Non‐invasive methods for the assessment of brown adipose tissue in humans

Abstract

Brown adipose tissue (BAT) is a recently rediscovered tissue in people that has shown promise as a potential therapeutic target against obesity and its metabolic abnormalities. Reliable non‐invasive assessment of BAT volume and activity is critical to allow its importance in metabolic control to be evaluated. Positron emission tomography/computed tomography (PET/CT) in combination with 2‐deoxy‐2‐[¹⁸F]fluoroglucose administration is currently the most frequently used and most established method for the detection and quantification of activated BAT in humans. However, it involves radiation exposure and can detect activated (e.g. after cold exposure), but not quiescent, BAT. Several alternative methods that overcome some of these limitations have been developed including different PET approaches, single‐photon emission imaging, CT, magnetic resonance based approaches, contrast‐enhanced ultrasound, near infrared spectroscopy, and temperature assessment of fat depots containing brown adipocytes. The purpose of this review is to summarize and critically evaluate the currently available methods that non‐invasively probe various aspects of BAT biology in order to assess BAT volume and/or metabolism. Although several of these methods show promise for the non‐invasive assessment of BAT volume and function, further research is needed to optimize them to enable an accurate, reproducible and practical means for the assessment of human BAT content and its metabolic function.

Introduction

Brown adipose tissue (BAT) is a relatively recently “rediscovered” tissue in adult humans, as a result of non‐invasive positron emission tomography/computed tomography (PET/CT) imaging, and has been proposed as a potential therapeutic target for obesity and its related metabolic perturbations. Brown adipocytes are morphologically and functionally distinct from white adipocytes (i.e. BAT adipocytes having smaller cell size, higher innervation and vascularization, multiple lipid droplets, and more mitochondria with high expression of uncoupling protein 1 (UCP1)) (Cannon & Nedergaard, 2004). UCP1 is a proton carrier protein located in the inner mitochondrial membrane and allows the entrance of protons produced during substrate oxidation from the electron transport chain back into the mitochondrial matrix bypassing the ATP‐synthase leading to heat production (thermogenesis) instead of chemical energy ATP (Cannon & Nedergaard, 2004). Upon activation with cold or other stimuli, BAT takes up glucose and lipids to maintain thermogenesis and presumably to refresh the lipid stores in BAT (Cannon & Nedergaard, 2004).

While the scientific interest in BAT has increased substantially since it was identified as active on PET/CT imaging, methods for the assessment of BAT are still in their infancy. The existence of functional BAT was first confirmed using 2‐deoxy‐2‐[¹⁸F]fluoroglucose (¹⁸F‐FDG)‐PET/CT where fat density tissue was identified with ¹⁸F‐FDG avidity in expected locations of BAT (Hany et al. 2002; Cohade et al. 2003b; Nedergaard et al. 2007). Subsequently, ¹⁸F‐FDG‐PET/CT phenotyping of BAT was informed by a combination with adipose tissue biopsies and molecular biology analysis (Cypess et al. 2009; van Marken Lichtenbelt et al. 2009; Virtanen et al. 2009). Ever since, several alternative non‐invasive BAT detection approaches have been proposed including different PET approaches, single‐photon emission imaging and single photon emission computerized tomography (SPECT), CT, magnetic resonance (MR)‐based approaches, contrast‐enhanced ultrasound (CEU), near infrared spectroscopy (NIRS) and temperature assessment of fat depots containing high amounts of brown adipocytes. The ideal method for the assessment of BAT in humans would: (i) involve no exposure to ionizing radiation, (ii) not require exposure to cold, (iii) be able to detect metabolically active and inactive BAT, (iv) have high resolution, reproducibility and specificity for BAT versus white adipose tissue (WAT), (v) be inexpensive and easy to perform, and (vi) have a broad field of view and deep penetration allowing the assessment and quantification of both shallow and deep BAT depots. The purpose of this review is to summarize and critically evaluate the advantages and limitations of the currently available non‐invasive methods for the assessment of BAT in people (Table 1).

Table 1.

Methods for the non‐invasive assessment of brown adipose tissue in humans

Method	Mechanism	Radiation activity administered	Cold stimulation enhances signal	Standardized criteria for assessment of BAT volume	Cost	Labour intensive	Other comments
PET tracers
18F‐FDG	Glucose disposal	2–10 mCi	Yes	Yes	High	Yes	−Quantification of activated BAT based on its ability to take up glucose when activated.
FTHA	FFA disposal	5 mCi	Yes	No	High	Yes	−Does not account for the utilization of intracellular lipids and circulating triglycerides. −Cyclotron in close proximity needed.
11C‐Acetate	Substrate for Krebs cycle, blood flow	5 mCi	Yes	No	High	Yes	+Assessment of BAT blood flow and oxidative capacity for circulating substrates. −Cyclotron in close proximity needed.
15O‐O, 15O‐CO, 15O‐H2O	O2 consumption	50–100 mCi	Yes	No	High	Yes	+Assessment of BAT metabolic rate and blood flow. −Cyclotron in very close proximity needed.
18F‐Dopamine	Dopamine analogue	1 mCi	Testing pending	No	High	Yes	+Assessment of BAT sympathetic innervation.
11C‐MRB	Ligand for NE transporter	10–20 mCi	No	No	High	Yes	+Assessment of BAT sympathetic innervation. −Cyclotron in close proximity needed.
68Ga‐DOTA‐NOC	Octreotide analogue	1.7–5.5 mCi	Testing pending	No	High	Yes	−Gallium generator required in close proximity.
SPECT tracers
MIBG	NE analogue	10 mCi	Yes	No	High	Yes	+Assessment of BAT sympathetic innervation.
99mTc‐MIBI	Blood flow, mitochondrial density	10–30 mCi	Yes	No	High	Yes
99mTc‐Tetrofosmin	Mitochondrial density	10–30 mCi	Testing pending	No	High	Yes
67Ga‐Citrate	Iron analogue	2–5 mCi	Testing pending	No	High	Yes
MRI/MRS
T1‐ and T2‐weighted MRI	Structural differences in BAT vs. WAT	None	Further testing pending	No	High	Yes	+Excellent soft tissue contrast. −Substantial overlap between BAT and WAT.
1H chemical shift	Fat/water content	None	Further testing pending	No	High	Yes	+Reports on chemical composition of tissue. −MRS voxels can be large. −Some mapping techniques require large assumptions about underlying MR spectrum.
BOLD MRI	Blood flow, oxy‐/deoxyhaemoglobin	None	Yes	No	High	Yes	+Proxy of BAT oxygenation/perfusion and metabolic status. −Sensitive to spurious magnet susceptibility artifacts. −Biophysical modelling is controversial.
CT	Tissue radiodensity	Variable	Further testing pending	No	High	No	+Anatomical characterization.
Contrast‐enhanced ultrasound	Perfusion	None	Yes	No	Low	Moderately
Near infrared spectroscopy	Blood flow, mitochondrial density	None	No	No	Low/moderate	No	−Limited depth of penetration and field of view. −Potential confounding from overlying tissues.
Infrared thermography	Heat production	None	Yes	No	Low	No	−Potential confounding from overlying tissues. −Assessment of superficial BAT.
Thermal probes	Heat production	None	No	No	Low	No	−Potential confounding from overlying tissues. −Assessment of superficial BAT.

BAT, brown adipose tissue; BOLD, blood‐oxygen‐level dependent; ¹¹C‐MRB, (S,S)‐¹¹C‐O‐methylreboxetine; CT, computed tomography; FFA, free fatty acid; ¹⁸F‐FDG, 2‐deoxy‐2‐[¹⁸F]fluoroglucose; FTHA,14‐(R,S)‐[¹⁸F]fluoro‐6‐thiaheptadecanoic acid; ⁶⁸Ga‐DOTA‐NOC, ⁶⁸Ga‐labelled 1,4,7,10‐tetraazacyclododecane‐1,4,7,10‐tetraacetic acid–1‐NaI³‐octreotide; MIBG, ¹²³I‐meta‐iodobenzylguanidine; MRI, magnetic resonance imaging; MRS, magnetic resonance spectroscopy; NE, noradrenaline; PET, positron emission tomography; SPECT, single‐photon emission imaging computerized tomography; ^99mTc‐MIBI, ^99mTc‐methylisobutylisonitrile; WAT, white adipose tissue; +, advantage; −, disadvantage.

Positron emission tomography

PET is the most frequently used imaging method for the assessment of BAT typically as an ¹⁸F‐FDG‐PET/CT scan. It requires the administration of a short‐lived positron emitting radiotracer, which specifically accumulates by binding to receptors or by transport and metabolism in tissues. While the main PET tracer used is ¹⁸F‐FDG, a variety of other PET radiotracers are becoming available. The radioactive decay of the positron emitter labelled tracers produces positrons, which travel a short distance in tissue and then collide with electrons in the tissues emitting two 511 keV photons and a neutrino. The dual photons (annihilation photons) exit the point of their production in opposite directions and can be detected by a ring of γ‐ray detectors as part of the PET scanner. Since the photons are emitted simultaneously, they are detected at nearly the same time by the detectors. This “coincidence” timing (electronic collimation), coupled with knowledge of the exact timing the photons are detected (Time of Flight information), after millions of events, allows for the relatively precise location, quantification of the site of position emission, and the number of radioactive molecules at a given site. Knowledge of body thickness and expected attenuation of photons is required and is most commonly provided by a low dose CT scan. In the early 2000s, the invention of hybrid PET/CT enabled the simultaneous assessment, with high resolution, of the radiodensity of the tissues, which allowed a better anatomical evaluation. More recently, PET–magnetic resonance imaging (MRI) has been proposed as an alternative method for the detection of BAT in people (Gariani et al. 2015). However, since MRI does not measure X‐ray attenuation, the difficulties in the precise quantification of tissue density via PET/MRI make quantitative assessments of BAT more challenging. PET acquisitions can be static or dynamic. Static acquisitions allow the image reconstruction from the count data of single time interval image sets that depict where the administered radiotracer has localized at a single time point (for example at 60–70 min post radiotracer injection). Dynamic acquisitions allow multiple sets of images of tissues to be obtained over time within a field of view and the application of kinetic models of radiotracer uptake which can provide information regarding the rate of various physiological processes (i.e. estimates of flow, uptake and substrate loss) in addition to absolute uptake of a radiotracer.

Although ¹⁸F‐FDG‐PET/CT revolutionized the discovery of functional BAT in people, leading to a paradigm shift in adipose tissue biology, it has numerous limitations: (1) a need for expensive radiological equipment and trained personnel; (2) a high cost; (3) exposure of participants/patients to ionizing radiation, which restricts its use in vulnerable populations and limits the implementation of studies that require multiple assessments of BAT; (4) requires standardized conditions (temperature, diet, medications, etc.); (5) the use of some tracers with a very short half‐time (t _1/2) requires the presence of a cyclotron in close proximity, further limiting access; (6) many of the most promising radiotracers are not US Food and Drug Administration (FDA) approved nor widely available; and (7) the kinetic modelling of dynamic PET data involves assumptions that have been primarily validated in tissues other than BAT and assume a steady state of BAT.

¹⁸F‐FDG

The ¹⁸F‐FDG‐PET is currently the most frequently used and best established method for the assessment of active BAT in people. The activated BAT takes up plasma glucose and ¹⁸F‐FDG. Once ¹⁸F‐FDG enters the cells it is phosphorylated by hexokinase but cannot be substantially further metabolized, and thus it can be easily visualized using PET. Due to its relative long t _1/2 (109 min), it does not require a cyclotron in close proximity. Figure 1 illustrates the increased ¹⁸F‐FDG uptake in the supraclavicular BAT (∼60 min post injection) using PET/CT after the patient underwent cold exposure.

Figure 1. Coronal (left) and transverse (right) 2‐deoxy‐2‐[¹⁸F]fluoro‐d‐glucose positron emission tomography‐computed tomography images from a volunteer after 6 h of cold exposure.

The intense orange colour in the supraclavicular area corresponds to brown adipose tissue. Reproduced from Chondronikola et al. (2014) with permission.

To date, many studies have used ¹⁸F‐FDG‐PET for the quantification of BAT. Many of these studies are retrospective including the review of ¹⁸F‐FDG‐PET scans conducted for diagnostic purposes during minimally controlled conditions (i.e. room temperature, duration of fasting). In fewer studies the ¹⁸F‐FDG‐PET scans were performed during standardized conditions. Age (Pfannenberg et al. 2010; Yoneshiro et al. 2011a), adiposity (Cypess et al. 2009; van Marken Lichtenbelt et al. 2009), male sex (Cypess et al. 2009; Jacene et al. 2011) and various pathological conditions (Hadi et al. 2007; Lahesmaa et al. 2014) have been associated with a lower amount of BAT. Season/outdoor temperature (Cohade et al. 2003a; Bahler et al. 2016; Yoneshiro et al. 2016), diet (Vosselman et al. 2013) and various medications (e.g. β agonists/blockers) (Cypess et al. 2015) are some of the modifiable factors that affect the amount of detectable BAT via ¹⁸F‐FDG‐PET. The criteria for the quantitative assessment of BAT via ¹⁸F‐FDG‐PET have recently been standardized with the BARCIST 1.0 (Chen et al. 2016). Hence, the lack of standardized criteria for almost a decade has led to a great variability in the identification and reporting of BAT‐related outcomes and encouraged qualitative assessment of BAT.

¹⁸F‐FDG‐PET has been used not only for the quantification of BAT, but also for the assessment of its role in glucose metabolism. Hyperglycaemia and diabetes are associated with lower amounts of detectable BAT (Jacene et al. 2011; Ouellet et al. 2011; Matsushita et al. 2014), whereas cold stimulates glucose accumulation/disposal in BAT (Orava et al. 2011; Ouellet et al. 2012) suggesting that BAT may contribute to the systemic regulation of glucose metabolism. The contribution of BAT in the systemic regulation of glucose metabolism is still debatable. Some studies reported that BAT minimally contributed to whole‐body plasma glucose utilization, which might be due to the short duration of cold exposure and the presence of mild shivering (Ouellet et al. 2012; Blondin et al. 2015). To address this gap, we studied individuals with high amounts of BAT (BAT+) or with no/minimal BAT (BAT−) during cold exposure and thermoneutral conditions using infusion of isotopically labelled glucose in conjunction with a hyperinsulinaemic euglycaemic clamp (Chondronikola et al. 2014). According to the results, cold increased whole‐body glucose disposal during basal and insulin‐stimulated conditions in the BAT+ group only, suggesting that BAT plays a significant role in the systemic regulation of glucose metabolism in humans. Further research is needed to dissect the direct and indirect contribution of BAT in glucose metabolism.

14‐(R,S)‐[¹⁸F]Fluoro‐6‐thiaheptadecanoic acid (FTHA)

FTHA (t _1/2 = 109 min) is the only PET tracer that has been used for the assessment of free fatty acid (FFA) metabolism in BAT. FTHA is metabolized via β‐oxidation and binds irreversibly to mitochondrial proteins (Ci et al. 2006; Bauwens et al. 2014). Although the validation of FTHA specifically for the assessment of BAT metabolism is pending, increased FTHA uptake in BAT in rodents coincides with increased expression of genes involved in lipid metabolism (Labbe et al. 2015). Cold increases FTHA uptake in human BAT (Ouellet et al. 2012; U Din et al. 2016). Hence, the BAT FTHA uptake accounted for only a small fraction of the total FFA turnover (Ouellet et al. 2012). Conversely, we recently reported that the amount of activated BAT is linked to adipose tissue lipolysis and whole‐body FFA oxidation independent of age and adiposity suggesting that BAT plays an important role in the regulation of systemic FFA metabolism (Chondronikola et al. 2016b). Although the current evidence supports the concept that BAT plays a role in the systemic lipid metabolism in people, future studies are needed to establish the role of BAT in circulating and intracellular triglyceride metabolism. Investigators considering the use of FTHA‐PET for the assessment of BAT lipid metabolism should recognize that FTHA uptake does not provide a universal measure of lipid metabolism, as it does not account for the utilization of intracellular lipids and lipoproteins.

¹¹C‐Acetate

¹¹C‐Acetate (t _1/2 = 20 min) is a short‐chain FFA, which is rapidly taken up from the circulation and metabolized to acetyl‐CoA to enter the Krebs cycle. ¹¹C‐Acetate‐PET has been used for the assessment of oxidative metabolism and perfusion. Cold has been shown to increase BAT perfusion and oxidative metabolism using ¹¹C‐acetate‐PET (Ouellet et al. 2012; Blondin et al. 2015), while individuals with diabetes demonstrate lower BAT oxidative capacity compared to healthy controls (Blondin et al. 2015). Although ¹¹C‐acetate‐PET can be useful for the assessment of BAT metabolism, it involves assumptions regarding the metabolic fate of the acetate in the cells and the contribution of acetate and Krebs cycle in the production of reducing equivalents and oxygen consumption (Ouellet et al. 2012). Moreover, its use for the assessment of oxidative metabolism has been validated in tissues other than BAT. Finally, its short t _1/2 limits the use of ¹¹C‐acetate to locations very proximate to a cyclotron.

¹⁵O tracers

¹⁵O tracers (t _1/2 = 2 min) have also been used for the assessment of BAT. Specifically, ¹⁵O‐CO has been used for the assessment of blood volume (due to its irreversible binding to haemoglobin), ¹⁵O‐H₂O for the assessment of tissue perfusion, and ¹⁵O‐O₂ has been used to assess the O₂ extraction from the circulation. Using ¹⁵O tracers, cold has been shown to increase BAT perfusion (Orava et al. 2011), while obesity blunts this response (Orava et al. 2013). Muzik et al. combined administration of ¹⁵O‐O₂, ¹⁵O‐CO and ¹⁵O‐H₂O with dynamic PET to estimate the contribution of BAT in energy metabolism (Muzik et al. 2013). According to the results, cold increased whole‐body energy expenditure (assessed via indirect calorimetry) and the metabolic rate of O₂ in BAT (assessed via ¹⁵O‐PET) in the BAT+ group, while BAT+ individuals also demonstrated higher metabolic O₂ rate in BAT even during thermoneutrality compared to BAT− individuals. Others reported similar results by using ¹⁵O‐O₂ and ¹⁵O‐H₂O‐PET (U Din et al. 2016). Cold exposure induces BAT activation and increases in whole‐body energy expenditure of 13–27% (van Marken Lichtenbelt et al. 2009; Yoneshiro et al. 2011b; Chondronikola et al. 2014), while the results from ¹⁵O‐PET studies indicate that BAT minimally contributed to the reported increase in energy expenditure (15–25 kcal day⁻¹) suggesting the contribution of BAT in energy expenditure is minimal. When interpreting the results of these studies, it is important to consider that chronic weight gain can result from even a small discordance in daily energy balance (e.g. an energy surplus of as little as 25 kcal day⁻¹ can result in a weight gain of 1 kg year⁻¹). Thus, even sporadic BAT activation for limited periods might have a significant cumulative impact on energy balance over time.

(S,S)‐¹¹C‐O‐Methylreboxetine (¹¹C‐MRB)

BAT is a densely innervated tissue, while noradrenaline (norepinephrine) is a major signal regulating metabolic activity in BAT (Cannon & Nedergaard, 2004). ¹¹C‐MRB (t _1/2 = 20.4 min) is a highly selective ligand for the noradrenaline transporter (NET) (Lin et al. 2012). NET is expressed in all adrenergic neurons and is responsible for the noradrenaline turnover in the synapses (Boschmann et al. 2002). In rodents, ¹¹C‐MRB is highly distributed in BAT even during room temperature, while cold only slightly increases ¹¹C‐MRB uptake in BAT (Lin et al. 2012). Similar results were also reported in people suggesting that ¹¹C‐MRB PET can be used for the assessment of BAT (Hwang et al. 2015). Since ¹¹C‐MRB crosses the brain–blood barrier and binds in adrenergic neurons, it could be used for the simultaneous assessment of BAT and probe central NET.

6‐[₁₈F]‐Fluorodopamine (¹⁸F‐FDA)

¹⁸F‐FDA (t _1/2 = 109 min) is a dopamine analogue which has also been used to assess the presence of BAT. Hadi et al. reviewed ¹⁸F‐FDA‐PET scans conducted for diagnosis of phaeochromocytoma for increased tracer uptake in fat depots linked to BAT (Hadi et al. 2007). Similar to the retrospective ¹⁸F‐FDG‐PET studies, the prevalence of increased ¹⁸F‐FDA BAT uptake was 17.9% suggesting that ¹⁸F‐FDA‐PET could be a useful approach for the assessment of BAT. No prospective studies have validated ¹⁸F‐FDA‐PET for the assessment of BAT.

⁶⁸Ga‐labelled 1,4,7,10‐tetraazacyclododecane‐1,4,7,10‐tetraacetic acid–1‐NaI3‐octreotide (⁶⁸Ga‐DOTA‐NOC)

⁶⁸Ga‐DOTA‐NOC (t _1/2 = 68 min) is an octreotide analogue that binds somatostatin receptors. In a retrospective review of diagnostic ⁶⁸Ga‐DOTA‐NOC‐PET scans, Kagna et al. reported increased ⁶⁸Ga‐DOTA‐NOC uptake in fat depots linked to BAT only in 1% of patients (Kagna et al. 2014). Considering the very low prevalence of increased uptake of ⁶⁸Ga‐DOTA‐NOC in BAT, it is unlikely that it can be used for the assessment of BAT.

Single‐photon emission imaging and single photon emission computerized tomography (SPECT)

Planar (2‐D) and SPECT (3‐D) imaging involves the administration of a γ‐emitting radiopharmaceutical that accumulates selectively to different body tissues based on radiotracer structure and tissue physiology. The accumulated radiopharmaceutical can be detected by a γ‐camera. SPECT is clinically used for the diagnosis/staging of malignancies (especially bone metastases), phaeochromocytoma, parathyroid disorders, infection, inflammatory disease and cardiac imaging among other applications. SPECT methods are becoming increasingly quantitative in their deployment with the more general distribution of SPECT/CT imaging, which allows for measurement of tissue attenuation. SPECT imaging uses radiotracers that have a longer t _1/2 and involve a higher radiation dose than most PET tracers. SPECT radiotracers do not emit two photons with decay and do not use coincidence imaging for detection of the radioactive decays. Moreover, SPECT has a lower resolution and quantification accuracy than PET. For the majority of SPECT tracers described below, the only studies available are retrospective. Therefore, prospective studies are needed to validate the ability of those tracers to assess BAT.

¹²³I‐meta‐Iodobenzylguanidine (MIBG)

MIBG (t _1/2 = 13.2 h) is a radiotracer approved by the FDA for the diagnosis of phaeochromocytoma and neuroblastoma, and for assessment of cardiac innervation. MIBG is accumulated in neurosecretory granules and secreted in the synapses, while its biodistribution depends on tissue innervation and the catecholamine production (Bombardieri et al. 2003; Fukuchi et al. 2004). Pre‐clinical studies show MIBG uptake increases in cold‐activated BAT by >3‐fold (Baba et al. 2007). Hadi et al. analysed diagnostic MIBG‐SPECT scans and reported that the prevalence of increased uptake in fat depots linked to BAT was 17.7% (Hadi et al. 2007). Further, Admiraal et al. performed ¹⁸F‐FDG‐PET and MIBG‐SPECT imaging during mild cold in healthy adults and reported that ¹⁸F‐FDG uptake in BAT correlated with MIBG uptake (Admiraal et al. 2013). MIBG‐SPECT can be used as an alternative marker to detect BAT and to assess its sympathetic innervation.

^99mTc‐Methylisobutylisonitrile (MIBI)

MIBI (t _1/2 = 6 h) is a lipophilic cation taken up in tissues with high blood flow and mitochondrial density. In rodents, MIBI bio‐distribution is higher in BAT compared to WAT, while adrenergic stimulation further increased MIBI uptake in BAT (Cypess et al. 2013). In retrospective studies, the prevalence of increased MIBI uptake in fat depots linked to BAT is 5.4–33% (Goetze et al. 2008; Cypess et al. 2013; Haghighatafshar & Farhoudi, 2016), while the presence of BAT in human adipose samples with increased MIBI uptake has been confirmed using gene expression analysis and immunohistochemistry (Cypess et al. 2013).

^99mTc‐Tetrofosmin

^99mTc‐Tetrofosmin (t _1/2 = 6 h) is also a lipophilic molecule binding to tissues with high mitochondrial density. In a retrospective study, Fukuchi et al. reported that the prevalence of increased ^99mTc‐tetrofosmin accumulation in fat depots linked to BAT was 17% (Fukuchi et al. 2003), suggesting that it may be an alternative tracer for the detection of the human BAT and the assessment of its mitochondrial function.

⁶⁷Ga‐Citrate

⁶⁷Ga‐Citrate (t _1/2 = 78.3 h) is an iron analogue and binds in transferrin and other iron‐containing structures. Rakheja et al. reviewed diagnostic SPECT scans and reported that the prevalence of increased ⁶⁷Ga‐citrate uptake in fat depots linked to BAT was 16% (Rakheja et al. 2013). These results indicate that ⁶⁷Ga‐citrate might be a biomarker for BAT, although the biological basis for such uptake is unclear and this method is not well established for the detection of BAT.

Computed tomography

CT uses rotating X‐rays and tomographic reconstructions to visualize internal body structures. Tissues with relatively low (vs. water), but higher than lung, tissue radiodensity correspond to adipose tissue (−190 to −10 Hounsfield units) (Chen et al. 2016). The variability in adipose tissue radiodensity could reflect differences in the cellular structure between BAT and WAT and/or the metabolic status of the tissue (e.g. depletion of intracellular lipids, increased perfusion). Retrospective studies have shown that ¹⁸F‐FDG‐PET‐detectable BAT has a higher radiodensity than WAT (Baba et al. 2010; Gupta et al. 2011; Hu et al. 2011; Ahmadi et al. 2013). Moreover, Baba et al. reported that BAT radiodensity increases during cold in rodents and humans (Baba et al. 2010) (Fig. 2), while others also reported similar findings (Ouellet et al. 2012; Hanssen et al. 2015; Chondronikola et al. 2016b). Limitations of CT for the detection of BAT include: (1) a need for radiological equipment and trained personnel, (2) high cost, (3) exposure to ionizing radiation, (4) restricted use in vulnerable populations, (5) an unclear performance as a sole metric of BAT presence and activation, and (6) a need for standardized conditions.

Figure 2. Assessment of brown adipose tissue (BAT) using computed tomography (CT).

A, sagittal CT and positron emission tomography–computed tomography of BAT (arrow) in rodent is greater under activated conditions than under control conditions. B, CT image of resected rodent BAT with surrounding WAT subjected to 4 h of cold stimulation shows greater CT density that is closer to water (top) than control kept at room temperature (bottom). Reproduced from Baba et al. (2010) with permission.

Magnetic resonance (MR)‐based approaches

BAT is chemically and structurally complex and thus lends itself well to non‐invasive, non‐ionizing interrogation with MR‐based techniques. Samples rich in nuclei that have a non‐zero spin, including ¹H (protons), ¹³C, ³¹P and ¹²⁹Xe, are readily studied via MR methods. Of particular interest in biological applications are ¹H spins, which are found in extremely high concentrations in vivo, mostly on water and lipid. When nuclei with non‐zero spin are placed inside a large external magnetic field, they will precess about the axis of the magnetic field. Through the clever design of experiments, precessing spins can be manipulated via radio‐frequency pulses to produce signals (characterized by frequencies, amplitudes, phases and decay rate constants) that are unique to the chemical composition, microenvironment and perfusion of the interrogated tissue. MR‐based techniques hold enormous promise in contributing to the study and understanding of BAT, though MR‐based detection/characterization of BAT is far from a “solved problem”.

T₁‐ and T₂‐weighted MRI

The earliest MR‐based studies of BAT by Osculati and Sbarbati et al., dating back to 1989, exploited microstructural differences between BAT and WAT such as lipid droplet size, water‐to‐fat ratio and density of iron‐rich mitochondria. Each of these would have a significant influence over bulk longitudinal (T₁) and transverse (T₂) relaxation time constants enabling identification of BAT on T₁‐ and T₂‐weighted images (Osculati et al. 1989, 1991; Sbarbati et al. 1997). Relaxation time constants have been precisely measured in recent years (Hamilton et al. 2011; Chen et al. 2012) and modern T₂‐weighted fast spin‐echo imaging techniques continue to be used to identify BAT in people (Chen et al. 2013). Though convenient due to their ubiquity and speed, the biophysical connection between T₁‐ and T₂‐weighted images and BAT composition and microstructure remains ambiguous. Further, there are commonalities between T₁‐ and T₂‐weighted signals from BAT and WAT, making their differentiation difficult with these simple imaging methods.

¹H chemical shift: MR spectroscopy, Dixon‐type imaging and fat fraction

The chemical shift phenomenon describes the small precession frequency shifts (in the order of parts per million) typically caused by the shielding/de‐shielding of non‐zero spin nuclei by proximal atoms (e.g. oxygen) within a molecule. This phenomenon has arguably been of greatest significance to MR‐based detection and characterization of BAT thus far. By exploiting the chemical shift phenomenon, MR spectroscopy (MRS) and MRI can readily map, characterize and measure the relative abundance of water and fat in BAT – a measure referred to as the fat fraction (FF). Alternatively, FF can be imaged/mapped using: (i) frequency selective imaging techniques, in which water and fat resonances are selectively saturated or excited to produce images of water and fat only, or (ii) the Dixon technique in which two images are collected: one in which water and fat are in‐phase with one another and one in which they are 180^o out of phase.

Significant work has been done towards characterizing the differential FFs and lipid compositions of WAT and BAT. In rodents, it has been established that the FF of BAT (∼20–50%) is typically smaller than that of WAT (∼70–90%) (Sbarbati et al. 1997; Lunati et al. 1999; Hu et al. 2010; Smith et al. 2013; Hamilton et al. 2011), while others have reported that the FF of BAT can be up to 80% (Hu et al. 2010). In 2012, Hu et al. conducted some critical experiments towards validating MR‐based FF measures in humans, the first of which confirmed histologically the MR‐based distinction between BAT and WAT in a case study of post‐mortem tissue (Hu et al. 2012). Since then, variations on the FF measurement have been applied in humans via MRS and MRI (Rasmussen et al. 2013; van Rooijen et al. 2013; Hu et al. 2013a,b, 2014; Deng et al. 2015; Kim et al. 2014; Ponrartana et al. 2014, 2016; Reddy et al. 2014; Lundstrom et al. 2015; Franssens et al. 2016; Gifford et al. 2016; Romu et al. 2016), a number of which join rodent studies in directly correlating the MR‐derived FF parameter with increased ¹⁸F‐FDG‐PET/CT uptake in BAT (Chen et al. 2013; Holstila et al. 2013; van Rooijen et al. 2013; Gifford et al. 2016). Further, a number of recent human and preclinical studies have interrogated changes in BAT FF after activation (Smith et al. 2013; van Rooijen et al. 2013; Grimpo et al. 2014; Gifford et al. 2016; Romu et al. 2016) – studies which show that activation of BAT reduces FF, presumably through the metabolism of the intercellular BAT triglyceride and/or the increase of BAT perfusion (increase in the water resonance amplitude). Taken as a whole, these studies suggest that interrogation of BAT with chemical‐shift MRS/MRI may be a very powerful tool in characterization of BAT in humans. Nevertheless, due to the dynamic nature of BAT (FF can change on relatively short time scales based on thermal/metabolic needs), the propensity for BAT and WAT to be mixed, and the large voxel size relative to cell size, one must take care to not over‐interpret BAT FF measures as an absolute measure of BAT FF or as an unequivocal identifier of BAT.

Blood‐oxygen‐level dependent MRI

BAT typically has a much greater capillary density compared to WAT. As such, MR techniques that are sensitive to blood flow and/or vascular volume have shown promise in identifying and characterizing the functional nature of BAT. Blood‐oxygen‐level dependent (BOLD) MRI, typically associated with functional MRI in the brain, is one such technique that is beginning to find useful application in the measure of BAT activity. At its core, BOLD MRI is sensitive to the susceptibility mismatch between intra‐ and extra‐vascular ¹H protons which is modulated by changes in the oxy/deoxy‐haemoglobin ratio, blood flow and vascular volume – all typically driven by changes in the metabolic needs of tissue. Thus, changes in the BOLD MRI signal are related to the metabolic demands of the tissue (Ogawa et al. 1990). In 2012, Khanna et al. showed, in mice, that activated BAT exhibits a strong BOLD effect (i.e. a transient decrease in T₂‐ and T₂ ^*‐weighted signal after noradrenaline stimulation) and may thus be a complementary or alternative tool to ¹⁸F‐FDG‐PET in detecting activated BAT (Khanna & Branca, 2012). This work was almost immediately expanded upon by Chen et al. (2013) and van Rooijen et al. (2013), and later by Gifford et al. (2016), all of whom correlated BOLD signal changes to ¹⁸F‐FDG‐PET uptake after BAT activation in human participants and noted the complementary value of MR‐measured fat fraction measures gathered during the same MR session as BOLD data. While still very much in its infancy, BOLD MRI holds immense promise as a tool for identifying and investigating BAT.

Contrast‐enhanced ultrasound (CEU)

Ultrasonography involves the use of an array of (typically piezoelectric) transducers, which produce sound waves in the ultrasonic frequencies (>1 MHz). These sound waves propagate into the body and reflect off of interfaces of large acoustic impedance (e.g. the interface between the amniotic fluid and the epidermis of a developing fetus). Acoustic waves, which are reflected back to the surface of the body, can then be transduced back to electrical signals, typically by the same piezoelectric transducers that generated the initial impulse. From the received signal, an image of the tissue can be constructed based on signal amplitudes and the travel time of the wave. CEU further involves the injection of a contrast material of high acoustic impedance (e.g. microbubbles). CEU has been used for the detection of the supraclavicular BAT against ¹⁸F‐FDG‐PET/CT in young healthy adults (Flynn et al. 2015). According to the results, cold increased perfusion in BAT, while BAT+ individuals demonstrated a higher perfusion rate in the supraclavicular fat depot during cold and warm conditions compared to the BAT− group. CEU is an attractive alternative approach for the assessment of BAT as it does not involve exposure to radiation and the equipment is widely available in clinical settings. Limitations of this method include the limited field of view and the small risk for cardiopulmonary complications related to the administration of the contrast material, one of which is now FDA approved. Further studies are needed to establish the validity of this method.

Near infrared spectroscopy (NIRS)

NIRS capitalizes upon of the physicochemical properties of haemoglobin (chromophore at near infrared light (700–900 nm)) to assess tissue perfusion. NIRS is sensitive to the differential spectral absorptions of oxy‐/deoxyhaemoglobin, and thus NIRS can be used as a surrogate measure of the oxygen extraction of a tissue and its metabolic state. The enormous power of NIRS is partly offset by its weak depth of penetration, limiting its use to superficial tissues. Since BAT has high vascularization and is relatively superficial, NIRS has been proposed as an alternative approach for the assessment of BAT. Muzik et al. reported that the ratio of the cold‐induced change in regional O₂ saturation (${\mathrm{rSO}}_2$) in BAT normalized for the rSO₂ in skeletal muscle was marginally lower in the BAT+ group compared with the BAT− group suggesting increased O₂ use relative to supply in BAT. The normalized ratio of the cold‐induced change in rSO₂ in BAT also correlated with the metabolic rate of O₂ in BAT via ¹⁵O‐PET (Muzik et al. 2013). When time‐resolved NIRS was used to assess the optical characteristics of the supraclavicular BAT against ¹⁸F‐FDG‐PET, BAT haemoglobin concentration and scattering coefficient (μ΄_s, an index of mitochondrial concentration) correlated with ¹⁸F‐FDG uptake (Nirengi et al. 2015). NIRS for the assessment of BAT does not, of necessity, require cold exposure or involve exposure to ionizing radiation, while the equipment is portable and relatively inexpensive. Conversely, NIRS is a user‐dependent method and cannot assess total BAT volume due to the limited field of view, while overlying tissues (skin, WAT, etc.) may affect the optical characteristics of the light pathway. Further studies are needed to address the limitations of NIRS and to further test its performance against PET/CT.

Skin temperature

Since the primary established function of BAT is thermogenesis, the assessment of the temperature of the skin overlying known BAT depots has been proposed as an index of BAT thermogenesis. Infrared thermography (IRT) and thermal skin probes are the two approaches that have been used to assess BAT. IRT involves the use of a camera that can detect the infrared electromagnetic radiation (9–14 μm) emitted and create images of the temperature distribution. Thermal skin probes are directly attached to the skin over the fat depot of interest (usually supraclavicular) and frequent temperature recordings are performed. Skin temperature over known BAT depots is higher compared to other non‐BAT related areas using IRT (Jang et al. 2014; Robinson et al. 2016) and thermal skin probes (Boon et al. 2014; van der Lans et al. 2016; Chondronikola et al. 2016a). IRT and thermal probes yield comparable results (Jang et al. 2014). Jang et al. validated IRT against ¹⁸F‐FDG‐PET/CT reporting that cold‐induced change in supraclavicular skin temperature >0.9°C can be used to detect BAT+ individuals (Jang et al. 2014). The advantages of those methods include their low cost and the lack of radiation exposure. Limitations of those thermographic methods include: (1) the potential confounding from overlying tissues and from the general skin cooling with some BAT activation approaches, and (2) the assessment of only superficial BAT depots. An additional limitation of the thermal probes is they will provide temperature measurements only for the area underneath the probe, while IRT has a broader field of view. Validation of these methods in other populations other than lean healthy individuals is still pending. Microwave radiometry, a non‐imaging method for detecting temperature deeper in the body has recently been shown feasible for detecting active BAT in a study of 19 normal volunteers. While feasible, the signals obtained were modest in size (Crandall et al. 2018).

Promising methods currently tested only in preclinical models

The development of non‐invasive approaches for the assessment of BAT is an area of intense scientific interest; as a result, numerous methods are currently under development. The following methods have shown enormous promise for the assessment of BAT content and/or metabolic function in humans.

¹⁸[F]‐Fluorobenzyltryphenyl phosphonium (¹⁸F‐FBnTP)‐PET

¹⁸F‐FBnTP (t _1/2 = 109 min) is a PET tracer that accumulates in mitochondria proportionally to their membrane potential (Madar et al. 2015). BAT is a tissue with a high number of mitochondria compared to WAT, and thus ¹⁸F‐FBnTP accumulates in high amounts in rodent BAT (Fig. 3), while adrenergic stimulation rapidly washes out the tracer from BAT (Madar et al. 2015). A major advantage of this PET tracer is its ability to detect BAT without cold exposure, while it can also provide information regarding its metabolic status.

Figure 3. ¹⁸F‐Fluorobenzyl tryphenyl phosphonium (¹⁸F‐FBnTP) uptake in interscapular brown adipose tissue at room temperature.

Co‐registered positron emission tomography (PET)–computed tomography (CT) images in transverse (upper panel), coronal (middle panel) and sagittal (lower panel) view, acquired in a rat at room temperature.¹⁸F‐FBnTP‐increased uptake in the interscapular area is confined to CT regions of low Hounsfield units (black arrows). Note the strong uptake of ¹⁸F‐FBnTP in BAT, similar to that seen in heart (blue arrow). Reproduced from Madar et al. (2015) with permission.

Contrast‐enhanced MRI

BAT typically has greater capillary density compared to WAT. As such, MR techniques that are sensitive to blood flow and/or vascular volume can be used to identify and characterize the functional nature of BAT. Specifically, exogenous paramagnetic (e.g. small gadolinium‐bearing chelates) or superparamagnetic (e.g. large iron oxide containing protein complexes) molecules can be used as contrast agents in conjunction with MRI to assess vascular volume and function, and other physiological parameters. Investigators have shown that signal enhancement after gadolinium injection was greater in adrenaline‐activated BAT compared to control BAT in rodents suggesting increased blood flow in activated BAT (Sbarbati et al. 2006). Further, iron oxide nanoparticles have been used to measure BAT perfusion (Chen et al. 2012), while iron oxide‐bearing lipid‐rich lipoprotein nanosomes have been shown to be preferentially taken up by activated BAT providing information regarding the contribution of BAT in lipoprotein metabolism (Jung et al. 2016).

Hyperpolarized ¹³C MR

Dissolution dynamic nuclear polarization of ¹³C, a technology which yields a transient but enormous gain in ¹³C signal to noise, has opened the possibility of direct observation and measurement of metabolic function in vivo (Ardenkjaer‐Larsen et al. 2003). Specifically, the conversion of intravenously administered hyperpolarized [1‐¹³C]pyruvate to [1‐¹³C]lactate can be directly observed in real time with MR. Given the differential metabolic flux through activated BAT versus inactive BAT/WAT, hyperpolarized ¹³C MR holds promise for identifying and characterizing the metabolic activity of BAT (Lau et al. 2014).

Hyperpolarized ¹²⁹xenon (¹²⁹Xe) MR

¹²⁹Xe is a non‐zero‐spin isotope of the noble gas xenon that can be hyperpolarized via spin‐exchange optical pumping and subsequently inhaled and visualized in vivo via MR. ¹²⁹Xe is considered a surrogate measure of tissue perfusion and it accumulates in the highly vascularized BAT in response to adrenergic stimulation via ¹²⁹Xe MR (Branca et al. 2014). Due to the temperature dependence of the ¹²⁹Xe chemical shift, ¹²⁹Xe MR can also be used to measure BAT thermogenesis.

Intermolecular zero quantum coherence

A fascinating and elegant application of fundamental MR principles to solve biological problems lies in the use of intermolecular quantum coherences. The technique can be considered sensitive to dipole–dipole interactions that occur over long distances (∼μm to mm) and the distance over which the technique is sensitive can be tuned using a magnetic field gradient. This approach has been successfully used to discern BAT from WAT in rodents and map BAT (Branca & Warren, 2011; Bao et al. 2013; Branca et al. 2013).

Summary and conclusion

The presence of significant amounts of active BAT in humans has only relatively recently been confirmed in adults. Thus, the methods for the detection and characterization of the human BAT in vivo are still under development. ¹⁸F‐FDG‐PET/CT after cold exposure is currently the most frequently used and best established method for the detection and quantification of activated BAT in people and the assessment of its metabolic activity. However, its limitations warrant further research to establish lower cost, valid and reproducible methods to detect and quantify BAT in humans with minimal or no use of ionizing radiation. Currently several approaches hold promise for the assessment of BAT (MR‐based approaches, PET/MRI, PET with tracers other than ¹⁸F‐FDG, etc.). Further research is needed to optimize the aforementioned approaches, translate methods currently tested only in preclinical models, and develop new approaches for the assessment of BAT in people. The development of novel methodological approaches will expedite the advancement of the current knowledge in the area of the human BAT and help investigators unveil its physiological significance in metabolic regulation in people.

Additional information

Competing interests

None declared.

Author contributions

M.C. and R.L.W. conceived the manuscript. M.C. and S.B. wrote the manuscript. M.C., S.B. and R.L.W. edited the manuscript and approved the final version of the manuscript. All persons designated as authors qualify for authorship, and all those who qualify for authorship are listed.

Funding

This work was supported by grants from the National Cancer Institute (USA) (U01 CA140204). M.C. is supported by a postdoctoral fellowship from the American Heart Association (17POST33060003).

Biographies

Maria Chondronikola is a Postdoctoral Research Scholar at the Washington University School of Medicine (WUSM) with a strong background in integrative physiology, nutrition and metabolism. Her current research focuses on understanding the role of BAT in metabolic function and health in humans.

Scott Beeman is an Assistant Professor at the WUSM. His research focuses on devising magnetic resonance‐based experimental and mathematical techniques to directly quantify physiology in vivo and applying these techniques to advance the understanding of metabolic diseases and their complications.

Richard L. Wahl is Professor and Chair of Radiology and the Director of the Mallinckrodt Institute of Radiology at the WUSM. He has been a leader in introducing and using PET in combination with CT and MRI to diagnose cancer and other diseases. These tools have allowed him and his colleagues to interrogate BAT non‐invasively with PET in humans and preclinical models.