The Relationship between Patients’ Serum Glucose Levels and Metabolically Active Brown Adipose Tissue Detected by PET/CT
- First Online:
- Cite this article as:
- Jacene, H.A., Cohade, C.C., Zhang, Z. et al. Mol Imaging Biol (2011) 13: 1278. doi:10.1007/s11307-010-0379-9
- 189 Views
To compare blood glucose levels in patients with or without “detectable” brown adipose tissue (BAT) using 2-deoxy-2-[18F]fluoro-d-glucose positron emission tomography/computed tomography (FDG PET/CT).
Nine hundred eight patients had PET/CT scans and were previously identified as having, or not having, FDG uptake in BAT. The original database was retrospectively reviewed for blood glucose level and body mass index (BMI) at the time of imaging. Blood glucose levels were compared between patients with or without FDG uptake in BAT, adjusting for age, sex, and BMI.
Fifty-six patients (6.2%) had FDG uptake in BAT. In the univariate analysis, patients without FDG uptake in BAT had a higher risk of glucose ≥100 mg/dL (odds ratio 3.4, 95% CI = 1.6-7.3; P = 0.0007). After adjustment for age, sex, BMI, and significant interaction of sex and BMI, patients without BAT tended to have a higher risk of glucose ≥100 mg/dL, although not statistically significant (odds ratio = 1.6, 95% CI = 0.7-3.6; P = 0.268).
Although causal relationships are not specified, the data suggest that BAT uptake, glucose levels, BMI, sex, and age are inter-related and the possibility that presence of “detectable” BAT is protective against diabetes and obesity. FDG PET/CT may be a vital tool for further investigations of diabetes and obesity.
Key wordsBrown fatBATGlucoseFDG“USA”-FatPETBrown adipose tissue
Brown adipose tissue (BAT) in mammals primarily serves to generate heat . By contrast, its counterpart white adipose tissue (WAT) is involved in caloric storage . Uncoupling protein-1 is expressed in BAT and it is important in conferring the thermogenic ability of BAT by uncoupling adenosine-5′-triphosphate production from the movement of protons down their concentration gradient in the inner mitochondrial membrane . The existence of BAT in hibernating rodents and human neonates is well established; however, its presence in adult humans remained controversial until its relatively recent description on 2-deoxy-2-[18F]fluoro-d-glucose positron emission tomography/computed tomography (FDG PET/CT) [2–7].
Cohade et al.  described foci of high metabolic activity localizing to fat density on CT scan in patients (mostly adults) undergoing FDG PET/CT imaging for oncologic evaluation, described as “USA-Fat” for its uptake mainly in the supraclavicular area of the body. The FDG uptake has since been described in other locations besides the supraclavicular area, for example the paraspinal region, mediastinum, neck, and upper abdomen [3, 8]. The high metabolic activity for FDG fuses to fat density on CT and is thus phenotypically due to active BAT. BAT has been shown to avidly accumulate FDG in animal studies [9–11].
In humans, FDG uptake in BAT is more common in females and younger patients and has been associated with cold exposure [3, 5–7, 12]. Studies have also suggested that patients with active BAT are thinner than those without active BAT [3, 5–7, 12]. In addition, pre-clinical and human studies have demonstrated that the use of beta-blockers is a negative predictor for the presence of active BAT detected on PET/CT [6, 11].
Mice without BAT are susceptible to insulin resistance and diet-induced diabetes and develop an obese phenotype [13–16]. BAT has been suggested to have a role in the regulation of insulin secretion and glucose homeostasis in animals [13–16]. Based on the animal data, we hypothesized that fasting glucose levels would be lower in adult humans with BAT uptake on FDG PET compared to those without BAT uptake. To test our hypothesis, we compared blood glucose levels in patients identified qualitatively as having active BAT on FDG PET/CT to patients with no “detectable” BAT uptake. We also considered other variables (age, sex, body mass index) that have also been shown to be associated with BAT uptake.
Materials and Methods
Nine hundred eight patients (445 females, 463 males, mean age 58 ± 15 years) underwent 1,016 consecutive FDG PET/CT scans between July 1, 2001 and June 30, 2002, all for the evaluation of known or suspected cancer. Seventy-five patients had more than one FDG PET/CT scan during the time period of the study and only the first scan was considered for the purpose of this investigation.
Permission to conduct this retrospective study was obtained from our Institutional Review Board under an exempt review. The requirement for informed consent was waived.
FDG PET/CT Scans
Patients fasted for a minimum of 4 h and had blood glucose levels <200 mg/dL measured immediately prior to the intravenous injection of a weight-based amount of FDG (8.14 MBq/kg [0.22 mCi/kg]).
After an approximately 60-min tracer uptake phase, a combined PET/CT scan (Discovery LS; GE Healthcare, Waukesha, WI, USA) was obtained from the mid-skull level to the mid-femur level. The details of the image acquisition and reconstruction techniques were previously published [2, 12].
Previously, all PET/CT scans were examined for the presence or absence of FDG uptake in BAT and the results reported . For the current study, the original patient database and electronic medical record was reviewed to determine patients’ blood glucose level just prior to the intravenous administration of FDG, height and weight. Body mass index (BMI) was determined for each patient using the following equation: weight (kg)/height (m2). For the patients with BAT uptake and a sub-group without BAT uptake (2:1 for no BAT uptake to BAT uptake) matched for age, sex, and BMI, medical records were also reviewed for the use of beta-blockers and steroids.
Results for continuous variables are expressed as mean ± standard deviation and categorical variables as frequency and percentage. For the primary analysis, glucose level was considered as a binary outcome based on the current boundary between normoglycemia (glucose <100 mg/dL) and impaired fasting glucose level (glucose ≥100 mg/dL) . The association between glucose group and BAT uptake, age, sex, and BMI was examined with the use of the logistic regression model. First, a univariate analysis was performed to assess the effects of individual variables on the risk of having glucose ≥100 mg/dL. Variables considered included BAT uptake (yes/no), age, sex (female/male), and BMI (underweight ≤18.5; normal = 18.5-24.9; overweight = 25-29.9; obesity ≥30) .
A multivariable logistic regression model was then applied to this cohort to assess the independent effects of risk factors with adjustment for potential confounders. A variable selection procedure was incorporated by stepwise selection while adhering to the hierarchical model building principal, that is, all effects contained in a significant interaction term must remain in the model, even if these effects are not conditionally significant. Potential interaction effects were considered between age and BAT, sex and BAT, BMI and BAT, age and BMI, and sex and BMI. As BAT uptake was of primary interest, it was included in the final model regardless of its outcome in the stepwise selection procedure. In this analysis, only two-way interaction was considered and data was modeled on probability of being at risk for diabetes (glucose level ≥100 mg/dL). Odds ratios and 95% confidence intervals are reported, which indicate the strength of the association between glucose group and BAT uptake, age, sex, and BMI.
We also explored the relationship between glucose level as a continuous variable and BAT uptake, age, sex, and BMI. For this analysis, a general linear model was employed and univariate and multivariable models were applied sequentially. The interaction effect (e.g., age and BAT, sex and BAT, BMI and BAT, age and BMI, sex and BMI) was assessed by including a cross-product term (two-way interaction) along with the associated main effects. Variables of significance based on the univariate models were retained in the multivariable models. As BAT uptake was of primary interest, it was considered in the final multivariable model regardless of its significance.
The analyses were performed with SAS (v. 9.2, SAS Institute, Cary, NC, USA). Throughout all analyses, statistical significance was considered at P value <0.05.
Univariate analysis of association between predictors and risk of glucose level ≥100 mg/dL
Glucose level ≥100 mg/dL (n = 317)
Glucose level <100 mg/dL (n = 590)
Odds ratio (95% CI)
BAT uptake, n (%)
Sex, n (%)
BMI, n (%)
Multivariable analysis of association between predictors and risk of glucose level ≥100 mg/dL
Odds ratio (95% CI)
No vs. yes
Age (unit of 10 years)
Sex × BMI
Obesity vs. normal weight
Overweight vs.normal weight
Underweight vs. normal weight
Obesity vs. overweight
Obesity vs. underweight
Overweight vs. underweight
Obesity vs. normal weight
Overweight vs. normal weight
Underweight vs. normal weight
Obesity vs. overweight
Obesity vs. underweight
Overweight vs. underweight
To further explore the independent effect of the predictors, a multivariable regression analysis was performed with a stepwise variable selection procedure, which included the main effects and all relevant two-way interactions. After controlling for age, sex, and BMI, BAT uptake did not remain an independent predictor of risk of glucose level ≥100 mg/dL. Age (P < 0.0001), sex (P = 0.941), BMI (P < 0.0001) and interaction of sex and BMI (P = 0.025) were retained in the model by stepwise selection.
A final model was fitted by adding BAT uptake to the above model as it was of primary interest to determine the strength of the association of BAT with risk of glucose level ≥100 mg/dL after controlling for confounders. In this final multivariable analysis (Table 2), age remained significant with older patients more associated with risk of glucose level ≥100 mg/dL (odds ratio for unit of 10 years 1.5, 95% CI 1.3-1.6, P < 0.0001). There was a significant interaction between sex and BMI, where the likelihood of having a glucose level ≥100 mg/dL was significantly higher in obese females than males (P for interaction = 0.027; Table 2). A trend was observed toward higher risk of glucose level ≥100 mg/dL in patients without BAT (odd ratio = 1.6, 95% CI = 0.7-3.6; P = 0.268), although not statistically significant after adjustment for age, sex, BMI, and significant interaction of sex and BMI.
Effects of independent predictors on glucose level as a continuous outcome variable
Sex × BMI
Medication histories were available for 35 patients with BAT uptake (median time from medication history to PET/CT scan is 9 days, range 0-169 days). For the control group without BAT uptake that was matched for age, sex, and BMI, median time between available medication history and PET/CT scan is 7 days (range 0-78 days). No patient (0%) with BAT uptake was on beta-blocker therapy, but eight of 73 (11%) in the control group were receiving beta-blocker therapy (P = 0.05, Fisher’s exact test). Seven patients were taking systemic steroids (n = 6 glucocorticoids, n = 2 sex hormones), seven were taking inhaled glucocorticoids, and two were taking systemic sex hormones and inhaled glucocorticoids. The steroid usage patterns between those with and without BAT uptake were not significantly different in regards to overall frequency of steroid usage (P = 0.15, Fisher’s exact test) and inhaled versus systemic steroids (P = 0.29, Fisher’s exact test).
Most studies evaluating BAT visualization on FDG PET and PET/CT have focused on BAT as a confounding factor for image interpretation [2, 3, 8, 19], but the advent of PET/CT alleviates most, but not all of the challenges differentiating co-existing hypermetabolism in BAT and tumor. The ability to visualize hypermetabolism in BAT on FDG PET/CT has perhaps, even more significant implications because animal studies have suggested a role of BAT in the development of obesity, insulin resistance and diabetes [13–16, 20, 21]. FDG PET/CT may be used as non-invasive imaging tool to study the role of BAT in adult humans.
In univariate analyses, blood glucose levels were lower in patients with active BAT on FDG PET/CT versus those without active BAT. Patients without BAT tended to a higher risk of glucose level ≥100 mg/dL with an attenuating strength, but this was not statistically significant after adjustment for other factors which did remain associated with higher glucose levels (age, sex, and BMI). All patients were required to have glucose levels <200 mg/dL prior to FDG PET imaging and it is possible that the range of glucose levels evaluated was not large enough to show the correlation between glucose and BAT uptake in the multivariable analyses.
Possibly, a more important explanation, and what we speculate to be true, may be that glucose, age, sex, BMI, and BAT uptake are too closely inter-related to separate their interactions. In our model, the effects of BAT uptake on glucose level appear to have been attenuated by the effects of age, sex and BMI. From our model, it appears that obese females without BAT are probably at the highest risk for higher glucose levels. These results, while similar to those of Cypess et al. , are not directly comparable because the outcome variable was BAT uptake, and not glucose level, in that study.
Although the available data thus far does not conclusively demonstrate causal relationships between BAT uptake, glucose levels, BMI, sex, and age, one interpretation is that the presence of active BAT is protective against diabetes and obesity. This interpretation is supported by substantial animal data showing that BAT-deficient mice develop obesity, hyperphagia, and insulin resistance in the absence of hypothalamic and serum corticosterone level abnormalities [13–16]. The relationship between BAT activity and serum glucose levels could be tested in prospective human studies using PET. It is quite possible the sample size we examined is simply too small to clearly show such a relationship.
Cypess et al.  recently showed that beta-blocker therapy was significantly associated with a lack of “activated” BAT on PET in a larger patient population . In our study, no patient with BAT was on beta-blocker therapy; however, generalizing the results of this sub-group analysis to the entire cohort is limited by patient numbers. Other animal models of obesity with and without a co-existing diabetic phenotype suggest impairment in the BAT adrenergic system [20–22]. Stimulation of brown adipocyte precursors to express uncoupling protein in WAT by a beta3 adrenergic agonist has been shown to be effective anti-obesity and anti-diabetic treatment in obese fa/fa rats . We speculate that functional BAT may contribute to a lean, non-diabetic human phenotype.
Williams and Kolodny  recently reported a lower frequency of BAT visualization on FDG PET/CT and lower glucose levels (just before PET) in patients prepared with high-fat, very low-carbohydrate diets (compared to patients who fasted). They suggested this may be due to an inhibitory effect of fatty acid loading on glucose metabolism. In contrast, animal data suggest that BAT is increased in rats fed a high-fat “cafeteria” diet and is involved in diet-induced thermogenesis . A major difference is that the animals were studied after 3 weeks of exposure to the high-fat diet whereas the patients’ exposure was less than 24 h. Our patients were fasting for at least 4 h prior to PET/CT, but we did not prescribe a specific preparatory diet.
The major limitations of our study are its retrospective nature and lack of control groups which did not allow compilation of complete clinical histories of documented cases of diabetes. The fasting serum glucose at the time of PET imaging was felt to serve as a surrogate and potentially found additional cases of undiagnosed diabetes. Other potential confounding factors that could not be controlled for were anxious states, cold weather or other cold conditions (i.e., air conditioning). In addition, the patients included in this study were referred for FDG PET/CT to evaluate for malignancy, and they could have metabolic states different from healthy adults.
Our study does not provide conclusive evidence that BAT is protective for the development of diabetes or obesity; however, our results along with other studies [5–7, 12] support this possibility. Further studies in humans are needed and FDG PET/CT may prove to be a vital tool for this purpose. Gaining a better understanding of the role of BAT in humans has potentially tremendous implications—potentially providing new treatment options for two major healthcare issues, diabetes and obesity.
Conflict of Interest
The authors declare that they have no conflict of interest.