Near-Infrared Time-Resolved Spectroscopy for Assessing Brown Adipose Tissue Density in Humans: A Review

Brown adipose tissue (BAT) mediates adaptive thermogenesis upon food intake and cold exposure, thus potentially contributing to the prevention of lifestyle-related diseases. 18F-fluorodeoxyglucose (FDG)–positron emission tomography (PET) with computed tomography (CT) (18FDG–PET/CT) is a standard method for assessing BAT activity and volume in humans. 18FDG–PET/CT has several limitations, including high device cost and ionizing radiation and acute cold exposure necessary to maximally stimulate BAT activity. In contrast, near-infrared spectroscopy (NIRS) has been used for measuring changes in O2-dependent light absorption in the tissue in a non-invasive manner, without using radiation. Among NIRS, time-resolved NIRS (NIRTRS) can quantify the concentrations of oxygenated and deoxygenated hemoglobin ([oxy-Hb] and [deoxy-Hb], respectively) by emitting ultrashort (100 ps) light pulses and counts photons, which are scattered and absorbed in the tissue. The basis for assessing BAT density (BAT-d) using NIRTRS is that the vascular density in the supraclavicular region, as estimated using Hb concentration, is higher in BAT than in white adipose tissue. In contrast, relatively low-cost continuous wavelength NIRS (NIRCWS) is employed for measuring relative changes in oxygenation in tissues. In this review, we provide evidence for the validity of NIRTRS and NIRCWS in estimating human BAT characteristics. The indicators (IndNIRS) examined were [oxy-Hb]sup, [deoxy-Hb]sup, total hemoglobin [total-Hb]sup, Hb O2 saturation (StO2sup), and reduced scattering coefficient (μs sup′) in the supraclavicular region, as determined by NIRTRS, and relative changes in corresponding parameters, as determined by NIRCWS. The evidence comprises the relationships between the IndNIRS investigated and those determined by 18FDG–PET/CT; the correlation between the IndNIRS and cold-induced thermogenesis; the relationship of the IndNIRS to parameters measured by 18FDG–PET/CT, which responded to seasonal temperature fluctuations; the relationship of the IndNIRS and plasma lipid metabolites; the analogy of the IndNIRS to chronological and anthropometric data; and changes in the IndNIRS following thermogenic food supplementation. The [total-Hb]sup and [oxy-Hb]sup determined by NIRTRS, but not parameters determined by NIRCWS, exhibited significant correlations with cold-induced thermogenesis parameters and plasma androgens in men in winter or analogies to 18FDG–PET. We conclude that NIRTRS can provide useful information for assessing BAT-d in a simple, rapid, non-invasive way, although further validation study is still needed.

tomography (CT) ( 18 FDG-PET/CT) is a standard method for assessing BAT activity and volume in humans. 18 FDG-PET/CT has several limitations, including high device cost and ionizing radiation and acute cold exposure necessary to maximally stimulate BAT activity. In contrast, near-infrared spectroscopy (NIRS) has been used for measuring changes in O 2 -dependent light absorption in the tissue in a non-invasive manner, without using radiation. Among NIRS, time-resolved NIRS (NIR TRS ) can quantify the concentrations of oxygenated and deoxygenated hemoglobin ([oxy-Hb] and [deoxy-Hb], respectively) by emitting ultrashort (100 ps) light pulses and counts photons, which are scattered and absorbed in the tissue. The basis for assessing BAT density (BAT-d) using NIR TRS is that the vascular density in the supraclavicular region, as estimated using Hb concentration, is higher in BAT than in white adipose tissue. In contrast, relatively low-cost continuous wavelength NIRS (NIR CWS ) is employed for measuring relative changes in oxygenation in tissues. In this review, we provide evidence for the validity of NIR TRS and NIR CWS in estimating human BAT characteristics. The indicators (Ind NIRS ) examined were [oxy-Hb] sup , [deoxy-Hb] sup , total hemoglobin [total-Hb] sup , Hb O 2 saturation (StO 2sup ), and reduced scattering coefficient (µ s ′ sup ) in the supraclavicular region, as determined by NIR TRS , and relative changes in corresponding parameters, as determined by NIR CWS . The evidence comprises the relationships between the Ind NIRS investigated and those determined by 18 FDG-PET/CT; the correlation between the Ind NIRS and cold-induced thermogenesis; the relationship of the Ind NIRS to parameters measured by 18 FDG-PET/CT, which responded to seasonal temperature fluctuations; the relationship of the Ind NIRS and plasma lipid metabolites; the analogy of the Ind NIRS to chronological and anthropometric data; and changes in the Ind NIRS following thermogenic food supplementation. The [total-Hb] sup and [oxy-Hb] sup determined by NIR TRS , but not

INTRODUCTION
Human adipose tissues are of a variety of types, such as white (WAT) and brown adipose tissue (BAT) (1). WAT is capable of depositing extra-energy as triglyceride droplets under conditions where energy intake is greater than its expenditure. In contrast, BAT promotes non-shivering thermogenesis to respond to decreases in core body temperature and, in contrast to WAT, is characterized by an abundance of mitochondria and vasculature. BAT has been extensively investigated in animals, and it has been determined that BAT-specific uncoupling protein (UCP)-1, mainly stimulated upon β 3 -adrenergic activation by cold and/or dietary intervention, enables BAT to dissipate free energy to heat by proton discharge through the inner mitochondrial membrane (2,3). BAT has drawn renewed attention since 2009, with several papers being published that report the existence of BAT deposits in adult humans (4)(5)(6)(7), which had previously been thought to be lost during the process of maturation. Human BAT is reported to be related to lower adiposity [body mass index (BMI), the percentage of whole body fat (%BF), and visceral fat area (VFA)] (6-9) and increased glucose sensitivity (10). In experimental studies, repeated exposure to cold environment enhanced the BAT activity and improved glucose tolerance in obese counterparts (11) and patients with type 2 diabetes mellitus (12) as well as in healthy individuals (9,13,14). Thus, increasing BAT activity or volume may aid in combatting obesity and chronic diseases, such as type 2 diabetes mellitus. It is well-known in humans that BAT can be evaluated by 18 F-fluorodeoxyglucose (FDG)-positron emission tomography (PET) with computed tomography (CT) ( 18 FDG-PET/CT) under cold-stimulated environments (3,4,6,15). However, 18 FDG-PET/CT has several limitations, including the enormous cost of the device and ionizing radiation exposure, and acute cold exposure-necessary to maximally stimulate BAT activity Abbreviations: adjStO 2 , adjusted supraclavicular StO 2 ; AR, β3-adrenergic receptor; BAT, brown adipose tissue; BAT-d, vascular or mitochondrial density in BAT; BMI, body mass index; CIT, cold-induced thermogenesis; CT, computed tomography; deoxy-Hb, deoxygenated Hb; FDG, 18 F-fluorodeoxyglucose; Hb, hemoglobin; NE, norepinephrine; NIRS, near-infrared spectroscopy; NIR CWS , NIR continuous-wave spectroscopy; NIR TRS , NIR time-resolved spectroscopy; oxy-Hb, oxygenated Hb; StO 2 , Hb O 2 saturation; PET, positron emission tomography; ROC, receiver operating characteristic; total-Hb, total hemoglobin; TRP, transient receptor potential channels; SUV max , maximal standardized uptake value; SUV mean , mean standardized uptake value; WAT, white adipose tissue; µ a , absorption coefficient; µ s ′ , reduced scattering coefficient. (16), which make a longitudinal 18 FDG-PET/CT study difficult and disrupt interventional research, specifically longitudinal ones in humans. Cold exposure is primarily required in 18 FDG-PET/CT studies to activate human BAT, and various protocols have been applied in the past. While standardized guidelines have recently been proposed, differences between protocols remain a significant obstacle to the comparison of observations from different studies (17).
Other non-invasive technologies have been utilized for evaluating BAT characteristics in humans, such as magnetic resonance imaging (MRI) (18,19), local skin thermal measurements (19), infrared thermography (20), and contrast ultrasound (21). A recent review on the detection of BAT using these non-invasive technologies can be found elsewhere (22). Regarding MRI technologies, proton-density fat fraction (PDFF) values are widely used to distinguish BAT from WAT (23,24). However, the PDFF range in the supraclavicular region widely varies among individuals, which makes the differentiation of BAT from WAT difficult, although several new technologies, such as the measurement of T2 * relaxation and diffusion-weighted imaging are under investigation (22). The local skin thermal measurements have been used for monitoring cold-induced temperature changes in the supraclavicular skin with infrared thermography (25,26). However, heat measurements could be influenced by the tissue conductive properties and thickness of the subcutaneous adipose tissue (27), and the individual vasomotor response (28); these evaluation obstacles should be solved in the future.
In addition, near-infrared spectroscopy (NIRS) is a relatively newly introduced methodology to monitor BAT properties (29). The basis for the application of NIRS to evaluate BAT properties is that the microvascular bed, as evaluated by total hemoglobin (Hb) concentration [total-Hb] sup in the supraclavicular region, is more abundant in BAT than in WAT (30). Furthermore, NIR time-resolved spectroscopy (NIR TRS ) may be used to assess the density of the microvasculature as well as mitochondrial content in BAT by measuring the reduced scattering coefficient (µ s ′ ), which reflects the in vitro mitochondrial content (31). BAT is a highly innervated tissue and is also highly perfused when exposed to cold (32). As the concentrations of oxygenated and deoxygenated Hb in the supraclavicular region ([oxy-Hb] sup and [deoxy-Hb] sup , respectively) are likely to change (especially [total-Hb] sup , which reflects blood volume), it could be a valid measure of BAT vasculature.
The purpose of this article is to provide evidence concerning the ability of NIRS to evaluate BAT characteristics in humans. In this review, we included studies examining BAT characteristics using NIRS in humans: most studies used NIR TRS (29,(33)(34)(35)(36)(37)(38)(39), a technology to quantify both absolute tissue absorption and scattering characteristics, while some utilized NIR continuous wave spectroscopy (NIR CWS ), an inexpensive technology that only provides relative values of tissue oxygenation (32,40). First, we present how NIRS functions to evaluate tissue oxygenation and blood volume. Then, we provide data indicating whether BAT characteristics can be evaluated using NIR CWS . The main body of the paper presents a series of evidence for NIR TRS to assess BAT characteristics. The evidence tested comprises (1) the relationship between parameters determined using NIRS and those measured by 18 FDG-PET/CT, (2) correlations between the NIRS parameters and cold-induced thermogenesis (CIT), (3) correspondence of the NIRS parameters to those reported using 18 FDG-PET/CT regarding chronological and anthropometric data, (4) the correspondence between NIRS parameters and those reported with 18 FDG-PET/CT in response to ambient temperature fluctuations, (5) the relationship between parameters determined using NIRS and plasma lipid metabolites, and (6) changes in NIRS parameters induced by supplementation with evidence-based thermogenic functional ingredients.

HOW NIRS FUNCTIONS AS EVALUATING TISSUE OXYGENATION AND BLOOD VOLUME
NIRS provides non-invasive monitoring of tissue oxygen and Hb dynamics in vivo (41)(42)(43)(44)(45). NIRS is able to monitor changes in O 2 -dependent light absorption in the heme in the red blood cells circulating in biological tissues (46). There are mainly three types of NIRS devices: NIR CWS , NIR TRS , phase modulation NIR spectroscopy (NIR PMS ), etc. (46)(47)(48). The most popular NIRS devices use NIR CWS that outputs only the qualitative tissue oxygenation. To calculate the changes in [oxy-Hb], [deoxy-Hb], [total-Hb], and Hb O 2 saturation (StO 2 ) using NIR CWS , a combination of multiple-wavelengths can be adopted in accordance with the Beer-Lambert law. The main reason why quantitative data cannot be provided as continuous NIR light path traveled through tissues is unknown (42)(43)(44)(45). However, spatially resolved NIRS (a type of NIR CWS ) is able to provide quantitative values considering several assumptions, although it is still unable to provide the tissue absorption and scattering properties.
On the other hand, NIR TRS and NIR PMS are more accurate, as they can quantify both tissue absorption and scattering characteristics. NIR TRS emits ultrafast (100 ps) light pulses from the skin surface and measures the photon distributions across the biological tissue with a 2-to 4-cm distance from the light emission. NIR TRS is able to quantitatively measure the absorption coefficient (µ a ), µ s ′ , and then calculates light path length, tissue [oxy-Hb], [deoxy-Hb], [total-Hb], and StO 2 (44,45,48). The validity of the signal obtained by NIR CWS and NIR TRS has been confirmed in an in vitro experiment using highly scattering Intralipid TM (43,44). Using this system, µ a in the NIR range was found to be strongly correlated with [total-Hb] (43,48). Furthermore, the study found a significant relationship between µ s ′ at 780 nm and the homogenized tissue mitochondrial concentration (31).

STUDIES USING NIR CWS
Prior to the NIR TRS study on human BAT, one study attempted to correlate oxygen dynamics using NIR CWS and BAT parameters (32). In this cross-sectional study, adult human subjects (25 subjects; 15 women and 10 men; mean age ± SD, 30 ± 7 years) were assigned into high-(BMI, 22.1 ± 3.1) and low-BAT groups (BMI, 24.7 ± 3.9) based on the levels of 18 F-FDG uptake in the cervical-supraclavicular region. It employed tripleoxygen PET scans (H 15 2 O, C 15 O, and 15 O 2 ) and daily energy expenditure measurements under resting and mild cold (15.5 • C) room conditions for 60 min using indirect calorimetry (32). They used a NIR CWS parameter, adjusted supraclavicular StO 2 (adjStO 2 ), a balance between oxygen supply and uptake. In the high-BAT group, there was a significant negative correlation between oxygen consumption determined by PET scans and adjStO 2 (p = 0.02, r 2 = 0.46) in the supraclavicular region at rest and after the exposure to cold, indicating increased oxygen uptake in highly active BAT (32). However, it detected a limited effect on the difference in adjStO 2 between the two groups (32). It should be noted that the study presented several limitations to consider when interpreting its results: (1) a non-individualized cooling protocol was used; (2) only one NIR CWS parameter, adjStO 2 , was used in the analysis; and (3) no kinetics data determined by the NIR CWS were provided.
Recently, in young healthy women, a study using the standardized cold exposure aimed to investigate the association between NIR CWS parameters in the supraclavicular and forearm regions and BAT capacity assessed by 18 FDG-PET/CT (40). Briefly, the subjects arrived at the laboratory (a room temperature of 19.5-20 • C) and wore a temperature-controlled water circulation cooling vest for 60 min, and the individual temperature to be exposed was determined, namely at ∼4 • C above the threshold of shivering, 48-72 h prior to the 18 FDG-PET/CT measurements. No association was found between any NIR CWS indicators and maximal standardized uptake value (SUV max ) and mean standardized uptake value (SUV mean ) of the radioactivity both under thermoneutral and cold conditions. Thus, NIR CWS would not be an appropriate technology to evaluate BAT capacity in this demographic. The lack of significant association between NIR CWS parameters is mainly due to differences in the instrumentation to that used in NIR TRS , which provides absolute values for tissue hemodynamics. Furthermore, NIR CWS permits an ∼15 mm depth of light penetration at a 30 mm input-output setups (44). However, the mean photon penetration would be deeper (∼20 mm at the 30-mm input-output setups) and wider when NIR TRS is used (49), which influences the differences in sensitivity between NIR CWS and NIR TRS with respect to BAT detection. Table 1 shows the relationship between [oxy-Hb] sup , [deoxy-Hb] sup ,  The correlation coefficients of parameters determined by NIR CWS and the uptake of 18 F-fluorodeoxy glucose (FDG) are presented under cold-exposed condition. [total-Hb] sup , StO 2sup , and adjStO 2sup determined by NIR CWS and 18 FDG-PET/CT parameters (SUV max and SUV mean ), which have been documented in previous studies (32,40). The only NIR CWS parameter found to be significantly correlated with a 18 FDG-PET/CT parameter, cold-induced oxygen uptake by BAT, is adjStO 2 , and only in the high-BAT group. Taken together, the studies using NIR CWS (32, 40) present potential limitations beyond the fact that it cannot evaluate the absorption and scattering properties of the tissue, including that it is more sensitive to changes in the skin blood flow than NIR TRS (32,40). Therefore, NIR CWS does not seem to be a valid measure for BAT function although emphasis should be placed in the need for further research examining this type of NIRS.

STUDIES USING NIR TRS Correlation Between Parameters Determined by NIR TRS and 18 FDG-PET/CT Parameters
It is speculated that, as BAT exhibits higher microvascular density and mitochondrial contents compared to WAT, NIR TRS can be used for assessing the density of microvascular or mitochondrial content in BAT (BAT-d) by measuring [total-Hb] sup and µ s ′ in the supraclavicular region (µ s ′ sup ), which reflects the in vitro mitochondrial content (31). It may be expected that BAT would exhibit higher values for [total-Hb] sup and µ s ′ sup than those exhibited by WAT. As NIR TRS measures the average tissue hemoglobin concentration in a volume of 4 cm 3 with a 3-cm optode separation (50), the 18  ′ sup under warm environment is significantly correlated with the SUV mean under cold environment but only in the supraclavicular fossa, a region of BAT located (29). Other parameters, except adjStO 2 specifically in the supraclavicular region, also showed significant correlation with SUV max and SUV mean under cold environment ( Table 2). Collectively, the  18 FDG-PET/CT. StO 2sup , however, proved to be inferior, and adjStO 2 was completely insensitive to changes in BAT activity ( Table 2). A 2-h cold exposure doubles the BAT blood flow (32), which appeared to be inconsistent with our observations. In an attempt to interpret this apparent discrepancy, we speculated that NIR TRS parameters are susceptible to the change in the volume but less sensitive to the change in the flow. The blood flow can be calculated by multiplying the blood flow velocity by the cross-sectional area of the vessel (the volume). There is presently a lack of NIR TRS -derived data concerning blood flow in BAT. Alternatively, while muscle blood flow increases by some 10-fold during peak exercise (51,52), [total-Hb], an indicator of blood volume, monitored by NIR TRS elevates only 1.1-fold (43). Thus, the change in blood volume measurable by NIR TRS is marginal compared to increases in blood flow velocity during metabolic activation. Collectively, both µ s ′ sup and [total-Hb] sup were evaluated using the NIR TRS technique can be applied to assess BAT-d in humans and are equivalent to the active BAT intensity or the BAT volume, as measured by 18 FDG-PET/CT under cold environment (29). Usually, to assign participants into high-BAT (BAT [+]) and low-BAT (BAT [-]) groups, a cutoff value of 2.0 for SUV mean is applied. The accuracy of [total-Hb] sup or µ s ′ sup in representing BAT activity was analyzed. Accordingly, the area under the receiver operating characteristic (ROC) curve was determined by SUV mean of 2.0 nearest to (0, 1) for µ s ′ sup and [total-Hb] sup (29). When 74.0 µM or 6.8 cm −1 was selected as the cutoff value, meaning that [total-Hb] sup or µ s ′ sup larger than 74.0 µM or 6.8 cm −1 , respectively, are regarded as BAT [+], ROC analysis yields results that are very good when compared to SUV mean (29).

Correlation Between Parameters Determined by NIR TRS and CIT
It is well-documented in rodents that the upregulation of the UCP-1 in brown adipocytes upon cold increases whole body oxygen consumption, termed as CIT. Although several authors have shown that CIT does not always reflect BAT activity (53), that the contribution of BAT thermogenesis to CIT is marginal (∼10 kcal/day when maximally activated) (54), and no correlation is found between BAT and CIT (55), the magnitude of CIT is related to the amount of BAT activity and/or volume (9,(56)(57)(58)(59). Thus, having already observed a significant correlation between NIR TRS parameters and SUV mean assessed by 18 (37). The participants sat for 20 min at 27 • C with a light clothing, and NIR TRS measurements were conducted for 5 min after fasting for 6-12 h. Then, the participants were tested at room temperature of  (37). In contrast, previous studies reported a significant correlation in the supraclavicular region between the adjStO 2 and oxygen consumption by BAT under cold environment and between µ s ′ sup and SUV max and SUV mean (29,32). It is noted that a personalized cooling protocol may be a better procedure to induce a CIT response personalized to each individual (60). Collectively, although the [total-Hb] sup , [oxy-Hb] sup , and [deoxy-Hb] sup are markers for BAT activity as evaluated by CIT, the adjStO 2 and µ s ′ sup become less sensitive to CIT ( Table 3).

Relationship Between NIR TRS Parameters in the Supraclavicular Region and Chronological and Anthropometric Data
18 FDG-PET/CT studies have revealed that a significant relationship exists between BAT activity and chronological and anthropometrical parameters (4,5,29,56). Cold-stimulated 18 FDG-PET/CT studies have shown that BAT activity negatively associated with age, sex, BMI, %BF mass, and VFA, and also that BAT was a significant independent determinant of glucose and HbA1c levels, after adjustment for age, sex, and body adiposity (10,56).
A cross-sectional study using NIR TRS demonstrated that [total-Hb] sup under warm environment was negatively associated with age and body adiposity in 413 Japanese individuals [a median age of 43.0 (33.0-58.0, interquartile range) years, BMI of 22.5 (20.7-24.5) kg/m 2 , and %BF of 26.8% (20.6-32.3%)] in winter (33) (Figure 1). With the exception of participants in their 20s, there were no sex-related differences in [total-Hb] sup among the groups tested (Figure 1). Multivariate analyses revealed that the %BF and VFA were significantly negatively correlated with [total-Hb] sup (33). The observation of the study was analogous to data acquired using 18

Changes in NIR TRS Parameters in the Supraclavicular Region in Response to Ambient Temperature Fluctuations
BAT increases in winter according to 18 FDG-PET/CT studies (4,(56)(57)(58)(59). However, one study reported that early winter showed higher BAT activity than late winter or early spring (61). A cross-sectional study (35) reported that [total-Hb] sup was higher in winter than in summer. It has also been reported that a lower average ambient temperature during the 4-6 weeks before the measurement day increases [total-Hb] sup (33). This finding is consistent with previous findings reporting that, while BAT activity rose during the winter, a few months are needed for the increase in BAT activity after a decrease in the air temperature (58). A longitudinal study using the same healthy subjects (men/women, 35/23; mean age, 37.4 years; mean BMI, 22.5 kg/m 2 ; BAT positive rate, 48%) in summer and winter under thermoneutral conditions revealed a significant increase in [total-Hb] sup , but not in the reference region or in the µ s ′ from any regions (37). It is unclear why µ s ′ sup region did not change between summer and winter. However, it is speculated that if triglyceride droplet in the supraclavicular area decreases in winter owing to the increase in BAT (or beiging), the µ s ′ should decrease because WAT (triglyceride droplet) obtains high scattering characteristics (62). Thus, even increasing the mitochondria content (the increase in µ s ′ ) would offset the decrease in WAT (the decrease in µ s ′ ), indicating that [total-Hb] sup may be a better indicator of BAT activity than µ s ′ sup . We demonstrated seasonal changes in other NIR TRS parameters, which supplement previously published findings in  (Table 5).
Collectively, the [oxy-Hb] sup , [total-Hb] sup , StO 2sup , and adjStO 2 can detect seasonal fluctuations of BAT-d, which is  (29,33). The values are presented as means ± standard error (SE), adjusting for body mass index, body fat ratio, and visceral fat area. © SPIE. Reproduced by permission of the publisher. Adopted from reference (33).

Correlation Between NIRS Parameters in the Supraclavicular Region and Lipid Metabolites
Finding blood biomolecules correlated with BAT characteristics would permit us to advance human BAT studies because PET/CT studies may be difficult owing to ionizing radiation and cold exposures. Studies on lipidomic profiles have clarified BAT and WAT characteristics according to muscle contractions or cold environment (11,12,14,16). BAT characteristics are related to unique profiles of lipid metabolites, such as the concentration of lysophosphatidylcholine-acyl (LysoPC-acyl) C16:0 in humans (63), and the concentration of phosphatidylethanolamine (PE) in the BAT and WAT was decreased in high-fat diet-fed mice (14). The relationships have been examined in the winter and summer between [total-Hb] sup , a parameter for evaluating BAT-d, measured using NIR TRS and plasma lipids in humans (38).  After the Q values were obtained by correcting false discovery rate, only androgens (testosterone, androstanedione, dehydroandrosterone, dehydroepiandrosterone, or epitestosterone) showed a significant (Q < 0.05) positive correlation with [total-Hb] sup in men in winter. Multivariate regression analysis revealed that [total-Hb] sup showed a significant correlation with androgens in men and VFA in women in winter. Notably, the [total-Hb] sup showed a significant relationship with androgens in winter in men but did not with any body-composition characteristics, such as whole body and visceral adiposity, which are generally associated with [total-Hb] sup . Although androgens deteriorated the BAT capacity in vitro (64), testosterone induced a preferable effect on BAT activity, body adiposity, and energy expenditure in animal models (65)(66)(67). Thus, BAT characteristics might be predicted by measuring plasma androgens as a biomarker in men in the winter. However, further detailed research is needed to discover biomarkers that predict BAT in women.

Changes in NIRS Parameters in the Supraclavicular Region by Thermogenic Functional Ingredients
Recent studies have demonstrated that BAT can improve health status and has a protective effect on lifestyle-related diseases (4-11, 13, 14). Consequently, research has been focused on finding methods for effectively enhancing BAT activity and/or mass (9,(11)(12)(13)(14). Developed strategies include cold acclimation (9,(11)(12)(13)(14) and acute treatment of β 3 -adrenergic receptor (AR) agonists in humans (68). However, cold exposure intervention would not be easy to apply to daily life (9), and β3-AR agonists may elicit unpreferable influence, including a risk for hypertension and increased susceptibility to arterial sclerosis (68). Recent investigations have revealed the mechanisms underlying the effects of thermogenic food ingredients. Pathways involved include the transient receptor potential channels (TRP)-BAT axis, a site of adaptive thermogenesis evoked by β-adrenoceptor activation (69). The TRP-BAT axis comprises the activation of cold-sensitive TRP channels located in peripheral tissues, such as the skin and intestines. The activation of TRP channels results in the signal delivery through the afferent nerve to the hypothalamus, which then evokes sympathetic nerve activation within BAT. This causes norepinephrine (NE) release, initiating β-adrenergic tracts to brown adipocytes and eliciting UCP1 upregulation and adaptive thermogenesis (69). In contrast to cold exposure intervention, functional food ingredients may be easily incorporated into daily life. This has been confirmed in animals and humans and includes capsinoids as TRP vanilloid 1 agonists, catechins as TRPA1/V1 agonists, and so on (70). Furthermore, they have the benefit of having no apparent side effects (9,34,36,69,70). Among thermogenic food ingredients, substances, such as capsiate are known to increase BAT activity (9,70). Previously, the effect of capsiate on [total-Hb] sup , determined by the NIR TRS , was examined (34). Twenty healthy individuals [capsiate group (n = 10) vs. placebo group (n = 10), 20.7 ± 1.2 years vs. 20.9 ± 0.9 years; BMI, 21.4 ± 1.8 vs. 21.9 ± 1.0 kg/m 2 ; %BF, 21.3 ± 7.6% vs. 22.9 ± 8.7%] were supplemented either with capsiate (9 mg/day) daily for 8 weeks or a placebo in a paralleled, double-blind manner, and [total-Hb] sup was measured during the treatment period, and for an 8-weeks follow-up period under thermoneutral conditions (34). The study also measured BAT activity with 18 FDG-PET/CT under cold-exposure conditions as previously reported (29). This was only done twice (not every 2 weeks), pre-and post-supplementation, to reduce participant exposure to ionizing radiation. The study demonstrated a parallel change in BAT-d (+46.4%, P < 0.05) pre-and postsupplementation, evaluated as [total-Hb] sup , or as BAT activity (+48.8%, P < 0.05) evaluated as the SUV max , a parameter of the BAT capacity, by 18 FDG-PET/CT, after the supplementation of thermogenic capsiate (Figures 2, 3). During the 8-weeks followup period, the [total-Hb] sup decreased both in the capsiate and placebo groups; the decrease was greater in the capsiate group (albeit not significantly, P = 0.07) compared to that of the placebo group.
Previous studies examined whether a catechins-rich green tea extract increases energy consumption in humans (71)(72)(73)(74). Animal studies have shown that catechin intake increases BAT, the effects of which were abolished when the β-blocker was administrated (75,76). Thus, we used NIR TRS under thermoneutral conditions to test the effect of sustained catechinrich ingredient (540 mg/day) intake on [total-Hb] sup and investigated potential associations between changes in [total-Hb] sup and body adiposity in 22 healthy women college students [catechin group (n = 10) vs. placebo group (n = 11), 21.1   (36). As for the µ s ′ , which was not documented in the previous study, it did not change during catechin ingestion. There was a significant negative relationship between the enhancement in [total-Hb] sup and the decrease in extramyocellular lipids, an indicator for possible insulin insensitivity (77), in the vastus lateralis muscle determined by proton-magnetic resonance spectroscopy (r = −0.66, P < 0.05).
After further analysis, some of which has not been documented in the previous study (34)

Limitations and Perspectives
The studies using NIRS contain several limitations. Several optical issues should be considered, as the multilayer, inhomogeneous tissue property created by skin, adipose tissue, and muscle may affect in vivo tissue scattering and absorption characteristics and modulation of optical path. In a study (39), the optical characteristics in the deltoid, abdominal, and supraclavicular regions were tested using NIR TRS . The results indicate that there are unique region-specific relationships between [total-Hb] and µ s ′ , suggesting that examining the [total-Hb]-µ s ′ relationship is a practical way to distinguish BAT from other tissues. It could be noted that due to the nature of optical measurements, the placement of the optodes for the NIR TRS must be always secure and in the same area, especially during longitudinal studies. Although NIR TRS is able to quantify tissue oxygen dynamics, the values are affected by optical characteristics underlying subcutaneous adipose tissue in the supraclavicular region, which varies depending on the body composition of subjects, thereby influencing NIR TRS measurements. The reason is that the values obtained is diluted by the lower [Hb] in the subcutaneous adipose tissue (78). The [total-Hb] sup values can be recalculated by considering the thickness of the adipose layer (79).
As no change in the [total-Hb] sup and µ s ′ sup was observed during 2-h conditions at 19 • C compared to baseline conditions at 27 • C (29), NIR TRS cannot detect changes in BAT characteristics responding to an acute cold exposure in nature because NIR TRS is insensitive to changes in the blood flow (33,(35)(36)(37). However, a newly developed NIR TRS system contains six wavelengths (760, 800, 830, 908, 936, and 976 nm), of which the latter three wavelengths are adopted to detect optical characteristics of lipids and water (80). This system could provide information on the changes in tissue water and lipid content in response to acute interventions, such as experimental cold exposure, which cannot be obtained using the conventional three-wavelength NIR TRS system. The new six-wavelength NIR TRS system could contribute further insight on the chronic as well as acute responsiveness of BAT metabolism in humans.
Finally, future studies should obtain further evidence to validate BAT evaluation using NIR TRS because 18 FDG-PET/CT measurements include several limitations. BAT mainly consumes intracellular lipids, as well as plasma non-esterified fatty acids and those derived from lipoproteins-whereas 18 FDG-PET/CT measures a glucose analog. The lack of standardization when quantifying BAT by 18 FDG-PET/CT is also a problem. Thus, additional experiments to reach this conclusion are required, such as (1) examining whether NIR TRS parameters actually represent the in vivo mitochondrial density of BAT, or are related to molecules implicated in the vascularization and thermogenesis of BAT [e.g., vascular endothelial-cell growth factor (VEGF), UCP-1, peroxisome proliferator-activated receptor γ coactivator 1-α (PGC1-α)]. This could be carried out by taking human biopsies from the supraclavicular area and examining the relationship between NIR TRS parameters and the molecular signature of this tissue; (2) using other radiotracers beyond 18 F-FDG, such as 15 O, H 15 2 O, C 15 O, or 11 C-acetate, which will allow to measure the real oxygen consumption, tissue perfusion, and metabolic activity of human BAT and which are more likely to represent the thermogenic nature or activity of this tissue than 18 F-FDG; (3) carrying out studies where the kinetics of NIR TRS are related to the kinetics of the metabolic activity of BAT (dynamic PET/CT); (4) using different cooling protocols, aiming to standardize the cooling stress to which individuals are submitted (avoiding potential biases in individual BAT activation); and (5) performing reliability studies to examine whether NIR TRS measures can be replicated in the short and long term.

CONCLUSION
Correlation coefficients are presented for parameters determined by NIRS and 18  , StO 2sup , and adjStO 2 was seen. Recently, androgens were found to show a significant positive correlation with [total-Hb] sup only in men in winter. Thus, BAT characteristics might be predicted by measuring plasma androgens as a biomarker in men in the winter.
We conclude that NIR TRS would be a useful non-invasive technology for assessing BAT-d, although further validation is still needed. Among the parameters evaluated by NIR TRS , the [oxy-Hb] sup as well as [total-Hb] sup would be applicable to assessing BAT characteristics in both cross-sectional and interventional studies.

AUTHOR CONTRIBUTIONS
TH, SN, SF, and YK collected the relevant literature and wrote the manuscript. SA, RK, MK, and TE assisted in illustrations, formatting, and collection of literature. NS, MM, MS, and TY coordinated and edited the relevant discussion on PET/CT measurements and BAT.

FUNDING
We have been granted by Japan Society for the Promotion of Science (15H03100 and 19H04061), which is an independent administrative institution, established by way of a national law for the purpose of contributing to the advancement of science in all fields of the natural and social sciences and the humanities.