Additional Supporting Information may be found in the online version of this article.
Author Contributions: JO, PN, TN, KHP and KAV conceived and carried out the experiments and analyzed the data. RP, AR and SE conceived the experiments and interpreted the data. TV was responsible for the PET tracer production. All authors were involved in writing the paper and gave final approval of the submitted and published versions.
Disclosure: SE is a member of the scientific advisory board and shareholder of Ember Therapeutics. The authors declare no other duality of interest related to this manuscript.
Funding agencies: This study was conducted within the Finnish Center of Excellence in Molecular Imaging in Cardiovascular and Metabolic Research supported by the Academy of Finland, the University of Turku, the Hospital District of Southwest Finland, and Åbo Akademi University. The work was supported by grants from the Finnish Medical, Instrumentarium Science, Novo Nordisk, Paulo and Diabetes Research foundations, and Helsinki University Hospital Research Funds. The work was also supported by the European Union (EU FP7 project 278373; DIABAT), and grants from the Swedish Research Council (2009-2590 and 2010-3281), Sahlgrenska's University Hospital (LUA-ALF), the Knut and Alice Wallenberg and Söderberg foundations, and the Swedish Foundation for Strategic Research through the Center for Cardiovascular and Metabolic Research. SE thanks Konung Gustaf V:s och Drottning Victorias Frimurarestiftelse for generous support, and all the authors acknowledge the staff of Turku PET Center and the Obesity Research Unit in Helsinki.
Inactive brown adipose tissue (BAT) may predispose to weight gain. This study was designed to measure metabolism in the BAT of obese humans, and to compare it to that in lean subjects. The impact of weight loss on BAT and the association of detectable BAT with various metabolic characteristics were also assessed.
Design and Methods
Using positron emission tomography (PET), cold- and insulin-stimulated glucose uptake and blood flow in the BAT of obese and lean humans were quantified. Further, cold-induced glucose uptake was measured in obese subjects before and after a 5-month conventional weight loss.
Mean responses in BAT glucose uptake rate to both cold and insulin stimulation were twice as large in lean as in obese subjects. Blood flow in BAT was also lower in obese subjects under cold conditions. The increase in cold-induced BAT glucose uptake rate after weight loss was not statistically significant. Subjects with cold-activated detectable BAT were leaner and had higher whole-body insulin sensitivity than BAT-negative subjects, irrespective of age and gender.
The effects of cold and insulin on BAT activity are severely blunted in obesity, and the presence of detectable BAT may contribute to a metabolically healthy status.
Brown adipose tissue (BAT) has the capacity to generate heat in response to cold, and this uncoupling of ATP synthesis in mitochondria is accompanied by an increase in energy expenditure (EE) . Energy for thermogenesis is mainly provided by fatty acids but glucose is also taken up by BAT . Rodent studies suggest that BAT deficiency is related to the development of obesity , and that BAT is beneficial for glucose homeostasis . The presence of functional BAT in humans has been confirmed with biopsies from the supraclavicular adipose tissue . Isolated brown adipocytes have also been found elsewhere, for instance, in perirenal adipose tissue . The BAT in lean humans has also been shown to be a highly insulin-sensitive tissue . However, the role of BAT in human obesity and insulin resistance is still equivocal, although it has been suggested that its activity is reduced in obesity [8, 9]. On the other hand, some rodent studies show that diet-induced weight gain is accompanied by an increase in lipid content of BAT , and that dietary restriction can either increase  or decrease  the mitochondrial activity of BAT. Further, only one human study addresses the effect of weight loss on BAT .
Our study was designed to address whether the metabolic responses of BAT to both cold and insulin stimulation are decreased in obese humans, and how cold-induced glucose uptake is changed after weight reduction. We also investigated detectable BAT activity in relation to various metabolic characteristics, including insulin sensitivity.
This study includes 36 obese subjects, 20 of whom participated in a 5-month weight loss program, and 27 lean subjects from a previous study  to serve as a control group (Figure 1A). All subjects were screened for medical history and metabolic status, as assessed on the basis of routine laboratory tests, a 2-h oral glucose tolerance test (OGTT), electrocardiograms, and measured blood pressure (BP). All subjects were nondiabetic and euthyroid (Table 1). The weight loss program (Supporting Information) consisted of individual and group-based diet and exercise counseling bi-monthly starting with a 6-week modified very-low-calorie diet phase . The study protocol was approved by the ethics committees of the hospital districts of Southwest Finland and Helsinki and Uusimaa, and the study was conducted according to the principles of the Declaration of Helsinki. Written informed consent was obtained from all subjects.
Table 1. Characteristics of lean and obese subjects
Lean (n = 27)
Obese (n = 36)
Values are means ± SD. P values are from age- and gender-adjusted unpaired t tests. The number of observations is given below the mean value if the measurement was not carried out in all subjects.
Energy expenditure (EE) was adjusted for fat-free mass (FFM).
Positron emission tomography (PET) imaging was performed after overnight fasting (Figure 1B). Whole-body EE, blood chemistry, BP, and heart rate (HR) were measured during the PET sessions (Supporting Information).
Glucose uptake in supraclavicular adipose tissue (BAT), different white adipose tissue (WAT) depots, and skeletal muscle was measured using 2-deoxy-2-[18F]fluoro-d-glucose ([18F]FDG) as a tracer during cold exposure in all subjects. On the cold exposure day, the subjects spent 2-h wearing light clothing in a room with an ambient temperature of 17°C before the PET imaging. During the imaging (ambient temperature of 23°C), cold exposure was induced by placing one foot intermittently (5 min in/5 min out) in water at a temperature of 8°C.
Blood flow in BAT, subcutaneous WAT, and skeletal muscle was measured using [15O]H2O as a tracer in conjunction with cold exposure in the 16 obese subjects who did not participate in the weight loss program, and in all 27 lean subjects. These subjects underwent a second PET imaging session on a separate day, during which glucose uptake and blood flow were measured either with (14 lean and 9 obese subjects) or without (12 lean and 7 obese subjects) insulin stimulation  in a warm environment. Two of the 20 subjects in the weight loss program withdrew from the study before the cold-induced glucose uptake was measured for the second time with PET (Figure 1A). Obese subjects were scanned with an ECAT EXACT HR+ scanner (Siemens/CTI, Knoxville, TN), and magnetic resonance imaging (MRI) was applied to obtain an anatomical reference image (Supporting Information).
Results are expressed as means ± SD, and a two-tailed P value <0.05 was considered significant. Differences between lean and obese subjects and individuals with and without detectable BAT were evaluated with the unpaired t test after adjusting for age and gender (residuals from linear regression analyses). The Mann–Whitney U test was used to assess differences in BAT mass between lean and obese subjects. Paired and unpaired t tests were applied to test the statistical significance between the PET results recorded under warm, cold, and insulin stimulation conditions. Partial Pearson correlations adjusted for age and gender were used to study associations between different metabolic responses. Multiple linear regression analysis was applied to test independent determinants of cold-activated detectable BAT (dependent variables were waist circumference, fat percentage, age, and gender). The following variables were log10-transformed before statistical testing: BAT glucose uptake rate, BAT blood flow, M-value, area under curve (AUC) of plasma insulin, plasma thyroid-stimulating hormone (TSH), and free plasma thyroxin (free T4). Statistical analyses were performed using Stata 11.0 and SPSS 20.0.0.
Differences in BAT metabolism between lean and obese humans
The cold-induced BAT glucose uptake rate was significantly higher in lean than in obese subjects (Figure 2A). Cold activation of BAT was detected in only 11 out of 36 (31%) obese subjects, while the respective ratio in the lean subjects was 19 out of 27 (70%). The subjects without detectable BAT activation were also included in the comparison of lean and obese subjects. The mass of BAT was greater in the lean than in the obese subjects (24 ± 24 vs. 14 ± 29 g, P = 0.009). Interestingly, the insulin-stimulated glucose uptake rate in BAT was more than twice as high in the lean as in the obese subjects (Figure 2A). To test whether this disparity in the effect of insulin was unique to BAT, we compared other tissues between the lean and the obese, and similar but less pronounced differences were found for insulin-stimulated glucose uptake rates of cervical (P = 0.019) and abdominal (P = 0.024) subcutaneous WAT, and for skeletal muscle (P = 0.0060), but not for visceral adipose tissue (VAT) (P = 0.42). We also found a positive association between cold-induced and insulin-stimulated BAT glucose uptake rates (all subjects: r = 0.58, P = 0.004; lean subjects: r = 0.47, P = 0.092; obese subjects: r = 0.28, P = 0.46) (Figure 2B).
Like the cold-induced glucose uptake rate, blood flow in BAT was twice as high in the lean as in the obese subjects during cold exposure (Figure 2C). Further, a positive correlation between glucose uptake rate and blood flow in BAT during cold exposure was found (all subjects: r = 0.73, P < 0.001; lean subjects: r = 0.82, P < 0.001; obese subjects: r = 0.40, P = 0.12) (Figure 2D). However, blood flow in BAT showed no difference between lean and obese subjects under warm or hyperinsulinemic conditions (Figure 2C). The subsequent determination of glucose extraction revealed that it was greater in the BAT of lean subjects only during insulin stimulation but not in cold exposure (Figure 2E).
Serum free fatty acids (FFA) were measured as an indicator of lipolysis, and their concentration was elevated in response to cold exposure (0.46 ± 0.18 vs. 0.69 ± 0.19 mmol l−1, P < 0.001) (Table 1). Cold had the opposite effect on plasma insulin (7.3 ± 5.8 vs. 4.0 ± 3.3 mU l−1, P < 0.001) without a significant difference between lean and obese subjects (Table 1). BP and HR were measured before and during cold exposure, and they were not significantly different, except for diastolic BP, which was lower at the end of the cold exposure in the lean than in the obese subjects (data not shown).
Characteristics associated with detectable BAT
Compared to males, females had a higher cold-induced BAT glucose uptake rate (3.1 ± 5.9 vs. 5.9 ± 5.6 μmol • (100 g)−1 • min−1, age-adjusted P < 0.001). BAT positivity in cold exposure was also significantly associated with young age. Therefore, the subsequent t tests between the BAT-positive and -negative subjects were adjusted for age and sex (Table 2). Subjects with detectable cold-activated BAT were significantly lighter, had lower body fat percentages and smaller waist circumferences. They were also more insulin-sensitive, had a larger increment of serum FFA during cold exposure and higher basal plasma TSH levels. However, whole-body EE under cold conditions was not higher in subjects with detectable BAT. The respiratory quotient (RQ) was also similar in both groups (Table 2). The cold-induced BAT glucose uptake rate correlated negatively with body mass index (BMI), body fat percentage, waist and hip circumferences (partial correlation adjusted for sex and age r = −0.39−(−0.50), P < 0.002). The correlation of large BAT activity with small waist circumference remained even after further adjustment for body fat percentage (r = −0.37, P = 0.0035). The cold-induced BAT glucose uptake rate also correlated with the M value (r = 0.35, P = 0.030) and AUC of glucose in OGTT (r = −0.30, P = 0.021) before but not after further adjustment for body fat percentage. In a multivariate regression model, age (β = −0.33, 95% CI −0.47−(−0.19) years, P < 0.001) and waist circumference (β = −0.18, 95% CI −0.33−(−0.03) cm, P < 0.017) remained significantly associated with cold-induced BAT glucose uptake rate, independent of sex and whole-body fat percentage (whole model R2 = 0.36, P < 0.001).
Table 2. Comparison of subjects with (+) and without (−) detectable BAT
Values are means ± SD. On the basis of detectable cold activation of BAT, altogether 63 subjects were classified into two groups, BAT-positive and BAT-negative. The number of observations is given below the mean value if the measurement was not carried out in all subjects.
P values for the unpaired t-tests were corrected for age and gender, starting from BMI.
Energy expenditure (EE) was adjusted for fat-free mass (FFM).
We also assessed the role of BAT in insulin stimulation. The M value correlated with the insulin-stimulated glucose uptake rate of BAT (Figure 3A) and slightly more modestly with insulin-stimulated glucose uptake rates in other adipose tissues (sex- and age-adjusted r = 0.48-0.52, P < 0.05). The insulin-stimulated glucose uptake rate in skeletal muscle was, as expected, highly significantly associated with the whole-body insulin-stimulated glucose uptake rate (Figure 3B). The glucose uptake rates of BAT and skeletal muscle correlated significantly during insulin stimulation (sex- and age-adjusted r = 0.50, P = 0.022).
Metabolic responses of BAT, WAT and skeletal muscle in obese humans
To investigate how different tissues in obese humans respond to cold and insulin stimulation, we determined glucose uptake rates not only in BAT (Figure 4A) but also in cervical and abdominal subcutaneous WAT, VAT, perirenal adipose tissue, and skeletal muscle. In the obese subjects, the mean BAT glucose uptake rates between warm and cold exposure did not differ significantly (P = 0.27) (Figure 4B); however, the BAT glucose uptake rate was high in some of the obese individuals during cold exposure (range: 0.3-19.1 μmol • (100 g)−1 • min−1). The glucose uptake rates in cervical and abdominal subcutaneous WAT, VAT, perirenal adipose tissue, and skeletal muscle were also unaffected by cold exposure (Figure 4B). Insulin increased glucose uptake significantly only in the skeletal muscle. In fact, the insulin-stimulated glucose uptake rate in the BAT of obese subjects was not significantly different from the insulin-stimulated glucose uptake rates in the cervical (P = 0.18) and abdominal subcutaneous WAT (P = 0.065), the perirenal (P = 0.18) and visceral adipose tissue (P = 0.076), whereas it was lower than the glucose uptake rate in skeletal muscle (P < 0.001). Compared to warm conditions, blood flow did not change in response to cold or insulin in any tissue studied in the obese subjects (Figure 4C), but insulin increased glucose extraction slightly in BAT and substantially in skeletal muscle (Figure 4D).
Effect of weight loss on BAT glucose uptake
The cold-induced supraclavicular BAT glucose uptake rate was not significantly different (P = 0.058) after a mean of 12.5% (range: 3.5-24.5%) weight loss, and no differences could be detected in other tissues either (Figure 4E). The mass of BAT was not increased as a result of weight loss (19 ± 32 vs. 18 ± 29 g, P = 0.72). However, whole-body glucose tolerance and insulin sensitivity improved after weight loss (Table 3). Effects of weight loss were especially large for HR, systolic and diastolic BP, all of which were lower in cold conditions after weight loss (Figure 5). Interestingly, in the six subjects (two males and four females) with the largest elevation in their BAT metabolism, defined as an increase of more than 1 μmol • (100 g)−1 • min−1 in the BAT glucose uptake rate after weight loss, the glucose uptake rate increased in their perirenal adipose tissue (1.7 ± 0.5 vs. 2.0 ± 0.4 μmol • (100 g)−1 • min−1, P = 0.042), while no higher glucose uptake was observed in the other organs studied. During the cold exposure before weight loss, the BAT nonactivators showed significant increases in both systolic (P < 0.001) and diastolic BP (P = 0.0038), whereas in the activators, BP did not change from warm to cold conditions. After weight loss the increase was milder and was only seen for systolic BP (P = 0.017) in the BAT nonactivators.
Table 3. Characteristics of subjects before and after weight loss
Before weight loss (n = 18)
After weight loss (n = 18)
Values are means ± SD. P values are from paired t–tests. The characteristics of the 18 subjects who completed the weight loss program are shown.
Energy expenditure (EE) was adjusted for fat–free mass (FFM).
In this study, we show that obese, as compared to lean subjects have significantly reduced metabolic responses in BAT to both cold and insulin stimulation. To our knowledge, this study is also the first non-retrospective study to show a sex-based difference in human BAT activity.
The low cold-induced glucose uptake in BAT in obesity was linked to reduced blood flow, whereas glucose extraction was similar to that in the lean subjects. While the findings of BAT blood flow are completely new, studies have previously shown, for instance, impairments of endothelium-dependent vasodilatation in the brachial artery in obesity , which improve after weight loss . We also found that BP increased in response to cold more before than after weight loss (Figure 5). On the other hand, HR decreased during cold significantly after but not before weight loss. Increase in BP during cold exposure was observed especially in the obese subjects without detectable BAT activation. Based on these findings, it can be postulated that sympathetic overactivity in obesity is related to impaired BAT metabolism, and that BAT activation in cold exposure requires an intensive increase in local blood flow, a response that is deactivated in obesity and reactivated with weight loss. On the other hand, a limitation in BP and HR measurements is that they do not assess sympathetic outflow specifically to BAT. It has been shown that the production of nitric oxide synthase in BAT is directly dependent on sympathetic activity, which modulates vasodilation . Thus, the measured responses in the sympathetic parameters of the cardiovascular system are not definitive evidence for the sympathetic overactivity. Measurements of the sympathetic activity directly in BAT or plasma catecholamines are required in the future.
Interestingly, a significant positive correlation with the M-value and BAT glucose uptake rate was found during hyperinsulinemia (Figure 3A), and the presence of detectable cold-activated BAT was also associated with higher M value (Table 2). However, it can be calculated that glucose uptake at whole-body level is, on average, 30- and 700-fold higher in skeletal muscles than in BAT during cold and insulin stimulation, respectively. On the other hand, it has been shown that transplanted BAT improves glucose homeostasis by increasing the insulin sensitivity of WAT and heart muscle in mice , suggesting that BAT may have endocrine functions. The molecular mechanisms of insulin resistance and hormonal functions of BAT in humans are therefore interesting areas for future studies.
The quantification of [18F]FDG-PET data in this report offers a precise determination of metabolic activity, because it provides an accurate metabolic uptake rate of glucose, independently of the size of the subject, plasma glucose concentration, and administered dose and plasma clearance of [18F]FDG. However, BAT has a key role in fatty acid handling . For logistical reasons and the limitation on radiation dose, we were not able to measure fatty acid utilization in this study, although differences in fatty acid consumption may also contribute to the variation observed in glucose uptake. On the other hand, the cooling was not a personalized protocol, and it can be speculated that it was not optimal to ignite BAT thermogenesis in some subjects, causing interindividual variation in BAT activity. However, it has been demonstrated that most morbidly obese humans fail to activate BAT even after personalized cooling has been applied, suggesting that some subjects lack active BAT altogether . It should be noted that lean and obese subjects were studied in separate time periods, although not significantly differently in terms of season, but this divergence did not cause variation in BAT activity. Namely, we could not establish any association between outside temperature and measured BAT glucose uptake (data not shown). The fact that two different PET scanners were used was also not relevant, since both the scanners have relatively similar spatial resolutions, and regions of interest were systematically outlined using a similar technique in all images. This approach substantially reduces the risk of bias. In summary, despite the disadvantages in the study, we conclude that our findings on decreased BAT metabolism in obese humans are genuine.
Pharmacological approaches used to activate human BAT have often been unsuccessful in the past  and recently . In the present study, we found a tendency for weight loss to increase cold-induced glucose uptake in BAT of obese humans, although this result did not reach statistical significance. This finding is in line with a study which suggests that morbidly obese subjects increase their BAT activity after bariatric surgery . Whole-body insulin sensitivity, as estimated with the Matsuda index , did improve after weight loss (Table 3), while no significant changes in cold-induced tissue glucose uptake were found (Figure 4E). This suggests that the impacts of weight loss on cold- and insulin-stimulated glucose disposal are not associated.
Interestingly, the subjects with the greatest increase in their BAT glucose uptake rate after weight loss, also had an increase in their glucose uptake rate in perirenal adipose tissue. Perirenal activity is sometimes coincidental with the apparent supraclavicular BAT uptake , and patients with catecholamine-secreting pheochromocytomas have BAT within their perirenal adipose tissue . Therefore, it seems reasonable to postulate that weight loss can increase cold-induced glucose uptake in perirenal BAT.
BAT positivity was not associated with higher EE in a population consisting of lean and obese subjects (Table 2). This may be due to the fact that glucose uptake, on which the detection of BAT was based, is not the principal source of energy of BAT , and thus, glucose uptake is not directly linked to EE. Interestingly, we observed an increase in serum FFA concentration in response to cold exposure. Our data suggest additionally that the increase in circulating FFAs due to cold exposure is higher in individuals with detectable BAT (Table 2). This can be interpreted in that the lipolytic physiological signals to the main sources of fatty acids, the subcutaneous WAT or VAT, probably have the same origin as the BAT activating signals, since concomitant release of fatty acids is needed to provide fuel for thermogenesis and subsequent refueling of human BAT . Further, we found that the cold-induced glucose uptake rate in BAT is inversely correlated with age and, even after correction of age and gender, to BMI, fat percentage and waist circumference. The association of low BAT activity to detrimental central fat distribution remained after adjusting for overall adiposity, suggesting that the metabolic conditions which promote abdominal fat deposition may in part be the same as those which endorse reduced activation of BAT, or that the presence of central adiposity is a sign of a general metabolically disadvantaged state, e.g. inflammation, which could also deactivate BAT .
In conclusion, we demonstrate that the metabolic responses of human BAT to activation by cold and insulin are severely impaired in obesity, and that the presence of cold-activated detectable BAT is associated with a metabolically beneficial phenotype.