Clinical measurements.

At all visit types, anthropometrics and vital signs are measured by standard methods. Height is measured on a wall-mounted stadiometer (screening only). The average of two measures is calculated (if > 0.5 cm discrepancy exists, a 3^rd measure is obtained). Heart rate, blood pressure and temperature are measured around the same time of day (morning) and after at least a 5-minute rest period by standard electronic methods. Weight measures are done in duplicate on a standing electronic scale and are to be within 0.1 kg (or a third measure is taken). Waist circumference is measured in duplicate after the participant removes their shoes. Briefly, the abdomen is exposed, and the participant stands with their feet hip-width apart and weight evenly distributed; after crossing their arms in front of them, study-staff locate their right upper hip bone and upper iliac crest and note the midaxillary line. A spring-loaded tape measure is placed around the participant’s abdomen at the level of the determined iliac crest. Two measurements are taken during exhalation. If they are > 0.5 cm different, a third measure is completed. Neck circumference is completed in duplicate within 0.5 cm, at the trunk of the neck. Averages of duplicate measures are calculated. Staff cross-train on measurement procedures to ensure similar approaches between staff and study visit sites. Body composition is measured by DXA (iDXA; GE Healthcare, Madison, WI) and abdominal visceral adipose tissue mass (VAT) determined with Encore software (version 14.10). A fasting blood draw is completed at all visits. Additional serial samples are obtained by indwelling IV catheter during Day 1 visits just prior to a standardized meal, then 30- and 60-min after the meal. Plasma and serum are aliquoted after standard processing and stored at −80°C for planned measures.

Magnetic resonance imaging and the standardized meal.

Participants undergo two MRI sessions as part of the Day 1 visit protocol. The pre-meal/fasting session includes quantitative structural imaging and fMRI while viewing visual food cues. The Fred Hutchinson Human Nutrition Laboratory prepares a standardized meal of macaroni and cheese (~40% carbohydrate, ~ 40% fat, ~ 15% protein) for the participant to provide ~33% of their estimated daily caloric need calculated by Mifflin and a standard activity factor of 1.2. Participants are asked to consume the meal within 10 minutes. The second MRI session includes fMRI of visual food cues and resting state, and occurs approximately 30 minutes after meal initiation.

MRI acquisitions. All scans are acquired with a 32-channel SENSE head coil on a 3-Telsa Philips Ingenia CX with dual Quasar gradients. In both sessions, a 306 volume T2*-weighted single-shot echo-planar imaging (EPI) time series is acquired during passive picture viewing. At the post-meal session, a 372 volume T2*-weighted single-shot EPI time series is acquired during rest (viewing of a crosshair). Both EPI acquisitions include: 51 ascending axial slices, 2.50 mm isotropic voxels, repetition time (TR)= 1200 ms, echo time (TE) = 30ms, flip angle = 70°, multi-band SENSE factor = 3. For distortion correction and registration a B0 field map (TR = 6.2ms; TE = 3.1ms; delta TE 1.0 ms; flip angle = 10°) and a compressed-sense 3D Magnetization-Prepared Rapid Gradient-Echo image with 176 sagittal slices (TR = 7.6, TE = 3.5 ms, flip angle = 7°, CS-SENSE factor = 3, matrix 240 x 240, 1 mm isotropic voxel) are acquired. A quantitative multi-slice/multi-echo T2-weighted sequence with 16 echoes (interecho spacing 10ms, slice thickness 2.0 mm-no gap) is acquired fasting. Coronal slices are acquired through the hypothalamus from the optic chiasm to the mammillary bodies.

fMRI paradigm. Details of our established fMRI protocol, procedures, stimuli, acquisition parameters, and processing are published elsewhere [36–38], final fMRI preprocessing protocols will be made publicly available online. Briefly, previously-validated stock photographs included high-calorie foods (e.g., pizza, desserts), low-calorie foods (e.g., fruits, vegetables) and non-food objects (e.g., household items, sports equipment). All food images are ready-to-eat, and non-food objects are easily recognizable. A subset of images are scrambled to be unrecognizable but retain similar color qualities. Each fMRI session includes a distinct set of 15 blocks with 10 images each presented for 2.4 seconds. Non-food blocks (N = 6) and scrambled image blocks (N = 3) are interleaved with high-calorie (N = 3) and low-calorie (N = 3) food blocks. The order is counterbalanced across participants (half viewed low-calorie images as the 1^st food block). All images are displayed on a dark gray background, were equal in size (600x400 dpi) and were matched for luminosity (F=(3,56)=0.26, P = 0.85) and food sub-category (e.g., comparable number of fruit and vegetable images) across blocks.

Participants are instructed to remain awake throughout the MRI sessions. Short videos are played during structural acquisitions. Prior to the fMRI tasks, participants are instructed to “think about how the food would taste if you ate it or how the object would feel to touch”. For the fMRI resting state, a white cross-hair is displayed on a black background and participants are asked to “focus on the cross, relax and let your mind wonder”. To assess attention during the visual food cue task, after each session participants are asked to respond yes or no if they recalled seeing a sub-set of 32 images in the scanner. Half of the images presented are distractor images and not displayed during the task.

fMRI analytic approach. As noted above, participants undergo fMRI before and after a standardized meal during which they view images of high-calorie food, low-calorie food, non-food objects and scrambled images. As a measure of central satiety response, planned initial analyses will measure activation within a set of a priori regions of interest (ROIs) established as markers of satiety [38,39] or predictors of food choice [38] (amygdala, insula, dorsal striatum, ventral striatum, medial orbitofrontal cortex, ventral tegmental area/substantia nigra) and shown by our group to be altered after behavioral obesity treatment in children [10]. A second set of inhibitory control ROIs will include the dorsolateral prefrontal cortex, inferior frontal gyrus, and posterior parietal lobe. These regions play important roles in exertion of dietary control [40] and value modulation [41], as well as response inhibition [42], and participate in functional executive-control networks [43,44] that are engaged to inhibit actions in accordance with immediate or long-term goals [40,42]. An additional ROI in the ventromedial prefrontal cortex will be included as it appears to be the downstream target of inhibitory control regions and may mediate computation of reward value during self-regulation [45,46]. ROIs will be created using anatomic-functional criteria [47]. Individual mean parameter estimates will be extracted from each ROI and then averaged. The primary contrast will be high-calorie food vs. object images. Comparison contrasts will be low-calorie food vs. objects and high-calorie vs. low-calorie food images. The use of ROIs provides an unbiased approach as well as transparency (they will be pre-determined and published) and reproducibility, critical elements to ensure rigor in fMRI studies [48]. Voxel-wise analyses of regions outside a priori ROIs will be considered exploratory.

Structural MRI of the hypothalamus analytic approach. We have validated T2 relaxation time as a marker of mediobasal hypothalamic gliosis in rodents [49] and humans [50]. Multi-slice/multi-echo T2-weighted sequences [50,51] are obtained at baseline and plateau. The coronal slice immediately posterior to the optic chiasm will be identified, regions of interest including the mediobasal hypothalamus and two reference regions (amygdala and putamen) will be placed bilaterally on high-resolution images, then transferred to the T2 parametric map and mean T2 relaxation time will be extracted for each region of interest.

Questionnaires and Behavioral Assessments.

Participants complete questionnaires prior to and at each in-person visit, and remotely at weeks 8, 16 and 24 of the behavioral weight loss program, then 9-, 12-, 15- and 18-months following the start of the program. Questionnaires assess medical history, sleep, physical activity, and eating behaviors. See Table 1 for a complete list of questionnaires, descriptions, and timing of assessments.

Download:

• PNG
  larger image
• TIFF
  original image

Table 1. Questionnaires.

https://doi.org/10.1371/journal.pone.0354678.t001

Behavioral Tasks. Participants complete the General Cognitive Index and Go/No-Go behavioral tasks in-person at baseline and remotely at 18-months. If applicable, both are also completed in-person at active weight loss, plateau and supplemental plateau visits. The General Cognitive Index (testmybrain.org, NIH-General Cognitive Index – Matrix Reasoning and Digit Symbol Matching) is completed on an online platform and measures general cognition. A Go/No-Go (testmybrain.org; TMB Gradual Onset Continuous Performance Test) task is completed to measure response inhibition.

Activity monitoring and measurement.

Participants are provided with a FitBit Inspire 3 device (Fitbit Inc., San Francisco, CA) that is linked to the study’s Fitabase (Fitabase, San Diego, CA) platform for monitoring and data collection. Participants are asked to wear the device continuously and sync it to their app daily throughout their study participation, beginning at the screening visit. The study captures data continuously from baseline through the end of the behavioral weight loss program, then in month-long units at each follow-up timepoint (9-, 12-, 15-, 18-months). Outcome measures include frequency of activity (min/d), activity intensity (moderate, vigorous, peak), total steps/day, average resting heart rate, average heart rate during activity, and heart rate variability.

Indirect calorimetry.

Indirect calorimetry using a metabolic cart (MGC diagnostics) is performed on Day 2 of the baseline and plateau visits, and at the active weight loss visit. Participants are instructed to remain still with minimal movement and without sleeping and a fitted mask or canopy is placed securely over the nose and mouth to ensure that no gas escapes during the test. The participant rests for up to 30 minutes in a supine position to acclimate to the apparatus, then an approximate 20-minute test is performed to achieve steady state. Data from the indirect calorimetry are used to calculate respiratory quotient (RQ) and resting energy expenditure (REE).

Biopsies.

Adipose and muscle biopsies are conducted by qualified personnel on day two of the Baseline and Plateau visits and at the active weight loss visit. Following injection of local anesthetic, adipose biopsies are collected using negative pressure from an incision site located in the lower quadrant of the abdominal area, 10–12 cm from the umbilicus. To provide superficial anesthetic, up to 10 mL of 1% lidocaine is injected into the dermis of the target area. Once the anesthetic has taken effect and the site is sufficiently numb, a < 0.5 cm scalpel incision is made through the adipose tissue, and pressure is applied with gauze to stop bleeding. A biopsy needle is passed through the incision into the subcutaneous fat, and negative pressure is applied to obtain the sample. Pressure is applied to the incision with gauze while the sample is removed from the biopsy needle and placed on ice. The wound is then closed with steri-strips while the adipose biopsy sample is weighed and portioned into two cryovials, both of which are immediately flash frozen using liquid nitrogen then stored at −80°C until analysis.

Also under local anesthetic, muscle biopsies are collected via an incision over the vastus lateralis muscle, 8–10 cm proximal to the knee, lateral to midline, and medial to the IT band. To provide superficial anesthetic, up to 5 mL of 1% lidocaine is injected into the dermis of the target area. After the initial anesthetic injection, 1% lidocaine is again injected and advanced into the quadricep muscle, distributing the anesthetic into the muscle belly. Once the anesthetic has taken effect and the site is sufficiently numb, a number 11 blade scalpel is used to make an incision through the skin, subcutaneous tissue, fascia, and into the vastus lateralis muscle. A biopsy needle is inserted into the incision, through the fascial layer and is advanced an additional 2–3 cm past to ensure the biopsy needle is positioned completely within the muscle belly. Negative pressure is applied to obtain the sample, then pressure is applied to the incision with gauze while the sample is removed from the biopsy needle and placed on ice. The wound is then closed with steri-strips while the muscle biopsy sample is weighed and portioned into four cryovials; one cryovial is immediately flash frozen using liquid nitrogen and the remaining samples are transported on ice for processing and analysis.

T-cell isolation and high-resolution respirometry.

Because peripheral blood mononuclear cells (PBMCs) represent a heterogeneous population of cells, T-cells are selected from freshly isolated PBMCs as previously described [61]. Three million cells are resuspended in supplemented DMEM (5 mM HEPES, 5 mM glucose, 1 mM sodium pyruvate, and 2 mM glutamate), and four million cells are resuspended in Mir05 (0.5 mM EGTA, 3 mM MgCl₂, 60 mM lactobionic acid, 20 mM taurine, 10 KH₂PO₄, 20 mM HEPES, 110 mM D-sucrose). High resolution respirometry is performed using an Oroboros O2K respirometer (Oroboros Instruments, Austria) in intact cells (Protocol 1) and permeabilized cells (Protocol 2). For Protocol 1, T-cells are injected into chambers containing supplemented DMEM. Once basal respiration becames stable, leak respiration is induced with 1mM oligomycin, maximal uncoupled respiration is stimulated with consecutive titrations of 0.5 mM 2-(trifluoromethoxy) phenyl hydrazinylidene-propanedinitrile (FCCP), and complex III is inhibited with 2.5 μM antimycin A to induce non-mitochondrial respiration. For Protocol 2, T-cells are injected into chambers containing Mir05 supplemented with 1 μM of TMRM to measure membrane potential as previously described [61]. Cells are permeabilized with 4 μg/mL of digitonin. Leak respiration and maximal mitochondrial membrane potential is stimulated with 5 mM succinate, then 10 mM glutamate, 1 mM malate, and 5 mM pyruvate. Adenosine di-phosphate (ADP, 1 mM – 4 mM) is then titrated to induce oxidative phosphorylation. Maximal uncoupled respiration is then stimulated by sequential 0.5 mM FCCP injections. Complex I is inhibited with 0.5 mM rotenone, and Complex III with 2.5 mM antimycin A. Non-mitochondrial respiration under antimycin A is subtracted from all respiration rates for statistical analysis.

Mitochondrial respiration in white adipose tissue (WAT) samples.

Approximately 100 mg of frozen WAT is used for the mitochondrial respiration assays. Samples are thawed on ice in mitochondrial assay solution (MAS), and the tissue is minced into smaller pieces with scissors. The fragments are collected and subsequently homogenized in ice-cold MAS. Mitochondria-enriched homogenates are obtained by differential centrifugation steps to remove the fat layer and other debris. Protein concentration is determined using the Pierce Rapid Gold BCA Protein Assay (Thermo Fisher Scientific).

Mitochondrial respiration in frozen samples is assessed using the Agilent Seahorse XFe24 Analyzer (Agilent Technologies, Santa Clara, CA) as previously described with minor modifications [62,63]. Homogenates containing mitochondria, and supplemented with cytochrome c, are loaded into Agilent Seahorse XF24 Cell Culture Microplates. For assessments of Complex I, II, and IV activities, substrates and inhibitors are injected subsequently as follows: NADH or Succinate + Rotenone (port A), rotenone + antimycin A (port B), TMPD + ascorbic acid (port C), and sodium azide (port D). Oxygen consumption rates are normalized to mitochondrial content.

Aim 1.

Primary outcomes are oxidative phosphorylation and maximal respiratory capacity in T-cells, and permeabilized muscle fibers, the latter with fatty acid oxidation and complex I substrate. Secondary outcomes include T-cell and permeabilized muscle fibers uncoupling (proton leak respiration), adipose tissue mitochondrial respiratory capacity and coupling, UCP-1 expression, and inflammatory and lipolytic markers. As a prelude to inferential testing, linear regression will assess change in each measure as a function of potential confounding baseline covariates including % body fat, age, and sex as well as resting energy expenditure. Linear mixed models (adjusted for baseline confounders if needed) will be used to test for a main effect of time (baseline, active weight loss, plateau), followed by Bonferroni-adjusted formal post hoc testing to assess differences between phases and evaluate if a weight loss plateau represents a distinct metabolic state. We will then explore evidence of cross talk among T-cells, muscle, adipose tissue, and brain measures of appetite (See Aim 2 below). For measures that are changed from baseline by weight loss, we will perform linear regression amongst outcomes of the change from baseline to active weight loss or weight loss plateau. Implications for weight maintenance will be explored via linear regression of % weight change from plateau to 18 mo. as a function of the change in primary and secondary metabolic measures at plateau phase from baseline, adjusting for time between plateau and end of follow-up. Secondary analyses will use values measured at plateau (not change). Stratified analyses will assess sex as a biological variable. A sample size of N = 40 at active weight loss and plateau provides 99% power to detect within-subject differences between baseline and plateau and end of follow up for primary outcomes.

Aim 2.

The primary analyses will evaluate differences between baseline and plateau phase in brain activation to visual food cues, specifically: 1) meal-induced change in mean extracted parameter estimates across all a priori satiety regions (high-calorie food vs. objects), 2) changes in pre-meal activation (high-calorie food vs. objects) from each of the four inhibitory control and reward modulation regions, and 3) associations of any findings with changes in Aim 1 metabolic outcomes. Outcomes for secondary analyses include mediobasal hypothalamic T2 relaxation time (vs. control regions) and meal-induced changes in salience network connectivity. Associations with weight regain post-plateau will be explored as will within-subject differences in regions outside a priori ROIs via fully adjusted voxel-wise comparisons. For within-subject comparisons, we will use linear mixed models that can accommodate repeated measures as well as missing data. Relationships between neurophysiologic measures and behavioral and hormonal correlates will be assessed using linear regression of change in behavioral correlates from baseline to plateau, as a function of each measure separately and together as covariates (i.e., changes in satiety response and inhibitory control activation), adjusting for time between plateau and baseline and, as needed, baseline covariates that are potential confounders such as sex, age, and % body fat. A sample size of N = 43 at plateau maintains 80% power to detect within subject differences in brain activation to visual food cues.

Aim 3.

The primary outcomes of interest being explored in Aim 3 are entry into a weight loss plateau (Yes/No) and time from baseline (beginning of behavioral weight loss program) to plateau entry (continuous). Entry into a weight loss plateau will be defined as the 1) day at which a subject met weight loss plateau criteria during real-time monitoring or 2) first day of a weight loss plateau identified via threshold regression modeling at study completion. A priori predictors of interest for Aim 3 are based on available literature and our own findings and include the slope of the first active weight loss period [64]; trends over time in craving for carbohydrates/starches, dietary restraint, and disinhibition [65]; changes in T-cell and muscle maximal respiratory capacity; fasting ghrelin [66]; and baseline central satiety response, mediobasal hypothalamic T2 relaxation time [67], leptin [28] concentrations, and HOMA-IR [68]. Other predictors (i.e., inhibitory control) may be tested based on the findings of Aims 1 and 2; remaining predictors will be considered exploratory. Analyses will include only participants who achieved active weight loss. Fully adjusted models will include baseline or change in % body fat. Sensitivity analyses will be completed to evaluate whether excluding participants not achieving adherence will have the same or better predictive capabilities. A sample size of N = 40 participants who achieve active weight loss provides 91% power to detect effect sizes of R² ≥ 0.36 at P < 0.05 for continuous outcomes.