Effects of resistance and aerobic exercise on physical function, bone mineral density, OPG and RANKL in older women

Abstract

This study compared the effects of a resistance training protocol and a moderate-impact aerobic training protocol on bone mineral density (BMD), physical ability, serum osteoprotegerin (OPG), and receptor activator of nuclear factor kappa B ligand (RANKL) levels. Seventy-one older women were randomly assigned to resistance exercise (RE), aerobic exercise (AE) or a control group (CON). Both interventions were conducted 3 times per week for 8 months. Outcome measures included proximal femur BMD, muscle strength, balance, body composition, serum OPG, and RANKL levels. Potential confounding variables included dietary intake, accelerometer-based physical activity (PA), and molecularly defined lactase nonpersistence. After 8 months, only RE group exhibited increases in BMD at the trochanter (2.9%) and total hip (1.5%), and improved body composition. Both RE and AE groups improved balance. No significant changes were observed in OPG and RANKL levels, and OPG/RANKL ratio. Lactase nonpersistence was not associated with BMD changes. No group differences were observed in baseline values or change in dietary intakes and daily PA. Data suggest that 8 months of RE may be more effective than AE for inducing favourable changes in BMD and muscle strength, whilst both interventions demonstrate to protect against the functional balance control that is strongly related to fall risk.

Research highlights

► RE increase BMD, muscle strength and balance. ► AE does not enhance BMD but improve balance. ► Serum OPG and RANKL levels do not change after 8 months of exercise.

Abbreviations

AE
aerobic exercise
ANOVA
one-way analysis of variance
AP
anterior–posterior
BMD
bone mineral density
CON
control group
COP
centre of pressure
CV
coefficient of variation
EA
elliptical area
ELISA
enzyme-linked immunosorbent assay
KE
knee extension
KF
knee flexion
ML
medial–lateral
MVPA
moderate to vigorous physical activity
OLS
one-leg stance
OPG
osteoprotegerin
PA
physical activity
RANKL
receptor activator of nuclear factor kappa B ligand
RE
resistance exercise
8-ft UG test
8-foot Up and Go Test

Keywords

Bone mass
Fall risk
Exercise
Age
Biomarkers

1. Introduction

Low bone mass and an increased risk of fracture rank high amongst the serious clinical problems faced by older adults. As such, the impact of this age-related condition extends beyond the significance of an increased prevalence, as severe individual and economic consequences of injurious falls can have profound implications for subsequent health, morbidity, functional independence, life quality and increased mortality of older people (Lane, 2006).

Many risk factors for falls have been identified, and increasing evidence has suggested that fall reduction programmes that involve systematic fall risk assessment and targeted interventions, exercise programmes and environmental and hazard-reduction programmes are the optimal approaches (Rubenstein, 2006). Importantly, lower extremity weakness as well as power and balance impairment is frequently reported as a risk factor that has the potential to be influenced with appropriate exercise prescription (Sherrington et al., 2008). As exercise may be an important way to reduce the incidence of this problem, recent systematic reviews have consistently shown that exercise can be used as a stand-alone intervention for fall prevention (Sherrington et al., 2008, Gillespie et al., 2009). Despite the positive effects seen in programmes that include strengthening, balance, and/or endurance training (Sherrington et al., 2008), and the benefit of aerobic exercise training (AE) or resistance exercise training (RE) as single interventions remains controversial, due mostly to the paucity of data. In fact, both types of activity are commonly prescribed and widely accepted, based on the variety of favourable adaptations that AE and RE in isolation can elicit in older adults (Chodzko-Zajko et al., 2009). Although the evidence that supports the notion that older adults can significantly increase their muscle strength and power after RE are overwhelming (Chodzko-Zajko et al., 2009), currently published data have not consistently shown that the use of RE alone improves balance in this population (Orr et al., 2008). Nevertheless, AE has been highlighted as the exercise regimen of choice for inducing several cardiovascular adaptations (Chodzko-Zajko et al., 2009); however its effectiveness in increasing muscle strength and balance is still under discussion.

In addition to its role in the prevention of falls, exercise is also considered to play a crucial role in bone modelling and remodelling (Borer, 2005). Animal studies have evaluated osteogenic responses to several exercise interventions, including running, swimming, jumping, climbing, and resistance training (Warner et al., 2006, Mori et al., 2003, Notomi et al., 2000). The results suggest that the exercise-induced osteogenic effect is site specific and dependent on the type of exercise and load applied to the bones. Most of the literature, usually based in animal models, published thus far supports the notion that greater strain magnitudes and unusual strain distributions provide the most effective stimuli for bone formation (Bailey and Brooke-Wavell, 2008). In support of this, RE has been recognised to be effective in stimulating an osteogenic response and elevating bone mineral density (BMD) in both young and old adults (Ryan et al., 2004). However, the isolated effects of AE on bone mass in older adults have been poorly investigated. Evidence regarding the effectiveness of this type of exercise in counteracting age-related declines in BMD has been controversial (Brooke-Wavell et al., 2001, Bonaiuti et al., 2002). Notably, the data suggest that the skeletal response to exercise is altered with age (Lanyon and Skerry, 2001). Actually, mechanical loading forces become less effective in eliciting an osteogenic effect with increasing age, suggesting a progressive loss of bone sensitivity to chemical and physical signals (Rubin et al., 1992). In addition, basic and clinical studies have established a consistent relationship between the osteoprotegerin (OPG)/receptor activator of the nuclear factor-kB (RANK)/RANK ligand (RANKL) system and skeletal health due to its critical role in bone remodelling (Boyce and Xing, 2008). OPG has an osteo-protective role in humans, protecting bones from excessive bone resorption via binding to RANKL and preventing it from binding to RANK (Boyce and Xing, 2008). In vitro and in vivo experiments have shown that mechanical stimulation can inhibit osteoclast formation and activity by changing the OPG/RANKL ratio in favour of OPG (Saunders et al., 2006, Rubin et al., 2003). However, the association between serum OPG/RANKL and the incidence of bone fractures and BMD has been inconsistent (Browner et al., 2001, Jorgensen et al., 2004, Stern et al., 2007). Although there is a great deal of basic research currently addressing the RANKL/RANK signalling pathway, less is known regarding how prolonged exercise may influence the release of soluble factors and if this reflects what is happening in bones. The present study aimed to compare the alterations in key factors associated with fracture risk, namely BMD, muscle strength and balance. We also tested whether there are changes in serum levels of OPG, RANKL and their ratios alter after an 8-month exercise training programme. The results obtained would contribute to a better understanding of how different exercise interventions can interact with the physiological systems associated with bone turnover and remodelling.

2. Materials and methods

2.1. Subjects and experimental design

Subjects were recruited through advertisements in Porto area newspapers for participation in this university-based study. A total of 90 Caucasian older women volunteered to participate in the study. The eligible subject pool was restricted to older women with the following characteristics: free of hormone therapy use for at least two years, aged 60–95 years, community-dwelling status, not engaged in regular exercise training in the last year, lack of use of any medication known to affect bone metabolism or to harm balance, postural stability and functional autonomy; and lack of diagnosed or self-reported neurologic disorders, disorders of the vestibular system, and cardiovascular, pulmonary, metabolic, renal, hepatic, or orthopaedic medical conditions that contraindicate participation in exercise. On the initial screening visit, all participants received a complete explanation of the purpose, risks, and procedures of the investigation and, after signing a written consent form, the past medical history and current medications of the subjects were determined. Nineteen subjects were excluded due to medical reasons, inability to be contacted or lack of willingness to participate in the study. Seventy-one subjects were randomised into one of three groups: resistance exercise training (n = 23, RE), aerobic exercise training (n = 24, AE), and a control group (n = 24, CON), using computer-generated random numbers. The technical assistant who provided the randomisation was not involved in the screening, testing, or training procedures. Participants were instructed to continue their daily routines and to refrain from changing their physical activity (PA) levels during the course of the experiment.

The baseline characteristics of the participants are given in Table 1. The study was carried out in full compliance with the Helsinki Declaration, and all methods and procedures were approved by the institutional review board.

Table 1. Baseline characteristics of the sample.

VariableRE group (n = 23)AE group (n = 24)CON group (n = 24)p-valuea
Age, y67.3 ± 5.270.3 ± 5.567.9 ± 5.90.17
Age at menarche, y13.3 ± 1.213.7 ± 1.412.8 ± 1.30.08
Age at menopause, y47.6 ± 3.548.3 ± 5.248.7 ± 3.60.86
Married, %63.455.073.70.48
Education, y9.1 ± 4.98.4 ± 3.47.4 ± 4.30.54
BMI, kg/m228.8 ± 4.627.5 ± 3.828.1 ± 3.50.61
Total body fat, %38.8 ± 4.439.2 ± 4.538.4 ± 4.60.84
Waist circumference, cm93.0 ± 10.589.1 ± 9.591.4 ± 8.70.38
Number of routine medications1.8 ± 1.82.7 ± 2.02.6 ± 1.60.42
History of, %
 Hypertension45.540.063.20.33
 Diabetes mellitus9.110.015.80.81
 Arthritis9.110.015.80.81
 Cigarette smoking18.210.010.50.77
Taking lipid-lowering agents, %27.35.021.10.20
Lactase persistence, %42.968.445.50.18
Energy intake, kcal/day1485.7 ± 360.31368.2 ± 241.41561.0 ± 334.10.13
Protein intake, g/day65.6 ± 13.964.5 ± 15.469.5 ± 15.20.52
Calcium intake, mg/day714.5 ± 358.4608.5 ± 248.8636.9 ± 280.90.54
Phosphorus intake, mg/day1013.5 ± 307.8965.4 ± 237.3979.2 ± 273.00.86
Vitamin D intake, μg/d1.8 ± 1.72.3 ± 1.42.0 ± 1.90.39
Coffee intake, mL/day66.5 ± 53.548.7 ± 41.543.7 ± 50.90.37
Time spend in MVPA, min/day93.2 ± 26.386.2 ± 32.178.8 ± 40.50.43
Daily count min− 1412.6 ± 117.9355.6 ± 113.0360.8 ± 161.60.39
Daily step count9000.9 ± 2544.68852.7 ± 2306.47905.1 ± 3323.10.40
Femoral neck, T-score− 1.6 ± 0.7− 1.8 ± 1.0− 1.6 ± 0.60.66
Total femur, T-score− 0.9 ± 1.0− 0.9 ± 1.0− 1.0 ± 0.70.91
a

One-way ANOVA for continuous variables; Chi-square test for categorical variables. RE: resistance exercise; AE: aerobic exercise; CON: control.

2.2. Measurements

Participants were tested on two occasions: the first assessment was conducted prior to the beginning of training and the second evaluation took place after eight months of training.

2.2.1. Bone and body composition

BMD was measured using dual-energy X-ray absorptiometry (DXA) (QDR 4500A, Hologic, Bedford, MA) at the proximal femur on the nondominant side using standard protocols. To minimise interobserver variation, the same investigator carried out all analyses. Bone phantoms were scanned daily, and coefficients of variation (CV) were verified before and during the experimental period to ensure assessment reliability.

Total body scans were taken using the same DXA instrument. All scans were performed by the same technician using standard procedures, as described in the Hologic user's manual. Scans were analysed for total lean mass, fat free mass and percent body fat mass. Fat free mass consists of lean mass and bone mineral mass. Lean mass (i.e., bone-free fat free mass) was included into the analysis as a surrogate measure of muscle mass. Because the exercise protocols were designed to improve bone tissue, fat free mass was included as a comprehensive measure expressing the pooled change in bone and lean mass.

To test the precision of our DXA scanner, repeated scans were performed on 15 healthy older adults. Each individual underwent three consecutive total-body and hip scans with repositioning. The CV (standard deviation/mean) for repeated measurements was 0.8% for total body BMD, 0.9% for femoral neck BMD and 1.1% for total femur BMD. CV-values for percent body fat, fat free mass, and lean body mass were 3.1%, 2.8%, and 1.1%, respectively.

Height and body mass were recorded using a portable stadiometer and balance weighing scales, respectively. Body mass index (BMI) was calculated using the standard formula: mass (kg)/height2 (m).

2.2.2. Muscular strength

The dynamic concentric muscle strength of the lower extremities, namely the knee flexion (KF) and extension (KE) muscle groups, was measured on an isokinetic dynamometer (Biodex System 4 Pro; Biodex, Shirley, NY). Strength measurements were carried out, in accordance with the manufacturer's instructions for KE/KF, at two angular velocities, 60°/s (1.05 rad s− 1) and 180°/s (3.14 rad s− 1). Each participant, after familiarisation with the machine, performed five maximal efforts at 180°/s and three at 60°/s with two minutes of rest between tests. The dynamometer angle reading was calibrated to the anatomic joint angle measured by a goniometer, and gravity corrections to torque were based on leg weight at 0° and calculated later by the equipment software. Prior to testing, subjects performed a five minute warm-up on a bicycle ergometer (Bike‐Max; Tectrix, Irvine, CA) at 45–60 W. During the test, participants were verbally encouraged to exert maximal muscular force. Peak torque, represented as a percentage normalised to body weight, was used for the statistical analyses.

2.2.3. Balance and mobility performance measures

Each subject performed two balance tests. Mobility/dynamic balance was assessed using the 8-foot Up and Go Test (8-ft UG test) (Rikli and Jones, 1999) and static balance was measured using the one-leg stance (OLS) (Bohannon, 1994). Before starting the tests, participants remained seated and rested for five minutes. In the 8-ft UG test, the score corresponded to the shortest time to rise from a seated position, walk 2.44 m (8 ft), turn, and return to the seated position, measured to the nearest 1/10th s. The OLS test involved standing upright as still as possible in a unassisted unipedal stance (on the nondominant leg) on a 40–60 cm force platform (Force Plate AM 4060–15; Bertec, Columbus, OH) with eyes open, head erect, and arms by the side of the trunk.

The OLS was timed in seconds from the time one foot was lifted from the floor to when it touched the ground or the standing leg. A longer time indicated better balance; the maximum time was set at 45 s. Two attempts were allowed, with one minute of rest between, and the best performance was used for force-platform-based analysis.

The signals from the force platform were sampled at 500 Hz. We used a personal computer to collect the data with the customised AcqKnowledge-based software (AcqKnowledge 3.9.1; Biopac, Goleta, CA). The data analysis was performed using MATLAB software (MATLAB 7.0; MathWorks, Natick, MA). Data from horizontal forces (Fy and Fx) and centre of pressure (COP) time-series were low-pass filtered with a zero-lag, fourth-order Butterworth filter with a cut-off frequency of 10 Hz.

The outcome variables were anterior–posterior (AP) and medial–lateral (ML) mean velocity (cm s− 1) of the COP; the elliptical area (EA) was calculated using the equation: √2σy × √2σx. Mean velocity was determined by dividing the total distance along the signal trajectory by the total recording time.

2.2.4. Blood sampling and serum measurements

Fasting venous blood samples were drawn between 7.30 and 9.30 a.m. After collection, blood samples were collected in tubes containing EDTA, and serum samples were clotted at room temperature for 90 min and were then centrifuged for 10 min at 1000 ×g. Samples were aliquoted and stored at − 80 °C until analysis. OPG and RANKL were determined by a commercial sandwich enzyme-linked immunosorbent assay (ELISA) according to the protocol of the manufacturer (Immunodiagnostic Systems Ltd, Boldon, UK and Cusabio Biotech, China, respectively). The same serum samples used for RANKL measurements were used for OPG measurements, and the assay was performed blind to the subject group. The detection limit was 0.140 pmol/L for the OPG assay and < 31.2 pg/mL for RANKL assay, with an intra- and inter-assay CV of < 10%.

2.2.5. Lifestyle behaviours and clinical status

A baseline self-administered questionnaire to assess the impact of present and past lifestyle choices was completed by interview to avoid misinterpretation of items and/or skipping of questions. The questionnaire included information regarding education; marital status; fall and fracture history; medical history; current medical conditions; medication use; current and past PA; age at menarche; menopause status; current and previous use of hormone replacement therapy; past dietary habits, including calcium intake; and current and past smoking.

2.2.6. Daily PA

The Actigraph GT1M accelerometer (Manufacturing Technology, Fort Walton Beach, FL) was used as an objective measure of daily PA, using a 15 second measurement interval (epoch). All participants agreed to wear an accelerometer for seven consecutive days and were instructed to wear the device over their right hip using an adjustable nylon belt. Exceptions included time spent sleeping and showering. Participants were asked to maintain usual activities. For data to be included in the analyses, participants were required to wear the accelerometer for at least four of the seven days. For both test periods (pre- and post-trainings), four files were corrupt, and six files had only two valid days. Those ten participants were contacted and agreed to wear the accelerometer again for seven days (one week later than the rest of the group). In total, pre- and post-training data from all participants were included in the analysis (90 files with seven valid days, 7 files with 6 valid days, and 11 files with 5 valid days).

The cut point was set at counts per minute ≥ 1041 (moderate to vigorous PA [MVPA]) which corresponded to a mean VO2 of 13 mL kg− 1 min− 1, based on the counts associated with a reference activity, which was walking at 3.2 km/h (Copeland and Esliger, 2009). The average daily moderate to vigorous PA, number of steps, and daily activity counts per minute (cpm) were analysed.

2.2.7. Nutritional assessment

Nutritional status was assessed using 4-day diet records over three weekdays and one weekend day. To ensure standardisation of the dietary records, a dietician gave individual instruction to the subjects concerning how to fill out the diet records and assess food serving sizes. Diet records were analysed using Food Processor Plus® (ESHA Research, Salem, OR), which uses the table of food components from the U.S. Department of Agriculture. Some traditional Portuguese dishes were added based on the table of Portuguese food composition. Total caloric, protein, calcium, phosphorous, vitamin D, and caffeine intakes were compared between the RE, AE and CON groups.

2.2.8. Lactose persistence status

The lactose persistence mutation, C/T − 13910, was genotyped by direct sequencing. A 359-bp fragment containing all mentioned mutations and located in the intron 13 of the MCM6 gene was amplified using primers 5′-GCAGGGCTCAAAGAACAATC-3′ (forward) and 5′-TGTTGCATGTTTTTAATCTTTGG-3′ (reverse). PCRs reactions contained 0.5 μM of each primer, 0.2 mM of each deoxynucleotide triphosphate (dNTP), 750 mM Tris–HCI (pH 8.8 at 25 °C), 200 mM (NH4)2SO4, 0.1% (v/v) Tween 20, 1.5 mM MgCl2 and 1 U Taq polymerase. The PCR profile consisted of the following: 94 °C for five minutes, 35 cycles of 94 °C for one minute, 58 °C for one minute and 72 °C for one minute, followed by 20 min of extension at 72 °C.

Sequencing reactions were carried out using the ABI Big Dye v3.1 Ready Reaction Kit and using the protocol specified by the manufacturer (Applied Biosystems, Foster City, CA). Products were run on an ABI PRISM 3130 × 1 sequencer and analysed in the ABI PRISM 3130 × 1 Genetic Analyser software (Applied Biosystems). The resulting chromatograms were inspected for the presence/absence of lactase mutations using the MEGA4.0 software (www.megasoftware.net) (Tamura et al., 2007).

DNA was obtained from buccal swabs using standard extraction methods.

2.3. Exercise protocol

2.3.1. Aerobic exercise

The AE group completed a 32-week endurance exercise training programme consisting of three sessions per week, with at least one day of rest between sessions. Each session lasted approximately 60 min, and all sessions were accompanied by appropriate music relevant to the required activity and participants' age. The exercise training consisted of stretching and warm-up exercises (10–15 min), dynamic aerobic activities (35–40 min) involving stepping, skipping, graded walking, jogging, dancing, aerobics and step choreographies, and cool-down/relaxation exercises (10 min). During the first six weeks, moderate intensity strength exercises were performed concentrically and eccentrically for approximately ten minutes for the hip flexors, extensors, and abductors; knee flexors and extensors; and ankle dorsiflexors and plantar flexors using body weight to ensure proper muscular resistance and to sustain the increments in training intensity. The initial exercise intensity was set at 50% to 60% of the subjects' heart rate reserve for the first two months; the target heart rate during exercise was continuously monitored by heart rate monitors (Polar Vantage XL, Polar Electro Inc., Port Washington, NY), and the rate of perceived exertion was assessed using Borg's 10-point psychometric scale (Borg et al., 1987). Exercise intensity was gradually increased from 65% to 85% of the heart rate reserve as adapted to the individual. Each session was led by three physical education instructors who specialised in PA for older adults and was supervised by the researchers.

2.3.2. Resistance exercise

RE sessions were performed three times per week on nonconsecutive days, with each session lasting approximately 60 min over a period of 32 weeks. All training sessions were conducted at University of Porto, Faculty of Sport Facilities and were supervised by three research assistants who were responsible for warm-up, cool down, and stretching exercises; the monitoring of correct lifting form; the appropriate amount of exercise and rest intervals; the maintenance of daily exercise logs; and the progression of the exercises. Subjects were also encouraged to exercise with a training partner to provide additional motivation. Each training session involved the following: (1) a standardised warm-up period (8–10 min) on a bicycle ergometer (Bike‐Max; Tectrix, Irvine, CA) and/or rowing ergometer (Concept II, Morrisville, VR) at low intensity and some stretching exercises; (2) specific resistance training period (30–40 min); and (3) a cool-down period (5–10 min) that included walking and stretching exercises. The RE protocol aimed to develop muscle mass and strength in the following muscle groups: (1) quadriceps (leg press and leg extension), (2) hamstrings (seated leg curl), (3) gluteal (hip abduction), (4) trunk and arms (double chest press, lateral raise and overhead press), and (5) abdominal wall (abdominal machine). Subjects exercised on variable resistance machines (Nautilus Sports/Medical Industries, Independence, VA). To minimise fatigue, the exercises for the upper/lower parts of the body were performed in a non-consecutive way, with a rest period of approximately two minutes between each set. Each repetition lasted three to six seconds, involving a period of at least two minutes between the two sets of 10–12 repetitions at 60–70% of 1RM. Training intensity was gradually increased during the first four weeks. Participants underwent a 2-week familiarisation period with the equipment and the exercises. The intensity of the training stimulus was initially set at 50% to 60% of one-repetition maximum (1RM), as determined at week 2, with a work range of two sets of 10 to 15 repetitions. Subjects then progressed from 75% to 80% of 1RM at a work range of six to eight repetitions (two sets) and remained at this level until the end of the programme. Training was continuously monitored by heart rate monitors (Polar Vantage XL, Polar Electro Inc., Port Washington, NY) and ratings of perceived exertion (Borg's 10-point psychometric scale) (Borg et al., 1987). 1RM tests were performed every two weeks for the first month and then every four weeks until the end of the programme. Between these tests, the load was increased for those subjects who were able to easily complete 12 or more repetitions for both sets.

Exercise compliance was defined as the number of exercise sessions reported divided by the number of maximum exercise sessions possible.

2.4. Statistical analysis

All statistical analyses were performed using PASW Statistics (version 18; SPSS, Inc., Chicago, IL) for Windows with a significance level of 0.05. Data were checked for distribution, and the means ± SD were calculated. Primary outcomes were changes from baseline in response to both 8-month interventions in balance, muscle strength, BMD and serum level of OPG and RANKL. Secondary outcomes included 8-month changes from baseline in dietary intake, daily PA, body composition (BMI, waist circumference, fat, fat-free mass, and lean mass), and the presence or absence of the lactase mutations. The results were analysed on an intention-to-treat basis, and missing data due to lack of follow-up (the method assumed data were missing at random) were replaced using the process of multiple imputation. This method has been adapted to the analysis of longitudinal data (Mazumdar et al., 1999). Potential differences amongst groups in baseline measurements were evaluated using one-way analysis of variance (ANOVA). Chi-squared tests were used for between-group comparisons of categorical variables at baseline. Pearson correlations were used to determine the relationship of potential confounding variables (e.g., lactase persistence, dietary intake, PA change, and fat mass change) with primary outcomes. Such confounding variables were then entered as covariates in the analysis of variance model as indicated. The delta percentage was calculated with the standard formula: % change = [(posttest score − pretest score) / pre-test score] × 100.

A two-way (group and time) factorial ANOVA, with repeated measures on one factor (time), was performed for differences in main effects and time by group interactions for each dependent variable. Main effects were considered when interactions were not significant. When significant interactions were found, Bonferroni post hoc tests were used to determine significant differences amongst mean values.

A power analysis based on a formulation of 75% power, an effect size of 0.5 for overall muscle strength, balance and BMD from previous studies, and a significance level of 0.05 for a one-tailed test deemed that a sample of 23 per group was sufficient to address the research questions.

3. Results

3.1. Recruitment

Of the 71 women aged 69.0 ± 5.3 (range 61–83) who underwent the initial assessment and randomisation, 44 were randomised to the following three groups: RE, n = 15; AE, n = 19; and CON, n = 20. One participant discontinued the intervention because of surgery, and five participants discontinued due to medical issues unrelated to the intervention. Two participants left the study due to unwillingness to participate, three due to loss of interest and six to personal reasons. As expected, dropout rates were higher in the exercise groups (8 RE, 5 AE) than the CON group (n = 4) because of the time commitment. However, no differences (p = 0.315) in dropout rates were observed between groups. Fig. 1 shows the number of participants at each stage of the study.

  1. Download : Download full-size image

Fig. 1. Flow of participants through the study.

3.2. Subject characteristics

Demographics and descriptive parameters of all groups are listed in Table 1. Of the participants, 73% of the participants were overweight, most of them had hypertension, and a small proportion had a history of cigarette smoking. On average, participants obtained 85 min of MVPA per day. The molecularly defined prevalence of lactase persistence (TT/TC genotypes) was similar for all groups. The prevalence of the TT and CT genotypes of the 13910 C/T polymorphism was 25.0% and 28.6%, respectively. There were no significant group differences in any baseline characteristic.

3.3. Compliance with intervention and adverse events

One-hundred percent compliance to the exercise sessions was set at 96 training sessions. Excluding dropouts, mean compliance to the RE sessions was 78.4% (61.6–95.9%), and for AE training, the mean compliance was 77.7% (64.2–96.8%). There were no exercise- or assessment-related (pre- and posttraining) adverse events.

In comparison to individuals who completed the trial, those who failed to provide follow-up data had no significant differences in any baseline measurements, including age, body weight, daily MVPA levels, strength, balance, or BMD.

3.4. Dietary intake

Total energy intake was similar amongst the groups at baseline and during the period of intervention. Energy intake was 1473 ± 318 kcal/day at baseline and 1499 ± 302 at eight months (p > 0.05 for all group changes). No group differences were apparent in baseline values (Table 1) or change in dietary protein, phosphorus, caffeine, calcium, and vitamin D intake in response to the interventions (data not shown). No significant difference in mean daily total calcium intake derived from milk and dairy products was evident amongst those with and without lactase persistence within each group.

3.5. Daily PA

No significant changes in MVPA level were observed at eight months. There was no significant interactive (p = 0.417) or main effect of group (p = 0.214) and time (p = 0.171) on changes in PA. Changes in MVPA were not related to changes in the primary outcomes.

3.6. Changes in body composition

No significant interaction occurred for BMI or waist circumference in response to exercise intervention. There was a main effect of time (p = 0.039) on waist circumference. Interactions were observed for lean mass (p = 0.026), fat free mass (p = 0.030), and percent fat mass (p = 0.028; Table 2), such that only the RE group significantly increased lean and fat free mass and decreased percentage fat mass, whereas no significant changes were observed for the AE and CON groups.

Table 2. Pre- and post-training values for body composition variables.

Resistance exercise groupAerobic exercise groupControl groupp (group)p (time)p (interaction)
VariablePre-trainingPost-trainingPre-trainingPost-trainingPre-trainingPost-training
BMI, kg/m228.8 ± 4.628.2 ± 3.927.5 ± 3.827.5 ± 3.328.1 ± 3.527.3 ± 2.00.6480.1070.377
WC, cm93.0 ± 10.591.2 ± 8.289.1 ± 9.586.7 ± 6.891.4 ± 8.790.8 ± 10.70.2930.0390.580
Lean mass, kg41.8 ± 8.644.6 ± 8.6a, b, c37.3 ± 5.237.2 ± 5.239.4 ± 5.038.2 ± 3.20.0060.3860.026
Fat free mass, kg43.6 ± 8.946.6 ± 9.1a, b, c39.0 ± 5.438.9 ± 5.641.1 ± 5.139.9 ± 3.40.0060.3780.030
Fat mass, %38.8 ± 4.435.2 ± 5.5a39.2 ± 4.538.4 ± 3.838.4 ± 4.637.8 ± 3.70.4020.0010.028

BMI: body mass index; WC: waist circumference.

a

Indicates a significant intra-group difference, p < 0.05.

b

Indicates a significant difference from AE Group at post test, p < 0.05.

c

Indicates a significant difference from CON Group at post test, p < 0.05.

3.7. Changes in balance and muscle strength

No significant difference between groups for the variables was apparent at baseline. There were significant interactions between group and time on all measurements of balance and strength (Table 3). Accordingly, both RE and AE groups improved the time to perform both balance tests, and a significant difference for post-training results was evident between exercise intervention groups and the CON group for EA and velocity values for ML-direction. However, only the AE group significantly decreased the mean velocity of the COP displacement for AP-direction. A significant decrease in 8 ft UG performance was observed for the control group; the trend indicated a decline in all balance and strength variables. Regarding muscle strength, only the RE group significantly increased their maximal KE and KF torques at both speeds.

Table 3. Pre- and post-training values for muscle strength, dynamic and static balance.

Resistance exercise groupAerobic exercise groupControl group
VariablePre-trainingPost-trainingPre-trainingPost-trainingPre-trainingPost-training
p (group)p (time)p (interaction)
8 ft UG, s5.5 ± 0.54.9 ± 0.3a5.9 ± 0.95.1 ± 0.6a6.0 ± 0.86.3 ± 1.2a< 0.001< 0.001< 0.001
OLS, s26.3 ± 13.231.7 ± 12.828.8 ± 14.932.9 ± 9.5b26.9 ± 16.222.3 ± 13.60.2210.3270.028
EA, cm27.4 ± 4.83.3 ± 1.0a, b7.2 ± 4.33.3 ± 1.3a, b7.2 ± 4.47.4 ± 4.40.063< 0.0010.001
AP velocity, cm s− 14.0 ± 1.13.5 ± 0.83.7 ± 1.13.0 ± 0.8a, b4.0 ± 1.14.4 ± 1.50.0250.0270.003
ML velocity, cm s− 14.7 ± 2.13.3 ± 0.8a, b4.3 ± 1.92.9 ± 0.8a, b4.8 ± 2.14.6 ± 2.40.085< 0.0010.041
KE PT/BW 180°/s, %76.2 ± 16.090.5 ± 15.1a84.5 ± 21.581.2 ± 24.981.3 ± 18.679.1 ± 19.30.8190.2440.013
KF PT/BW 180°/s, %50.5 ± 18.361.6 ± 14.9a, b47.2 ± 12.851.7 ± 11.850.6 ± 15.049.9 ± 11.10.2110.0100.047
KE PT/BW 60°/s, %123.0 ± 29.8140.7 ± 31.1a137.7 ± 36.5132.5 ± 28.2134.4 ± 27.3129.6 ± 28.60.9150.4240.010
KF PT/BW 60°/s, %74.6 ± 23.494.4 ± 24.5a, b71.8 ± 19.077.1 ± 19.368.6 ± 20.066.6 ± 20.20.0260.0030.003

8 ft UG: 8-foot Up and Go Test; OLS: one leg stance; EA: elliptical area; AP: anterior–posterior; ML: medial–lateral; KE: knee extension; KF: knee flexion; PT: peak torque; BW: body weight.

a

Indicates a significant intra-group difference, p < 0.05.

b

Indicates a significant difference from CON Group at post test, p < 0.05.

3.8. Changes in BMD

At baseline, there were no significant differences amongst the groups in BMD at any site measured (Table 4). There were significant interactions between group and time on BMD at the trochanter (p = 0.005) and total hip (p = 0.034). Accordingly, the RE group significantly increased BMD by 2.9% (0.020 g/cm2) at the trochanter and 1.5% (0.013 g/cm2) at the total hip (Fig. 2). No significant changes in BMD were observed for the AE and control groups (p > 0.05). There was no significant interaction or main effects of group and time on serum OPG and RANKL levels or the OPG/RANKL ratio (all p > 0.05; Table 4, Fig. 2).

Table 4. Eight-month changes for BMD, OPG, RANKL and OPG/RANKL.

Resistance exercise groupAerobic exercise groupControl groupp (group)p (time)p (interaction)
VariablePre-trainingPost-trainingPre-trainingPost-trainingPre-trainingPost-training
Femoral neck, g/cm20.684 ± 0.0820.676 ± 0.0900.657 ± 0.1050.660 ± 0.1110.678 ± 0.0560.676 ± 0.0650.7190.6410.553
Troch, g/cm20.646 ± 0.0950.666 ± 0.106a0.638 ± 0.0990.641 ± 0.0980.628 ± 0.0380.621 ± 0.0460.4480.0120.005
Inter, g/cm21.035 ± 0.1681.047 ± 0.1641.022 ± 0.1411.020 ± 0.1420.990 ± 0.0850.980 ± 0.1130.3680.3340.343
Total hip, g/cm20.859 ± 0.1240.873 ± 0.132a0.848 ± 0.1250.849 ± 0.1240.831 ± 0.0650.824 ± 0.0820.4680.0060.034
OPG, pmol/L5.46 ± 1.245.42 ± 1.007.58 ± 3.917.21 ± 4.4310.10 ± 6.249.06 ± 4.940.0620.1170.420
RANKL, pmol/L171.56 ± 86.26152.73 ± 63.76180.78 ± 85.77188.32 ± 89.99157.49 ± 63.25171.11 ± 64.800.6610.8800.052
OPG/RANKL0.037 ± 0.0120.039 ± 0.0100.056 ± 0.0440.048 ± 0.0360.067 ± 0.0400.058 ± 0.0360.2220.0540.185

BMD = bone mineral density, Troch = trochanter, Inter = intertrochanteric region.

a

Indicates a significant intra-group difference, p < 0.05.

  1. Download : Download full-size image

Fig. 2. Percentage of changes from baseline in serum OPG, RANKL an OPG/RANKL ratio, and bone mineral density of the proximal femur in response to exercise or placebo (control) over 8 months. Values are mean ± SEM. RE: resistance exercise; AE: aerobic exercise; COM: control group.

In the collective sample (n = 71), changes in caffeine intake were significantly related to changes in trochanter BMD (r = − 0.26, p = 0.045). Nevertheless, the interaction of trochanter BMD remained significant (p = 0.003) after controlling for changes in caffeine intake. Changes in percent fat mass were significantly related to changes in intertrochanteric region BMD (r = 0.27, p = 0.035). The lack of a significant interaction between group and time, and the main effects on the intertrochanteric region remained unchanged after adjusting for change in total percent fat mass.

4. Discussion

Age-related functional changes, including reduced balance, gait ability and muscle strength, have been consistently related to fall risk. Given that low BMD along with the above-mentioned functional declines combine to make older adults, especially women, much more prone to bone fractures, it is reasonable to determine whether BMD, balance and strength might significantly increase after long-term exercise training and what type of exercise could induce the most pronounced effects in elderly women. Moreover, the regulation of osteoclastic activity is critical for understanding bone changes induced by exercise (mechanical load). OPG and RANKL have been shown to be important regulators of osteoclastogenesis, although few comprehensive efforts have been made to characterise the effects of long-term exercise on serum expression of both cytokines. Overall, data from the present study have shown that RE increases the BMD at the trochanter and total hip, balance and strength and that these effects are more pronounced than after AE in older women. No changes were observed in OPG and RANKL after eight months of exercise.

The use of exercise as a possible prevention strategy for prevention of fractures in elderly people has previously been hypothesised (Vogel et al., 2009). However, results have been discordant, depending on the type, intensity, duration of exercise, and on participants' age and functional status. To detect relevant biomechanical changes in postural stability, force platform-based measures were obtained during the OLS test. Although the effectiveness and validity of force platforms to assess postural balance in older people have been established (Pajala et al., 2008), few studies have documented the results of force platform balance test data. The present work confirmed that both resistance and aerobic exercise resulted in increased static and dynamic balance. For instance, favourable changes in postural sway, including decreased EA and slower velocity of COP displacement have previously been reported after exercise training (Messier et al., 2000). In addition, instability and age are expected to increase the EA and the velocity of COP trajectories (Abrahamova and Hlavacka, 2008, Latash et al., 2003). These results are somewhat surprising, as several studies have failed to support exercise-related benefits on balance, although these studies generally assess balance after single interventions without a specific balance/proprioceptive-related training component (Manini et al., 2007, Henwood and Taaffe, 2006). Furthermore, we observed that only RE had a significant effect on maximal knee extension and flexion strength. Although there is consensus that older adults can substantially increase their strength and power after RE (Chodzko-Zajko et al., 2009), the effectiveness of AE is questionable, with a number of studies showing significant improvements (Misic et al., 2009, Nalbant et al., 2009) and others reporting no evidence of muscle strength or power increase (Haykowsky et al., 2005, Tarpenning et al., 2006). Together, these findings reinforce the notion that exercise training has the potential to reduce fall risk.

The mechanical loading of bone through exercise has been investigated thoroughly for its potential to positively alter structural variables, including bone mass (Bailey and Brooke-Wavell, 2008). This change in bone as a result of exercise has been attributed to strain (defined as the fractional change in the dimension of a bone in response to a changing load) (Kohrt et al., 2004), which represents the key intermediate variable, and to its effect on cells by directly changing their dimensions or indirectly impacting intralacunar pressure, shear stresses, or charged fluid flow (Lanyon, 1996). In addition, accumulating evidence suggests that high strain rates and unusual strain distributions are positively related to osteogenic response (Bailey and Brooke-Wavell, 2008). In the present study, the RE-based intervention significantly increased BMD at the trochanter and total hip. Furthermore, no significant changes resulted after eight months of AE and a non-significant trend towards diminished bone density in the control group was observed. Together, these observations suggest that aerobic training protocols, which include concentric and eccentric muscle actions and loading impacts although at a lower intensity than with RE, produce modest effects on BMD in older women. In fact, despite being an important training mode, especially for the induction of cardiovascular and metabolic changes (Chodzko-Zajko et al., 2009), previous studies using aerobic training protocols have reported conflicting results regarding BMD. Although Silverman et al. (2009) found a significant improvement of 2% at the femoral neck in postmenopausal women after a 24-week walking programme, a number of studies have shown limited bone density improvements in postmenopausal women after AE (Palombaro, 2005, Martyn-St James and Carroll, 2008). One possible explanation is that aerobic protocols based only on walking activities, which lack lateral and twisting movements, do not represent a unique stimulus to bone. Conversely, our aerobic protocol included more diverse activities, such as jogging, skipping, step climbing/descending, dancing, aerobics and step choreographies. In fact, the data from Kohrt et al. (1997) demonstrated that an exercise programme including walking, jogging and stair climbing resulted in significant increases in BMD of the whole body, lumbar spine, femoral neck, and Ward's triangle. Conversely, our results suggest that the strain levels induced by the present AE were not sufficient to improve bone mass.

Our results on the RE-induced elevation in BMD are consistent with prior studies that have similarly reported the effectiveness of exercise in promoting an osteogenic response in elderly adults (Ryan et al., 2004, Bemben and Bemben, 2010). However, there is still some controversy regarding its osteogenic potential in older adults. Previous studies have described a lack of significant alterations in BMD at the proximal femur or lumbar spine after progressive resistance exercise programmes (Rhodes et al., 2000, Stengel et al., 2005). To date, several studies have focused on the impact of resistance exercise interventions on bone mass in premenopausal women (Martyn-St James and Carroll, 2006). Nevertheless, some of them have also reported a lack of BMD response to resistance training (Singh et al., 2009, Nakata et al., 2008). Given the heterogeneity of women's responses at different ages (likely due to a deterioration of the ability of older bone cells to perceive these physical signals or a failure of their capacity to respond) and the fact that oestrogen withdrawal is associated with increased remodelling intensity (Lanyon and Skerry, 2001), inconsistent results between studies amongst pre-, postmenopausal and older women can be anticipated. However, whilst strategies to increase bone mass in young premenopausal women can involve high-impact exercises, such as vertical jumping, exercises that introduce high strain levels to the skeleton are difficult to perform with advancing age due to the high risk of traumatic fractures, stress injuries and arthritic complications. In such circumstances, the optimal exercise prescription for older people should meet other paramount needs, including being feasible, safe, acceptable, and cost-effective. The positive effect observed at the total femur and trochanter is probably related, at least partially, with the inclusion of movements, such as hip abduction that stimulates the gluteus medius and minimus, as both insert on the greater trochanter of the femur, and are also assisted by the lateral rotator group which inserts on or near the greater trochanter of the femur. The hip flexors (iliacus and psoas major) play an important role in the leg press exercise, and may also have an osteogenic effect at the femur as it connects to the lesser trochanter.

In this study, serum OPG level, serum RANKL levels and the OPG/RANKL ratio did not change significantly after an 8-month RE and AE training programme. Previous data have shown that mechanical stimulation (i.e., dynamic flow-induced shear stress) induced in vitro inhibits osteoclastogenesis through an upregulation of OPG and a downregulation of RANKL (Kim et al., 2006). Saunders et al. (2006) also found that mechanical stimulation via substrate deformation significantly increases soluble OPG levels by osteoblastic cells. Despite the above-mentioned results from in vitro studies and other promising results of long-distance running effects on BMD via the OPG/sRANKL system (Ziegler et al., 2005), there are no data available in the literature concerning the effectiveness of common exercise modalities, such as RE and AE on serum OPG and RANKL levels amongst older adults. Our results show that eight months of RE and AE training do not elicit significant changes in either biomarkers. Although it is well known that OPG blocks the differentiation of pre-osteoclastic cells into osteoclasts by inhibiting the binding of RANK to RANKL and, consequently, reduces osteoclastic bone resorption (Boyle et al., 2003), not all mechanical stimulation studies have shown an increase in OPG with osteoblast stimulation (Liegibel et al., 2002). Moreover, Esen et al. (2009) reported no significant changes in OPG levels after a 10-week walking programme in middle-aged men. An increased expression of RANKL may be involved in the excessive bone resorption observed in osteoporosis, thus a down-regulation of RANKL should prevent bone loss. There is evidence from in vitro studies that mechanical load may induce down-regulation of RANKL (Rubin et al., 2003, Rubin et al., 2000, Lau et al., 2010). In contrast to these studies, our findings showed that exercise did not favourably affect RANKL levels. One study observed exercise-induced RANKL changes, reporting that a 10-week high intensity walking programme significantly reduced RANKL in middle-aged men (Esen et al., 2009). Conversely, in a previous study using human cell lines, mechanical stimulation did not affect RANKL (Saunders et al., 2006). The reasons for this dissimilarity in results, apart from the disparity in setting (in vitro and in vivo studies), may stem from the differences in mechanical stimulation mode, intervention length, and physiological factors, such as cyclic variations, age and gender.

A major limitation of this study is the lack of concurrent measures of the magnitude of the applied loads based on calculations of the ground reaction forces (GRF). Moreover, we used an accelerometer cut point of 1041 cpm which is substantially lower than the cut point of 1952 cpm that is typically used for moderate activity in younger adults (Freedson et al., 1998). Indeed, using the former cut point the mean MVPA decreases to 32 min. Although the 1041 cpm may be more appropriate to our sample age, this cut point was obtained from a small sample of older adults and no vigorous activity was included in the calibration protocol, which may overestimate the time spent in moderate activity. The mechanical competence of bone is a function not only of its intrinsic material properties (mass, density and stiffness), but also of its structural properties (size, shape and geometry). DXA is the method most commonly used to measure BMD area (g/cm2) because of its speed, precision, low radiation exposure and availability of reference data (Watts, 2004); however, this two-dimensional skeletal outcome represents only one part of overall bone strength. Finally, our sample size may have been too small to detect significant changes in all variables.

A key strength of our study is the novel data it provides regarding the relationship between exercise training modes and OPG levels in this particular population. Moreover, this study has considered the possible influence of critical confounding variables, such as daily PA levels objectively measured by accelerometers, nutrition and lactose persistence (genetically defined by the C/T − 13910 genotype). Lactase deficiency has been associated with BMD (Obermayer-Pietsch et al., 2004) and reduced intake of calcium (Bacsi et al., 2009) and could represent a genetic risk factor for bone fractures for older adults (Enattah et al., 2005). However, no associations were found between lactase deficiency status and reduced BMD or calcium intake and BMD changes after training.

In conclusion, the results indicate that 8-month RE, but not AE, can induce significant bone adaptation in older women without significantly affecting OPG and RANKL levels. We have also demonstrated that both exercise training regimens elicited significant gains in balance. The results presented here suggest that higher workloads may be necessary in AE programmes to increase bone mass. Although these findings provide some clue into the potential for exercise to reduce fracture risk in community-dwelling older women, additional data are needed to validate and build upon our findings using additional outcome measures.

Acknowledgments

The authors thank Dr Conceição Gonçalves, Nádia Gonçalves, Margarida Coelho and Joana Campos for their kind support in biochemical assays, and Norton Oliveira for carrying out isokinetic strength tests. This research was funded by the Portuguese Foundation of Science and Technology, grant FCOMP-01-0124-FEDER-009587-PTDC/DES/102094/2008. E. A. Marques, F. Wanderley and J. Mota are supported by grants from Portuguese Foundation of Science and Technology (SFRH/BD/36319/2007, SFRH/BD/33124/2007, and SFRH/BSAB/1025/2010 respectively).

References

View Abstract
Copyright © 2011 Elsevier Inc. All rights reserved.