Abstract
Aims:
To determine whether loss of muscle mass (approximated using fat free mass [FFM]) is associated with risk for type 2 diabetes mellitus (T2DM) in Hispanic/Latino adults in the United States.
Methods:
Participants were Hispanic/Latino adults (18–74-year-olds) who completed Visit 2 of the Hispanic Community Health Study/Study of Latinos (HCHS/SOL; multi-site, prospective cohort study; 6.1-year follow-up) and did not have T2DM at baseline (n=6264). At baseline and Visit 2, FFM was measured using bioelectrical impedance analysis and fasting glucose, HbA1c, and fasting insulin were measured by examiners. Diabetes was defined according to American Diabetes Association criteria. Survey-weighted Poisson regression models examined the association of percent change in relative FFM (%ΔFFM) with incident prediabetes and T2DM. Survey-weighted multivariable regression models examined associations of %ΔFFM with changes in glucose and insulin measures.
Results:
Relative FFM declined by 2.1% between visits. %ΔFFM was inversely associated with incident prediabetes (p-for-trend=0.001) and with changes in glucose and insulin measures (p-for-trend<0.0001). Findings were null, except for HOMA-IR, after adjustment for changes in adiposity measures. Associations were generally stronger for individuals with baseline overweight/obesity.
Conclusions:
Reducing loss of FFM during adulthood may reduce prediabetes risk (primarily insulin resistance), particularly among individuals with overweight/obesity.
Keywords: type 2 diabetes, adiposity, prospective studies, sarcopenia, Hispanic
INTRODUCTION
Prediabetes and type 2 diabetes mellitus (T2DM) affect 9.3% and 33.9% of adults in the United States (US), respectively (1), and minority race/ethnic groups face an even higher burden of these diseases. Specifically, the age-adjusted prevalence of diagnosed or undiagnosed diabetes for Hispanic/Latino adults is 76% greater than that for non-Hispanic Whites (16.4% vs. 9.3%) (1). While the factors that place Hispanic/Latino adults at an elevated risk for T2DM are not fully elucidated (2), body composition may play a role.
Loss of skeletal muscle mass with age (a condition termed sarcopenia when it occurs in combination with loss of muscle function (3)) is greater in Mexican American versus non-Hispanic Whites (4,5) and may promote impaired glucose tolerance. Skeletal muscle promotes insulin sensitivity through the proteins (myokines) it secretes (6) and is a major site for glucose consumption and storage (7). For example, in a euglycemic hyperinsulinemia state, skeletal muscle accounts for approximately 80% of postprandial glucose uptake (7,8) and non-oxidative glucose storage (i.e., glycogen synthesis) (7,9). Sarcopenia has been associated with prediabetes and T2DM in cross-sectional studies (10–12); however, few prospective studies have been conducted (13–17), and there is a call for additional longitudinal work (10,18).
Existing prospective studies have primarily focused on how baseline skeletal muscle mass is associated with T2DM at a later point in time (13,16,17), rather than looking at how changes in skeletal muscle mass are associated with T2DM risk. Additionally, these studies have primarily focused on non-Hispanic White and Asian populations. To our knowledge, only one study has examined how changes in skeletal muscle mass are associated with incident T2DM, and the authors reported an inverse association between an increase in muscle mass and incident T2DM in non-Hispanic White adults (15). Additionally, one other study examined how changes in skeletal muscle mass are associated with changes in risk factors for T2DM (e.g., fasting glucose, insulin resistance) (14), but the study did not adjust for potential confounders, examined changes over a one-year period only, and was limited to Korean adults living in Korea (14). Previous research has also indicated that age, sex, and obesity status influence the association between skeletal muscle mass and T2DM, but there is a need to clarify the directionality (11–13,17).
The objective of this study was thus to determine how loss of skeletal muscle mass (approximated using fat free mass [FFM]) is associated with risk for prediabetes and T2DM in a large sample of Hispanic/Latino adults living in the US. FFM, rather than skeletal muscle mass, was used in this analysis due to the Hispanic Community Health Study/Study of Latinos (HCHS/SOL) (19,20) assessing body composition with wrist to ankle bioelectrical impedance analysis (BIA) – a non-invasive measure of body composition well-suited for large epidemiologic studies (21,22). While FFM can only approximate skeletal muscle mass (~50% of FFM is skeletal muscle mass), it is a clinically relevant proxy for skeletal muscle mass (23). We hypothesized that FFM loss compared to gain over a 6-year follow-up period would be associated with a higher incidence of prediabetes and T2DM and greater fasting glucose, HbA1c, and HOMA-IR. We further hypothesized stronger associations among older adults, males, and individuals with obesity.
METHODS
Subjects.
HCHS/SOL is a prospective, community-based cohort study of self-identified Hispanic/Latino adults (aged 18–74 years) who were selected using a stratified, two-stage probability sampling design across four US communities (Bronx, New York; Chicago, Illinois; Miami, Florida; San Diego, California), supported by a Coordinating Center at the University of North Carolina at Chapel Hill (19,20). Persons on active duty military service, not currently living at home, planning to move from the area in the next six months, physically unable to attend the clinic examination, or unable to complete the study questionnaires in English or Spanish were considered ineligible (19,20).
Between March 2008 and June 2011, the HCHS/SOL study examined 16,415 Hispanic/Latino adults (baseline) (24). A follow-up clinic examination (Visit 2) was conducted between October 2014 and December 2017 in 11,623 Hispanic/Latino adults. Between these two visits, 30-minute annual interviews were conducted via telephone (19). An overview of the data collection timeline is provided in Figure 1. All participants provided informed consent, and the study had institutional review board approval from each institution participating in the study and the Coordinating Center.
Fig. 1.
Timeline of data collection in HCHS/SOL for T2DM and anthropometrics. BIA Bioelectrical Impedance Analysis.
Materials and Methods.
Baseline and Visit 2 clinical examinations included anthropometric measurements, a fasting blood sample, an oral glucose tolerance test (OGTT), and interviewer-administered questionnaires in the participant’s language of preference (19). Protocols described below were used for both examinations. At the approval of the PI, participants were offered abbreviated clinic or in-home examinations for Visit 2 if they were unable to make it to the clinic (e.g., due to being frail, disabled, or cognitively impaired or providing dependent care for another person). Modifications to protocols for these visits (n=338) are noted below.
Anthropometrics.
Height was measured to the nearest 1.0 cm using a wall-mounted stadiometer, and weight was measured to the nearest 0.1 kg with participants in light clothing using a digital scale (Tanita Body Composition Analyzer, TBF-300A; Tanita Corporation, Tokyo, Japan). At in-home Visit 2 exams, weight was the only anthropometric collected. Body mass index (BMI) was calculated as weight (kg) divided by height (m2). Individuals were classified as underweight (<18.5 kg/m2), normal weight (18.5 to <25 kg/m2), overweight (25 kg/m2 to <30 kg/m2), or obese (≥30 kg/m2) (25). Because few individuals were classified as underweight, the underweight and normal weight groups were combined. Waist circumference (WC) was measured to the nearest 0.1 cm at the uppermost lateral border of the right ilium using a measuring tape.
Fat mass and fat free mass (FFM) were measured in kilograms at the same time as total body weight using the BIA function of the Tanita scale in all participants except for those with a pacemaker, defibrillator, other implanted electronic device, or prosthetic limb. FFM was obtained directly from the BIA machine, and the same BIA machine was used for Visit 1 and Visit 2. The brand, model, and built-in equation to derive FFM did not change across study visits.
A measure of FFM relative to overall body weight (henceforth referred to as “relative FFM”) was computed by dividing FFM by total body mass. Percent change in relative FFM (henceforth denoted as “%ΔFFM”) was calculated as ([relative FFMbaseline- relative FFMVisit 2]/relative FFMbaseline)*100. Sarcopenia was defined as having a FFM ≥2 standard deviations below the sex-specific mean FFM of healthy, 20–29-year-old Mexican Americans assessed in the Third National Health and Nutrition Examination Survey (NHANES III) (26,27). Muscle function was not assessed in HCHS/SOL and thus was not used to define sacropenia. Fat mass index (FMI) was calculated as fat mass/height-squared (kg/m2).
Laboratory measures.
Blood specimens were drawn on all individuals in the morning under fasting conditions (at least 8 hours) and following a 2-hour OGTT (19). Individuals were not eligible for the OGTT if they reported having/being treated for diabetes, being on kidney dialysis, having partial removal of the stomach or small intestine, having fasted for less than 8 hours, or if their fasting glucose was greater than 150 mg/dL. All specimens were processed on site and stored at −70°C. For in-home Visit 2 exams, OGTT was not conducted, and blood specimens were only collected if the biospecimen could arrive at the field center laboratory within 120 minutes from the blood draw. Fasting glucose was not reported from in-home Visit 2 exams due to the time between blood draw and analysis. The University of Minnesota’s Advanced Research and Diagnostic Laboratory performed all laboratory assays.
Glucose was measured in EDTA plasma on a Roche Modular P Chemistry Analyzer (Roche Diagnostics Corporation) using a hexokinase enzymatic method (Roche Diagnostics, Indianapolis, IN 46250). Glycosylated hemoglobin (HbA1c) was measured in EDTA whole blood using a Tosoh G7 Automated HPLC Analyzer (Tosoh Bioscience, Inc, South San Francisco, CA 94080). Insulin was measured in serum using a commercial ELISA assay (Mercodia AB, Uppsala, Sweden) through October 28, 2009 and in serum on a Roche Elecsys 2010 Analyzer (Roche Diagnostics Corporation) using a sandwich immunoassay method (Roche Diagnostics, Indianapolis, IN 46250) from October 29, 2009 until the study end date. The homeostatic assessment of insulin resistance (HOMA-IR) was calculated as (fasting insulin [μU/mL]*fasting glucose [mg/dl])/405 (28).
Prediabetes and T2DM.
Prediabetes and T2DM were defined in accordance with the American Diabetes Association using measured lab values (29). Prediabetes was defined as having a fasting glucose value in the range of 100–125 mg/dL, a post-OGTT glucose value of 140–199 mg/dL, or an HbA1c in the range of 5.7 to 6.5% (39 to 48 mmol/mol). T2DM was defined as having a fasting glucose value ≥126 mg/dL, a post-OGTT glucose value ≥200 mg/dL, an HbA1c ≥6.5% (≥48 mmol/mol), or use of diabetic medication. Medication use was based on participants self-reporting whether any medications used in the past four weeks were for high blood sugar or diabetes. Prediabetes and T2DM status were set to missing only if all diabetes-related laboratory values and diabetic medication use data were missing. Otherwise, available lab and medication use information was used to determine diagnosis.
Physical activity.
Physical activity was measured at baseline using an Actical accelerometer (version B-1, model 198–0200-03; Philips Respironics; Bend, Oregon), which has acceptable technical reliability for counts (30,31). Participants were instructed to undertake usual activities for 7 days while wearing the monitor above the iliac crest, and to remove it only for swimming, showering, and sleeping. The Actical was programmed to record counts per one-minute epochs. Adherence to the physical activity protocol was defined as a minimum of 3 days with ≥10 hours/day of wear-time, with approximately three-quarters of the sample meeting this criterion (32). For the present analyses, average counts per minute per day was used to summarize physical activity.
Self-reported covariates.
Age, sex, Hispanic/Latino background, nativity, educational attainment, and annual household income were self-reported at baseline.
Statistical analysis.
The base analytic sample size for all analyses was 6264 adults, as described in Figure 2. Analyses adjusted for FMI, as described below, excluded an additional 35 individuals from the base sample for having percent changes in FMI that were greater than 3 standard deviations from the mean. Similarly, for analyses adjusted for BMI and waist circumference, 103 individuals were excluded from the base sample for having percent changes in BMI and waist circumference that were greater than 3 standard deviations from the mean.
Figure 2.
Flow diagram of HCHS/SOL participant exclusion criteria for each prospective analysis of percent change in relative FFM and incident prediabetes, incident T2DM, change in glucose measures, and change in insulin measures.
T2DM type 2 diabetes mellitus, V1 Visit 1 (baseline), V2 Visit 2, OGTT oral glucose tolerance test, FFM fat free mass, SD standard deviation, FM fat mass
The main exposure in all models was %ΔFFM. Four sex-specific %ΔFFM categories were derived for each analytic sample separately (Supplementary Table 1). Specifically, individuals whose %ΔFFM was ≥0 were split into two groups based on the sex-specific median percent increase in relative FFM. Individuals whose %ΔFFM was <0 were also split into two groups according to the sex-specific median percent decrease in relative FFM. In order to estimate the p-for-trend across these four categories, individuals were assigned the median scores within each of these four groups, and the resulting variable was treated as a continuous measure.
To determine the incidence rate ratios of prediabetes and T2DM according to %ΔFFM, Poisson regression models were used with the outcome as prediabetes or T2DM at Visit 2 (yes/no). Given that individuals could have been diagnosed with T2DM at any point between baseline and Visit 2, the same Poisson regression models were also run (as a sensitivity analysis) after restricting the sample to individuals who were at risk for a first-time T2DM diagnosis at Visit 2 (n=5360). Participants who self-reported a diagnosis of T2DM during any of the annual follow-up calls or at baseline or Visit 2 were excluded in these sensitivity analyses. To examine associations between %ΔFFM and changes in fasting glucose, HbA1c, and HOMA-IR, separate multivariable linear regression models were used. All aforementioned models were re-run after excluding women who reported being pregnant at any time during the HCHS/SOL study (n=395), given that pregnancy and postpartum weight retention influence current and future body composition (33).
All models were adjusted for age group, sex, Hispanic/Latino background, study center, nativity, education, income, and accelerometer average counts per minute per day. Years between baseline and Visit 2 (range: 3.4 to 9.6 years) was included as an offset variable in Poisson regression models and as a covariate in multivariable regression models. Diabetic medication use at Visit 2 was included as a covariate in the models examining fasting glucose, HbA1c, and HOMA-IR as the outcomes. These models were not adjusted for baseline lab values to avoid introducing bias (34).
All models were re-run after adjustment for measures of adiposity with the aim of examining associations between %ΔFFM and each outcome independent of overall and central adiposity. Specifically, each model was adjusted for either 1) baseline FMI, 2) percent change in FMI, 3) baseline BMI and waist circumference, or 4) percent change in BMI and waist circumference. Interactions between %ΔFFM and age group, sex, and baseline BMI category were additionally examined in separate models that were not adjusted for FMI or BMI and waist circumference.
All analyses accounted for stratification and for clustering by primary sampling units and were weighted to adjust for sampling probability of selection and nonresponse with the use of complex survey procedures. Due to the large amount of missing accelerometer data, analyses accounted for missing data using inverse probability weighting (IPW) (35,36). All analyses were conducted in SAS software version 9.4 (SAS Institute), and Poisson and multivariable regression models were run using SUDAAN version 11.0.
RESULTS
An overview of the socio-demographic and body composition measures of the HCHS/SOL target population by %ΔFFM groups is provided for the overall sample in Table 1 and stratified by sex in Supplementary Tables 2 and 3. Groups 1–4 are labeled in order of increasing percent change in relative FFM. Group 1 experienced the greatest mean percent decrease in relative FFM over time (−8.6%), and Group 4 experienced the greatest mean percent increase in relative FFM over time (7.5%). Individuals most frequently reported being of Mexican descent (39.6% to 42.0%) and living in low socioeconomic status households (56.5% to 60.9% with household income <$30.000). Compared to the population represented by Group 4, the population represented by Group 1 had a higher proportion of 18–44-year-olds (77.7% vs. 60.1%) and females (53.1% vs. 48.6%), a lower BMI at baseline (27.7 kg/m2 vs. 31.0 kg/m2), and a greater percent increase in BMI over follow-up (11.4% vs. −6.7%). Overall, only eight participants (<1%) met the criteria for sarcopenia.
Table 1.
Baseline socio-demographics, baseline physical activity, and body composition measures of the target population of HCHS/SOL by percent change in relative FFM groups (n=6264)
| Group 1 (n=1910) | Group 2 (n=1910) | Group 3 (n=1222) | Group 4 (n=1222) | |
|---|---|---|---|---|
| %ΔFFM (Mean [SE]) | −8.6 (0.2) | −2.0 (0.04) | 1.4 (0.04) | 7.5 (0.2) |
| Socio-demographics (%) | ||||
| Age group, years | ||||
| 18–44 | 77.7 | 65.5 | 59.2 | 60.1 |
| ≥45 | 22.3 | 34.5 | 40.8 | 39.9 |
| Sex | ||||
| Female | 53.1 | 55.1 | 46.8 | 48.6 |
| Male | 46.9 | 44.9 | 53.2 | 51.4 |
| Hispanic/Latino background | ||||
| Central American | 7.9 | 8.9 | 7.3 | 7.8 |
| Cuban | 20.6 | 19.6 | 21.9 | 15.8 |
| Dominican | 9.4 | 9.0 | 8.5 | 7.6 |
| Mexican | 39.6 | 42.0 | 35.0 | 41.9 |
| Mixed/Other | 5.3 | 4.0 | 5.5 | 3.4 |
| Puerto Rican | 13.0 | 10.9 | 15.2 | 18.4 |
| South American | 4.2 | 5.5 | 6.7 | 5.2 |
| Nativity | ||||
| Born in United States | 26.9 | 19.0 | 22.7 | 27.0 |
| Born elsewhere | 73.1 | 81.0 | 77.3 | 73.0 |
| Education | ||||
| No HS diploma or GED | 27.2 | 28.0 | 33.7 | 34.4 |
| HS diploma or GED | 29.8 | 29.3 | 28.3 | 25.5 |
| >HS diploma or GED | 43.0 | 42.7 | 38.0 | 40.2 |
| Household income | ||||
| ≤$15,000 | 27.5 | 26.0 | 26.2 | 32.3 |
| >$15,000 to <$30,000 | 30.6 | 30.5 | 30.9 | 28.6 |
| ≥$30,000 | 31.0 | 32.5 | 32.7 | 32.2 |
| Not reported | 10.9 | 11.0 | 10.1 | 7.0 |
|
Physical activity, (Mean [SE]), average counts/min/day |
194.1 (5.2) | 189.8 (6.2) | 189.2 (6.6) | 177.7 (6.9) |
| Body composition (Mean [SE]) | ||||
| Baseline BMI, kg/m2 | 27.7 (0.2) | 28.7 (0.2) | 28.8 (0.3) | 31.0 (0.2) |
| %ΔBMI | 11.4 (0.3) | 2.7 (0.2) | −1.5 (0.2) | −6.7 (0.3) |
| Baseline FMI, kg/m2 | 8.6 (0.1) | 9.7 (0.2) | 9.6 (0.2) | 11.5 (0.2) |
| %ΔFMI | 41.0 (1.3) | 7.8 (0.3) | −5.2 (0.5) | −19.7 (0.6) |
| Baseline WC, cm | 93.1 (0.5) | 95.4 (0.5) | 95.9 (0.7) | 100.9 (0.7) |
| %ΔWC | 8.2 (0.3) | 2.9 (0.2) | 0.3 (0.3) | −3.8 (0.3) |
HS high school, %ΔFFM percent change in fat free mass/weight from baseline to Visit 2, BMI body mass index, %Δ percent change from baseline to Visit 2, FMI fat mass index, WC waist circumference
All statistics represent frequency (as a percentage), unless otherwise noted, and are weighted with the sampling weights.
Incidence of prediabetes and T2DM
Table 2 shows the incidence rate ratios for prediabetes and T2DM according to %ΔFFM groups. Percent change in relative FFM was associated with incidence of prediabetes (p-for-trend=0.001) but not incidence of T2DM. Compared to individuals with the greatest percent increase in relative FFM, those with the greatest percent decrease in relative FFM had a 1.46 (95% CI: 1.13, 1.89) times higher incidence of prediabetes. Adjustment for FMI or BMI and waist circumference at baseline strengthened the association for incident prediabetes (p-for-trend<0.0001), but all findings were null once adjusted for percent change in these measures.
Table 2.
Incidence rate ratios (95% CI) for prediabetes and type 2 diabetes mellitus at Visit 2 according to percent changes in relative FFM (%ΔFFM) from baseline to Visit 2 of HCHS/SOL
| %ΔFFM | Incident prediabetes | Incident type 2 diabetes mellitus |
|---|---|---|
| No adjustment for FMI, BMI, or WC | (n=3040) | (n=6264) |
| Overall F-test | ** | - |
| Group 1 | 1.46 (1.13, 1.89)** | 0.94 (0.68, 1.30) |
| Group 2 | 1.26 (0.99, 1.61) | 1.11 (0.82, 1.52) |
| Group 3 | 1.10 (0.83, 1.46) | 1.02 (0.75, 1.39) |
| Group 4 | 1.00 | 1.00 |
| p-for-trend | 0.001 | 0.72 |
| Adjusted for FMI baseline | (n=3010) | (n=6229) |
| Overall F-test | *** | - |
| Group 1 | 1.76 (1.36, 2.27)*** | 1.38 (1.00, 1.93) |
| Group 2 | 1.44 (1.13, 1.83)** | 1.47 (1.08, 2.01)* |
| Group 3 | 1.25 (0.94, 1.66) | 1.33 (0.99, 1.78) |
| Group 4 | 1.00 | 1.00 |
| p-for-trend | <0.0001 | 0.06 |
| Adjusted for %Δ FMI | (n=3010) | (n=6229) |
| Overall F-test | - | - |
| Group 1 | 1.07 (0.69, 1.65) | 1.24 (0.53, 2.89) |
| Group 2 | 0.82 (0.55, 1.22) | 1.22 (0.63, 2.34) |
| Group 3 | 0.89 (0.63, 1.24) | 0.86 (0.53, 1.39) |
| Group 4 | 1.00 | 1.00 |
| p-for-trend | 0.19 | 0.59 |
| Adjusted for BMI baseline and WC baseline | (n=2971) | (n=6161) |
| Overall F-test | ** | - |
| Group 1 | 1.64 (1.26, 2.13)** | 1.27 (0.92, 1.77) |
| Group 2 | 1.45 (1.13, 1.86)** | 1.43 (1.05, 1.95)* |
| Group 3 | 1.26 (0.94, 1.68) | 1.30 (0.97, 1.74) |
| Group 4 | 1.00 | 1.00 |
| p-for-trend | <0.0001 | 0.17 |
| Adjusted for %ΔBMI and %ΔWC | (n=2971) | (n=6161) |
| Overall F-test | - | - |
| Group 1 | 0.98 (0.73, 1.33) | 1.13 (0.69, 1.84) |
| Group 2 | 0.95 (0.72, 1.25) | 1.30 (0.83, 2.02) |
| Group 3 | 0.94 (0.71, 1.24) | 1.09 (0.76, 1.55) |
| Group 4 | 1.00 | 1.00 |
| p-for-trend | 0.57 | 0.58 |
p<0.05,
p<0.01,
p<0.0001,
- Not significant at p<0.05
%ΔFFM percent change in fat free mass/weight, FMI fat mass index, %Δ percent change, BMI body mass index, WC waist circumference
Group 1: greatest % decrease in relative fat free mass; Group 4: greatest % increase in relative fat free mass
Models adjusted for age group, sex, Hispanic/Latino background, site, nativity, education, household income, and physical activity.
The association between %ΔFFM and incidence of prediabetes was modified by baseline BMI category (p=0.008; Supplementary Table 4). There was a strong association among individuals who were overweight or obese but a null association among individuals who were normal weight. For example, among individuals with obesity, those who had the greatest percent decrease in relative FFM had a 1.90 (95%: 1.28, 2.84) times higher incidence of prediabetes compared to individuals who had the greatest percent increase in relative FFM.
In sensitivity analyses for incident T2DM—in which individuals who self-reported diabetes during baseline, Visit 2, or any annual follow-up call were excluded—, the overall association between %ΔFFM and incident T2DM was also null (Supplementary Table 5). However, there was effect modification by baseline BMI category (p=0.05) such that the association was significant among individuals with obesity only (p-for-trend=0.003).
Changes in glucose and insulin measures
Table 3 shows the association between %ΔFFM and changes in fasting glucose, HbA1c, and HOMA-IR. Percent change in relative FFM was associated with changes in all measures (p-for-trend<0.0001). Compared to individuals who had the greatest percent increase in relative FFM, individuals who had the greatest percent decrease in relative FFM had mean increases in fasting glucose, HbA1c, and HOMA-IR that were greater by 3.52 mg/dL (95% CI: 2.51, 4.54), 0.14 (95% CI: 0.11, 0.17) percentage points (1.55 [95% CI: 1.21, 1.89] mmol/mol), and 1.92 (95% CI: 1.67, 2.18) units, respectively. Adjustment for FMI or BMI and waist circumference at baseline strengthened all associations. Adjustment for percent change in these measures resulted in null associations for all measures except HOMA-IR (p-for-trend<0.0001), for which associations were only slightly attenuated.
Table 3.
Predicted mean changes (95% CI) in glucose and insulin measures according to percent changes in relative FFM from baseline to Visit 2 of HCHS/SOL
| %ΔFFM | Δfasting glucose | ΔHbA1c, % | ΔHbA1c, mmol/mol | HOMA-IR |
|---|---|---|---|---|
| No adjustment for FMI, BMI, or WC | (n=5980) | (n=5980) | (n=5980) | (n=6018) |
| Overall F-test | *** | *** | *** | *** |
| Group 1 | 3.52 (2.51, 4.54)*** | 0.14 (0.11, 0.17)*** | 1.55 (1.21, 1.89)*** | 1.92 (1.67, 2.18)*** |
| Group 2 | 2.17 (1.19, 3.15)*** | 0.09 (0.06, 0.12)*** | 0.94 (0.62, 1.26)*** | 1.16 (0.92, 1.39)*** |
| Group 3 | 1.88 (0.74, 3.01)** | 0.06 (0.03, 0.09)** | 0.65 (0.30, 0.99)** | 0.68 (0.44, 0.91)*** |
| Group 4 | 0.00 | 0.00 | 0.00 | 0.00 |
| p-for-trend | <0.0001 | <0.0001 | <0.0001 | <0.0001 |
| Adjusted for FMI baseline | (n=5947) | (n=5947) | (n=5947) | (n=5985) |
| Overall F-test | *** | *** | *** | *** |
| Group 1 | 4.19 (3.16, 5.22)*** | 0.17 (0.13, 0.20)*** | 1.80 (1.46, 2.14)*** | 2.05 (1.78, 2.31)*** |
| Group 2 | 2.62 (1.63, 3.60)*** | 0.10 (0.07, 0.13)*** | 1.07 (0.76, 1.37)*** | 1.26 (1.01, 1.51)*** |
| Group 3 | 2.35 (1.18, 3.52)** | 0.08 (0.04, 0.11)*** | 0.80 (0.45, 1.14)*** | 0.78 (0.53, 1.03)*** |
| Group 4 | 0.00 | 0.00 | 0.00 | 0.00 |
| p-for-trend | <0.0001 | <0.0001 | <0.0001 | <0.0001 |
| Adjusted for %Δ FMI | (n=5947) | (n=5947) | (n=5947) | (n=5985) |
| Overall F-test | - | - | - | *** |
| Group 1 | −1.06 (−3.30, 1.18) | 0.00 (−0.07, 0.07) | 0.00 (−0.81, 0.82) | 0.91 (0.41, 1.40)** |
| Group 2 | −2.12 (−4.27, 0.02) | −0.02 (−0.08, 0.04) | −0.26 (−0.93, 0.42) | 0.34 (−0.11, 0.79) |
| Group 3 | −0.56 (−2.16, 1.04) | −0.01 (−0.06, 0.03) | −0.16 (−0.69, 0.36) | 0.21 (−0.16, 0.57) |
| Group 4 | 0.00 | 0.00 | 0.00 | 0.00 |
| p-for-trend | 0.62 | 0.30 | 0.32 | <0.0001 |
| Adjusted for BMI baseline and WCbaseline | (n=5882) | (n=5882) | (n=5882) | (n=5921) |
| Overall F-test | *** | *** | *** | *** |
| Group 1 | 4.02 (3.00, 5.04)*** | 0.15 (0.12, 0.19)*** | 1.68 (1.33, 2.02)*** | 1.97 (1.70, 2.24)*** |
| Group 2 | 2.60 (1.62, 3.57)*** | 0.10 (0.07, 0.12)*** | 1.02 (0.71, 1.33)*** | 1.22 (0.98, 1.47)*** |
| Group 3 | 2.35 (1.20, 3.49)** | 0.07 (0.04, 0.10)*** | 0.74 (0.40, 1.09)*** | 0.74 (0.49, 0.99)*** |
| Group 4 | 0.00 | 0.00 | 0.00 | 0.00 |
| p-for-trend | <0.0001 | <0.0001 | <0.0001 | <0.0001 |
| Adjusted for %ΔBMI and %ΔWC | (n=5882) | (n=5882) | (n=5882) | (n=5921) |
| Overall F-test | - | - | - | ** |
| Group 1 | −0.13 (−1.45, 1.20) | 0.01 (−0.03, 0.05) | 0.08 (−0.39, 0.55) | 0.77 (0.45, 1.10)*** |
| Group 2 | −0.40 (−1.72, 0.93) | 0.01 (−0.03, 0.05) | 0.06 (−0.37, 0.49) | 0.51 (0.22, 0.80)** |
| Group 3 | 0.51 (−0.71, 1.72) | 0.02 (−0.01, 0.06) | 0.22 (−0.18, 0.63) | 0.40 (0.12, 0.67)** |
| Group 4 | 0.00 | 0.00 | 0.00 | 0.00 |
| p-for-trend | 1.00 | 0.76 | 0.87 | <0.0001 |
p<0.05,
p<0.01,
p<0.0001,
- Not significant at p<0.05
%ΔFFM percent change in fat free mass/weight, HbA1c hemoglobin A1c, HOMA-IR homeostatic model assessment of insulin resistance, FMI fat mass index, %Δ percent change, BMI body mass index, WC waist circumference
Group 1: greatest % decrease in relative fat free mass; Group 4: greatest % increase in relative fat free mass
Models adjusted for age group, sex, Hispanic/Latino background, site, nativity, education, household income, physical activity, years between baseline and Visit 2, and diabetic medication use
The association between %ΔFFM and changes in HbA1c and HOMA-IR was modified by baseline BMI category (p-for-interaction=0.006 [0.014 for HbA1c, mmol/mol] and <0.0001, respectively). Although all p-for-trend were <0.0001, the strongest associations were among individuals with obesity (Supplementary Table 6).
Additional effect modification and sensitivity analyses
In analyses for incident prediabetes and T2DM and for changes in glucose and insulin measures, there was no effect modification by sex or age group (see Supplementary Tables 7 and 8 for sex-stratified analyses). Exclusion of women who were pregnant at any point during HCHS/SOL enrollment did not significantly alter any findings (Supplementary Table 9).
DISCUSSION
To our knowledge, this is the first study in Hispanics/Latinos to investigate the prospective association of changes in FFM with incident prediabetes and T2DM and changes in glucose and insulin measures. In the target population of Hispanic/Latino adults living in four US cities, we found that, compared to individuals who had the greatest percent increase in relative FFM, individuals who had the greatest percent decrease in relative FFM during the 6-year follow-up period had a higher incidence of prediabetes and greater increases in all glucose and insulin measures. Associations of changes in relative FFM with incidence of prediabetes and with changes in HbA1c and HOMA-IR were stronger among individuals who were overweight or obese compared to normal weight at baseline.
Findings from the current study contribute to the limited literature on prospective associations between changes in FFM and incidence of prediabetes and diabetes and changes in glucose and insulin measures. The results for incident prediabetes and T2DM in the present study are in contrast with the one previous study that examined this relationship in non-Hispanic White adults (15). Specifically, we observed an inverse association between %ΔFFM and incident T2DM only when analyses were stratified by baseline BMI category and restricted to individuals who did not self-report T2DM during any follow-up or clinic visit. Further, we reported an inverse association between %ΔFFM and incident prediabetes (the previous study found no association), with this association being strongest among individuals who were overweight and obese.
A potential explanation for these discrepancies is that associations with incident prediabetes and T2DM may differ for FFM compared to skeletal muscle mass (the measure used in the previous study (15)). In HCHS/SOL, the BIA device was a four-electrode, single frequency tool that, despite being the best available BIA measure at the time of study design, was unable to provide estimates of skeletal muscle mass directly (37). FFM thus represents a composite of the body’s skeletal muscle mass, bones, organs, and water content (38). Recent developments in BIA technology have resulted in segmental, eight-electrode, multifrequency BIA devices which measure limbs and trunk independently, rather than the traditional wrist to ankle measure used in HCHS/SOL (38). This allows for estimates of skeletal muscle mass through summation of the mass of the four limbs (38). While skeletal muscle mass (and the water it stores) accounts for approximately half of FFM (45% in females and 49% in males) (23), it is still possible that differences in these measures explain our discordant findings. Further, BIA itself is less precise than gold standard measures such as DXA, with discrepancies in these measures depending on the individual’s level of adiposity (39).
Despite these limitations, FFM estimates from BIA are still a relevant proxy for skeletal muscle mass given there is a direct, linear relationship between FFM and skeletal muscle mass (23). BIA provides clinically meaningful approximates of skeletal muscle mass in adults (23) and youth (40) and is an appealing alternative to gold standard measures such as DXA in both large epidemiological studies and clinical practice (23). This is due to BIA being simple, quick, and non-invasive (21) and requiring only inexpensive, portable equipment (22). Thus, in future research and clinical practice, researchers and clinicians should not rule out using BIA devices, but may consider using newer BIA devices to obtain more direct measures of skeletal muscle mass.
Despite inconsistencies between our findings and previous findings for T2DM and prediabetes, our findings for changes in insulin and glucose measures are consistent with those from the one previous study which assessed these associations using skeletal muscle mass instead of FFM (14). The association between %ΔFFM and HOMA-IR in the present study was also independent of changes in FMI or BMI and waist circumference, indicating a strong relationship between relative FFM and insulin resistance. Further, similar to our analysis for incident prediabetes and for the sensitivity analyses on incident T2DM, associations between %ΔFFM and changes in HbA1c and HOMA-IR were modified by baseline BMI category, with stronger associations among individuals who were overweight or obese.
We are the first to report that the association between %ΔFFM and changes in insulin and glucose measures are moderated by baseline BMI category. We hypothesized that individuals with both reduced FFM and obesity would be at an elevated risk for T2DM due to the ratio of adipokines/inflammatory cytokines to myokines being highly disproportionate in this group. Specifically, while individuals (of any weight) who lose muscle mass experience reduced secretion of myokines, in turn promoting insulin resistance and inflammation (6), individuals who have obesity additionally experience 1) accumulation of visceral fat stores that secrete adipokines and inflammatory cytokines and 2) accumulation of ectopic fat in the muscle and liver (41). These characteristics of obesity result in insulin resistance in the muscle and liver and systemic inflammation (41), increasing the risk for T2DM among individuals with reduced muscle mass and obesity beyond that of individuals with reduced muscle mass but no obesity (11). Despite the consistency of our findings with this explanation and with previous studies (11,17), there is some evidence of a stronger association between lower relative skeletal muscle mass and incident T2DM in individuals who are normal weight compared to overweight/obese (12,13). It is possible that the discrepancy in these previous findings is due to studies using baseline relative FFM (rather than percent change in relative FFM) and skeletal muscle mass (instead of FFM) to examine their associations. Thus, future prospective studies examining changes in %Δskeletal muscle mass in association with risk for T2DM are needed to clarify this association.
We had additionally hypothesized that age group and sex would modify the association between %ΔFFM and risk for T2DM, but we did not see evidence of effect modification. We had expected that the association between FFM loss and T2DM would be stronger in older individuals given aging is associated with accumulation of intramuscular fat deposits (42) and a decrease in the size of type II muscle fibers (43), both of which promote insulin resistance. We also hypothesized stronger associations among males, consistent with a previous study that found a stronger association between relative skeletal muscle mass and 3-year incident T2DM in adults (13). However, given our null findings and the mixed findings in the literature (11–13,17), there is a need for additional research to examine the moderating role of age group and sex in associations between FFM and T2DM. Further, additional research should examine why younger Hispanic/Latino adults in our sample were more likely to experience FFM loss than FFM gain over follow-up (as shown in Table 1) and whether this explains the lack of effect modification by age. Cross-sectional research has shown that the rate of skeletal muscle mass loss after 27 years of age differs according to race/ethnicity, with Hispanic males experiencing the most rapid declines in skeletal muscle mass with age. However, longitudinal research in Hispanic/Latino adults is needed to confirm these differences in body composition changes (4,44).
Strengths and Limitations
A major strength of our study is the use of a large cohort of Hispanic/Latino adults of diverse backgrounds. We had objective measures of prediabetes and T2DM, as well as of key covariates, including physical activity—an important determinant of body composition for which previous studies have either relied on self-report measures (13,17) or have not adjusted for at all. Additionally, we used a measure of FFM that the lay public can easily understand (i.e., expressing FFM relative to total body weight rather than to height-squared) (14).
The main limitation of this study is the reliance on FFM instead of skeletal muscle mass. Given how closely fat mass and FFM are related (relative FM + relative FFM = 1), it was challenging to isolate changes in FFM as the reason for observed associations. We addressed this concern by adjusting for baseline measures and changes in FMI or BMI and waist circumference. However, these measures were highly correlated with %ΔFFM (correlation coefficients of −0.84, −0.78, and −0.55 for percent change in FMI, BMI, and waist circumference, respectively). Thus, it is difficult to determine whether the reason for the largely null results after adjustment for percent changes in FMI or BMI and waist circumference was issues of covariance or the role of adiposity. Future studies using more precise measures of body composition are needed to replicate these findings. We were also limited by the low prevalence of sarcopenia in the HCHS/SOL sample, which prevented us from examining the sarcopenic obesity phenotype. However, our examination of effect modification by baseline BMI category allowed us to address the joint effect of obesity and low FFM on T2DM risk. Additionally, while our sensitivity analysis aimed to elucidate true incident cases of T2DM, we recognize that many individuals likely had undiagnosed T2DM prior to Visit 2 (1). Future work is thus needed which examines associations between FFM at an earlier point in time and later T2DM to clarify the temporality of the association.
Conclusions
In this study of Hispanic/Latino adults, we found that individuals who lost the most relative FFM over a 6-year period had a higher incidence of prediabetes and greater increases in fasting glucose, HbA1c, and HOMA-IR compared to individuals who gained the most relative FFM. Reducing loss of FFM during adulthood may reduce prediabetes risk (primarily insulin resistance), particularly among individuals with overweight/obesity.
Supplementary Material
ACKNOWLEDGEMENTS:
Funding: The Hispanic Community Health Study/Study of Latinos was supported by contracts from the NHLBI to the University of North Carolina (N01-HC65233), University of Miami (N01-HC65234), Albert Einstein College of Medicine (N01-HC65235), the University of Illinois at Chicago (HHSN268201300003I), Northwestern University (N01-HC65236), and San Diego State University (N01-HC65237). The following Institutes/Centers/Offices contributed to the HCHS/SOL through a transfer of funds to NHLBI: National Center on Minority Health and Health Disparities, the National Institute of Deafness and Other Communications Disorders, the National Institute of Dental and Craniofacial Research, the National Institute of Diabetes and Digestive and Kidney Diseases, the National Institute of Neurologic Disorders and Stroke, and the Office of Dietary Supplements. Additional support was provided by the Life Course Methodology Core (LCMC) at Albert Einstein College of Medicine and the New York Regional Center for Diabetes Translation Research (P30 DK111022-8786 and P30 DK111022) through funds from the National Institute of Diabetes and Digestive and Kidney Diseases. Support for the lead author was provided by an NHLBI training grant (T32-HL144456). The content is solely the responsibility of the authors and does not necessarily represent the official views of the NHLBI or the National Institutes of Health.
Footnotes
Conflict of interest: The authors declare no conflict of interest.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
References
- 1.Centers for Disease Control and Prevention. National Diabetes Statistics Report, 2017. Atlanta, GA: Centers for Disease Control and Prevention, U.S. Dept of Health and Human Services; 2017. [Google Scholar]
- 2.Spanakis EK, Golden SH. Race/Ethnic Difference in Diabetes and Diabetic Complications. Curr Diab Rep. 2013;13(6):814–23. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Santilli V, Bernetti A, Mangone M, Paoloni M. Clinical definition of sarcopenia. Clin Cases Miner Bone Metab. 2014;11(3):177–80. [PMC free article] [PubMed] [Google Scholar]
- 4.Silva AM, Shen W, Heo M, Gallagher D, Wang Z, Sardinha LB, et al. Ethnicity-Related Skeletal Muscle Differences Across the Lifespan. Am J Hum Biol. 2010;22(1):76–82. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Baumgartner RN, Koehler KM, Gallagher D, Romero L, Heymsfield SB, Ross RR, et al. Epidemiology of sarcopenia among the elderly in New Mexico. Am J Epidemiol. 1998;147(8):755–63. [DOI] [PubMed] [Google Scholar]
- 6.Eckardt K, Görgens SW, Raschke S, Eckel J. Myokines in insulin resistance and type 2 diabetes. Diabetologia. 2014;57(6):1087–99. [DOI] [PubMed] [Google Scholar]
- 7.DeFronzo RA, Tripathy D. Skeletal Muscle Insulin Resistance Is the Primary Defect in Type 2 Diabetes. Diabetes Care. 2009;32(Suppl 2):S157–63. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Thiebaud D, Jacot E, DeFronzo RA, Maeder E, Jequier E, Felber JP. The effect of graded doses of insulin on total glucose uptake, glucose oxidation, and glucose storage in man. Diabetes. 1982;31(11):957–63. [DOI] [PubMed] [Google Scholar]
- 9.DeFronzo RA, Jacot E, Jequier E, Maeder E, Wahren J, Felber JP. The Effect of Insulin on the Disposal of Intravenous Glucose: Results from Indirect Calorimetry and Hepatic and Femoral Venous Catheterization. Diabetes. 1981;30(12):1000–7. [DOI] [PubMed] [Google Scholar]
- 10.Khadra D, Itani L, Tannir H, Kreidieh D, El Masri D, El Ghoch M. Association between sarcopenic obesity and higher risk of type 2 diabetes in adults: A systematic review and meta-analysis. World J Diabetes. 2019;10(5):311–23. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Srikanthan P, Hevener AL, Karlamangla AS. Sarcopenia exacerbates obesity-associated insulin resistance and dysglycemia: findings from the National Health and Nutrition Examination Survey III. PLoS One. 2010;5(5):e10805. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Moon S-S. Low skeletal muscle mass is associated with insulin resistance, diabetes, and metabolic syndrome in the Korean population: the Korea National Health and Nutrition Examination Survey (KNHANES) 2009–2010. Endocr J. 2014;61(1):61–70. [DOI] [PubMed] [Google Scholar]
- 13.Hong S, Chang Y, Jung H-S, Yun KE, Shin H, Ryu S. Relative muscle mass and the risk of incident type 2 diabetes: A cohort study. PLoS One. 2017;12(11):e0188650. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Kim G, Lee S-E, Jun JE, Lee Y-B, Ahn J, Bae JC, et al. Increase in relative skeletal muscle mass over time and its inverse association with metabolic syndrome development: a 7-year retrospective cohort study. Cardiovasc Diabetol. 2018;17(1):23. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Maliszewska K, Adamska-Patruno E, Goscik J, Lipinska D, Citko A, Krahel A, et al. The Role of Muscle Decline in Type 2 Diabetes Development: A 5-Year Prospective Observational Cohort Study. Nutrients. 2019;11(4):E834. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Park BS, Yoon JS. Relative skeletal muscle mass is associated with development of metabolic syndrome. Diabetes Metab J. 2013;37(6):458–64. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Larsen BA, Wassel CL, Kritchevsky SB, Strotmeyer ES, Criqui MH, Kanaya AM, et al. Association of Muscle Mass, Area, and Strength With Incident Diabetes in Older Adults: The Health ABC Study. J Clin Endocrinol Metab. 2016;101(4):1847–55. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Roden M. Future of muscle research in diabetes: a look into the crystal ball. Diabetologia. 2015;58(8):1693–8. [DOI] [PubMed] [Google Scholar]
- 19.Sorlie PD, Avilés-Santa LM, Wassertheil-Smoller S, Kaplan RC, Daviglus ML, Giachello AL, et al. Design and implementation of the Hispanic Community Health Study/Study of Latinos. Ann Epidemiol. 2010;20(8):629–41. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Lavange LM, Kalsbeek WD, Sorlie PD, Avilés-Santa LM, Kaplan RC, Barnhart J, et al. Sample design and cohort selection in the Hispanic Community Health Study/Study of Latinos. Ann Epidemiol. 2010;20(8):642–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Diaz EO, Villar J, Immink M, Gonzales T. Bioimpedance or anthropometry? Eur J Clin Nutr. 1989;43(2):129–37. [PubMed] [Google Scholar]
- 22.Dehghan M, Merchant AT. Is bioelectrical impedance accurate for use in large epidemiological studies? Nutr J. 2008;7:26. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Jensen B, Braun W, Geisler C, Both M, Klückmann K, Müller MJ, et al. Limitations of Fat-Free Mass for the Assessment of Muscle Mass in Obesity. Obes Facts. 2019;12(3):307–15. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Arredondo EM, Sotres-Alvarez D, Stoutenberg M, Davis SM, Crespo NC, Carnethon MR, et al. Physical Activity Levels in U.S. Latino/Hispanic Adults: Results From the Hispanic Community Health Study/Study of Latinos. Am J Prev Med. 2016;50(4):500–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Hales CM, Carroll MD, Fryar CD, Ogden CL. Prevalence of Obesity Among Adults and Youth: United States, 2015–2016. NCHS Data Brief, no 288. Hyattsville, MD: National Center for Health Statistics. 2017. [PubMed] [Google Scholar]
- 26.Chumlea WC, Guo SS, Kuczmarski RJ, Flegal KM, Johnson CL, Heymsfield SB, et al. Body composition estimates from NHANES III bioelectrical impedance data. Int J Obes Relat Metab Disord. 2002;26(12):1596–609. [DOI] [PubMed] [Google Scholar]
- 27.Muscaritoli M, Anker SD, Argilés J, Aversa Z, Bauer JM, Biolo G, et al. Consensus definition of sarcopenia, cachexia and pre-cachexia: joint document elaborated by Special Interest Groups (SIG) “cachexia-anorexia in chronic wasting diseases” and “nutrition in geriatrics.” Clin Nutr. 2010;29(2):154–9. [DOI] [PubMed] [Google Scholar]
- 28.Matthews DR, Hosker JP, Rudenski AS, Naylor BA, Treacher DF, Turner RC. Homeostasis model assessment: insulin resistance and beta-cell function from fasting plasma glucose and insulin concentrations in man. Diabetologia. 1985;28(7):412–9. [DOI] [PubMed] [Google Scholar]
- 29.Association AD. Diagnosis and Classification of Diabetes Mellitus. Diabetes Care. 2010;33(Supplement 1):S62–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Esliger DW, Tremblay MS. Technical reliability assessment of three accelerometer models in a mechanical setup. Med Sci Sports Exerc. 2006;38(12):2173–81. [DOI] [PubMed] [Google Scholar]
- 31.Welk GJ, Schaben JA, Morrow JR. Reliability of accelerometry-based activity monitors: a generalizability study. Med Sci Sports Exerc. 2004;36(9):1637–45. [PubMed] [Google Scholar]
- 32.Evenson KR, Sotres-Alvarez D, Deng Y, Marshall SJ, Isasi CR, Esliger DW, et al. Accelerometer Adherence and Performance in a Cohort Study of US Hispanic Adults. Med Sci Sports Exerc. 2015;47(4):725–34. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Gunderson EP. Childbearing and Obesity in Women: Weight Before, During, and After Pregnancy. Obstet Gynecol Clin North Am. 2009;36(2):317–32. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Glymour MM, Weuve J, Berkman LF, Kawachi I, Robins JM. When is baseline adjustment useful in analyses of change? An example with education and cognitive change. Am J Epidemiol. 2005;162(3):267–78. [DOI] [PubMed] [Google Scholar]
- 35.Seaman SR, White IR. Review of inverse probability weighting for dealing with missing data. Stat Methods Med Res. 2013;22(3):278–95. [DOI] [PubMed] [Google Scholar]
- 36.Palta P, McMurray RG, Gouskova NA, Sotres-Alvarez D, Davis SM, Carnethon M, et al. Self-reported and accelerometer-measured physical activity by body mass index in US Hispanic/Latino adults: HCHS/SOL. Prev Med Rep. 2015;2:824–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Tanita Corporation. TBF-300A Total Body Composition Analyzer. Tanita Corporation of America; website https://www.tanita.com/en/tbf-300a/. 2018. Accessed September 10, 2019. [Google Scholar]
- 38.Lemos T, Gallagher D. Current body composition measurement techniques. Curr Opin Endocrinol Diabetes Obes. 2017;24(5):310–4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Leon B, Jenkins S, Pepin K, Chaudhry H, Smith K, Zalos G, et al. Insulin and Extremity Muscle Mass in Overweight and Obese Women. Int J Obes (Lond). 2013;37(12):1560–4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Beeson WL, Batech M, Schultz E, Salto L, Firek A, DeLeon M, et al. Comparison of body composition by bioelectrical impedance analysis and dual-energy X-ray absorptiometry in Hispanic diabetics. Int J Body Compos Res. 2010;8(2):45–50. [PMC free article] [PubMed] [Google Scholar]
- 41.Hardy OT, Czech MP, Corvera S. What causes the insulin resistance underlying obesity? Curr Opin Endocrinol Diabetes Obes. 2012;19(2):81–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Addison O, Marcus RL, LaStayo PC, Ryan AS. Intermuscular Fat: A Review of the Consequences and Causes. Int J Endocrinol. 2014:309570. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.Cleasby ME, Jamieson PM, Atherton PJ. Insulin resistance and sarcopenia: mechanistic links between common co-morbidities. J Endocrinol. 2016;229(2):R67–81. [DOI] [PubMed] [Google Scholar]
- 44.He Q, Heo M, Heshka S, Wang J, Pierson RN, Albu J, et al. Total body potassium differs by sex and race across the adult age span. Am J Clin Nutr. 2003July;78(1):72–7. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.


