch and others (40)
|
Clozapine (n = 12) and clozapine + fluvoxamine
(n = 11)
|
The increase in plasma leptin levels was significantly stronger and faster
during the combined therapy.
The BWG was similar in both treatment groups.
|
The observed elevation in leptin levels during the combined therapy cannot
be explained by the changes in BW.
|
|
Bromel and others (42)
|
Clozapine (n = 12)
|
Serum leptin levels and BW significantly increased.
|
None
|
|
Kraus and others (43)
|
Clozapine (n = 11) olanzapine (n = 8) or haloperidol (n = 13)
|
Leptin levels significantly increased in patients who gained BW during
clozapine or olanzapine administration. During haloperidol administration,
no changes were observed in BW or in leptin levels.
|
None
|
|
Baptista and others (44)
|
Sulpiride in healthy volunteers: 12 women and 7 men
|
In men, BW and leptin levels significantly increased. In women, neither
BW nor leptin were significantly affected.
|
None
|
|
Baptista and others (45)
|
A combination of typical APs (haloperidol, fluphenazine, chlorpromazine)
in psychotic women and a group of age- and BMI-matched healthy volunteers
|
Leptin levels did not differ in the 2 groups. Obese subjects (patients
and controls) displayed higher leptin levels than those with normal BMI.
|
None
|
|
Herran and others (46)
|
Conventional depot APs
(n = 17); conventional oral APs (n = 26); risperidone
(n = 5); clozapine (n =5); olanzapine (n =7). Control group: healthy subjects
matched by sex, age and BMI
|
Leptin levels did not differ between patients and control subjects in the
whole sample.
In patients with atypical APs, leptin levels were significantly higher
in the olanzapine group, intermediate in the clozapine group, and lower
in the risperidone group (P = 0.03).
|
Clozapine-treated patients displayed lower leptin levels than expected
for their BMI.
Olanzapine appears to induce a greater increase in leptin, independently
of BWG.
|
|
Eder and others (47)
|
Olanzapine
|
A significant increase in BW, leptin levels, and percentage of body fat
was observed during olanzapine administration.
|
None
|
|
Hagg and others (48)
|
Clozapine (n = 41) conventional APs (n = 62) healthy control subjects
(n
= 189)
|
After adjustment for sex, BMI, and other variables, hyperleptinemia was
independently associated with treatment with clozapine
(P < 0.001) or conventional
APs (P < 0.001).
|
In women, leptin levels were not significantly associated with treatment
with conventional APs.
|
|
AP = anti[psychotic; BW = body weight; BMI = Body Mass Index; BWG = Body
weight gain; TNF-a = tumor necrosis factor-a
|
Melkerson and Hulting speculate that clozapine and olanzapine may induce
leptin resistance at the hypothalamic level and thus impair the ability
of leptin to properly regulate BW (8, p 210). This is an interesting suggestion;
however, there are no published empirical data on this topic. In addition,
leptin resistance appears to be a general trait in primary obesity (11).
It is thus unclear how leptin resistance could be a specific effect of
the above-mentioned APs. In any event, if this is confirmed, it would rather
support the contention that leptin prevents rather than promotes AP-induced
BWG.
In their study of patients treated with clozapine alone or clozapine plus
fluvoxamine, Hinze-Selch and others reported the anomaly that the observed
results in leptin levels cannot be explained by the changes in BW (40).
Specifically, the increase in the leptin plasma levels was significantly
stronger and faster in patients on combined therapy, even though the data
on BW and BMI did not follow the same pattern. However, a detailed evaluation
of their Table 5 shows that the rate of BW change was different in the
2 groups. Subjects in the combined treatment group gained 1.2 kg in the
first 2 weeks, whereas the net change during that period for subjects receiving
clozapine alone was 0 kg. (In fact, the clozapine group lost 0.9 kg in
the first week). This is reflected in the leptin levels: the net change
in the first 2 weeks was 10 and 2.1 ng/ml for the combined and monotherapy
groups, respectively. For the combined and monotherapy groups, the net
BW change for the whole 6 weeks of treatment was 5 and 3.3 kg, respectively;
the net leptin-level change was 10.6 and 3.9 ng/ml, respectively. When
assessed in isolation, the changes in both variables appear congruent and
proportional. However, in the between-group comparisons, only leptin achieved
statistical significance. The authors emphasize that finding to further
support the postulated dissociation between leptin and BW. Still, it may
be argued that the difference in total BW gain (1.7 kg), while statistically
nonsignificant, may be physiologically relevant for leptin regulation.
In fact, we found that, after a BW gain of just 0.61 kg, leptin levels
doubled in healthy men treated with the AP sulpiride (44). Lastly, we wonder
whether a different statistical analysis might have produced other results—for
example, if leptin levels had been analyzed separately in patients who
gained BW and those who did not or if the data had been analyzed by paired
comparisons within the same subjects (that is, before-and-after tests).
This appears particularly relevant when the high variability in leptin
levels is considered.
Results obtained with the cytokines are difficult to interpret in regard
to BW regulation. In this study, TNF-a and soluble interleukin-2 receptor
(sIL-2r) levels were significantly higher in the monotherapy group (receiving
clozapine alone), which tended to gain less BW. This finding, in isolation,
allows speculation that those cytokines rather prevented or lessened the
drug effects on BW. The soluble cytokine receptors (sTNFRp55 and sTNFRp75)
significantly increased within each group, but the between-group comparisons
were nonsignificant. However, in another study of 10 olanzapine-treated
patients (in which 7 were observed to gain weight significantly), the same
authors reported no changes in IL-6, sIL-2r and TNF-a levels (53), although
sTNFRp55 and sTNFRp75 did increase significantly. In their paper, Schuld
and others did not intend to reconcile their findings with olanzapine with
those obtained by their research group in a study with clozapine: TNF-a
, sIL-2r, sTNFRp55, and sTNFRp75 levels and BW significantly increased
during clozapine monotherapy (40), but only sTNFRp55 and sTNFRp75 increased
(along with BW) during olanzapine treatment. Thus, we are left with the
difficult suggestion (not stated by the authors) that TNF-a and sIL-2r
may induce BWG in clozapine-treated, but not in olanzapine-treated, subjects.
This is of course possible, but few data appear to support it. Again, we
wonder what might be the results if cytokines were compared between subjects
who gained BW and those who did not.
To reconcile several disparate findings about the role of TNF-a in BW regulation,
Basile and others (9) adhere to the proposal that this cytokine may display
anabolic effects, and hence it may promote obesity, when it displays physiological
increase. Conversely, when TNF- levels are elevated at supraphysiological
levels, this cytokine may rather display catabolic effects and thus promote
BW loss, as in cachexia. Basile and others further speculate that this
hypothesis may explain an interesting observation: the higher the BMI before
starting AP treatment, the smaller the amount of BWG (for example, during
olanzapine administration [54]). A problem remains, however: the catabolic
effects of TNF-a appear to be observed in people whose BMI is higher than
45. This is an extreme degree of obesity that may exclude a patient from
clinical trials. In fact, only 10 out of 186 subjects had a BMI > 40 in
an olanzapine study (Lilly, personal communication, 2002). Hence, most
patients with high basal BMI (> 27.6) were well below the threshold value
of 45 and still gained little BW. In any case, this interesting proposal
deserves further investigation.
In another recent study, Basile and others reported that patients who had
an A/A genotype for the G-308A polimorphism of the TNF-a gene and displayed
high cytokine levels reported more BWG during clozapine administration
than did patients with other genotypes (and lower TNF-a levels) (9). This
important result supports the notion that TNF-a may promote obesity in
AP-treated patients. As discussed above, however, and given the many contradictory
findings in this field, future studies must address the alternative explanation
that such high TNF-a levels may represent a compensatory response to prevent
additional BWG in predisposed organisms.
Finally, the anomaly reported by Hagg and others—lack of association between
treatment with conventional APs and high leptin levels—may be explained
by a lack of power in the linear regression model used, as stated by the
authors (48). Hagg and others speculate that interaction between conventional
APs and androgens might contribute to the observed sex differences, since
leptin levels correlate directly to testosterone levels in healthy nonobese
women. Thus, abnormal androgen levels should be detected in women after
long-term treatment with typical APs. However, this prediction was not
confirmed in a recent study (45).
1 | 2| 3 | 4
|
