Volume 35, Issue 3 p. 430-439
Original Article
Free Access

Effect of Aerobic or Resistance Exercise, or Both, on Bone Mineral Density and Bone Metabolism in Obese Older Adults While Dieting: A Randomized Controlled Trial

Reina Armamento-Villareal

Reina Armamento-Villareal

Division of Endocrinology, Diabetes, and Metabolism, Baylor College of Medicine, Houston, TX, USA

Center for Translational Research on Inflammatory Diseases (CTRID), Michael E DeBakey VA Medical Center, Houston, TX, USA

Search for more papers by this author
Lina Aguirre

Lina Aguirre

Medicine Care Line, New Mexico VA Health Care System, Albuquerque, NM, USA

Department of Internal Medicine, University of New Mexico School of Medicine, Albuquerque, NM, USA

Search for more papers by this author
Debra L Waters

Debra L Waters

Department of Internal Medicine, University of New Mexico School of Medicine, Albuquerque, NM, USA

Department of Medicine, School of Physiotherapy, University of Otago, Dunedin, New Zealand

Search for more papers by this author
Nicola Napoli

Nicola Napoli

Division of Bone and Mineral Diseases, Washington University School of Medicine, St. Louis, MO, USA

Search for more papers by this author
Clifford Qualls

Clifford Qualls

Department of Mathematics and Statistics, University of New Mexico School of Medicine, Albuquerque, NM, USA

Search for more papers by this author
Dennis T Villareal

Corresponding Author

Dennis T Villareal

Division of Endocrinology, Diabetes, and Metabolism, Baylor College of Medicine, Houston, TX, USA

Center for Translational Research on Inflammatory Diseases (CTRID), Michael E DeBakey VA Medical Center, Houston, TX, USA

Address correspondence to: Dennis T Villareal, MD, Baylor College of Medicine, Michael E DeBakey VA Medical Center, 2002 Holcombe Avenue, Houston, TX 77030, USA. E-mail: [email protected]Search for more papers by this author
First published: 04 December 2019
Citations: 38

ABSTRACT

Weight loss therapy of older adults with obesity is limited by weight loss–induced decrease in bone mineral density (BMD), which could exacerbate ongoing age-related bone loss and increase the risk for fractures. Therefore, it is recommended that weight loss therapy of older adults with obesity should include an intervention such as regular exercise to reduce the concomitant bone loss. However, the most appropriate exercise types to combine with weight loss therapy in this older population is unknown. In a randomized controlled trial, we performed a head-to-head comparison of aerobic or resistance exercise, or both, during matched ~10% weight loss in 160 older adults with obesity. We measured changes in BMD (total hip, femoral neck, trochanter, intertrochanter, one-third radius, lumbar spine) and bone markers. Changes between groups were analyzed using mixed-model repeated measures analyses of variance. After 6 months of intensive lifestyle interventions, BMD decreased less in the resistance group (−0.006 g/cm2 [−0.7%]) and combination group (−0.012 g/cm2 [−1.1%]) than in the aerobic group (−0.027 g/cm2 [−2.6%]) (p = 0.001 for between-group comparisons). Serum C-telopeptide, procollagen type 1 N-propeptide, and osteocalcin concentrations increased more in the aerobic group (33%, 16%, and 16%, respectively) than in the resistance group (7%, 2%, and 0%, respectively) and combination group (11%, 2%, and 5%, respectively) (p = 0.004 to 0.048 for between-group comparisons). Multiple regression analyses revealed that the decline in whole body mass and serum leptin were the independent predictors of the decline in hip BMD (multiple R = 0.45 [p < .001]). These findings indicate that compared with aerobic exercise, resistance and combined aerobic and resistance exercise are associated with less weight loss–induced decrease in hip BMD and less weight loss–induced increase in bone turnover. Therefore, both resistance and combined aerobic and resistance exercise can be recommended to protect against bone loss during weight loss therapy of older adults with obesity. (LITOE ClinicalTrials.gov number NCT01065636.) © 2019 American Society for Bone and Mineral Research. Published 2019. This article is a U.S. Government work and is in the public domain in the USA.

Introduction

The continued increasing prevalence of obesity in older adults in the US and other more developed countries is a major public health concern exacerbated by the age-related decline in physical function that results in frailty, decrease in quality of life, and increase in nursing home admissions.1, 2 However, the appropriate treatment approach for obesity in older adults remains controversial because of the apprehension that weight loss may potentially worsen the age-related decline in muscle mass and also bone mass that could increase the risk for fractures.3-8 Therefore, it has been recommended that weight loss therapy in older adults should include an intervention such as regular exercise to prevent bone and muscle loss.2

We previously reported that weight loss plus combined aerobic and resistance exercise attenuated but did not eliminate bone loss induced by weight loss.9, 10 The physiologic adaptations to aerobic and resistance exercise are distinctly different—some studies have shown that aerobic exercise may interfere with strength gains from resistance exercise when the two types of training are performed together (interference effect).11-14 It is not known whether the interference effect from concurrent training may also apply to the bone-adaptive response to specific exercises during weight loss therapy of older adults with obesity. In fact, little is known about the comparative efficacy of different exercise types in preventing the weight loss–induced reduction of bone mineral density (BMD). Therefore, we performed a head-to-head comparison of aerobic or resistance exercise, or both, in dieting older adults with obesity to test the hypothesis that resistance exercise would prevent bone loss more than aerobic exercise or combined aerobic and resistance exercise. The data reported in this article were obtained using the same subject group that participated in the Lifestyle Intervention Trial in Obese Elderly (LITOE).15

Materials and Methods

Overview

The LITOE was a randomized controlled trial (RCT) with the main aim to evaluate the relative efficacy of several exercise modes in improving physical function and reversing frailty in older adults with obesity. The principal results showed that weight loss plus combined aerobic and resistance exercise was the most effective in improving functional status of older adults with obesity.15 The present study reports secondary analyses of the trial examining changes in BMD, bone turnover, and bone-active hormones, as prespecified in the protocol.

Study population

We conducted the LITOE at the University of New Mexico School of Medicine and New Mexico Veterans Affairs Health Care System from April 2010 through June 2015 as previously reported.15 The study was approved by the university's institutional review board and was monitored by an independent data and safety monitoring board. We recruited volunteers from the community through advertisements, and informed consent was obtained from each participant. All potential subjects underwent a comprehensive medical examination before participation. Details of the inclusion and exclusion criteria have been described.15 Briefly, inclusion criteria were: older age (aged ≥65 years), obesity (body mass index ≥30 in kg/m2), sedentary lifestyle (regular exercise <1 hr/week), stable body weight (change in body weight of ≤2 kg in the preceding year), on stable medications (≥6 months) before enrollment, and mild to moderate frailty.16, 17 Exclusion criteria included severe cardiopulmonary disease, musculoskeletal or neuromuscular impairments that precluded exercise training, diagnosis of dementia, and BMD T-scores of <−2.3 at the lumbar spine or proximal femur. Subjects who were treated with bone-acting drugs (eg, bisphosphonates, glucocorticoids, sex-steroid compounds) during the previous year were also excluded from participation. The effects of the weight loss plus exercise interventions on frailty, body composition, specific physical functions, quality of life, and some measures of BMD have been reported.15

Design

We randomized study participants, with stratification for sex, for a 26-week period, to one of the following four groups: (1) control group that did not participate in a weight-management or exercise intervention; (2) aerobic group that participated in a weight-management program and aerobic exercise training; (3) resistance group that participated in a weight-management program and resistance exercise training; and (4) combination group that participated in a weight-management program and combined aerobic and resistance exercise training. All exercise sessions were supervised by exercise physiologists at our facility. The randomization algorithm was generated and maintained by the study biostatistician, who did not interact with the participants.

The control group was instructed not to participate in external weight loss or exercise programs. However, this group attended educational sessions about a healthful diet during monthly visits. The aerobic group was prescribed a balanced diet that provided an energy deficit of 500 to 750 kcal per day. Participants met weekly with a dietitian for dietary adjustments and behavioral therapy, with the goal to achieve a weight loss of approximately 10% at 6 months. They also participated in aerobic exercise training sessions three times weekly; the sessions were ~60 minutes long and included 10 minutes of flexibility exercise, followed by 40 minutes of aerobic exercise, and 10 minutes of balance exercise. The aerobic exercises included walking on a treadmill, stationary cycling, and stair climbing. Participants exercised at ~65% of their peak heart rate, which was gradually increased to 70% to 85%. The resistance group participated in the same weight-management program as the aerobic group, as well as resistance exercise training three times weekly; the sessions were ~60 minutes long and included 10 minutes of flexibility exercise, followed by 40 minutes of resistance exercise and 10 minutes of balance exercise. The resistance training consisted of nine upper-body and lower-body exercises using weight-lifting machines. The initial sessions were 1 to 2 sets of 8 to 12 repetitions at 65% of the one-repetition maximum (1-RM; the maximum weight a participant can lift, in one attempt), which was increased progressively to 2 to 3 sets at ~85% of the 1-RM. The combination group participated in the same weight-management program as the other exercise groups, as well as combined aerobic and resistance exercise training sessions three times weekly. The sessions were 75 to 90 minutes long and included 10 minutes of flexibility exercise, followed by 30 to 40 minutes of aerobic exercise, 30 to 40 minutes of resistance exercise, and 10 minutes of balance exercise. To ensure sufficient physiological response from both exercise interventions and test the interference effect,13 the aerobic and resistance training were balanced between groups; the longer duration of exercise in the combined group allowed the participants to perform an amount of aerobic exercise that was equivalent to that of the aerobic group and an amount of resistance exercise that was equivalent to that of the resistance group.15, 18 All participants received supplements to ensure an intake of ~1500 mg/d of calcium and ~1000 IU/d of vitamin D. Additional details about the interventions including compliance data and exercise adaptations have been reported in the primary paper.15

Outcome assessments

Body composition and bone mineral density

We measured whole body mass, lean mass, fat mass, and bone mineral density of the whole body, proximal femur, lumbar spine, and distal radius with the use of dual-energy X-ray absorptiometry (Lunar DPX [General Electric, Madison, WI, USA] or Discovery A [Hologic, Waltham, MA, USA] scanner), as described previously.17, 19 For each participant, baseline and follow-up scans were obtained with the use of the same instrument.15

Serum markers of bone metabolism and bone-active hormones

We obtained venous blood samples in the morning after subjects had fasted for at least 12 hours. The blood samples were centrifuged to obtain the serum specimens and frozen at –80°C until batch analyses. We used electrochemiluminescence immunoassay kit to measure (CTX) (Elecsys ß-Crosslaps, Cobas e601, Roche Diagnostics, Indianapolis, IN, USA; coefficient of variation [CV] 2.1%) as a marker of bone resorption, and enzyme-linked immunosorbent assay kit to measure osteocalcin (Metra OSC; Quidel, San Diego, CA, USA; CV 4.4%) as a marker of bone formation. We used a radioimmunoassay kit to measure serum N-terminal propeptide of type I procollagen (PINP) (Orion Diagnostica, Espoo, Finland; CV 5.2%) as an additional marker of bone formation. We also used radioimmunoassay kits to measure serum estradiol (Ultra-sensitive estradiol DSL-4800; Diagnostic Systems Laboratories, Inc., Webster, TX, USA), leptin (Leptin HL-81 K; Linco Research, Inc., St. Charles, MO, USA), and 25-hydroxyvitamin D (25(OH) D) (DiaSorin, Stillwater, MN, USA). We used enzyme-linked immunosorbent assay kit to measure sclerostin (TECOmedical, San Diego, CA, USA), adiponectin (Linco), and chemiluminescence kit to measure intact parathyroid hormone (VITROS ECi/ECiQ Immunodiagnostic Systems, Raritan, NJ, USA). The CVs for these hormone measurements were all less than 10%. Blood samples at 6 months were obtained 24 to 36 hours after the last bout of exercise.

Peak oxygen consumption and muscle strength testing

We assessed peak oxygen consumption (VO2peak) during graded treadmill walking by indirect calorimetry as described previously.17 We determined 1-RMs for upper- (biceps curl, bench press, and seated row) and lower-body exercises (knee extension, knee flexion, and leg press) and used the sum to calculate the total 1-RM.9, 15

Follow-up assessments

All baseline assessments were repeated at 6 months. The personnel who conducted the assessments were blinded to the study-group assignments.

Statistical analyses

The samples size calculated for the main outcome of this study had more than 80% power to detect a clinically meaningful 2.3 ± 2.4% difference in the change in total hip BMD (based on preliminary data) among the groups, at an alpha level of 5%.

We performed intention-to-treat analyses using SAS software (SAS version 9.4, Cary, NC, USA). We compared baseline characteristics using analyses of variance or Fisher's exact tests. We tested longitudinal changes between groups with mixed-model repeated-measures ANOVAs, adjusting for baseline values and sex. The primary focus of the analyses was the 6-month change in outcome in the four groups. Within the framework of the mixed model, when the overall p value for the interaction between group and time was less than 0.05, prespecified contrast statements were used to test five hypotheses: that changes in the aerobic group would differ from those in the control group, that changes in the resistance group would differ from those in the control group, that changes in the aerobic group would differ from those in the resistance group, and that changes in the combination group would differ from those in the aerobic group and from those in the resistance group. We also performed analyses testing for within-group changes using mixed-model repeated-measures ANOVA. We first used Pearson's correlation to examine relationships among changes in selected variables and changes in total hip BMD. We then used stepwise multiple linear regression analysis (forward elimination method, validated by backward elimination method) to identify which among the selected variables were the important independent predictors to the changes in total hip BMD. Multicollinearity among the predictors was estimated with the variance inflation factor (VIF). Because of high collinearity with change in whole body mass (VIF = 5.1), change in fat mass was excluded from the stepwise procedures.

We performed sensitivity analyses to validate the statistical results obtained using multiple imputation for missing fitness data (which confirmed a similar pattern of results). Data for change scores and percentage changes are presented as least-squares-adjusted means (±SE). All statistical tests were two-tailed, and p < 0.05 was considered statistically significant.

Results

The CONSORT diagram has been reported previously (Supplemental Fig. S1).15 Briefly, 160 participants were randomized and 141 (88%) completed the study. Nineteen participants discontinued the intervention (4 in control, 5 in aerobic, 5 in resistance, and 5 in combination group) owing to personal or medical reasons but were included in the intention-to-treat analyses. The four groups did not differ in baseline characteristics, including in age, sex, race or ethnicity, body weight, body mass index (BMI), and BMD T-score (Table 1). Diet compliance was 96% (interquartile range [IQR] 87 to 100) in the aerobic group, 100% (IQR 90 to 100) in the resistance group, and 97% (IQR 89 to 100) in the combination group. Exercise compliance was 96% (IQR 84 to 100) in the aerobic group, 98% (IQR, 81 to 100) in the resistance group, and 93% (IQR 83 to 100) in the combination group.

Table 1. Baseline Characteristics of the Participantsa
Control Aerobic Resistance Combination
(n = 40) (n = 40) (n = 40) (n = 40)
Age (years) 70 ± 5 70 ± 4 70 ± 5 70 ± 5
Sex (n [%])
Male 12 (30) 14 (35) 15 (37) 16 (40)
Female 28 (70) 26 (65) 25 (63) 24 (60)
Race (n [%])b a
White 36 (90) 36 (90) 33 (83) 35 (88)
Black 1 (3) 0 (0) 3 (7) 2 (5)
Other 3 (7) 4 (10) 4 (10) 3 (7)
Ethnicity (n [%])b
Hispanic or Latino 13 (33) 13 (33) 12 (30) 12 (30)
Not Hispanic or Latino 27 (67) 27 (67) 27 (67) 28 (70)
Unknown 0 (0) 0 (0) 1 (3) 0 (0)
Marital status (n [%])
Single 8 (20) 4 (10) 9 (22) 5 (12)
Married 15 (37) 25 (63) 22 (55) 23 (58)
Divorced 9 (23) 7 (17) 6 (15) 8 (20)
Widowed 8 (20) 4 (10) 3 (8) 4 (10)
Height (m) 163.4 ± 11.6 164.8 ± 15.3 166.5 ± 11.9 165.6 ± 9.2
Weight (kg) 97.5 ± 17.6 97.6 ± 15.4 101.7 ± 18.2 98.4 ± 17.5
Body mass index (kg/m2) 36.7 ± 5.0 35.9 ± 4.4 36.7 ± 5.8 35.8 ± 4.5
Bone mineral density (T-score)
Lumbar spine 0.6 ± 1.7 0.4 ± 1.6 0.6 ± 1.7 0.4 ± 2.6
Total hip 0.5 ± 1.0 0.3 ± 0.8 0.2 ± 1.1 0.4 ± 1.1
Femoral neck –0.7 ± 0.9 –0.8 ± 0.8 –1.1 ± 1.1 –0.9 ± 1.1
  • There were no significant differences among groups (all p > 0.05). Intervention groups: aerobic = weight management and aerobic training; resistance = weight management and resistance training; combination = weight management and combined aerobic and resistance training.
  • a Plus-minus values are means ±SD.
  • b Race and ethnicity were self-reported.

As reported, body weight decreased similarly in the aerobic group (−9.0 ± 0.6 kg [9% decrease], resistance group (−8.5 ± 0.5 kg [9% decrease]), and combination group (−8.5 ± 0.5 kg [9% decrease]) but not in the control group (−0.9 ± 0.5 kg [<1% decrease]).15 However, despite equal weight loss, BMD at the total hip decreased less in the resistance group (−0.006 ± 0.004 g/cm2 [0.7% decrease]) and combination group (−0.012 ± 0.004 g/cm2 [1.1% decrease]) than in the aerobic group (−0.027 ± 0.004 g/cm2 [2.6% decrease]) (Table 2 and Fig. 1). The changes in total hip BMD between the resistance and combination groups were not different. Similar findings were observed for changes in BMD at the femoral neck, trochanter, and intertrochanteric sites (Fig. 1). There were no significant differences from baseline and between groups in changes in BMD at the one-third radius, lumbar spine, or whole body. Lean mass decreased less in the resistance group (−1.0 ± 0.3 kg [2% decrease]) and combination group (−1.7 ± 0.3 kg [3% decrease]) than in the aerobic group (−2.7 ± 0.3 kg [5% decrease]) as reported.15 Fat mass decreased similarly in the aerobic group (−6.3 ± 0.3 kg [16% decrease]), resistance group (−7.3 ± 0.4 kg [17% decrease]), and combination group (−7.0 ± 0.5 kg [17% decrease]). Moreover, in response to the aerobic training, VO2peak increased more in the aerobic group (3.3 ± 0.3 ml/kg/min [18% increase]) and combination group (3.1 ± 0.3 ml/kg/min [17% increase]) than in the resistance group (1.3 ± 0.3 [8% increase). In response to resistance training, total 1-RM strength increased more in the resistance group (49 ± 5 kg [19% increase]) and combination group (48 ± 5 kg [18% increase] than in the aerobic group (5 ± 5 kg [4% increase]).

Table 2. Effect of Specific Exercise Modes Added to Diet-Induced Weight Loss on Bone Mineral Densitya
Outcome variables Control (n = 40) Aerobic (n = 40) Resistance (n = 40) Combination (n = 40) p valueb
Group-time interaction Aerobic vs. Control Resistance vs. Control Aerobic vs. Resistance Combination vs. Aerobic Combination vs. Resistance
Total hip (g/cm2)
Baseline 1.031 ± 0.025 1.018 ± 0.019 1.047 ± 0.022 1.010 ± 0.025
Change at 6 months 0.002 ± 0.004 −0.027 ± 0.004* −0.006 ± 0.004 −0.012 ± 0.004*** <0.001 <0.001 0.37 0.005 0.04 0.43
Femoral neck (g/cm2)
Baseline 0.806 ± 0.023 0.812 ± 0.022 0.804 ± 0.020 0.797 ± 0.027
Change at 6 months −0.001 ± 0.003 −0.020 ± 0.003* −0.003 ± 0.003 −0.008 ± 0.003 0.005 0.001 0.63 0.004 0.03 0.44
Trochanter (g/cm2)
Baseline 0.739 ± 0.118 0.757 ± 0.148 0.762 ± 0.122 0.773 ± 0.024
Change at 6 months 0.006 ± 0.007 −0.035 ± 0.007* −0.006 ± 0.007 -0.016 ± 0.007*** 0.01 0.001 0.12 0.08 0.25 0.54
Intertrochanter (g/cm2)
Baseline 1.244 ± 0.033 1.259 ± 0.024 1.252 ± 0.031 1.228 ± 0.003
Change at 6 months 0.006 ± 0.007 −0.035 ± 0.007** −0.006 ± 0.007 −0.016 ± 0.007***¶ 0.03 0.004 0.37 0.04 0.14 0.55
One-third radius (g/cm2)
Baseline 0.661 ± 0.023 0.651 ± 0.015 0.684 ± 0.098 0.655 ± 0.016
Change at 6 months 0.001 ± 0.001 −0.001 ± 0.001 −0.0020 ± 0.001 -0.001 ± 0.002 0.70
Lumbar spine (g/cm2)
Baseline 1.141 ± 0.033 1.118 ± 0.022 1.144 ± 0.033 1.157 ± 0.033
Change at 6 months 0.010 ± 0.006 0.002 ± 0.006 0.008 ± 0.006 0.008 ± 0.005 0.49
Whole body (g/cm2)
Baseline 1.134 ± 0.023 1.118 ± 0.022 1.115 ± 0.021 1.120 ± 0.023 0.94
Change at 6 months 0.001 ± 0.005 −0.003 ± 0.005 0.005 ± 0.005 0.002 ± 0.005
  • Intervention groups: aerobic = weight management and aerobic training; resistance = weight management and resistance training; combination = weight management and combined aerobic and resistance training.
  • a Plus-minus values for the change scores are the least-square adjusted means ±SE from the repeated measures analyses of variance; plus-minus values for the baseline values are the observed means ±SE.
  • b p values for the changes from baseline to 6 months in between-group comparisons were calculated with the use of mixed-model-repeated-measures analyses of variance (with baseline values and sex as covariates) and are reported when the overall p value was <0.05 for the interaction among the four groups over time.
  • *p < 0.001, **p < 0.01, and ***p < 0.05 for the comparison of the value at the follow-up time with the baseline value within the group, as calculated with the use of mixed-model repeated-measures analysis of variance.
Details are in the caption following the image
Mean percentage changes from baseline in bone mineral density at the (A) total hip, (B) femoral neck, (C) trochanter, (D) intertrochanter, (E) lumbar spine, and (F) one-third radius during the interventions. *p < 0.05 for the comparison of the value from the control group; †p < 0.05 for the comparison of the value from the aerobic group. I bars indicate standard errors.

Serum CTX concentrations increased more in the aerobic group (0.124 ± 0.018 μg/L [33% increase] than in the combination group (0.043 ± 0.017 μg/L [11% increase], whereas they did not significantly change in the resistance group (0.027 ± 0.018 μg/L [7% increase]) and in the control group (−0.005 ± 0.017 μg/L [1% decrease]) (Table 3 and Fig. 2). Serum PINP concentrations increased in the aerobic group (6.9 ± 1.6 μg/L [16% increase]) but did not significantly change in the resistance group (−0.8 ± 1.7 μg/L [2% decrease]), combination group (0.7 ± 1.6 μg/L [2% increase]), or control group (−2.6 ± 1.6 μg/L [6% decrease]). Likewise, serum OSC concentrations increased in the aerobic group (1.2 ± 0.3 ng/L [16% increase]) but did not significantly change in the resistance group (−0.0 ± 0.3 ng/L [0% decrease]), combination group (−0.4 ± 0.3 ng/L [5% decrease]), or control group (−0.8 ± 0.3 ng/L [9% decrease]).

Table 3. Effect of Specific Exercise Modes Added to Diet-Induced Weight Loss on Bone Markers and Hormonesa
Outcome variables Control (n = 40) Aerobic (n = 40) Resistance (n = 40) Combination (n = 40) p valueb
Group-time interaction Aerobic vs. Control Resistance vs. Control Aerobic vs. Resistance Combination vs. Aerobic Combination vs. Resistance
CTX (μg/L)
Baseline 0.381 ± 0.028 0.382 ± 0.022 0.359 ± 0.025 0.393 ± 0.024
Change at 3 months −0.009 ± 0.016 0.060 ± 0.019 0.007 ± 0.017 0.015 ± 0.017
Change at 6 months −0.005 ± 0.017 0.124 ± 0.018* 0.027 ± 0.018 0.043 ± 0.017 0.004 <0.001 0.10 0.005 0.008 0.81
P1NP (μg/L)
Baseline 45.4 ± 2.8 44.6 ± 2.0 41.7 ± 2.0 44.7 ± 2.5
Change at 3 months −1.5 ± 1.5 3.4 ± 1.7 −0.5 ± 1.7 −0.3 ± 1.6
Change at 6 months −2.6 ± 1.6 6.9 ± 1.6* −0.8 ± 1.7 0.7 ± 1.6 0.048 0.001 0.47 0.01 0.03 0.64
OSC (ng/L)
Baseline 8.1 ± 0.3 7.7 ± 0.3 7.6 ± 0.2 8.4 ± 0.3
Change at 3 months −0.8 ± 0.3 0.2 ± 0.3 −0.3 ± 0.3 −0.6 ± .0.3
Change at 6 months −0.8 ± 0.3 1.2 ± 0.3* −0.0 ± 0.3 −0.4 ± 0.3 0.01 <0.001 0.13 0.03 0.001 0.30
25-hydroxyvitamin D (ng/mL)
Baseline 26.6 ± 1.8 25.7 ± 1.6 27.5 ± 1.5 27.9 ± 1.9
Change at 6 months 3.0 ± 0.9*** 5.1 ± 1.0** 3.3 ± 0.9** 3.4 ± 0.9*** 0.61
Parathyroid hormone (pg/mL)
Baseline 55.2 ± 4.1 48.1 ± 3.1 48.8 ± 3.2 58.1 ± 3.3
Change at 6 months −1.8 ± 2.5 −5.6 ± 2.5 4.4 ± 2.4 0.9 ± 2.6 0.21
Sclerostin (ng/mL)
Baseline 0.82 ± 0.04 0.80 ± 0.04 0.83 ± 0.04 0.79 ± 0.04
Change at 6 months 0.01 ± 0.02 0.00 ± 0.02 0.02 ± 0.02 −0.06 ± 0.02 0.37
Adiponectin (ng/mL)
Baseline 13124 ± 1214 10782 ± 1064 11760 ± 895 12930 ± 1486
Change at 6 months –197 ± 384 1400 ± 414** 1726 ± 413** 1753 ± 396*** 0.03 0.04 0.01 0.70 0.69 0.99
Leptin (pg/mL)
Baseline 49.9 ± 4.1 50.2 ± 4.7 43.9 ± 5.0 42.7 ± 3.8
Change at 6 months –1.3 ± 1.6 −18.3 ± 1.6* −14.0 ± 1.5* -14.0 ± 1.6* <0.001 <0.001 <0.001 0.12 0.12 0.98
Estradiol (pg/mL) 
Women
Baseline 15.6 ± 1.2 17.4 ± 1.0 17.9 ± 2.3 17.1 ± 1.7
Change at 6 months 0.3 ± 0.5 -3.4 ± 0.6*** −3.5 ± 0.6** −3.2 ± 0.6** 0.001 <0.001 <0.001 0.84 0.84 0.69
Men
Baseline 24.9 ± 2.0 23.9 ± 1.1 25.2 ± 1.9 23.2 ± 2.2
Change at 6 months 1.2 ± 0.8 -3.6 ± 0.7*** −2.9 ± 0.7*** −3.3 ± 0.7*** 0.02 0.006 0.01 0.71 0.77 0.94
  • Intervention groups: aerobic = weight management and aerobic training; resistance = weight management and resistance training; combination = weight management and combined aerobic and resistance training.
  • CTX = C-terminal telopeptide of type 1 collagen; P1NP = intact N-terminal propeptide of type 1 procollagen; OSC = osteocalcin.
  • a Plus-minus values for the change scores are the least-square adjusted means ±SE from the repeated measures analyses of variance; plus-minus values for the baseline values are the observed means ±SE.
  • b p values for the changes from baseline to 6 months in between-group comparisons were calculated with the use of mixed-model-repeated-measures analyses of variance (with baseline values and sex as covariates) and are reported when the overall p value was <0.05 for the interaction among the 4 groups over time.
  • *p < 0.001, **p < 0.01, and ***p < 0.05 for the comparison of the value at the follow-up time with the baseline value within the group, as calculated with the use of mixed-model repeated-measures analysis of variance.
Details are in the caption following the image
Mean percentage changes from baseline in serum (A) C-terminal telopeptide of type 1 collagen (CTX), (B) intact N-terminal propeptide of type I procollagen (PINP), and (C) osteocalcin (OSC) during the interventions. *p < 0.05 for the comparison of the value from the control group; †p < 0.05 for the comparison of the value from the aerobic group. I bars indicate standard errors.

Serum 25-hydroxyvitamin D concentrations increased in all groups (control 3.0 ± 0.9 ng/mL [7% increase], aerobic 5.1 ± 1.0 ng/mL [11% increase], resistance 3.3 ± 0.9 ng/mL [8% increase], and combination 3.4 ± 0.9 ng/mL [8% increase]) (Table 3). There were no significant changes in serum PTH or sclerostin concentrations in any group. Serum adiponectin concentrations increased in all intervention groups (aerobic 1400 ± 414 ng/mL [13% increase], resistance 1726 ± 413 ng/mL [15% increase], combination 1753 ± 396 ng/mL [14% increase]), whereas they did not significantly change in the control group (−197 ± 384 ng/mL [2% decrease]). On the other hand, serum leptin decreased in all intervention groups (aerobic −18.3 ± 1.6 pg/mL [11% decrease], resistance −14.0 ± 1.5 pg/mL [8% decrease], combination −14.0 ± 1.6 pg/mL [8% decrease]) but did not significantly change in the control group (−1.3 ± 1.6 pg/mL 6 [3% decrease]). Serum estradiol decreased in women in all intervention groups (aerobic −3.4 ± 0.6 pg/mL [19% decrease], resistance −3.4 ± 0.6 pg/mL [20% decrease], combination −3.2 ± 0.6 pg/mL [18% decrease]) but did not significantly change in the control group (0.3 ± 0.5 pg/mL [2% increase]). Likewise, serum estradiol decreased in men in all intervention groups (aerobic −3.6 ± 0.8 pg/mL [15% decrease], resistance −2.9 ± 0.8 pg/mL [12% decrease], combination −3.3 ± 0.8 pg/mL [14% decrease]) but did not significantly change in the control group (1.2 ± 0.8 pg/mL [5% increase]).

Changes in whole body mass (r = 0.41; p < 0.001), fat mass (r = 0.39; p < 0.001), lean mass (r = 0.24; p = 0.009), CTX (r = −0.33; p < 0.001), PINP (r = −0.31; p = 0.001), osteocalcin (r = −0.25; p = 0.008), adiponectin (r = −0.29; p = 0.002), and leptin (r = 0.18; p = 0.03) correlated with changes in total hip BMD (Supplemental Table S1). However, in the stepwise multiple regression analyses, only changes in whole body mass (β = 0.54) and leptin (β = 0.21) remained significant in the final model, explaining 20% of the variance in changes in total hip BMD (multiple R = 0.45; p < 0.001) (Table 4).

Table 4. Final Model in the Stepwise Multiple Regression Analysis Identifying Predictors of Changes in Total Hip Bone Mineral Densitya
Multiple R = 0.45; p < 0.001
β p Value
Change in whole body mass 0.544 <0.001
Change in leptin 0.246 0.03
  • a Values entered in the model were change in whole body mass, change in lean mass, change in C-terminal telopeptide of type 1 collagen, change in intact N-terminal propeptide of type I procollagen, change in osteocalcin, change in adiponectin, and change in leptin.

Discussion

The results of the current study indicate that resistance exercise, and the combination of aerobic and resistance exercise, resulted in attenuated weight loss–induced bone loss at the total hip and femoral neck compared with aerobic exercise. Compared with the significant increase in bone turnover markers in the aerobic exercise group, markers of bone turnover were not significantly increased in both the resistance and combined aerobic and resistance exercise group, the changes of which were not significantly different from the control group. As expected with weight loss, there were significant decreases in leptin and estradiol and increase in adiponectin in the weight loss groups compared with control.

An important limitation to improve the physical and metabolic function of older adults with obesity with weight loss therapy is reduction of BMD, which could increase the risk of fractures.2, 7, 8, 20 In fact, it has been reported that after long-term weight loss, bone loss continues despite some degree of weight regain21, 22 arguing against the notion that the loss of BMD with weight loss is a physiologic response to loss of total body mass. With weight regain, there is fat regain, but this is not accompanied by bone regain.21, 22 This bone loss on top of ongoing age-related bone loss may result in osteoporosis, especially among those who have lower BMD at the start of weight loss. In addition, although obesity is associated with increased BMD, it is associated with increased risk for frailty,17, 23 an independent risk factor for falls and fractures.24, 25 Moreover, results from epidemiological studies suggest that though the prevalence of fractures decreases as BMI moves from underweight to normal, an increase in BMI from overweight to obese is actually associated with a higher risk for fractures after accounting for BMD in the model.26 In the Osteoporosis Fractures in Men (MrOS) study, men with obesity were five times more likely to experience hip fracture than normal-weight men, after controlling for BMD.26 Our group has shown that despite a positive correlation between BMD and BMI, a negative correlation between BMD and adiposity exists, ie, participants in the highest tertile of body fat had the lowest BMD.27 The physiological mechanism may be that with increase in body fat, there is an increase in systemic inflammation leading to frailty and poor bone quality, thus contributing to an increased risk for fractures.28

We reported in a different trial that weight loss by itself was associated with improvement in physical function and amelioration of frailty whether exercise was part of the intervention or not.9 Also in that different trial, exercise (combined aerobic and resistance) with weight loss attenuated but did not eliminate the weight loss–induced bone loss.9, 10 This led us to investigate whether the type of exercise affects the bone response to lifestyle intervention in the current trial. We hypothesized that due to the potential interference between the specific adaptations to aerobic and resistance exercise when performed together,11, 13, 14 resistance exercise might be more effective than aerobic exercise or combined aerobic and resistance exercise in preserving BMD during weight loss. However, while some previous studies reported interference between the cardiovascular adaptations to aerobic exercise and strength adaptations to resistance exercise,11, 13, 14 we were able to show that combined aerobic and resistance exercise improved VO2peak to the same extent as aerobic exercise alone and strength to the same extent as resistance exercise alone.15 Therefore, combined aerobic and resistance exercise resulted in the greatest improvement among the interventions in physical function and amelioration of frailty in older adults with obesity in the current trial.15 In this secondary analysis focusing on bone outcomes with similar degree of weight loss, the aerobic group had a higher degree of weight loss–induced bone loss at the total hip and femoral neck compared with resistance and combination groups who had attenuated bone loss in these important skeletal sites. Therefore, the combination of aerobic and resistance exercises not only resulted in optimum improvement in physical function compared with aerobic alone or resistance alone but, importantly, also better preservation of BMD (compared with aerobic alone), suggesting the absence of interference between the bone adaptations to specific exercises during combined exercise. These findings highlight the importance of recommending weight loss plus combined aerobic and resistance exercise for the overall health of older adults with obesity.

The modest loss of BMD at the total hip in the combined group, which was not significantly different from that of the resistance only group, supports the equivalence of both exercise training interventions on weight loss–induced bone loss. This finding is consistent across other sites of the proximal femur such as the femoral neck, trochanter, and intertrochanter. The equivalence of resistance and combined groups is also supported by the bone marker data showing no difference in changes in bone markers between the resistance and combined group, ie, either a slight increase or reduction compared with the significant increase in bone resorption and formation markers in the aerobic group. The aerobic group lost ~2.5% of total hip BMD, which is comparable to the bone loss we previously reported in the weight loss alone group,9, 10 indicating that aerobic exercise likely had little effect in preserving BMD during weight loss therapy. In fact, the degree of hip BMD loss we observed in the aerobic group is also comparable with the results of a meta-analyses on the effect of diet-induced weight loss in the general overweight or obese population.29 Similar to our current findings in which resistance exercise attenuated but did not prevent the bone loss associated with weight loss, a recent study likewise showed that exercise was unable to prevent weight loss–induced bone loss at the hip in older adults with obesity undergoing community-based exercises for 30 months; however, there was attenuated bone loss among those assigned to resistance compared with aerobic training.30 Unlike our study, there was no combined aerobic and resistance exercise group, which is the recommended exercise program by the American College of Sports Medicine and American Heart Association for maintaining overall health for adults.31

It is important to note that at 6 months, bone turnover had not returned toward baseline compared with our prior weight loss studies of longer duration.32, 33 This is perhaps because the current study was only for 6 months, not allowing enough time for bone turnover markers to stabilize. In addition, we did not find differences in the changes in sclerostin levels, which we previously reported to increase with weight loss.34 However, the levels only increased in the group that had weight loss from diet alone but not in the group that had weight loss plus exercise.34 In the current study, all weight loss groups also participated in an exercise program consisting of either aerobic or resistance or both.

Consistent with our previous report in a different trial,10 there appears to be a differential effect on bone loss, ie site-specific, and only found at weight-bearing hip but not at non-weight-bearing radius or lumbar spine. As mechanical loading is important for maintaining bone balance by stimulating bone formation through increasing proliferation of osteoblasts and osteocytes,35 unloading is associated with bone loss, which could be more pronounced in the highly loaded bone such as the hip. The femurs are the main body parts used for locomotion and the loss of lean mass with weight loss could have a greater impact on bones subjected to constant loading by reduced mechanical strain on the hip, which was not experienced by the relatively less loaded bone such as the spine.

The precise mechanisms underlying bone loss with weight loss are unknown and are thought to be multifactorial, including mechanical loading and hormonal factors.36-38 Indeed, in our participants undergoing weight loss plus different exercise types and supplemented with calcium and vitamin D, the decline in whole body mass was the most important independent predictor of the decrease in hip BMD (β = 0.54) in the stepwise multiple regression analyses. Thus, weight loss in our participants likely contributed to bone loss by decreasing mechanical strain on the skeleton,39 consistent with our findings of bone loss especially at the weight-bearing hip. On the other hand, among the hormone levels that changed during the interventions, the decline in leptin secreted from adipocytes was the additional independent determinant of the decline in hip BMD (β = 0.25). Leptin may act peripherally by increasing osteoprotegerin levels leading to binding of RANKL, resulting in inhibition of osteoclastogenesis.40, 41 Thus, the elevated levels of leptin in individuals with obesity are osteoprotective and the reduction of leptin levels induced by weight loss in our participants contributed to the increase in bone resorption, highlighting important fat-bone interactions during weight loss therapy.42 Given that the final model in the multiple regression analyses explained only 20% of the variance in hip BMD, further studies are needed to clarify the mechanisms for the weight loss–induced bone loss in older adults with obesity.

The strengths of our study include the randomized controlled trial study design, the high adherence rate to our intensive lifestyle interventions, and the comprehensive assessments of BMD and bone metabolism. The limitations include that we did not assess bone quality, for example, using finite element analyses to measure bone strength43 and high-resolution peripheral quantitated computed tomography to measure bone microarchitecture.44 It is possible that weight loss plus exercise training in older men and women with obesity could preserve or improve bone quality despite the decline in BMD.6 Another limitation is that the short duration of the study precluded the assessment of the effect of our interventions on falls and fractures. However, a study with fracture outcomes would need a very large sample size and is therefore unlikely. Future studies might assess whether the aggregate benefits of weight loss plus combined aerobic and resistance exercise on physical function and possibly on bone quality could lower the risk of fractures despite the decline in hip BMD. Our study was not large enough to examine potential sex differences in BMD responses. We controlled for the effect of sex by including it as a covariate in the mixed-model ANOVA.

In conclusion, the results of the present study indicate that compared with aerobic exercise, resistance and combined aerobic and resistance exercise are associated with less weight loss–induced decrease in hip BMD and less weight loss–induced increase in bone turnover. Therefore, weight loss plus combined aerobic and resistance exercise not only optimally improves physical function15 but also protects against bone loss during weight loss therapy of the increasing number of older adults with obesity in the US and other more developed countries with similar obesity profiles to the US.

Disclosures

The authors declare no conflicts of interest.

Acknowledgments

We thank the participants for their cooperation, Kenneth Fowler for study coordination, Brandy Martinez and Erik Faria for exercise training, and Ronni Farris and Reed Vawter for weight loss therapy. Supported by grants R01-AG031176, UL1-TR000041, P30-DK020579, and CX-00096. We also thank the members of the Alkek Foundation for their support. The findings reported in this article are the result of work supported with resources and the use of facilities at the New Mexico VA Health Care System and Michael E DeBakey VA Medical Center. The content is solely the responsibility of the authors and does not necessarily represent the views of the National Institutes of Health and/or of the US Department of Veterans Affairs or the United States Government.

Authors’ roles: Study design: DTV, RA, and CQ. Study conduct: DTV, LA, RA, and CQ. Data analysis: DTV, LA, RA, NN, DW, and CQ: Data interpretation: DTV, LA, RA, NN, DW, and CQ. Revising manuscript content: DTV, LA, RA, NN, DW, and CQ. Approving final version of manuscript: DTV, LA, RA, NN, DW, and CQ. DTV takes responsibility for the integrity of the data analysis.