Imaging of Brown Adipose Tissue: State of the Art

We review the basic biology and existing approaches to image brown adipose tissue (BAT), including techniques currently used in research applications as well as newer imaging approaches based on the expanding knowledge of the developmental origins and physiology of BAT.

Abstract

The rates of diabetes, obesity, and metabolic disease have reached epidemic proportions worldwide. In recent years there has been renewed interest in combating these diseases not only by modifying energy intake and lifestyle factors, but also by inducing endogenous energy expenditure. This approach has largely been stimulated by the recent recognition that brown adipose tissue (BAT)—long known to promote heat production and energy expenditure in infants and hibernating mammals—also exists in adult humans. This landmark finding relied on the use of clinical fluorine 18 fluorodeoxyglucose positron emission tomography/computed tomography, and imaging techniques continue to play a crucial and increasingly central role in understanding BAT physiology and function. Herein, the authors review the origins of BAT imaging, discuss current preclinical and clinical strategies for imaging BAT, and discuss imaging methods that will provide crucial insight into metabolic disease and how it may be treated by modulating BAT activity.

^© RSNA, 2016

Introduction

The past 5 years have seen unprecedented advances in the understanding of adipocyte physiology, with important insights into the metabolic contribution of brown adipocytes in adult mammals. The existence of brown adipose tissue (BAT) in eutherian (placental) mammals and human infants has long been known (1). Although cervical and supraclavicular areas of increased metabolism had been documented in the radiology literature (2), the recent explosion of interest in BAT metabolism is driven by the observation that metabolically active BAT can be observed with imaging in a significant fraction of the adult population (3–5). The existence of BAT in adult humans has led to substantial interest in the therapeutic potential of heat production by this tissue (thermogenesis) as a means to increase whole-body energy expenditure in obesity. Furthermore, the recent description that cold exposure can induce certain adipose cells toward a phenotype similar to BAT (6) has further broadened the implications of BAT metabolism for the treatment of obesity and insulin resistance. In light of the important role that BAT is thought to play in normal and pathologic energy expenditure, it is crucial that practicing radiologists understand the physiologic role of BAT and the evolving imaging methods to assess its presence and function.

Imaging-based approaches have been crucial for studying the presence and regulators of human BAT. Initial studies described an association between the presence of catecholamine-secreting tumors (pheochromocytoma) and brown fat tumors (hibernoma) (7), as well as the ability of pheochromocytomas to induce BAT overgrowth in remote sites (8), foreshadowing the now-well-known effect of cold and β3-adrenergic receptor signaling in activating BAT (9). However, the crucial observations regarding the presence of BAT in adults came with the advent of fluorine 18 (⁸F) fluorodeoxyglucose (FDG) positron emission tomography (PET) and the identification of hypermetabolic activity in the cervical and supraclavicular regions (10). This was initially thought to reflect radiotracer uptake by muscle, but subsequent anatomic coregistration with ¹⁸F FDG PET/computed tomography (CT) definitively established the existence of hypermetabolic fat in adults (2).

Although ¹⁸F FDG PET/CT has clearly been useful in such proof-of-concept human studies, this approach has several important limitations, including heterogeneity of response, sensitivity to experimental or environmental factors, insensitivity to fatty acid–mediated metabolism (the preferred energy source for BAT), and confounding variables for standardized uptake value (SUV)–based quantitation (11). For these reasons, efforts are currently under way to develop alternative methods for BAT imaging, with the hope that this will allow noninvasive, quantitative measures of mass and activity. Current imaging approaches being investigated include alternative radionuclide-based approaches, advanced magnetic resonance (MR) imaging techniques, including spectroscopy and the use of hyperpolarized contrast media, and ultrasonographic (US) contrast agents.

Herein, we review the basic biology of BAT and the existing approaches used for imaging it, including techniques currently used in research applications as well as newer imaging approaches based on the expanding knowledge of the developmental origins and physiology of BAT. This review should provide the practicing radiologist with an understanding of the importance of BAT beyond confounding clinical ¹⁸F-FDG imaging studies and will discuss how dedicated BAT imaging can contribute to the evaluation of metabolic disease.

The Biology of BAT

Adipose tissue has been traditionally categorized as either white or brown, a distinction that reflects the histologic, anatomic, developmental, and functional differences between their respective adipocytic lineages (Fig 1) (12). From a bioenergetic standpoint, white adipose tissue (WAT) is primarily known for its role in energy storage in the form of unilocular triglyceride droplets, whereas brown adipocytes are highly specialized for heat production. This specialization is reflected by dense packing of mitochondria and multilocular intracellular lipid droplets, and crucially by expression of the mitochondria-associated uncoupling protein 1 (UCP1). UCP1 mediates biochemical uncoupling of electron transport by allowing proton leakage across the inner mitochondrial membrane, thus short-circuiting the usual linkage of electron transport to adenosine triphosphate production. Instead, the chemical energy generated by cellular respiration in brown adipocytes is dissipated in the form of heat, which allows thermoregulation by mammals in cold environments. It is estimated that fully activated BAT can produce up to 300 W/kg, whereas most other tissues produce only 1 W/kg (13).

Figure 1:

Distinct developmental origins of brown, white, and beige adipocytes. Unlike skeletal muscle and BAT, which arise from a common Pax7⁺/Myf5⁺ stem cell lineage (left), white and beige adipocytes appear to arise from Pax7⁻/Myf5⁻ cell lineage (right). (Reprinted, with permission, from reference 12.)

Although the thermoregulatory function of BAT in small and hibernating mammals has been known for decades, the recent resurgence in interest on this tissue stems mainly from its potential to mitigate metabolic disease, specifically obesity and type 2 diabetes. Indeed, there is overwhelming evidence from genetic models demonstrating that increased amounts or activity of BAT can improve whole-body metabolism (12,14,15). The relevance of these observations was initially debated, given the limited amount of BAT in adult humans. However, the recent description of inducible “browning” of white fat has given new significance to these earlier studies (12).

It is now known that a certain population of UCP1-positive adipocytes within WAT can be induced to produce heat when stimulated by cold or peroxisome proliferator–activated receptor gamma agonists such as rosiglitazone (16,17). The developmental origin of this unique cell type, which has been termed “brite” (a portmanteau for “brown in white”) fat, “beige” fat, or “induced WAT,” appears to be distinct from that of classic BAT, which develops from a mesodermal population characterized by expression of muscle lineage markers Myf5 and Pax7 (18). Despite this distinct developmental origin, beige adipocytes express other BAT effector genes such as Cidea and Pgc1a, in addition to Ucp1. Likewise, it is clear that beige adipocytes can participate in thermogenesis (6).

Studies of WAT, BAT, and beige fat naturally led to the question of whether these pathways could be harnessed for metabolic benefit. Increased activity of BAT or beige fat leads to improvement in glucose tolerance and insulin sensitivity, and the observation that overexpression of Pgc1a was sufficient to convey many of these beneficial effects led to the identification of a cleaved version of the membrane protein FNDC5 as a potential mediator of browning and improved metabolism induced by exercise (19). Many other secreted inducers of browning have also been identified, including FGF21 (20,21), natriuretic peptides (22), and Bmp8b (23), which now represent attractive targets for drug development in obesity and diabetes (24). The development and application of such targeted therapeutics will, however, require the establishment of imaging modalities to allow quantitative readout of thermogenic fat mass as well as activity (Table 1). In this review, “BAT imaging” refers to methods that reveal both classic BAT and beige cells, as current techniques are unable to help discriminate between these tissue types. Selective imaging of beige adipocytes is an important challenge that will be the subject of future investigations.

Table 1.

Current Approaches to BAT Imaging

Note.—BOLD = blood oxygenation level dependent, FBnTP = fluorobenzyl triphenyl phosphonium, MRB = (S,S)-O-methylreboxetine, THA = 14(R, S)-fluoro-6-thiaheptadecanoic acid.

Prevalence of BAT in Humans

The prevalence of BAT in human subjects has been debated, with estimates derived from both retrospective and prospective studies (Table 2). Most retrospective studies have attempted to estimate BAT prevalence on the basis of analysis of PET/CT scans obtained under nonstandardized environmental and imaging conditions, an approach that allows large cohorts to be analyzed but has significant variability in degrees of BAT activation (40). Such studies have generally estimated the prevalence of active BAT to be approximately 5% (3,40), with variation in this percentage also attributable to differences in cohorts among studies.

Table 2.

Factors That Affect the Prevalence of BAT in Human Retrospective Studies

Note.—Future prospective studies with controlled BAT activation will provide more reliable insights into the true effect of the above factors, confirming those that in fact play a relevant role in the presence of human BAT.

In contrast, prospective studies of BAT prevalence generally use cold exposure to fully activate BAT metabolism, thus boosting the probability of detection. Such analyses have led to an estimated prevalence of 40%–100% in the general population (32,41,42). However, numerous factors have been described as contributing to the appearance of metabolically active (ie, ¹⁸F-FDG–avid) BAT in humans. These include age (more frequently observed in younger subjects), sex (more frequent in women), and body mass index (less frequent in obese subjects) (3,25–27).

Imaging of BAT

PET Imaging

¹⁸F-FDG PET/CT.—BAT imaging originated in the field of oncology imaging, specifically with use of ¹⁸F-FDG PET/CT for the assessment of glucose uptake in tumors. FDG is a glucose analog that has selective uptake by glucose transporters (glucose transporter protein 1 or glucose transporter protein 4 [Glut-4]) and is trapped intracellularly by means of the action of hexokinase. It became apparent from early studies that “spurious” FDG uptake was often seen in areas with no discernible tumor, a finding usually attributed to muscle uptake in patients who were shivering. This represented a potential diagnostic dilemma until it was found that this uptake could be pharmacologically inhibited by means of pretreatment with benzodiazepines (eg, diazepam) or nonselective β-adrenergic blockers (eg, propranolol) (10,43). Separate studies demonstrated that this uptake often coincided with areas of low attenuation on coregistered CT scans, with Hounsfield unit measurements indicative of fat density (44). This was speculated to represent FDG uptake in BAT, which is consistent with earlier autopsy studies that indicated that BAT was both present and widely distributed in adult humans (1,2). Although these studies led to efforts in using ¹⁸F-FDG PET/CT to estimate the prevalence and quantity of BAT, it was only in 2009 that a series of seminal articles definitively established that a specific pattern of cervical and upper thoracic FDG uptake in fact represented BAT (3–5,32,45). The years since the publication of these studies have seen an explosion of interest in the use of ¹⁸F-FDG PET/CT to investigate the physiologic characteristics of BAT, being the dominant imaging modality used in the field.

It is important to appreciate the complex biology underlying ¹⁸F-FDG uptake to properly evaluate studies by using ¹⁸F-FDG PET/CT to investigate BAT function. Irrespective of tissue type, the overall cellular uptake of ¹⁸F-FDG is dependent on two main factors: expression and activity of glucose transporters. Classic brown adipocytes demonstrate both high constitutive expression of Glut-4 as well as inducible expression of glucose transporter protein 1 and are therefore competent to uptake ¹⁸F-FDG (46,47). Glut-4 is expressed in all insulin-sensitive tissues (skeletal muscle, heart, WAT, and BAT) and is under the control of complex regulatory mechanisms including insulin-dependent transporter translocation to the cell surface (48). In keeping with these findings, BAT demonstrates strong insulin sensitivity with regard to glucose uptake (49). In addition to insulin-dependent cell surface translocation, Glut-4 expression in BAT can also be directly regulated by other factors known to control BAT activation. This includes β-adrenergic signaling, which has been shown in animal studies to be required for cold-induced BAT activation and to increase expression of Glut-4 and UCP1 in BAT (50,51). Although both insulin and cold and/or β-adrenergic signaling promote ¹⁸F-FDG uptake, they appear to do so by independent mechanisms, as cold (but not insulin) induces increased BAT perfusion that underlies its functions in energy dissipation and thermoregulation (49). Conversely, β-adrenergic signaling appears to be critically important not only for normal cold-induced activation of classic BAT, but also for the pathologic “browning” of white fat seen in cancer-induced cachexia (52). Finally, it is important to keep in mind that glucose is not the only or even the main energy source for BAT, which also demonstrates avid uptake of free fatty acids from the circulation (53).

The insulin sensitivity of BAT naturally makes ¹⁸F-FDG PET/CT uptake critically dependent on many factors known to affect ¹⁸F-FDG uptake in other tissues, such as skeletal muscle, heart, and tumors. This includes fasting status, diabetes, muscle activity, and medications such as β-blockers (3,25,27). As noted earlier, uptake is influenced by a host of patient-specific demographic and anthropometric factors, including age, sex, and body mass index (Table 2) (26,28,54). This makes proper and reproducible patient selection crucial in the assessment of the overall prevalence and distribution of BAT.

However, a most important factor is properly differentiating between presence and activity of BAT. Because ¹⁸F-FDG is taken up primarily in active BAT, a lack of uptake can be due to either lack of BAT tissue or low levels of BAT activation. This dichotomy is clear from studies demonstrating large amounts of cold-activated BAT in patients who show no such BAT activation when exposed to warm conditions (55). This is in keeping with the observation that UCP1-positive brown adipocytes can be histologically identified in the supraclavicular fat regardless of ¹⁸F-FDG uptake (56). Such effects may contribute to substantial intrasubject variability, an issue that has yet to be systematically addressed. Indeed, in a study of oncology patients undergoing serial ¹⁸F-FDG PET/CT for tumor follow-up, less than 10% of patients with detectable BAT on an initial scan had detectable BAT on a second study (25). Clearly, BAT activation status is both a complex and crucial factor for interpreting the results of ¹⁸F-FDG PET/CT studies, especially those performed retrospectively.

In addition to complexities related to the mechanism of ¹⁸F-FDG uptake in BAT, quantitating the results of these studies has proved to be challenging. This largely reflects a lack of standardization within the field as to how PET images are collected, cutoffs for SUVs to be considered positive for BAT, and CT attenuation criteria for separating fat from nonfat soft tissue. By convention, BAT is usually defined as metabolically active tissue (maximum SUV greater than a given threshold) localized to fat density (as determined with Hounsfield unit measurement) on coregistered CT scans. The threshold maximum SUV used in studies most commonly varies between 1.0 and 2.0, whereas Hounsfield unit cutoffs range from −10 to −250 (40). These two parameters can be cross-correlated to identify BAT on the basis of both its higher FDG uptake and its increased Hounsfield unit values compared with those of WAT, with the latter effect attributed to increased perfusion and higher intracellular water content (57,58). One potential use of such approaches is in automated segmentation of BAT from fused PET/CT images (59). Although promising, these efforts are currently limited by factors including the lower spatial resolution of PET in comparison with CT, misregistration of PET and CT images, and partial volume effects resulting in overestimation of BAT volume. Moreover, the specificity of CT in the detection of BAT remains to be determined.

Despite its methodologic limitations, ¹⁸F-FDG PET/CT has proved to be an invaluable tool in BAT studies. Early studies focused on defining the prevalence and distribution of classic BAT, with widely varied results depending on experimental factors such as those discussed earlier. This has led to attempts to address this heterogeneity by studying more standardized populations (generally healthy, young, nonobese volunteers) in which BAT is also activated before imaging with a variety of cooling techniques. These studies have led to much higher estimates of the overall prevalence of detectable supraclavicular BAT, approaching 100% in studies in which tailored protocols are used to induce maximal activation, including cooling methods and β3-adrenergic receptor agonists (Figs 2–4; Table 3) (9,42,63). Beyond the classic supraclavicular depot, BAT uptake has also been demonstrated with ¹⁸F-FDG PET/CT in a variety of additional sites, including cervical, mediastinal, upper thoracic, and paravertebral locations (3,40). Recent studies have also used ¹⁸F-FDG PET/CT to investigate the relationship between BAT and a variety of pathologic states, including diabetes, obesity, human immunodeficiency virus lipodystrophy, bone loss, and cancer-associated cachexia (61,62,64,65). Among other findings, these studies have revealed an association of BAT mass and activity with bone mineral density, inhibited BAT activation in obesity, and an intermediate BAT phenotype in the dorsocervical fat of patients with human immunodeficiency virus lipodystrophy.

Figure 2a:

Images in healthy 21-year-old female volunteer (body mass index: 19 kg/m²) show BAT activated by means of a cooling vest protocol (subject wore a cooling vest 1 hour before and 1 hour after FDG administration). (a) Axial CT scan at level of T1 vertebral body for anatomic reference shows supraclavicular fat (arrow). (b) Superimposed ¹⁸F-FDG PET scan shows FDG uptake in fat-attenuation regions, including supraclavicular (white arrow), paravertebral (black arrow), and intermuscular (arrowheads) regions. (c) Axial CT image at level of humeral heads for anatomic reference. (d) Fused ¹⁸F-FDG PET/CT scan shows FDG-avid adipose tissue in axillary (white arrow), paravertebral (black arrow), and upper mediastinal (arrowhead) regions. (e) Coronal reformatted CT scan for anatomic reference shows supraclavicular and axillary fat. (f) Fused ¹⁸F-FDG PET/CT scan shows FDG uptake in fat-attenuation areas, including supraclavicular (white arrow) and paravertebral (black arrow) regions. (g) Coronal multiple intensity projection image shows extensive uptake consistent with BAT throughout cervical, supraclavicular (white arrow), axillary, and paravertebral (black arrow) regions.

Figure 4a:

BAT activation in humans with β3-adrenergic receptor agonist mirabegron. Twelve subjects underwent ¹⁸F-FDG PET after administration of either placebo or 200-mg mirabegron. (a) ¹⁸F-FDG PET images in 21-year-old man after administration of placebo (left) and mirabegron (right). There are multiple sites of BAT uptake after mirabegron administration. (b) Graph shows results from all 12 patients. There is a significant increase in detectable BAT after mirabegron administration. Patient identification numbers are listed on graph. (Reprinted, with permission, from reference 9.)

Table 3.

Current Approaches to Activate Human BAT for ¹⁸F-FDG PET/CT

Figure 2b:

Images in healthy 21-year-old female volunteer (body mass index: 19 kg/m²) show BAT activated by means of a cooling vest protocol (subject wore a cooling vest 1 hour before and 1 hour after FDG administration). (a) Axial CT scan at level of T1 vertebral body for anatomic reference shows supraclavicular fat (arrow). (b) Superimposed ¹⁸F-FDG PET scan shows FDG uptake in fat-attenuation regions, including supraclavicular (white arrow), paravertebral (black arrow), and intermuscular (arrowheads) regions. (c) Axial CT image at level of humeral heads for anatomic reference. (d) Fused ¹⁸F-FDG PET/CT scan shows FDG-avid adipose tissue in axillary (white arrow), paravertebral (black arrow), and upper mediastinal (arrowhead) regions. (e) Coronal reformatted CT scan for anatomic reference shows supraclavicular and axillary fat. (f) Fused ¹⁸F-FDG PET/CT scan shows FDG uptake in fat-attenuation areas, including supraclavicular (white arrow) and paravertebral (black arrow) regions. (g) Coronal multiple intensity projection image shows extensive uptake consistent with BAT throughout cervical, supraclavicular (white arrow), axillary, and paravertebral (black arrow) regions.

Figure 2c:

Images in healthy 21-year-old female volunteer (body mass index: 19 kg/m²) show BAT activated by means of a cooling vest protocol (subject wore a cooling vest 1 hour before and 1 hour after FDG administration). (a) Axial CT scan at level of T1 vertebral body for anatomic reference shows supraclavicular fat (arrow). (b) Superimposed ¹⁸F-FDG PET scan shows FDG uptake in fat-attenuation regions, including supraclavicular (white arrow), paravertebral (black arrow), and intermuscular (arrowheads) regions. (c) Axial CT image at level of humeral heads for anatomic reference. (d) Fused ¹⁸F-FDG PET/CT scan shows FDG-avid adipose tissue in axillary (white arrow), paravertebral (black arrow), and upper mediastinal (arrowhead) regions. (e) Coronal reformatted CT scan for anatomic reference shows supraclavicular and axillary fat. (f) Fused ¹⁸F-FDG PET/CT scan shows FDG uptake in fat-attenuation areas, including supraclavicular (white arrow) and paravertebral (black arrow) regions. (g) Coronal multiple intensity projection image shows extensive uptake consistent with BAT throughout cervical, supraclavicular (white arrow), axillary, and paravertebral (black arrow) regions.

Figure 2d:

Images in healthy 21-year-old female volunteer (body mass index: 19 kg/m²) show BAT activated by means of a cooling vest protocol (subject wore a cooling vest 1 hour before and 1 hour after FDG administration). (a) Axial CT scan at level of T1 vertebral body for anatomic reference shows supraclavicular fat (arrow). (b) Superimposed ¹⁸F-FDG PET scan shows FDG uptake in fat-attenuation regions, including supraclavicular (white arrow), paravertebral (black arrow), and intermuscular (arrowheads) regions. (c) Axial CT image at level of humeral heads for anatomic reference. (d) Fused ¹⁸F-FDG PET/CT scan shows FDG-avid adipose tissue in axillary (white arrow), paravertebral (black arrow), and upper mediastinal (arrowhead) regions. (e) Coronal reformatted CT scan for anatomic reference shows supraclavicular and axillary fat. (f) Fused ¹⁸F-FDG PET/CT scan shows FDG uptake in fat-attenuation areas, including supraclavicular (white arrow) and paravertebral (black arrow) regions. (g) Coronal multiple intensity projection image shows extensive uptake consistent with BAT throughout cervical, supraclavicular (white arrow), axillary, and paravertebral (black arrow) regions.

Figure 2e:

Images in healthy 21-year-old female volunteer (body mass index: 19 kg/m²) show BAT activated by means of a cooling vest protocol (subject wore a cooling vest 1 hour before and 1 hour after FDG administration). (a) Axial CT scan at level of T1 vertebral body for anatomic reference shows supraclavicular fat (arrow). (b) Superimposed ¹⁸F-FDG PET scan shows FDG uptake in fat-attenuation regions, including supraclavicular (white arrow), paravertebral (black arrow), and intermuscular (arrowheads) regions. (c) Axial CT image at level of humeral heads for anatomic reference. (d) Fused ¹⁸F-FDG PET/CT scan shows FDG-avid adipose tissue in axillary (white arrow), paravertebral (black arrow), and upper mediastinal (arrowhead) regions. (e) Coronal reformatted CT scan for anatomic reference shows supraclavicular and axillary fat. (f) Fused ¹⁸F-FDG PET/CT scan shows FDG uptake in fat-attenuation areas, including supraclavicular (white arrow) and paravertebral (black arrow) regions. (g) Coronal multiple intensity projection image shows extensive uptake consistent with BAT throughout cervical, supraclavicular (white arrow), axillary, and paravertebral (black arrow) regions.

Figure 2f:

Images in healthy 21-year-old female volunteer (body mass index: 19 kg/m²) show BAT activated by means of a cooling vest protocol (subject wore a cooling vest 1 hour before and 1 hour after FDG administration). (a) Axial CT scan at level of T1 vertebral body for anatomic reference shows supraclavicular fat (arrow). (b) Superimposed ¹⁸F-FDG PET scan shows FDG uptake in fat-attenuation regions, including supraclavicular (white arrow), paravertebral (black arrow), and intermuscular (arrowheads) regions. (c) Axial CT image at level of humeral heads for anatomic reference. (d) Fused ¹⁸F-FDG PET/CT scan shows FDG-avid adipose tissue in axillary (white arrow), paravertebral (black arrow), and upper mediastinal (arrowhead) regions. (e) Coronal reformatted CT scan for anatomic reference shows supraclavicular and axillary fat. (f) Fused ¹⁸F-FDG PET/CT scan shows FDG uptake in fat-attenuation areas, including supraclavicular (white arrow) and paravertebral (black arrow) regions. (g) Coronal multiple intensity projection image shows extensive uptake consistent with BAT throughout cervical, supraclavicular (white arrow), axillary, and paravertebral (black arrow) regions.

Figure 2g:

Images in healthy 21-year-old female volunteer (body mass index: 19 kg/m²) show BAT activated by means of a cooling vest protocol (subject wore a cooling vest 1 hour before and 1 hour after FDG administration). (a) Axial CT scan at level of T1 vertebral body for anatomic reference shows supraclavicular fat (arrow). (b) Superimposed ¹⁸F-FDG PET scan shows FDG uptake in fat-attenuation regions, including supraclavicular (white arrow), paravertebral (black arrow), and intermuscular (arrowheads) regions. (c) Axial CT image at level of humeral heads for anatomic reference. (d) Fused ¹⁸F-FDG PET/CT scan shows FDG-avid adipose tissue in axillary (white arrow), paravertebral (black arrow), and upper mediastinal (arrowhead) regions. (e) Coronal reformatted CT scan for anatomic reference shows supraclavicular and axillary fat. (f) Fused ¹⁸F-FDG PET/CT scan shows FDG uptake in fat-attenuation areas, including supraclavicular (white arrow) and paravertebral (black arrow) regions. (g) Coronal multiple intensity projection image shows extensive uptake consistent with BAT throughout cervical, supraclavicular (white arrow), axillary, and paravertebral (black arrow) regions.

Figure 3:

BAT activation induced by means of cold exposure but not sympathomimetic treatment. Coronal ¹⁸F FDG PET (left column), CT (middle column), and fused PET/CT (right column) images in healthy volunteer after control saline treatment (top row), ephedrine treatment (middle row), or cold treatment by means of a cooling vest (bottom row). Arrowheads indicate tracer uptake specifically in response to cold treatment. (Reprinted, with permission, from reference 41.)

Figure 4b:

BAT activation in humans with β3-adrenergic receptor agonist mirabegron. Twelve subjects underwent ¹⁸F-FDG PET after administration of either placebo or 200-mg mirabegron. (a) ¹⁸F-FDG PET images in 21-year-old man after administration of placebo (left) and mirabegron (right). There are multiple sites of BAT uptake after mirabegron administration. (b) Graph shows results from all 12 patients. There is a significant increase in detectable BAT after mirabegron administration. Patient identification numbers are listed on graph. (Reprinted, with permission, from reference 9.)

Going forward, ¹⁸F-FDG PET/CT will undoubtedly continue to evolve and play a dominant role in BAT imaging. With regard to image acquisition, it is likely that further improvements in instrumentation will continue to deliver higher spatial resolution and decreased overall radiation doses. More sophisticated imaging protocols are likely to become prevalent, such as the use of dynamic rather than static PET imaging. This involves serial imaging beginning at the time of tracer administration and allows the calculation of glucose uptake rates and time-activity curves that are less sensitive to confounding factors that affect steady-state glucose metabolism (42,66). Beyond technical innovations related to image acquisition, the most impactful improvements for ¹⁸F-FDG PET/CT may come in the realm of improved methods for standardized image quantitation and analysis. For instance, although current studies rely heavily on single-location maximum SUV as a readout of both BAT presence and activity, advances in automated segmentation will likely spur a movement toward the use of volumetric measurements such as mean SUV or total glycolytic activity (product of volume and mean SUV) as superior quantitative measures of BAT function (59).

Other PET/single photon emission CT tracers.—Although ¹⁸F-FDG has proved to be the early workhorse of BAT imaging, a variety of other PET tracers have been investigated as sources of potentially valuable tissue contrast mechanisms. One of the earliest such tracers was ¹⁸F- or ¹⁴C-fluorobenzyltriphenyl phosphonium, a probe that accumulates in proportion to cellular membrane potential. Because mitochondria maintain an elevated membrane potential and are enriched in brown adipocytes, fluorobenzyltriphenyl phosphonium accumulates to high levels in inactive BAT (67). The downside of this approach is that tracer accumulation decreases in activated BAT, making it best suited for use as a marker of BAT volume rather than activity. This is also true for (S,S)-O-[¹¹C]methylreboxetine, a PET tracer that labels sites of sympathetic innervation by binding to the norepinephrine transporter. Although no published data are yet available regarding ¹¹C-methylreboxetine use in humans, preclinical data suggest that accumulation of this tracer is relatively high in inactive BAT (higher uptake relative to heart than for FDG); however, little increase is seen upon BAT activation (68).

Conversely, a substantial increase in uptake in activated BAT can be seen when using tracers responsive to nutrient uptake and metabolic status. Thus, both 14 (R,S)-[¹⁸F]fluoro-6-thiaheptadecanoic acid (a marker of fatty acid uptake) and ¹¹C-acetate (a marker of oxidative activity) show increased uptake in BAT after activation in healthy human volunteers (69). Increases in activated BAT uptake have also been seen with ¹⁵O-O₂ and ¹⁵O-CO, both markers of oxygen consumption and, therefore, oxidative rate (60). In addition to fatty acid uptake and oxidative function, the crucial parameter of perfusion has also been quantitated by using dynamic PET imaging with ¹⁵O-H₂O (70). It should also be noted that BAT imaging with PET is not limited to small molecule approaches. New cell surface markers of brown adipocytes continue to be identified (71,72) and may be amenable to development as tracers by using immuno-PET approaches already validated for cell surface markers in other systems (73).

Finally, it should be noted that several other clinically used SPECT tracers, including technetium 99m (^99mTc)-methoxyisobutylisonitrile (sestamibi) and iodine 123 (¹²³I)-metaiodobenzylguanidine, have been reported to depict BAT in both human and animal studies (74–76). Sestamibi is a perfusion agent that accumulates in cells with high cellular and mitochondrial membrane potentials, and the overall prevalence of detectable human BAT with use of ^99mTc-sestamibi is approximately 6%–7%, which is similar to that observed in studies that used ¹⁸F-FDG PET (77). Although increased uptake of ^99mTc-sestamibi is seen in activated BAT, the change is small in comparison with PET agents such as ¹⁸F-FDG (76), making it less suited for studies of BAT activity.

MR Imaging

The histologic differences between WAT and BAT create an opportunity to augment the signal obtained from BAT by virtue of its intrinsic contrast mechanisms. In particular, the multilocular lipid content, dense mitochondrial packing, and relatively higher capillarity within BAT collectively lead to significantly higher water content and increased magnetic susceptibility, raising the possibility of selective BAT imaging with use of MR imaging–based approaches. The lack of ionizing radiation makes MR imaging a particularly attractive modality, as it opens the door to repeated measurements in patients of all age groups.

One of the first MR imaging–based approaches attempted for BAT imaging used MR spectroscopy and chemical shift imaging. These techniques exploit the differential precession frequencies of protons between lipidic and aqueous environments to achieve water-fat separation (78). Early studies revealed that BAT could be successfully imaged by using this method, both ex vivo and in vivo (79,80). Subsequent studies further elucidated characteristic signatures of BAT by using MR spectroscopy, including lower levels of unsaturated triglycerides relative to WAT as well as expected lower fat and higher water fractions (Fig 5) (81). A related approach that used multiecho Dixon-based fat-water separation likewise demonstrated the possibility of separating BAT and WAT on the basis of relative fat fraction (82). In experimental animals, this approach successfully helped identify interscapular BAT and reduced BAT fat fraction with lower ambient temperature (82) and was subsequently used in human studies.

Figure 5a:

Images in healthy 22-year-old female volunteer (body mass index: 28 kg/m²) after cold activation of BAT by using a cooling vest protocol. (a) Coronal Dixon fat-only MR image for anatomic reference shows cervical and supraclavicular fat (arrow). (b) Fused ¹⁸F-FDG PET/MR image shows increased FDG uptake in areas corresponding to fat (arrow), consistent with BAT. (c) Axial Dixon fat-only MR image for anatomic reference. (d) Fused ¹⁸F-FDG PET/MR image shows increased FDG uptake in supraclavicular (white arrow) and intermuscular (black arrow) fat consistent with BAT. (e) Axial Dixon water-only MR image at same level shows pixel values indicating relatively higher water content in FDG-avid supraclavicular fat (arrow) compared with subcutaneous fat (*).

Figure 5b:

Images in healthy 22-year-old female volunteer (body mass index: 28 kg/m²) after cold activation of BAT by using a cooling vest protocol. (a) Coronal Dixon fat-only MR image for anatomic reference shows cervical and supraclavicular fat (arrow). (b) Fused ¹⁸F-FDG PET/MR image shows increased FDG uptake in areas corresponding to fat (arrow), consistent with BAT. (c) Axial Dixon fat-only MR image for anatomic reference. (d) Fused ¹⁸F-FDG PET/MR image shows increased FDG uptake in supraclavicular (white arrow) and intermuscular (black arrow) fat consistent with BAT. (e) Axial Dixon water-only MR image at same level shows pixel values indicating relatively higher water content in FDG-avid supraclavicular fat (arrow) compared with subcutaneous fat (*).

Figure 5c:

Images in healthy 22-year-old female volunteer (body mass index: 28 kg/m²) after cold activation of BAT by using a cooling vest protocol. (a) Coronal Dixon fat-only MR image for anatomic reference shows cervical and supraclavicular fat (arrow). (b) Fused ¹⁸F-FDG PET/MR image shows increased FDG uptake in areas corresponding to fat (arrow), consistent with BAT. (c) Axial Dixon fat-only MR image for anatomic reference. (d) Fused ¹⁸F-FDG PET/MR image shows increased FDG uptake in supraclavicular (white arrow) and intermuscular (black arrow) fat consistent with BAT. (e) Axial Dixon water-only MR image at same level shows pixel values indicating relatively higher water content in FDG-avid supraclavicular fat (arrow) compared with subcutaneous fat (*).

Figure 5d:

Images in healthy 22-year-old female volunteer (body mass index: 28 kg/m²) after cold activation of BAT by using a cooling vest protocol. (a) Coronal Dixon fat-only MR image for anatomic reference shows cervical and supraclavicular fat (arrow). (b) Fused ¹⁸F-FDG PET/MR image shows increased FDG uptake in areas corresponding to fat (arrow), consistent with BAT. (c) Axial Dixon fat-only MR image for anatomic reference. (d) Fused ¹⁸F-FDG PET/MR image shows increased FDG uptake in supraclavicular (white arrow) and intermuscular (black arrow) fat consistent with BAT. (e) Axial Dixon water-only MR image at same level shows pixel values indicating relatively higher water content in FDG-avid supraclavicular fat (arrow) compared with subcutaneous fat (*).

Figure 5e:

Images in healthy 22-year-old female volunteer (body mass index: 28 kg/m²) after cold activation of BAT by using a cooling vest protocol. (a) Coronal Dixon fat-only MR image for anatomic reference shows cervical and supraclavicular fat (arrow). (b) Fused ¹⁸F-FDG PET/MR image shows increased FDG uptake in areas corresponding to fat (arrow), consistent with BAT. (c) Axial Dixon fat-only MR image for anatomic reference. (d) Fused ¹⁸F-FDG PET/MR image shows increased FDG uptake in supraclavicular (white arrow) and intermuscular (black arrow) fat consistent with BAT. (e) Axial Dixon water-only MR image at same level shows pixel values indicating relatively higher water content in FDG-avid supraclavicular fat (arrow) compared with subcutaneous fat (*).

One limitation of the Dixon method is that it cannot be used to probe BAT content in mixed voxels; that is, situations where the voxel of interest contains both BAT and WAT could prevent quantification of lipid abundance. To overcome this limitation, a spectroscopy-based approach that uses intermolecular zero-quantum coherence transitions between lipid and water spins has been described (83). This approach relies on the fact that, as a consequence of smaller lipid droplet size, protons in water and lipids are in close proximity to one another within BAT, which therefore demonstrates maximal intermolecular zero-quantum coherence signal due to equal admixture of water- and fat-derived spins. The relevant spatial correlation scale of this technique is defined by the user and adjusted to the same order as the approximate diameter of an adipocyte. The authors demonstrate that this technique can be used to detect a water-methylene intermolecular zero-quantum coherence peak specific for BAT, and, as a result of the correlation length scale, the technique is intrinsically insensitive to field inhomogeneities arising over larger scales. Further work from the same group demonstrated that this technique could be used in animals to generate BAT-specific coherence images, or BATSCI, which closely correlate with ¹⁸F-FDG PET signal after BAT stimulation (84). The application of this approach in human subjects therefore represents an interesting avenue for further research.

Although spectroscopic and Dixon-based techniques clearly can help identify BAT-derived signal, these approaches are technically challenging, necessitating specialized pulse sequences and data analysis platforms. In principle, the simplest MR imaging–based approach for BAT imaging would use conventional (ie, clinically available) sequences. Recently, it was reported that BAT could be identified in small animals by using fast spin-echo sequences on the basis of the observation that signal intensity varied between BAT and WAT as a function of water content (85). With use of this approach, the authors demonstrated correlation with dissected BAT mass. Subsequently, the same group demonstrated that BAT could be detected in human subjects by using a water saturation approach that incorporates both T2-weighted spin-echo as well fast spin-echo sequences (86). This approach correlated well with the results of Dixon-based water-fat separation as well as with retrospective analysis of patient-matched ¹⁸F-FDG PET/CT.

In addition to measuring BAT mass, MR imaging–based metrics and methodologies are also needed for quantifying the activity of BAT. The enriched vascular network within BAT raises the possibility that the increased cellular metabolism seen during activation might be accompanied by increased blood flow. This is known to be the case in the brain, where increase in activity-related blood flow has been extensively used in functional MR imaging by means of BOLD imaging, which exploits the differential paramagnetic properties of oxyhemoglobin and deoxyhemoglobin to generate contrast. In fact, BAT activation by the β3-adrenergic agonist CL-316,243 can be detected by using a functional MR imaging approach in small animals (85). To extend this work, the same group subsequently investigated whether BOLD signal changes could be observed with functional MR imaging in human subjects subjected to cold challenge. Volunteers were exposed to 13°–16°C for 60 minutes while wearing a vest circulating chilled water during MR imaging. In response to cold exposure, an increase in BOLD signal within BAT was noted, which was in fact substantially larger than that seen in the brain during conventional functional MR imaging (86) (Fig 6). This raises the possibility that BOLD-based MR techniques may have substantial utility for imaging BAT activation in dynamic studies.

Figure 6:

Detection of BAT activation in three human subjects with functional MR imaging. Significant BOLD signal changes (increasing from red to yellow) were detected after cold challenge (13°–16°C) and superimposed on anatomic images. Green circles indicate areas of cold-activated BOLD signal. (Reprinted, with permission, from reference 86.)

Hyperpolarized MR imaging is another approach that may have utility in BAT imaging. Owing to the relatively low degree of polarization inducible by MR magnets, conventional MR imaging is limited to signal generated by proton spins in water or lipids. Hyperpolarized MR imaging uses in vitro nuclear spin polarization, providing orders of magnitude improvement in the signal-to-noise ratio, as well as access to new substrates as contrast agents (87). For instance, breakdown of hyperpolarized ¹³C-pyruvate into bicarbonate and lactate was recently demonstrated in norepinephrine-activated interscapular BAT in rats (88). Noble gases in particular have been used as hyperpolarized contrast agents owing to their inert nature and, consequently, lower potential for biologic effects. An extensive literature now exists on pulmonary imaging with use of ¹²⁹Xe, which has proved to be a safe and effective tracer in humans (89). A recent study, moreover, reported the use of hyperpolarized ¹²⁹Xe gas to detect BAT activation, yielding a more than 15-fold increase in hyperpolarized xenon uptake within BAT during stimulation (90), which suggests that this approach represents a viable future approach for human BAT imaging. Intriguingly, use of the same agent to directly track BAT heat production by exploiting the temperature dependence of xenon chemical shift has been reported (90). The excitement surrounding hyperpolarized MR imaging must, however, be tempered by acknowledgment that this remains an experimental technique. Widespread adoption of hyperpolarized MR imaging will depend on the more general availability of hyperpolarizers and/or ready access to the polarized agents themselves, which generally have short half-lives.

As with most MR imaging–based approaches, the next generation of experimental implementations will almost certainly involve combined PET/MR imaging. The advantages of such an approach are obvious, as it captures both the molecular readout of PET with the sophisticated tissue-level analysis allowed by MR imaging (Fig 5). Indeed, there has already been an early stage report regarding the use of ¹⁸F-FDG PET/MR imaging to detect BAT activation in response to either noradrenaline or adenosine A_2A receptor agonist treatment in mice (91). While this preclinical implementation appears promising, the true future promise of the technique lies in generating increasingly specific readouts of BAT mass and activation, both with PET and MR imaging. One can therefore imagine a future hybrid imaging modality that combines imaging of a BAT-specific cell surface receptor with PET to inform on mass with a BOLD MR imaging readout of activation.

US Imaging

Beyond PET and MR imaging, a variety of imaging techniques have shown promise for BAT imaging, most of which remain predominantly in the preclinical stage. The technique that seems best suited for clinical application is US. This is due to several key factors: it is relatively inexpensive, relies on widely available equipment and expertise, does not involve ionizing radiation, and can be used to dynamically interrogate numerous anatomic areas where BAT has been found. Because of these advantages, it also holds potential for serial imaging, a crucial point for longitudinal studies involving pharmacologic interventions or natural history of disease. Although targeted tissue contrast techniques for US are still largely in the experimental stage, there is a large body of literature on the use of microbubbles for vascular contrast imaging (92). Early stage work in animals has indeed demonstrated the feasibility of microbubble contrast-enhanced US for BAT imaging (93). This approach uses brief pulses of high-energy ultrasound to locally destroy intravenously administered microbubbles, followed by dynamic imaging to monitor blood inflow (quantitated as reacquisition of microbubble-mediated contrast).

With use of this technique, BAT blood flow can be reliably quantitated and that flow increases after treatment with norepinephrine, which implies that BAT activation can be monitored noninvasively by using contrast-enhanced US (93). This is supported by the observation that a norepinephrine-induced increase in BAT perfusion was blunted in mice lacking UCP1, which also fails to induce oxygen consumption and thermogenesis in response to norepinephrine (94). In follow-up work, the same group demonstrated that total BAT mass could also be derived by using steady-state imaging during continuous norepinephrine and microbubble administration (95). Contrast-enhanced US–derived BAT mass correlated well with mass at necropsy, and the technique demonstrated reasonable intra- and interobserver variability. Overall, although this approach is limited to more superficial BAT depots owing to inherent technical limitations (eg, penetration depth), contrast-enhanced US is promising as a potentially clinically relevant technique allowing quantitation of both BAT volume and activation (Fig 7). This approach may become even more attractive if further advances in targeted US contrast eventually allow molecular targeting of contrast agents.

Figure 7:

Contrast-enhanced US scans of, A, B, control or, C, D, obese db/db mice before (A, C) and after (B, D) intravenous infusion of BAT-activating factor norepinephrine (NE) show limited BAT activation in db/db mice. Dotted line indicates outline of left BAT lobe. White bar = 2.5 mm. (Reprinted, with permission, from reference 95.)

Other Imaging Modalities

Thermometry.—Another technique that has shown potential for clinical imaging is direct thermometry. Because the principal purpose of classic BAT is heat generation, it makes sense that heat measurement should be a direct reflection of BAT function. Indeed, studies in human subjects have demonstrated that thermal imaging can depict increases in supraclavicular temperature in response to cold challenge and that this increase is specific to the supraclavicular fossa (96,97). The ability to detect this increase is somewhat surprising given that most of the generated heat is likely to be dissipated to the circulation by means of BAT blood flow, and it remains to be seen if this will represent a challenge for further clinical applications. Although MR imaging can in principle be used to quantify thermogenic activity by using MR thermometry techniques, to our knowledge no studies have attempted BAT detection with this approach.

Preclinical techniques: near infrared fluorescence (peptide), fatty acid uptake, UCP1-luc.—Several optical approaches to BAT imaging have shown promise in animal models and, despite unclear translational potential, will doubtlessly have value in future preclinical studies. This includes optical imaging of ¹⁸F-FDG itself by means of Cerenkov luminescence, demonstrating the expected accumulation in interscapular BAT (98). Other optical approaches include imaging of BAT vasculature by using a near-infrared coupled peptide probe, which allowed visualization of both inactive and activated interscapular BAT in animal studies (99). This probe notably also labeled subcutaneous beige fat, which has proved to be difficult with use of other techniques. Importantly, the conjugation chemistry that was used is compatible with radioisotope labeling and thus provides a way for clinical translation should this probe also prove specific in humans.

Finally, bioluminescence approaches have been used to image BAT and beige fat in transgenic mice. Covalent coupling of the luciferase substrate luciferin to an unbranched (62) fatty acid produced an imaging probe (FFA-SS-luc) that is specifically taken up in tissues that express fatty acid transporters, including skeletal muscle, heart, WAT, and BAT. BAT could then be imaged in transgenic mice expressing luciferase in all tissues (53). A more direct approach was taken in a recent study in which luciferase was placed under the direct control of the UCP1 promoter, generating mice in which luciferase is expressed in all cells expressing UCP1 (100). These animals proved useful for monitoring brown and beige fat with in vivo molecular imaging, as well as for generation of cell lines that can be used for screening to identify compounds capable of inducing UCP1 production.

Future Directions

The long-term goal in clinical BAT imaging is to develop noninvasive methods for detecting both BAT quantity and function with use of safe and clinically relevant techniques. Such approaches will further our understanding of the role of BAT in health and disease and also provide a prognostic marker of metabolic disease. Accomplishing this goal—in addition to selective imaging of beige fat—will require advances in BAT basic science (identification of specific and abundant markers), as well as in imaging sciences (improved PET standardization and quantitation, further exploration of MR imaging– and US-based methods, and selective BAT activation methods). Contributions will also likely come from better understanding the human genetic variation underlying differences in BAT abundance and activity. As with other diseases, the potential value of combining imaging and serum biomarkers should also be explored. Ultimately, the goal should be the development of a simple and noninvasive method for comprehensive BAT assessment that might provide an important biomarker of overall metabolic risk.

Conclusion

The obesity epidemic has made the development of new diagnostic and therapeutic approaches for metabolic disease an urgent priority. BAT imaging has the potential to contribute greatly in this regard and has made great strides from its origins in clinical nuclear medicine with ¹⁸F-FDG PET. Although a wide variety of techniques are now available for both preclinical and clinical use, it is clear that each has limitations and none will individually suffice for all facets of BAT investigation. It is crucial that progress continues in each modality, which will ensure that imaging techniques continue to play a central role in revealing the biology of BAT and its relevance for human metabolic disease.

Essentials

• ■ Brown adipose tissue (BAT) regulates basal and inducible energy expenditure in humans and is composed of cells that contain numerous lipid droplets and mitochondria.

• ■ BAT converts energy stores into heat by expressing uncoupling protein 1 (UCP1), which uncouples cellular respiration from energy production; clusters of UCP1-positive cells are also found amid white adipose tissue after cold exposure and are known as “beige” or “brite” adipocytes.

• ■ Imaging of BAT has relied on its fluorine 18 (¹⁸F) fluorodeoxyglucose (FDG) avidity, making ¹⁸F FDG PET/CT the dominant modality for measuring its presence and quantity.

• ■ Next-generation approaches for BAT imaging include specific PET tracers, MR imaging, and US-based strategies to determine BAT mass and activity.

• ■ Imaging of BAT will be essential as investigators attempt to harness the biology of this tissue for the treatment of obesity, type 2 diabetes, and metabolic syndrome.