Intranasal Insulin Reduces Body Fat in Men but not in Women

Participants in this study were 40 healthy normal weight subjects (16 women, 24 men). They were all nonsmokers and none were taking medication, except that all women were taking oral contraceptives. All relevant illnesses were excluded by clinical examination and routine laboratory tests. Subjects abstained from alcohol, caffeine, and food for 12 h before the testing. They gave written informed consent before the experiments, and the study protocol was approved by the local ethics committee. Experiments were performed in a double-blind manner.

We randomly assigned subjects to either the insulin or placebo group (12 men and 8 women in each group). Both groups were comparable in age (24.6 ± 1.3 and 25.8 ± 1.2 years, respectively) and BMI (22.6 ± 0.3 and 22.7 ± 0.4 kg/m², respectively) in the pretesting examination. During 2 weeks of baseline, all subjects received placebo. During the succeeding 8-week treatment phase, subjects were administered insulin or placebo intranasally four times a day: in the morning, around noon, and in the evening (each ∼30 min before mealtime) and before going to bed. Each dose consisted of 0.4 ml insulin (40 IU; Insulin Actrapid; Novo Nordisk, Mainz, Germany) or vehicle (HOE 31 dilution buffer for H-Insulin; Aventis Pharma, Bad Soden, Germany) administered within four 0.1-ml puffs (two per nostril), amounting to 1.6 ml (160 IU) of insulin or vehicle per day. Sprays were stored at ∼5°C and replaced every week. To ensure compliance, subjects kept a diary about their intake routine and were told that irregular intake would be detected via urine sampling.

Test sessions (scheduled between 0700 and 0900) took place at the beginning (session A) and end (session B) of the baseline phase and at the end of the treatment phase (session C). Subjects were kept unaware of the study aims by embedding examinations into the assessment of cognitive parameters, not described here in detail. Posttreatment interviews ensured all subjects had remained unaware of the treatment effects. In the beginning of a session, subjects were administered placebo (session A), insulin or placebo (B), and again placebo (C). Thus, session A was used to familiarize subjects with the intake routine, and sessions B and C allowed for the examination of acute and long-term treatment effects, respectively.

Immediately after substance administration, we weighed subjects and measured their body composition by standard bioelectrical impedance analysis (frequencies of 1, 5, 50, and 100 Hz; BIA 2000-M; Data Input, Frankfurt, Germany) indicating body fat, total body water, intracellular water, extracellular water, lean body mass, and body cell mass (Eurobody software; Data Input). Waist circumference was also measured, and subjects completed a questionnaire on their eating behavior (18). Subjects rated their hunger on a 10-point scale 60 min after insulin administration and at the end of the session. Thus we obtained difference values indicating the gradient of hunger feelings. Between ratings, resting energy expenditure was registered using indirect calorimetry (Deltatrac II, MBM-200 Metabolic Monitor; Datex-Engström Deutschland, Achim, Germany). Heart rate variability was measured simultaneously for an interval of 20 min (12-lead simultaneous electrocardiogram, sampling rate 1,000 Hz; ECG Lab Version 2.0; Meigaoyi, Beijing, China). Weekly, at 0800, we weighed subjects and sampled blood to determine leptin, insulin, and glucose concentrations. Levels of epinephrine and norepinephrine were taken from 12-h (2000–0800) nocturnal urine samples and measured by standard high-performance liquid chromatography.

To control for possible side effects, we monitored various other parameters in the weekly examinations, including plasma concentrations of ACTH, blood pressure, heart rate, and routine laboratory measurements (serum electrolytes; creatinine; HDL, LDL, and total cholesterol; and triglycerides). In a follow-up examination performed 4–5 months after cessation of treatment, we measured body weight, body composition, and blood parameters of the male subjects.

Blood samples were centrifuged immediately, and the plasma was stored at −20°C. Concentrations of leptin, insulin, and ACTH were assessed using standard radioimmunoassays (Human Leptin RIA KIT, Linco Research, St. Charles, MO; Pharmacia Insulin RIA100, Pharmacia & Upjohn, Uppsala, Sweden; Lumitest ACTH, Brahms Diagnostica, Hennigsdorf, Germany). Glucose was measured in fluoride plasma (Aeroset; Abbott, Wiesbaden, Germany).

Comparisons between the effects of insulin and placebo were based on ANCOVA, with a group factor representing the treatment condition. In additional analyses, sex was included as a group factor, and treatment effects were evaluated separately for the male and female subgroups. Significant interaction effects were followed by pairwise one-sided contrasts. Respective values of session A (start of the baseline phase) served as covariates for parameters that could be subject to the acute influence of insulin (e.g., heart rate variability); values of session B (start of the treatment phase) were used as covariates in the assessment of long-term influence on body weight and related variables. All data are presented as means ± SE. P ≤ 0.05 was considered significant.

There were no overall differences in body weight, hormonal concentrations of interest, or any other measurements between the two treatment conditions due to the fact that men and women displayed striking differences in their response to intranasal insulin treatment, with only the men losing weight. Statistical analyses confirmed highly significant treatment × sex interactions for body weight [F(1,35) = 8.91, P ≤ 0.005], BMI [F(1,35) = 9.87, P ≤ 0.003], and the ratio of extracellular mass to body cell mass [F(1,35) = 7.10, P ≤ 0.02] after 8 weeks of treatment. Consequently, separate analyses were performed for the groups of men and women.

In the men, insulin treatment reduced body fat content [F(1,21) = 4.65, P ≤ 0.05], body weight [F(1,21) = 4.50, P ≤ 0.05] (Fig. 1), and BMI [F(1,21) = 4.33, P ≤ 0.05]. Waist circumference was also decreased [F(1,21) = 3.98, P ≤ 0.05] (Table 1). Over the 8-week period, the insulin-treated men lost 1.28 ± 0.71 kg of their body weight (versus a slight gain of 0.45 ± 0.49 kg for the placebo group) and 1.38 ± 0.59 kg of body fat (versus a gain of 0.57 ± 0.61 kg for the placebo group). The BMI of insulin-treated men declined by 0.38 ± 0.21 kg/m² (versus a gain of 0.13 ± 0.14 for the placebo group), and their waist circumference decreased by 1.63 ± 1.17 cm (versus a gain of 0.62 ± 0.88 for the placebo group). During the final examination after 8 weeks of treatment, hunger ratings were decreased [F(1,18) = 6.99, P ≤ 0.02]. Heart rate appeared to be enhanced, but this effect was not significant [F(1,19) = 3.15, P = 0.09].

Leptin levels in the male subjects of the insulin group dropped during treatment, reaching significance compared with the placebo group after 8 weeks [2.06 ± 0.29 vs. 2.86 ± 0.25 ng/ml; F(1,18) = 4.33, P ≤ 0.05]. Insulin and glucose levels did not differ between the groups after 8 weeks of treatment (Table 1) or when average values during the entire treatment period were compared [8.87 ± 1.65 vs. 10.85 ± 1.65 μU/ml, F(1,21) = 0.70, P > 0.41; 86.69 ± 0.92 vs. 88.22 ± 0.84 mg/dl; F(1,19) = 1.50, P > 0.24]. Also, epinephrine and norepinephrine levels were not affected [8.93 ± 1.05 vs. 6.76 ± 1.00 μg/l, F(1,18) = 2.06, P > 0.17; 22.11 ± 2.17 vs. 23.28 ± 2.17 μg/l, F(1,17) = 0.13, P > 0.72, respectively], excluding pronounced increases in sympathetic tone after insulin treatment. Comparisons of all other parameters, including resting energy expenditure and heart rate variability (expressed as the ratio of low- to high-frequency power) failed to reveal significant effects of long-term insulin administration in the male subjects (Table 1). The acute effects of 40 IU of insulin were restricted to an increase in diastolic blood pressure [70.96 ± 2.35 vs. 63.96 ± 2.35 mmHg, F(1,21) = 4.43, P ≤ 0.05].

In contrast to the effects in men, in the insulin-treated women, body weight increased as soon as 1 week after initiation of treatment [insulin, 61.85 ± 0.27 and placebo, 61.00 ± 0.27 kg; F(1,13) = 4.92, P ≤ 0.05] and remained constantly elevated thereafter (Fig. 2). The increase in body weight [F(1,13) = 7.49, P ≤ 0.02] and BMI [F(1,13) = 8.74, P ≤ 0.01] after 8 weeks appeared to be due to a rise in extracellular water [F(1,13) = 5.10, P ≤ 0.04] (Table 2, Fig. 2), in conjunction with an increased ratio of extracellular mass to body cell mass [F(1,13) = 4.49, P ≤ 0.05] and tendencies toward augmented total body water [F(1,13) = 3.65, P = 0.08] and lean body mass [F(1,13) = 3.28, P = 0.09]. The supplementary analysis of plasma sodium indicated a trend toward increased levels in the insulin-treated women over the 8-week period [139.41 ± 0.34 vs. 138.46 ± 0.34 mmol/l; F(1,13) = 3.57, P = 0.08], whereas corresponding values in men remained unaffected [139.87 ± 0.23 vs. 139.96 ± 0.23 mmol/l, F(1,21) = 0.08, P > 0.78; F(1,35) = 5.69, P ≤ 0.05 for treatment × sex]. There were no treatment effects on plasma hormone and glucose concentrations after 8 weeks of treatment (Table 2) or when average values during the treatment period were compared [leptin: 12.16 ± 0.57 vs. 11.96 ± 0.57 ng/ml, F(1,13) = 0.06, P > 0.81; insulin: 13.70 ± 1.82 vs. 11.54 ± 1.95 μU/ml, F(1,12) = 0.66, P > 0.43; glucose: 80.15 ± 1.46 vs. 83.90 ± 1.36 mg/dl, F(1,12) = 3.44, P > 0.09]. Also, average catecholamine levels were comparable [epinephrine: 6.50 ± 0.96 vs. 4.83 ± 0.96 μg/l, F(1,13) = 1.52, P > 0.24; norepinephrine: 20.00 ± 1.95 vs. 15.47 ± 1.95 μg/l, F(1,13) = 2.70, P > 0.13]. All other parameters, including those after acute insulin administration, remained unchanged in the female subjects.

In the follow-up examination of the men, conducted 17.6 ± 0.52 weeks after cessation of treatment, the differences between the insulin and placebo groups had vanished: body fat [15.24 ± 0.63 vs. 16.11 ± 0.70 kg; F(1,17) = 0.86, P > 0.32], body weight [80.47 ± 0.96 vs. 80.03 ± 1.06 kg; F(1,17) = 0.10, P > 0.76] and BMI [23.82 ± 0.30 vs. 23.75 ± 0.33 kg/m²; F(1,17) = 0.03, P > 0.87] were comparable in the insulin and placebo groups, respectively. There seemed to be enhanced levels in plasma leptin and insulin levels in the insulin group, but this difference failed to reach significance [plasma leptin: 4.00 ± 0.35 vs. 3.13 ± 0.40 ng/ml, F(1,11) = 2.59, P = 0.14; plasma insulin: 6.09 ± 0.79 vs. 4.15 ± 0.83 μU/ml, F(1,16) = 2.77, P = 0.12]. None of the other parameters displayed any significant difference [glucose: 89.04 ± 2.45 vs. 87.95 ± 2.57 mg/dl, F(1,14) = 0.09, P > 0.76]. Interviews conducted after the final examinations did not reveal any hints at side effects of the treatment. In all, four subjects of the (total) insulin group and five subjects of the placebo group believed that they had been treated with an active agent.

Prolonged intranasal administration of insulin in healthy humans reduced body weight, body adiposity, and leptin levels in the group of men compared with placebo treatment, whereas it induced weight gain due to enhanced water storage in the women. The weight loss in the insulin-treated men vanished 4–5 months after treatment ended. Our results corroborate the importance of CNS insulin signaling in the control of body weight, and confirm reports from animal experiments indicating a fundamental difference in central insulin sensitivity between men and women (19). It could be argued that the lack of body weight reduction in the women was due to a hampered nose-to-brain insulin transport, so that the substance did not reach the target sites in female subjects. However, in our previous studies assessing CSF accumulations after intranasal administration of 40 IU insulin, a distinct increase in CSF insulin concentrations after substance administration was observed in all three women of that study (averaging 1.84 ± 0.67 μU/ml within the first 40 min) (15). This increase also appeared to be comparable with that seen in the men (averaging 1.81 ± 1.02 μU/ml), although increases appeared to be rather variable, independent of the subject’s sex. Also, measurements of declarative memory conducted in the present study revealed beneficial effects of 8 weeks of insulin administration in all treated subjects (20). This outcome likewise supports the view that brain uptake of intranasally administered insulin in the women was as efficient as in the men. Sex differences were, in fact, restricted to insulin effects on body weight regulation.

In animal studies, male rats decreased food intake after intracerebroventricular insulin infusion and had lost substantial weight after only 24 h of treatment, whereas age- and weight-matched female rats remained unaffected. Leptin infusion yielded a reverse pattern, with a greater impact being seen on the female rats (19). Our finding of different central nervous sensitivity to insulin also fits with this picture, as insulin treatment reduced waist circumference in men; visceral fat content is correlated with endogenous insulin secretion and is more prevalent in obese men than women, who carry more subcutaneous fat (21). The mechanism underlying the observed sex difference is unclear. No such difference has been found for downstream neuronal mediators of the catabolic effects of insulin and leptin (19,22). Thus, the observed sex difference may have its source in differentially regulated insulin and leptin signaling, an explanation that appears plausible given that in mouse brain, insulin signaling is essential for reproductive function (9).

In the women, insulin surprisingly increased extracellular water, as reflected by an increased ratio of extracellular mass to body cell mass, associated with rapidly increased total body weight and augmented BMI. A confounding effect of the menstrual cycle can be safely ruled out as all women were taking oral contraceptives and each woman was examined at identical times in the cycle. Body water distribution may have been altered by renal sodium reabsorption, which is known to be modulated by peripheral insulin sensitivity (23). A trend toward increased plasma sodium levels was indeed revealed in the insulin-treated women. Intranasal insulin may have mediated the effects on water storage by enhancing systemic insulin sensitivity, which depends on intact hypothalamic insulin signaling (24). It is also noteworthy that with reference to body weight, the women received a higher dosage of intranasal insulin than the men, which could have enhanced this type of effect.

The decline in body weight and adiposity after insulin administration in the male subjects may have stemmed mainly from reduced daily food intake. Insulin’s reducing influence on hunger has been inferred from previous studies where, under euglycemic conditions, rated hunger rose more slowly during infusion of higher versus lower dosages of insulin (25). Studies of long-term central nervous insulin administration in several species have indicated a rather slowly evolving attenuation of 24-h food ingestion, extending up to several weeks (7,26). These observations would fit the assumption that even a small decrease in hunger motivation, of which an individual may not necessarily be aware, can lead to reduced caloric intake. The effect of intranasal insulin on body weight in the men emerged after several weeks of treatment, a finding that supports the concept of a gradual impact of elevated brain insulin levels on the regulation of actual hunger and satiety. Correspondingly, rated hunger was not affected by acute insulin administration but decreased after 8 weeks of treatment. This observation does not exclude the possibility that effects on body weight could occur even sooner with higher CSF concentrations of the compound, although this might be difficult to achieve with intranasal insulin administration in humans. Central nervous insulin administration in animals is reported to not only inhibit food intake but also stimulate thermogenesis via enhanced sympathetic neuronal outflow to thermogenic tissue (5,27). The signs of increased sympathetic tone found in our study (i.e., acutely increased diastolic blood pressure and a trend toward elevated heart rate after long-term treatment) were too weak to justify corresponding conclusions. Likewise, catecholamine levels, resting energy expenditure, and heart rate variability were not elevated after insulin administration.

The central nervous pathways between hypothalamic nuclei and the caudal brainstem, where neural, endocrine, and duodenal nutrient and adiposity signals converge to terminate single meals, have been extensively reviewed (2–4,10). Insulin and leptin regulate body weight through joint downstream signaling via melanocortins, establishing a tonic catabolic output to avert excessive weight gain (28). In our group of men, weight loss was associated with decreased leptin concentrations. After the treatment period, the discontinued insulin in conjunction with diminished leptin signaling presumably permitted prevailing activation of anabolic pathways so that body weight and fat returned to pretreatment levels. Regaining weight might ultimately lead to counterregulation, as was expressed in the slight increase in insulin and leptin levels in the follow-up examination.

In conclusion, the present data strongly support the assumption that in humans insulin is a prominent adiposity signal in the brain and exerts catabolic effects on body weight and adiposity. Because insulin is pharmaceutically produced in large quantities, its intranasal administration may represent an economic but effective therapeutic means of reducing body weight in obese men. Besides impaired central nervous sensitivity to insulin and leptin (29), decreased blood-to-brain transport of both messengers possibly plays a crucial role in the pathogenesis of obesity (30). Reduced CSF levels of insulin in obese animals have been repeatedly reported (31,32), and data gathered in our laboratory point to a similarly degraded level of insulin in the CSF of obese men (33). Thus, enhancing central nervous insulin concentrations by intranasal administration appears to be a promising approach in the campaign against obesity.

Aero Pump GmbH (Hochheim, Germany) generously provided us with precision nasal air pumps.

We thank A. Hatke for her expert technical assistance and C. Otten and H. Ruf for their invaluable laboratory work.