Impaired Cold-Stimulated Supraclavicular Brown Adipose Tissue Activity in Young Boys With Obesity

Pediatric obesity is a growing worldwide problem. The World Health Organization has suggested that the prevalence of obesity in children has reached “alarming proportions” (1). Childhood obesity is linked to unfavorable health outcomes including type 2 diabetes, nonalcoholic fatty liver disease, and other cardiovascular disease risk factors (2) in childhood and adulthood. Interventions to treat childhood obesity have generally focused on changes in health behaviors to achieve a negative energy balance. These interventions have had only modest efficacy and have been challenging to sustain, highlighting the need to identify novel approaches (3).

Brown adipose tissue (BAT) contributes to energy expenditure through its ability to produce heat by uncoupling mitochondrial oxidative phosphorylation through uncoupling protein 1 (UCP1) (4) and possibly other futile cycles (5). BAT is detectable in the supraclavicular (SCV) region of the neck in children, adolescents (6–8), and adults (9–11). Cold exposure has been reported to be the most powerful stimulus for BAT activity, and it enhances BAT oxidative metabolism (12). Cold-induced uptake of ¹⁸F-labeled fluorodeoxyglucose (¹⁸F-FDG uptake) in the SCV region, measured with positron emission tomography and computed tomography (PET-CT) imaging, is the most common way to measure BAT activity (4). Studies using this methodology report that adults with obesity and with diabetes have lower cold-stimulated BAT activity than those without (13–15). In children, PET-CT scans obtained for oncological follow-up have been used to examine BAT ¹⁸F-FDG uptake under ambient conditions (6–8). In these reports, ¹⁸F-FDG uptake was lower in children with obesity (6), higher during puberty (8), and directly related to the child’s muscle volume (7,8). However, in adults, imaging BAT in ambient conditions reveals 2–7% BAT-positive individuals vs. 70–100% with cold exposure (16). Furthermore, BAT is identified in biopsies from individuals who do not have detectable ¹⁸F-FDG uptake (17), reinforcing the need for cold exposure to accurately assess BAT metabolic activity.

Due to exposure to ionizing radiation, no studies have been done with PET-CT to assess cold-stimulated BAT in children. In the only studies with examination of cold exposure in children, SCV skin temperature was measured with infrared thermography, demonstrating an increase in skin temperature (18–20) that was blunted in children with higher BMI (19) or increased liver fat (21). However, infrared thermography has significant limitations as a tool to evaluate BAT, given that BAT is not the sole source of heat in the SCV region during cold exposure (4) and that the variability in the thickness of the subcutaneous fat overlying BAT in the SCV region may alter results due to differences in heat conduction (22). Therefore, alternative methods to assess BAT activity in children are required.

One method developed to measure BAT activity without exposure to ionizing radiation or the influence of superficial subcutaneous fat involves the measurement of proton density fat fraction (PDFF) using MRI (4). The method is based on the observation that the primary fuel of activated BAT during cold exposure is endogenous triglycerides (23). Consistent with these findings, only the change in fat fraction in response to cold exposure correlated with SCV ¹⁸F-FDG uptake (24,25). Additionally, whole-body cold exposure, known to stimulate BAT activity in adults (15,23,26), leads to reductions in SCV PDFF within 10 min and reaches a nadir within the first hour of exposure (11).

Importantly, BAT PDFF was histologically verified and correlated with UCP1 expression (27–30) and is lower in adults with obesity (24) or nonalcoholic fatty liver disease (31). While studies in infants (32) and children have found that lower SCV PDFF measured at ambient temperature is associated with reduced weight gain (33), lower BMI, hepatic fat, and abdominal obesity (34), and changes in blood glucose (35,36), SCV PDFF measured at ambient temperature is not directly associated with changes in BAT activity assessed with FDG-PET imaging (24). Therefore, whether cold-stimulated changes in SCV PDFF are associated with obesity in children has not yet been investigated.

In the current study, we examined cold-stimulated changes in SCV PDFF in prepubertal boys. Our objectives were to 1) evaluate change in SCV PDFF with a 1-h whole-body cold exposure (18°C), 2) compare the cold-stimulated change in SCV PDFF between boys with and without overweight/obesity, and 3) examine the relationship of cold-stimulated change in SCV PDFF with adiposity in young boys.

Boys aged 8–10 years were recruited to this two-visit, cross-sectional study. Participants were recruited through public announcements and from the Growing Healthy pediatric weight-management clinic at McMaster Children’s Hospital between February 2018 and August 2019. The study was approved by the Hamilton Integrated Research Ethics Board (HiREB). All participants provided informed assent; informed consent was provided by the child’s guardian. Exclusion criteria included contraindications for MRI (claustrophobia, implanted metal, metallic injuries, recent tattoo, or weight >300 lbs), prior bariatric surgery or liver transplant, and use of medications likely to influence BAT activity or metabolic health (β-adrenergic, steatogenic, antihyperglycemic, antidepressant, antipsychotic, anxiolytic, thyroid hormone, and antiemetic: 5HT3 antagonists or serotonergic drugs) (11). We intentionally chose this initial group to be very similar in age, young enough to be prepubertal but likely old enough to withstand the cold stimulus and cooperate with the MRI examinations without sedation.

Participants were asked to fast for at least 8 h, to avoid intake of caffeine for at least 12 h, and to abstain from serotonin-rich foods (e.g., banana, tomato, kiwi, walnut, avocado, pineapple, and plum) for at least 24 h prior to each study visit (37). Visits 1 and 2 (Supplementary Fig. 1) took place in the morning and were scheduled at least 8 days apart but ideally within 1 month of each other.

Visit 1 took place at the McMaster Children’s Hospital (Hamilton, Ontario, Canada). Weight (in kilograms) was measured by trained research personnel using an electronic platform scale (BMI scale model 882; Seca, Hamburg, Germany) to the nearest 0.1 kg. Height (centimeters) was measured with a wall-mounted stadiometer (height measuring rod model 240; Seca) to the nearest 0.1 cm. BMI was calculated as weight in kilograms divided by the square of height in meters. BMI z score was calculated with the WHO AnthroPlus anthropometric calculator, with the World Health Organization defining normal weight according to a z score <1 and overweight and obese z score ≥1. Waist circumference (centimeters) was measured at the midpoint between the costal margin and the iliac crest at the end of expiration with a weighted measuring tape (Pull Type Spring Scale; OHAUS, Parsippany, NJ) set at 750 g. Weight, height, and waist circumference were each measured three times, and the average was used. Body composition (body fat [percentage], fat mass [kilograms], lean mass [kilograms]) was assessed with a DXA scanner (Lunar PRODIGY Advance 8743; GE Healthcare, Waukesha, WI). All scans were reviewed by one individual (K.M.M.) to ensure the consistency of the regions of interest (ROIs).

An oral glucose tolerance test (1.75 g/kg to a maximum of 75 g glucose) was undertaken with blood samples drawn with subjects fasting and after 2 h, and glucose was measured in heparinized plasma samples with the ARCHITECT c4000 or c4100 (Abbott, Chicago, IL). The fasting lipid profile (fasting serum total cholesterol, HDL cholesterol [HDL-C], triglycerides) was measured with ARCHITECT C16 (Abbott) in the Clinical Laboratory at Hamilton Health Sciences (Hamilton, ON, Canada), and LDL cholesterol was calculated with the Friedewald equation (38). Fasting serum insulin levels were measured with the Immunometric (noncompetitive) methodology on the VITROS XT 7600 Microwell (Ortho Clinical Diagnostics, Rochester, NY). HOMA of insulin resistance (HOMA-IR) was calculated as follows: fasting insulin (mU/L) × fasting glucose (mmol/L)/ 22.5. QUICKI was calculated as 1 divided by the sum of log fasting insulin (mU/L) and log fasting glucose (mg/dL). HOMA-IR and QUICKI are presented herein as surrogate measures for insulin sensitivity (39).

Visit 2 took place at the Imaging Research Centre at St. Joseph’s Healthcare Hamilton (Hamilton, ON, Canada). It involved MRI of the SCV BAT, liver, and abdomen before and after a standardized cold exposure protocol. On arrival, participants wore a standardized cotton tank top and shorts and were acclimated to room temperature for 30 min. The temperature in the room was recorded (Wireless Forecast Station with Pressure History, model no. WS-9037U-IT; La Crosse Technology, La Crosse, WI). The cooling protocol was based on published methodology (11,23,40), with a modification in the length of cold exposure from 3 h to 1 h, consistent with our findings in young adults (11). Participants were fitted with a high-density liquid-conditioned suit (two pieces with zippers and open access neck) (Med-Eng, Ottawa, Ontario, Canada), and 18°C water was circulated with a temperature- and flow-controlled circulation bath (Isotemp 6200 R28; Fisher Scientific, Waltham, MA). During cold exposure, participants stayed in a hospital bed in a semireclined position with access to a device to watch a movie but were not permitted to type or do any activities that required movement. Shivering activity and skin temperature were monitored during cold exposure as detailed below.

For estimation of shivering during cold exposure, surface electrodes were placed on the trapezius, vastus lateralis, and vastus medialis muscles. Electromyography (EMG) signals were recorded at baseline (before the start of cold exposure) and continuously during cold exposure with surface EMG (Trigno Wireless System, Delsys, Natick, MA, with Trigno Snap Lead Sensor connected to prejelled Norotrode 20 Bipolar SEMG electrodes, Myotronics, Kent, WA). For minimization of voluntary muscle activity throughout the cold exposure, the participants were asked to avoid voluntary movements during the recording periods.

Raw EMG signals were collected at a sampling rate of 2,148 Hz, filtered for removal of spectral components <20 Hz and >500 Hz as well as 60 Hz contamination and related harmonics, and analyzed with custom-designed MATLAB algorithms (MathWorks, Natick, MA). The shivering intensity of individual muscles was determined from root mean square (RMS) values calculated from raw EMG signals with a 50-ms overlapping window (50%). In brief, baseline RMS values (15-min RMS average measured before cold exposure) were subtracted from cold-exposure (shivering) RMS values.

After the baseline MRI scans, continuous evaluation of skin temperature was done via 15 autonomous temperature sensors (41) (Thermochron iButtons, no. DS1922L; Maxim Integrated, San Jose, CA) that were placed over the abdomen, biceps, cheek, chest, foot, forearm, forehead, front and back of the lower leg, hamstring, hand, occiput, quadriceps, and upper and lower back on the left side of the participants’ body. Skin temperature was monitored every minute for 10 min before and during the cold protocol. The arithmetic mean skin temperature from the 15 iButtons was calculated for each participant (42), and then the change in mean skin temperature (ΔT_skin) was calculated as follows: ΔT_skin = postcold T_skin (mean of last 10 min) – precold T_skin (mean of 10 min before the start of cold exposure).

Afferent (water entering the suit) and efferent (water exiting the suit) temperature were recorded during the cold exposure at 15 s intervals with a dynamic range of 16 bit with use of a data logger (PowerLab; ADInstruments, Sydney, Australia) connected to thermocouples (TMQSS-020G-2; OMEGA Engineering, Stamford, CT) fixed to the inlet and outlet manifolds. The difference between the efferent and afferent temperatures (ΔT) was calculated from data collected during the last 10 min of the cold exposure.

MRI was performed with a 3-Tesla whole-body scanner (Discovery MR750; GE Healthcare). SCV measurements of PDFF were acquired with iterative decomposition of water and fat with echo symmetry and least-squares estimation (IDEAL-IQ) pulse sequences, which are confounder-corrected three-dimensional gradient multiecho MRI sequences. MRI-PDFF measures the ratio of the total density of fat mobile protons to the total density of fat and water mobile protons (43) and differentiates BAT from the surrounding white adipose tissues (WAT) and muscles based on their physical properties. Thus, BAT-containing depots show lower PDFF than WAT (44,45). IDEAL-IQ provides an accurate measure of tissue triglyceride content by using multiple spectral modeling of adipose tissues while accounting for T2* decay (43). SCV measurements of PDFF were acquired with a head/neck/spine coil, with an additional attachment that provided signal from the anterior portion of the chest, and axial images were taken from the C2/C3 disc to the T4/T5 disc (slice thickness 3 mm, 50 slices, flip angle 4°, echo time (TE) 1.3 ms, repetition time (TR) 8.4 ms, field of five (FOV) 340 mm, image resolution 1.52 × 1.42 × 3 mm, acceleration factor 2, scan time 2.4 min). Six distinct image contrasts were generated: water-only, fat-only, in-phase, out-of-phase, corrected PDFF, and R2∗ images.

Segmentation and analysis of all MRIs were conducted with AnalyzePro (version 1; Biomedical Imaging Resource, Mayo Clinic, AnalyzeDirect, Overland Park, KS) by one reader (B.A.A.). The SCV region was segmented (the adipose tissue bound by the sternocleidomastoid medially, trapezius posteriorly, and clavicle inferiorly between the C5–C6 and T1–T2 disks) as previously described (11,31,46) (Fig. 1A and B). The image analysis was consistent with that described in our previous publications (11,31) but is described briefly here. A fat mask was generated from the fat-only images and applied to the PDFF map at the C7-T1 disc to isolate the adipose tissues and exclude background noise from the MRI images. Then a 30–100% fat fraction threshold was applied to the PDFF images. The ROI was manually drawn over the SCV region as defined above. A onetime two-dimensional (2D) erosion (1 × 3 voxels) was applied to the ROIs to correct for partial volume effects. T2∗ images were generated from the R2∗ data using the following relationship: T2∗ = 1/R2∗. We applied a T2∗ mask to the fat fraction map to differentiate BAT from WAT by isolating voxels with a T2∗ between 2 and 25 ms. The lower range of 2 ms was selected due to the MRI sequence’s limitation in detecting very low T2∗ values, and the higher range was based on a published report that tissue with a T2∗ value of ≥26 ms consists mostly of muscle, fluid, or white adipocytes (47). All voxels that satisfied the above criteria were averaged and classified as SCV PDFF.

For evaluation of BAT activity accounting for the precold SCV PDFF, the percent change in SCV PDFF was calculated as [(precold SCV PDFF − postcold SCV PDFF)/(precold SCV PDFF)] * 100 and is noted as cold-induced percent decline in SCV PDFF—a measure of BAT activity (31). For comparison with SCV BAT, the PDFF of the subcutaneous adipose tissue (SAT) in the posterior neck region (neck SAT) was analyzed at three MRI slices, C5–C6, C6–C7, and T1–T2, with manual tracing and the criteria that were used for SCV BAT (Fig. 1B).

ROIs were drawn over the entire liver using with a tool that “snaps” to the edges of regions where changes in voxel intensities are high. The in-phase image was used as a reference when the boundaries of the liver were not clearly defined in the fat-fraction image. The ROIs were then postprocessed with a 2D erosion (3 × 3 voxels) to correct for partial volume effects. All voxels that satisfied the above criteria were averaged and classified as hepatic fat (Fig. 1C). MRI-PDFF is a well-established, noninvasive, quantitative, and accurate means to measure hepatic fat content (48).

Abdominal SAT PDFF was calculated according to our previously published protocol (31) from a single slice at L4–L5 obtained in a single breath hold pre- and postcold scans (IDEAL-IQ, slice thickness 5 mm, 32 slices, flip angle 3°, TE 6.3 ms, TR 1.0 ms, FOV 360 mm, image resolution 2.25 × 1.97 × 5 mm, acceleration factor 2 × 2, scan time 14 s). A manual tracing tool in AnalyzePro was used to segment abdominal SAT. Voxels with low PDFF values (<30%) were excluded to isolate adipose tissues from muscle. We applied a 2D (3 × 3 voxels) to attenuate partial volume effects. All voxels that satisfied the above criteria were averaged and classified as SAT PDFF (Fig. 1D).

Visceral adipose tissue (VAT) area was measured from a single slice located at the level of the umbilicus that has been previously validated for CT measurement of VAT (48). The images were obtained from a single breath hold LAVA Flex precold scan (LAVA Flex, slice thickness 5 mm, 32 slices, flip angle 3°, TE 1.3 ms, TR 4.2 ms, FOV 360 mm, spatial resolution 1.13 × 1.29 × 5mm, acceleration factor 1.5, scan time 12 s). LAVA Flex is a three-dimensional gradient dual-echo MRI sequence that generates four distinct image contrasts: water only, fat only, in phase, and out of phase. Due to its nature, the LAVA Flex pulse sequence provides a higher resolution image with sharper tissue boundaries compared with IDEAL-IQ. For postimaging processing, a fat mask was applied for exclusion of nonadipose tissues from analysis, while SAT and VAT were separated semiautomatically based on seed points. Manual tracing was used to include parts that may have been missed from the seeding. Retroperitoneal adipose tissue was excluded manually from VAT. Finally, the ROIs were postprocessed with use of 2D erosion (3 × 3 voxels) for attenuation of partial volume effects (Fig. 1E).

Our primary objective was to examine the relationship between BAT activity and weight-related measures controlling for shivering activity. Based on the work of Harris (50), the suggested number of participants should equal 30 + m (where m is the number of independent variables). Thus, a sample size of 34 participants was needed to have sufficient power to address the primary question. We also wanted to compare the cold-induced changes in SCV PDFF between those with normal BMI and boys with overweight/obesity. However, as no studies of cold-induced changes had previously been done in children, we were unable to determine the required sample size, so this was an exploratory aim.

SPSS Statistics (version 27; IBM, North Castle, NY) was used for all statistical analyses. We assessed normality by following the procedures outlined by Tabachnick and Fidell (51) where variables with Z_skewness or Z_kurtosis ≥|3.29|were classified as nonnormal. Nonparametric tests were used for analyses if data were nonnormally distributed. Data are presented as mean ± SEM or median (quartile 1, quartile 3) for skewed variables. Paired Student t test was used to examine the effect of cold exposure on the MRI outcomes (before and after cold exposure). Independent-samples t test and Mann-Whitney U test were used to compare differences of normally distributed and skewed variables, respectively, between boys with normal weight and those with overweight/obesity. Paired Student t test was used to evaluate the changes in PDFF of different fat depots from before to after cold exposure. Pearson correlation coefficient was used to assess the association between normally distributed variables. Spearman correlation coefficient was used to assess the association between skewed variables and for nonlinear monotonic relationships. Multivariate regression models were used to examine whether cold-induced percent decline in SCV PDFF (independent variable) was associated with significant correlates (dependent variables) (BMI, waist circumference, and VAT) after confounding factors (shivering in the trapezius muscle at 60 min of cold exposure) were controlled for. A two-tailed P value of <0.05 was considered significant.

The clinical data set generated/analyzed during the current study has not been deposited in a public repository because the consent form signed by study participants noted that individualized data would not be published. Grouped data are available on reasonable request. All requests should be directed to and will be fulfilled by the corresponding author. No applicable resources were generated or analyzed during the current study.

Of the 98 participants who inquired about the study, 35 were enrolled and attended visit 1 (Fig. 2). Of these, 33 attended visit 2 after a median of 12 days (range 8–43), and 26 completed both the pre- and postcold MRI scans. Seven participants were unable to complete the MRI measurements due to claustrophobia. The cold exposure was 60 min in 25 children and was terminated at 38 min in 1 child on the child’s request. The cold stimulus was similar between boys with and without overweight/obesity as measured with the temperature difference between outlet and inlet temperature of the cooling suit (P = 0.362) and the decline in skin temperature (P = 0.333) (Table 1).

Of the 26 boys with pre- and postcold scans, 10 boys were obese and 3 were overweight based on BMI z score. As noted in Table 1, boys with overweight/obesity had, as expected, higher body weight, waist circumference, total body adiposity, hepatic fat (median [quartile 1, quartile 3] 4.6% [3.5%, 8.7%] vs. 3.0% [2.5%, 3.2%]; P < 0.001), and VAT (mean ± SEM 24.3 ± 4.5 vs. 4.7 ± 0.6 cm²; P = 0.001) than those with normal BMI, respectively. Although there were no differences between groups in fasting or 2-h blood glucose, boys with overweight/obesity had lower indices of insulin sensitivity (HOMA-IR, QUIKI) compared with boys with normal BMI. One boy with obesity had elevated 2-h blood glucose (12.3 mmol/L).

Boys with overweight/obesity had higher precold SCV PDFF compared with those with normal BMI (mean ± SEM 72.7% ± 2.2% vs. 56.6% ± 1.4%, respectively; P < 0.001) (Fig. 3A). With cold exposure, SCV PDFF declined in boys with normal BMI but not in boys with overweight/obesity (Fig. 3A). The absolute cold-induced decline in SCV PDFF tended to be lower in boys with overweight/obesity compared with normal BMI (1.1% ± 0.5% [range −2.0 to 4.2] vs. 2.7% ± 0.7% [−1.1 to 7.1]; P = 0.085) (Fig. 3B). The relative percent decline in SCV PDFF was less in boys with overweight/obesity compared with the normal BMI group (1.6% ± 0.8% [range −3.0 to 6.1] vs. 4.7% ± 1.2% [−1.9 to 12.0]; P = 0.044) (Fig. 3C).

Importantly, there was no cold-induced change in PDFF within the neck SAT or abdominal SAT in either group (Fig. 4A and B) suggesting that the differences observed in PDFF within the BAT are unlikely due to artifactual MRI phantoms as a result of temperature changes.

The cold-induced percent decline in SCV PDFF was inversely related to BMI (r = −0.39, P = 0.047), waist circumference (r = −0.48, P = 0.014), and VAT (ρ = −0.47, P = 0.014), but the relationships with BMI z score, total body fat, and hepatic fat were not statistically significant (r = −0.37, P = 0.065, ρ= −0.28, P = 0.167, and ρ = −0.34, P = 0.088, respectively) (Fig. 5).

The percent decline in SCV PDFF was unrelated to shivering measured in the vastus lateralis or medialis muscles at any time point during the cold exposure. However, boys with normal BMI had higher EMG shivering signals in the trapezius and the vastus lateralis muscles compared with boys with overweight/obesity (P = 0.035 and 0.034, respectively) (Supplementary Table 1). Boys with greater decline in SCV PDFF also had more shivering in the trapezius after 20, 30, 40, and 60 min of cold exposure (Supplementary Fig. 2).

For evaluation of whether the cold-induced change in SCV PDFF predicts the anthropometric and body composition variables independent of the shivering intensity in the trapezius, a multivariate regression analysis was done. As demonstrated (Supplementary Table 2), cold-induced percent decline in SCV PDFF controlling for shivering activity in the trapezius muscle was not related to BMI (P = 0.078) or VAT (P = 0.064).

Using cold-induced percent decline in SCV PDFF to measure BAT activity, we have demonstrated that young boys with overweight/obesity have lower BAT activity than age- and sex-matched control subjects with normal BMI. Further, as this is the first evaluation of cold-induced changes in SCV PDFF in children, we have demonstrated the feasibility of carrying this out in a standardized manner (fixed cold temperature for all the participants) as is done in adults (11). Our findings are consistent with previous studies in adults with use of a similar methodology (11,26) and suggest that even at a young age, boys with overweight/obesity have lower cold-induced BAT activity. Further, there is an inverse relationship between cold-induced BAT activity and both VAT and waist circumference.

Boys with normal BMI in this cohort had more variability in their cold-induced BAT activity (percent decline in PDFF ranges −1.9% to 12%) than boys with overweight/obesity (−3.0% to 6.1%). Of the 13 boys with normal BMI, 4 had particularly high BAT activity, contributing to the significant relationships with total body adiposity and VAT volume. We were not able to identify distinct phenotypic characteristics of these boys. Further, we do not know the significance of low “BAT activity” in boys with normal BMI. As none of these boys had metabolic abnormalities or visceral fat accumulation, this relationship could not be truly examined. Lower SCV PDFF (without cold stimulation) in infants predicted less gain in body fat over the subsequent months (33). Little is known about the trajectory of cold-stimulated BAT activity in children, and longitudinal studies should be undertaken to determine the relationship between low BAT activity and metabolic health over time. These observations highlight the importance of future work with a larger sample of boys with variable body composition.

The use of the liquid-conditioned suit in this study enabled a direct measurement of heat production with measurements of the inlet-outlet temperatures (52,53), and these did not differ between groups, suggesting that the cold stress was the same in both groups. We noted that boys with normal BMI had greater shivering than boys with overweight/obesity in the trapezius and the vastus lateralis muscles only. These differences in shivering noted may suggest that children with overweight/obesity either 1) relied on other nonshivering heat-generating mechanisms or 2) were able to produce an equivalent amount of heat by recruiting a smaller relative proportion of motor units and muscle fibers. With our measurements for shivering we are unable to adjust for muscle mass recruited in shivering.

Boys with overweight/obesity also had a higher precold SCV PDFF than those with normal BMI. This is consistent with previous reports suggesting children with obesity have higher SCV PDFF at ambient temperature (46,54) and the finding that lower BMI in preschool children was related to lower ambient SCV PDFF (34). The mean precold SCV PDFF in our cohort is higher than the mean found in a group of slightly younger children (mean ± SD 64.7% ± 2.1% vs. 57.2% ± 5.2%, respectively) (35), but this likely stems from differences in the extent of obesity; 2 of 63 children in the study of Andersson et al. (35) had obesity, 1 had severe obesity, and 19 were overweight in contrast to 6 boys (17%) with severe obesity (BMI z score ≥3) in our study. Ambient SCV PDFF in the boys with normal BMI is comparable with that in children in the study of Andersson et al. (56.6% ± 1.4%).

The absolute reduction in SCV PDFF in this cohort is lower than what has been reported in young adults by us and others (11,55). Similarly, the percent change in SCV PDFF in this cohort is lower than what we measured in young adult males (3.2% ± 0.8%, n = 26, vs. 6.6% ± 0.9%, n = 23, respectively). Furthermore, boys with normal BMI have lower precold SCV PDFF compared with young adult males with normal BMI we previously studied (56.5 ± 1.4, n = 13, vs. 64.6 ± 1.4%, n = 16). These observations raise questions about the trajectory of BAT activity over time from childhood to adulthood. Although BAT is most abundant in infancy (56) and declines with time, little is known of the trajectory of BAT or the factors that influence it. This study supports the feasibility of conducting such studies in the future.

We examine herein cold-induced BAT activity using MRI in children; this has not been previously investigated. Although we have demonstrated the feasibility of this approach, future studies will be needed to determine whether low BAT activity prospectively predicts future adiposity or metabolic health. Further, we cannot generalize our findings to females, other age-groups, or youth who have entered puberty. Ambient ¹⁸F-FDG uptake in PET-CT scans obtained for cancer surveillance is higher in those who have entered puberty (6,57–59). It will be important to reexamine this finding using cold-stimulated BAT activity, as ambient ¹⁸F-FDG uptake in adults (24) is a poor measure of BAT activity. To better understand the relationship of metabolic health with BAT activity, larger studies in children with metabolic abnormalities will be required. Consideration should be given to the measurement of energy expenditure during cold stimulation and the relationship of this to BAT activity—something we were unable to do for technical reasons.

Cold-stimulated SCV PDFF (a measure of BAT activity) in young boys with overweight/obesity is lower than in age- and sex-matched control subjects with normal BMI. Further, we have demonstrated the feasibility of delivering standardized whole-body cold exposure to young children in a manner that is compatible with the evaluation of the characteristics of BAT with MRI. This early work will enable further examination of these relationships in studies in girls and children of other ages such that we may better understand the factors impacting BAT activity in the early years of life, the trajectory of BAT activity over time, and the relationship of BAT activity with obesity-related health indicators. Together, these data will inform the hypothesis that BAT may be a promising therapeutic target for obesity-related illnesses.

Acknowledgments. The authors thank the participants and their families. The authors also thank the MRI technologists at the Imaging Research Centre at St. Joseph’s Healthcare Hamilton. The authors thank Prasiddha Parthasarathy and Emily Hutchings (McMaster University) for their contributions toward data collection and management. The authors are also thankful for the computer server support provided by the laboratory of Dr. Andrew McArthur (McMaster University) and the McMaster Service Lab and Repository.

Funding. This study is funded by grants from the Canadian Institutes of Health Research (grant 144625-1) and McMaster University (Faculty of Health Sciences, the Boris Family Award). B.A.A. holds the Lau Family Scholarship for Science and Engineering and was funded by the Ontario Graduate Scholarship. N.V. holds the Canadian Institutes of Health Research Graduate Scholarship and was funded by the Ontario Graduate Scholarship. A.C.C. holds the Canada Research Chair in Molecular Imaging of Diabetes and research funding from the Canadian Institutes of Health Research, Fonds de recherche Québec – Santé. M.G.S. is funded by the Canadian Institutes of Health Research, Genome Canada, and the W. Garfield Weston Foundation and holds a Tier 1 Canada Research Chair in Interdisciplinary Microbiome Research. Z.P. has received research funds from the Canadian Institutes of Health Research. G.R.S. receives funding from the Canadian Institutes of Health Research (201709FDN-CEBA-116200), Diabetes Canada Investigator Award (DI-5-17-5302-GS), a Tier 1 Canada Research Chair, and the J. Bruce Duncan Endowed Chair in Metabolic Diseases. K.M.M. holds research funding from the Canadian Institutes of Health Research, Heart and Stroke Foundation of Canada, McMaster Children’s Hospital Foundation, and McMaster University.

Duality of Interest. D.P.B. holds the GSK Chair in Diabetes of Université de Sherbrooke that has been created in part through a donation of $1 million by GSK to Université de Sherbrooke. D.P.B. has received honoraria/consulting fees from Boehringer Ingelheim. A.C.C. has participated in advisory boards for the companies Amgen, uniQure, Merck, Janssen, Novo Nordisk, Novartis, HLS Therapeutics Inc., TVM Capital Life Science, AstraZeneca, and Eli Lilly and made one conference sponsored by AstraZeneca. Z.P. has received honoraria for advice and speaking from Abbott, AstraZeneca/Bristol-Myers Squibb, Boehringer Ingelheim/Eli Lilly, Janssen, Merck, Novo Nordisk, Pfizer, and Sanofi. He has received research funds from Amgen, AstraZeneca/Bristol-Myers Squibb, Lexicon, Merck, Novo Nordisk, and Sanofi. G.R.S. is a co-founder and shareholder of Espervita Therapeutics, a company developing new medications for liver cancer; receives research funding from Espervita Therapeutics, Esperion Therapeutics, Poxel, and Novo Nordisk; and receives honoraria/consulting fees from AstraZeneca, Eli Lilly, Esperion Therapeutics, Poxel, and Merck. K.M.M. has received research funds from AstraZeneca and is an advisory board member for Novo Nordisk and Akcea Therapeutics Canada. No other potential conflicts of interest relevant to this article were reported.

Author Contributions. B.A.A., F.J.O., G.R.S., and K.M.M. conceptualized the study. B.A.A., N.V., F.J.O., and E.G. conducted the experiments. B.A.A., F.J.O., D.P.B., N.B.K., N.P.S., M.D.N., F.H., A.C.C., Z.P., G.R.S., and K.M.M. contributed to the study methodology. B.A.A., N.V., D.P.B., and E.G. ran the formal analysis. N.B.K., M.D.N., F.H., A.C.C., G.R.S., and K.M.M. provided resources. B.A.A., G.R.S., and K.M.M. wrote the manuscript. All of the authors edited the final version of the manuscript. B.A.A. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.