1.

Introduction

Active brown adipose tissue (BAT) has been detected by ${}^{18}\mathrm{F}$-fluorodeoxyglucose (FDG)–positron emission tomography (PET) combined with computed tomography (CT)^1–4 in adults and neonates, as well as animals. It has been reported that human BAT activity is related to cold-induced thermogenesis,⁵ body weight,⁵ and glucose tolerance.^6,7 Furthermore, daily cold exposure increased the BAT activity in healthy individuals,^5,8,9 obese individuals,¹⁰ and patients with type 2 diabetes.⁷ Thus, BAT is expected to be a potential tool to combat obesity and lifestyle-related diseases.

Most epidemiological studies related to BAT have been informed by work in oncology and used ${}^{18}\mathrm{FDG}\text{-}\mathrm{PET}/\mathrm{CT}$ in thermoneutral conditions. These data showed that BAT activity was higher in women, younger individuals, and lean patients, as well as under conditions of lower ambient temperatures (AmT).^11–17 However, since these data were collected for the diagnosis of cancer, data on healthy adults were lacking, and the mean age was biased toward patients older than 60 years of age. In addition, the measurements were conducted under thermoneutral conditions, wherein the FDG uptake might have been underestimated.^1,18 Indeed, the ${}^{18}\mathrm{FDG}\text{-}\mathrm{PET}/\mathrm{CT}$-positive BAT ratio is variable; in studies involving cancer patients without cold exposure, the ratio was found to be extremely low ($\sim 5\%$).^{11–17,19,20} However, in younger healthy individuals, who were exposed to cold, the ratio was much higher ($\sim 50\%$ to 100%).^3,6 With regards to the effect of AmT on the activation of BAT, the results are conflicting with one study demonstrating that the rise in the BAT activity was delayed by a few months in association with a beginning in the decline in the AmT¹¹ and the other study stating that BAT activity was higher in early winter than late winter or early spring.²¹

Matsushita et al.⁶ and Yoneshiro et al.²² showed that functional BAT could be epidemiologically evaluated using cold-stimulated ${}^{18}\mathrm{FDG}\text{-}\mathrm{PET}/\mathrm{CT}$, in relatively healthy younger adults. They demonstrated that BAT activity was also negatively correlated with age, body fatness, and glucose metabolism. However, there were no sex differences in the BAT activity;^6,22 this was inconsistent with results of the aforementioned oncology studies, in which experimental cold exposure was not utilized. Thus, the difference in the sensitivity, between men and women, to mild outdoor cold exposure might be prominent in oncology studies.

In general, the near-infrared time-resolved spectroscopy (${\mathrm{NIR}}_{\mathrm{TRS}}$) method can evaluate the total hemoglobin concentration [total-Hb], which is an index of the blood volume or tissue vasculature. Therefore, this method can be used for vascular density evaluations, wherein BAT is expected to exhibit a higher density than that exhibited by white adipose tissue. Recently, we demonstrated, in humans, that the [total-Hb], determined by ${\mathrm{NIR}}_{\mathrm{TRS}}$, under thermoneutral conditions, is positively correlated with cold-induced ${}^{18}\mathrm{FDG}\text{-}\mathrm{PET}/\mathrm{CT}$ parameters, only in the supraclavicular region, which potentially contains BAT deposits.²³ As we found no difference between the [total-Hb] at 27°C and at the end of a 2-h-long exposure to cold conditions at 19°C, we were able to measure the [total-Hb] for the estimation of the vascularity of BAT, using ${\mathrm{NIR}}_{\mathrm{TRS}}$, without conditions of cold exposure. Collectively, the ${\mathrm{NIR}}_{\mathrm{TRS}}$ method is expected to be suitable for evaluating BAT density (BAT-d) in men and women, which is equivalent to the active BAT intensity or the amount of BAT, as determined by ${}^{18}\mathrm{FDG}\text{-}\mathrm{PET}/\mathrm{CT}$ with cold exposure.^23–26 As the ${\mathrm{NIR}}_{\mathrm{TRS}}$ method was introduced only in 2015, there are no data on the relationship of BAT with age, sex, anthropometrical parameters, and environmental factors. To verify the usefulness of ${\mathrm{NIR}}_{\mathrm{TRS}}$, there is a need to accumulate evidence on analogous data with regards to ${}^{18}\mathrm{FDG}\text{-}\mathrm{PET}/\mathrm{CT}$ with cold exposure. It is also important to obtain new evidence on human BAT-d, through epidemiological studies, using ${\mathrm{NIR}}_{\mathrm{TRS}}$, as otherwise, ${}^{18}\mathrm{FDG}\text{-}\mathrm{PET}/\mathrm{CT}$ measurement would be impossible, owing to several limitations such as the lack of accessibility to the relevant device and radiation exposure.

Thus, the purpose of this cross-sectional study was to confirm the relationship between BAT-d, determined through the use of ${\mathrm{NIR}}_{\mathrm{TRS}}$, and anthropometrical parameters, such as body fatness, to determine the differences in BAT-d, in terms of age and sex. We also aimed to identify influencing factors, such as seasonal fluctuations in the AmT on BAT-d, in Japanese individuals, across a wide age range.

2.

Methods

Volunteers were recruited through advertisements on posters or direct contact in three different areas—Hokkaido, Shiga Prefecture, and Tokyo, Japan. After the participants arrived at the laboratory, the following parameters were measured: BAT-d, height, body weight, percentage of body fat (%BF), visceral fat area (VFA), subcutaneous fat thickness in the supraclavicular region, systolic blood pressure (SBP), diastolic blood pressure (DBP), and heart rate (HR). The room temperature was regulated from 23°C to 27°C by the use of an air conditioner and heater. In addition, data on the AmT and day length, in each region, were obtained from the Japan Meteorological Agency²⁷ and the National Astronomical Observatory of Japan,²⁸ respectively. The study design and protocols were approved by the institutional review boards of Ritsumeikan University and Tokyo Medical University, in accordance with the ethical principles contained in the Declaration of Helsinki. Written informed consent was obtained from all the participants. These studies were conducted in the cold weather season (December, January, February, and March) from 2013 to 2017.

2.2.

Measurement of Anthropometric and Circulatory Parameters

Body mass index (BMI) was calculated as follows: body weight in kilograms divided by the square of the height in meters ($\mathrm{kg}/{\mathrm{m}}^2$). We calculated the body surface area using the DuBois equation: body surface area (${\mathrm{m}}^2$) = height (cm)^0.725 $\times$ body weight (kg) $0.425\times 71.84\times {10}^{-4}$. The %BF was estimated by a dual-energy x-ray absorptiometry scan (Lunar Prodigy; GE Healthcare, Buckinghamshire, United Kingdom) or the multifrequency bioelectric impedance method (Inbody 720 Body Composition Analyzer; Biospace, Seoul, Republic of Korea). The VFA, at the abdominal level of L4-L5, was estimated using 1.5-T magnetic resonance imaging (Signa HDxt; GE Healthcare, Buckinghamshire, United Kingdom) or a bioelectrical impedance analysis (EW-FA90; Panasonic, Osaka, Japan). The subcutaneous fat thickness at supraclavicular region was measured by B-mode ultrasonography (Vscan Dual Probe; GE Vingmed Ultrasound AS, Horten, Norway). The subcutaneous fat thickness was measured by the investigator using the attached distance measuring system and calculated as the mean value for three points (Fig. 1). SBP, DBP, and HR were measured by an automated sphygmomanometer (HBP-9020; Omron Healthcare, Kyoto, Japan).

Fig. 1

A typical image of the subcutaneous fat layer in the supraclavicular region This is an ultrasonic image of the supraclavicular region in a 27-year-old man. The subcutaneous fat layer was measured using B-mode ultrasonography (Vscan Dual Probe; GE Vingmed Ultrasound AS, Horte, Norway), whereas the thickness was measured using the attached distance measuring system by an investigator and calculated as the mean value for three points. The mean thickness in this region is 0.26 cm.

3.

Results

The [total-Hb] in the supraclavicular region was measured in a total of 413 participants using ${\mathrm{NIR}}_{\mathrm{TRS}}$, during the cold weather season, in Japan. The participants had a median age of 43.0 (33.0 to 58.0) years, BMI of $22.5(20.7\mathrm{to}24.5)\mathrm{kg}/{\mathrm{m}}^2$, and %BF of 26.8% (20.6% to 32.3%) (Table 1). As shown in Table 1, the BAT-positive ratio was 33.9% in all the participants (140/413), and it tended to significantly decrease with age; it was more than 50% in those in their twenties (43/75) but less than 20% in those in their sixties (6/40) and those older than 70 years (10/56).

To assess the impact of BAT-d, age, and sex on body fatness, we conducted univariate and multivariate analyses. The univariate regression analysis showed that BAT-d, age, and sex significantly was correlated with BMI, %BF, and VFA, respectively (Table 2). The multivariate regression analysis revealed that BAT-d and sex were determinants of BMI, %BF, and VFA, but age did not affect BMI (Table 2).

Table 2

Independent associations of age, sex, and BATs with body fatness.

After controlling for BMI, %BF, and VFA, we observed chronological and sex differences in the BAT-d. As shown in Fig. 2, the BAT-d of participants who were in their sixties was significantly lower than that of those in their twenties. A significant interaction (sex $\times$ chronological group) was noted, as was the presence of a significant main effect in the chronological group but not in terms of sex (Fig. 2). A significant sex difference was observed only among those in their twenties (Fig. 2). The BAT-d of participants older than 50 years was significantly decreased compared to that of men in their twenties (Fig. 2).

Fig. 2

Chronological and sex differences in terms of BAT. The [total-Hb] in the supraclavicular region potentially contains BAT. The values are expressed as means ± SE, adjusting for BMI, body fat ratio, and VFA.

To assess the factors influencing BAT-d, univariate and multivariate regression analyses were conducted. The univariate analysis showed sex, age, the mean AmT 6 weeks before the measurement day, and the average AmT during the 4 and 6 weeks before to be significant determinants of BAT-d; however, the multivariate regression analysis revealed that age and the average AmT during the 4 and 6 weeks before the measurement day remained determinants of BAT-d (Table 3). Although some of the AmT indices were significantly correlated with BAT, there was no significant correlation between BAT and day length (Table 3).

Table 3

Independent associations of age, sex, outdoor temperature, and day length with BAT.

As a significant result, following multivariate regression analysis, we investigated the relation between BAT-d and the average AmT during the 4 and 6 weeks before the measurement day after controlling for %BF and VFA. The BAT-d, when measured at temperatures lower than 3.7°C of the average AmT, during the 4 and 6 weeks before the measurement day, was significantly higher than that measured at temperatures higher than 5.0°C (Fig. 3).

Fig. 3

The relation between BAT and the average temperature at four and six weeks before the measurement day. The [total-Hb] in the supraclavicular region potentially contains BAT. Analysis of covariance was conducted, adjusting for body fat ratio and VFA.

4.

Discussion

In the present cross-sectional study, we applied ${\mathrm{NIR}}_{\mathrm{TRS}}$ to evaluate the BAT-d in 413 Japanese individuals, aged 20 to 85 years. Our study has three main findings. First, we confirmed that BAT-d was negatively correlated with body fatness. Second, older age was associated with decreased BAT-d, with positive activity rates of BAT observed in those in their fifties and sixties, and those above 70-years old at 25.0%, 15.0%, and 17.9%, respectively. Third, the average ambient lower temperature, during the 4 and 6 weeks prior to the measurement day, under free-living situations, was associated with an elevation in the BAT-d. In particular, the [total-Hb], at an average AmT lower than 3.7°C, during the 4 and 6 weeks prior to the measurement day, was higher than that observed at temperatures higher than 5.0°C.

The positive rate of BAT observed in our study (33.9%), as determined in the same manner as the ${\mathrm{NIR}}_{\mathrm{TRS}}$ measurements,²³ was comparable or slightly lower than that obtained in a Japanese study, which used ${}^{18}\mathrm{FDG}\text{-}\mathrm{PET}/\mathrm{CT}$ with cold exposure (41.4%).²² The reason for the relatively lower rate could be the fact that our study included elderly adults, with a median age of 43.0 years, whereas the previously conducted study predominantly included those in the age range of 20 to 40 years (median age of 26.0 years).⁶ In fact, the positive rates of BAT observed in the participants in their twenties (57.3% versus 53.0%), thirties (46.1% versus 39.5%), and forties (29.1% versus 26.9%) were similar to those noted in another study.²² The finding of the positive rate of BAT among those in their fifties (25.0%), sixties (15.0%), and adults over the age of 70 years (17.9%) is new, in Japanese settings.

We showed that body fatness parameters were negatively correlated with BAT-d, which is consistent with the results of various studies that used ${}^{18}\mathrm{FDG}\text{-}\mathrm{PET}/\mathrm{CT}$, regardless of the presence of cold exposure.^{6,14–17,19,20} Although we should consider the possibility that the BAT-d was affected by subcutaneous fat thickness,²⁶ the subcutaneous fat thickness in the supraclavicular region is within a very narrow range and has little effect on the measured parameters. For instance, the fat thickness in the supraclavicular region ranged from 0.6 to 2.5 mm when evaluated by B-mode ultrasonography in this study. This corresponds to an attenuation of the NIR signal by $<10\%$ of the assumed 0-mm fat layer thickness.³² After the elimination of the effect of confounder factors, using stepwise multiple regressions, the relationship between BAT-d and BMI, %BF, and VFA was obtained.

Findings on the sex differences in terms of BAT are controversial.^{6,13–17,19,21} BAT activity was found to be higher in women, in many oncology studies.^13–17,20 However, no sex differences were found in ${}^{18}\mathrm{FDG}\text{-}\mathrm{PET}/\mathrm{CT}$ studies with cold exposure.^6,22 The oncology studies had very large sample sizes but did not include cold exposure, whereas the studies using ${}^{18}\mathrm{FDG}\text{-}\mathrm{PET}/\mathrm{CT}$ with cold exposure had relatively small sample sizes. In this study, we evaluated BAT-d in 413 individuals, using ${\mathrm{NIR}}_{\mathrm{TRS}}$, which is equivalent to ${}^{18}\mathrm{FDG}\text{-}\mathrm{PET}/\mathrm{CT}$ with cold exposure. As a result, the fact that BAT-d is higher in women was not confirmed. Further studies are necessary to assess sex differences in terms of BAT.

This is the first study to demonstrate that the average ambient lower temperature, during the 4 and 6 weeks prior to the measurement day, under free-living situations, is associated with an elevation in the BAT-d. Laboratory studies showed that daily cold exposure (10°C to 17°C, 2 to $6\mathrm{h}/\mathrm{day}$), when dressed in unnaturally light clothing (t-shirt with underwear), for 10 days to 6 weeks, promoted BAT recruitment,^5,8–10 and the results of these previous studies might provide a hint for the consideration of our result. Cohade et al.¹¹ showed a rise in the monthly prevalence of BAT activity during the winter, but the rise was delayed by a few months, in relation to the beginning of a decrease in the AmT. Kim et al.²¹ showed that BAT activity was higher in early winter than later winter or early spring. The temperature at which some cold exposure studies were carried out at can be equivalent to a degree in our cold season. However, since the type of clothing could have varied and the duration of the daily cold exposure in unknown, it is difficult to compare the results of our study with those of other studies. Despite the undetermined duration of the daily cold exposure needed to increase functional BAT activity/mass, in this study, we found that the [total-Hb], at an average AmT lower than 3.7°C, during the 4 and 6 weeks prior to the measurement day, was higher than that at temperatures higher than 5.0°C. This observation suggests that BAT-d increases during the 4 and 6 weeks after the average AmT drops to $<4°\mathrm{C}$ or 5°C. A larger variation in the [total-Hb], at a lower temperature, indicates differences in the individual sensitivity to the exposure to cold environments; participants with a low BAT-d could be insensitive while differing sensitivities could have been present among those with a high BAT-d. Because we could not acquire relevant longitudinal data, we are uncertain of the effects of an increase in the outside temperature on BAT-d. However, according to a thermogenic capsinoid supplementation study conducted to increase BAT-d,²⁴ the cessation of supplementation during an increase in the outside temperature decreased BAT-d. Therefore, we speculate that BAT-d would gradually decrease several months later, when the outside temperature increases during spring. Further studies, with a longitudinal design and larger sample size, are needed, along with the monitoring of the degree and duration of daily cold environments, to determine the critical AmT cutoff value needed to recruit functional BAT under free-living situations.

There was no significant correlation between BAT density and day length, in this study. Contrary to our results, one previously conducted study showed, in its univariate analysis, that a shorter day length was positively correlated with BAT activity.¹² However, that study did not include cold exposure before ${}^{18}\mathrm{FDG}\text{-}\mathrm{PET}/\mathrm{CT}$ measurements;¹² therefore, the AmT, which is related to day length, on the measurement day, may be confounded. Although Ouellet et al.¹⁵ also demonstrated, through an univariate analyses, the presence of a significant relationship between day length and BAT activity, the relationship was not significant in the multivariate analyses.

The validity of evaluating BAT-d using ${\mathrm{NIR}}_{\mathrm{TRS}}$ has been verified in previously conducted studies.^23,24 We reported a significant relationship between BAT-d, as evaluated by the [total-Hb] using ${\mathrm{NIR}}_{\mathrm{TRS}}$, and BAT activity, as evaluated by ${}^{18}\mathrm{FDG}\text{-}\mathrm{PET}/\mathrm{CT}$. We have already shown that, in younger individuals, the [total-Hb] can be increased, and that the degree of the increase in BAT-d and BAT activity is almost similar (48.8% and 46.4%, respectively) after daily thermogenic capsinoid ingestion,²⁴ which is known to increase BAT-d and/or mass. We also found that the [total-Hb] tended to decrease during the 8-week follow-up period, after capsinoid treatment.²⁴

Precise determination of the volume of interest is difficult using the NIR spectroscopic (NIRS) method. On the basis of a continuous wave NIRS study in the milk model by Chance et al,³³ if we assume the volume of interest to be a rotation body of the ellipse revolving across 180 deg or a hemisphere with a 3-cm optode separation, it can be estimated to be $\sim 4{\mathrm{cm}}^3$. However, as a banana-like pattern has been reported for photon migration,³⁴ the above assumption might be inappropriate. Moreover, according to a recent study, the mean depth of the light penetration with a 3-cm optode separation using ${\mathrm{NIR}}_{\mathrm{TRS}}$ would be $\sim 2\mathrm{cm}$,³⁰ which is greater than half of the emitter–detector separation reported in many previous studies.^29,33,34 Although ${\mathrm{NIR}}_{\mathrm{TRS}}$ is noninvasive, uncomplicated, less expensive than PET/CT, and free of radiation exposure, for the evaluation of tissue oxygenation, in humans,²³ further studies on the volume measured using ${\mathrm{NIR}}_{\mathrm{TRS}}$ are needed.

Although several studies have reported an increase in the BAT activity in younger individuals, the question of whether BAT activity can be increased in individuals with a lower BAT-d and/or advancing age remains unresolved.^35–37 In this study, we found that even some individuals over their fifties had BAT-positive rates of 15% to 25%, and this suggests the possibility to maintain and/or increase the BAT-d even in relatively older individuals. However, the effect of genetic predispositions cannot be ruled out for individuals with a higher BAT-d.

This study has several limitations. It included a relatively small number of participants, although some other studies on BAT also involved only a small sample size. The presence of selection bias and variation in the age distribution cannot be ruled out. The cross-sectional design has the inherent disadvantage of defining causality. We did not evaluate BAT activity using the well-established, standardized ${}^{18}\mathrm{FDG}\text{-}\mathrm{PET}/\mathrm{CT}$ method, although several studies have validated the use of ${\mathrm{NIR}}_{\mathrm{TRS}}$.

In conclusion, the results of this cross-sectional study showed that a decrease in BAT-d is associated with advancing age and the lower AmT was associated with an increase in the BAT-d through the use of ${\mathrm{NIR}}_{\mathrm{TRS}}$, in a wide age range of participants. Our results are consistent with those of previously conducted ${}^{18}\mathrm{FDG}\text{-}\mathrm{PET}/\mathrm{CT}$ studies. In particular, the average AmT, which was lower than 4°C or 5°C during the 4 and 6 weeks prior to the measurement day, might be a determining factor for increasing BAT-d. As the study sample of this cross-sectional and exploratory study was too small to draw a conclusion, further studies are needed to identify the factors that affect BAT-d, which take into consideration genetic predisposition and environmental factors, such as clothing, physical activity levels, and nutritional factors.