Vascular adaptation to deconditioning and the effect of an exercise countermeasure: results of the Berlin Bed Rest study. J Appl Physiol 99: 1293–1300, 2005. First published June 2, 2005; doi:10.1152/japplphysiol.00118.2005.—Deconditioning is a risk factor for cardiovascular disease. The physiology of vascular adaptation to deconditioning has not been elucidated. The purpose of the present study was to assess the effects of bed rest deconditioning on vascular dimension and function of leg conduit arteries. In addition, the effectiveness of resistive vibration exercise as a countermeasure for vascular deconditioning during bed rest was evaluated. Sixteen healthy men were randomly assigned to bed rest (BR-Ctrl) or to bed rest with resistive vibration exercise (BR-RVE). Before and after 25 and 52 days of strict horizontal bed rest, arterial diameter, blood flow, flow-mediated dilatation (FMD), and nitroglycerin-mediated dilatation were measured by echo Doppler ultrasound. In the BR-Ctrl group, the diameter of the common femoral artery decreased by 13 ± 3% after 25 and 17 ± 1% after 52 days of bed rest (P < 0.001). In the BR-RVE group this decrease in diameter was significantly attenuated (5 ± 2% after 25 days and 6 ± 2% after 52 days, P < 0.01 vs. BR-Ctrl). Baseline blood flow did not change after bed rest in either group. After 52 days of bed rest, FMD and nitroglycerinmediated dilatation of the superficial femoral artery were increased in both groups, possibly by increased nitric oxide sensitivity. In conclusion, bed rest deconditioning is accompanied by a reduction in the diameter of the conduit arteries and by an increased reactivity to nitric oxide. Resistive vibration exercise effectively attenuates the diameter decrease of leg conduit arteries after bed rest.
 

echo Doppler ultrasound; flow-mediated dilatation; bed rest deconditioning
 

PHYSICAL INACTIVITY OR DECONDITIONING is an independent risk factor for atherosclerosis and cardiovascular disease (3, 38). In a prospective observational study, improvement of physical fitness decreased cardiovascular mortality risk by 51% (3). Endothelial dysfunction plays an important role in the pathogenesis of cardiovascular disease and is directly related to cardiovascular mortality (29). Cross-sectional studies have demonstrated a lower vascular dimension (24) and endothelial function (19) in sedentary subjects compared with exercisetrained individuals. This may reflect either downregulation by physical inactivity or upregulation by exercise training. Although changes in blood flow after deconditioning have been observed in humans, data on vascular dimension and endothelial function are scarce. Moreover, the underlying physiological mechanism of vascular adaptation to deconditioning in humans has not been elucidated.
 

Longitudinal deconditioning intervention studies have demonstrated the detrimental effects of bed rest on muscle function (2, 5), bone density (5), and orthostatic tolerance (13, 23). Previous studies on vascular adaptation to deconditioning interventions are restricted to flow measurements and have mainly focused on the arm vascular bed (15, 47, 48). In most of these studies, the effect of physical inactivity on blood flow is confounded by the effects of head-down tilt on plasma volume (15, 47, 48). Because of their role in standing and locomotion, the legs more accurately reflect the intense deconditioning during bed rest. Vascular remodeling as a result of deconditioning will be reflected in changes in vascular dimension. Moreover, endothelial function is of paramount importance for vascular remodeling and in the pathogenesis of cardiovascular disease. Therefore, the purpose of the present study was to assess the effect of horizontal bed rest deconditioning on vascular dimension of leg and arm conduit arteries and on endothelial function of a leg conduit artery.
 

Exercise training has been shown to improve vascular dimension and endothelial function in longitudinal intervention studies (20, 33, 34), and exercise has been propagated as a countermeasure for orthostatic intolerance in both space travelers (12) and hospitalized patients (13). Therefore, a second purpose of the study was to evaluate the effectiveness of exercise as a countermeasure for vascular adaptation to bed rest. Resistive vibration exercise has recently emerged as a training modality that increases oxygen uptake (42), leg blood flow (28), muscle strength (43, 51), and bone density (51). As such, we hypothesized that resistive vibration exercise would counteract the vascular changes induced by bed rest deconditioning.
 

METHODS
 

Subjects
 

Sixteen healthy men (age 34 ± 2 yr) participated in this study and represent a subpopulation of the Berlin Bed Rest study. All subjects were screened with a medical history and physical examination and did not have any medical problems. None of the subjects suffered from diabetes or cardiovascular disease or used any medication. Cholesterol and triglyceride levels were in the normal range (Table 1).

 

Smoking was not used as an exclusion criterion, but smoking was prohibited during the bed rest trial. Subjects were randomly assigned to bed rest (BR-Ctrl) or bed rest with resistive vibration exercise (BR-RVE). All subjects gave their written, informed consent. The Ethics Committee of the Medical School of the Free University Berlin has approved the Berlin Bed Rest study and the present experiment within it.
 

Procedures
 

The vascular characteristics of all subjects were measured three times: 2 days before, and 25 and 52 days after bed rest deconditioning (BR-2, BR25, and BR52, respectively).
 

Bed rest protocol. After the initial series of experiments, subjects were placed at complete horizontal bed rest. All personal hygiene activities were performed in supine position. Subjects were housed in a dedicated clinical ward of the University Hospital Benjamin Franklin. The subjects were monitored with video cameras to guarantee compliance with the bed rest protocol. In addition, the monitoring with force transducers of the vertical forces generated by the subjects ensured strict bed rest and avoidance of powerful movements. The diet of the subjects was controlled carefully.
 

Resistive vibration exercise. The subjects who were randomly assigned to the BR-RVE group were exposed to resistive vibration exercise (RVE) twice daily for 30 min (8 min pure exercise time), with the exception of Sundays and Wednesday afternoons. RVE was performed with a device that was specifically manufactured for the Berlin Bed Rest study with modifications for the use during supine bed rest (Galileo Space, Novotec, Pforzheim, Germany). The subjects attached themselves to the device with four supporting belts (an image of RVE is available as Fig. 5 in the online version of this article). The subjects pushed their feet against the device’s footplate and pulled on the hand and hip belts; this caused elongation of the springs and generated a force resisting the body extension. Force transducers at the end of each spring yielded the platform reaction force. The vibration of the footplate was elicited by an eccentric rotation of a mass that was phase shifted by 180° for the right and left part of the footplate. By virtue of that construction (preset vibration frequency), the acceleration of the eccentrically rotating mass changes with vibration frequency. Hence, the greater the vibration frequency, the greater the peak platform reaction force elicited on the exercising subjects. At the beginning of each training session, the length of the supporting belt was adjusted so that a certain resting platform reaction force was created in full knee extension. Then, four different exercise units were performed. 1) Squatting exercise: knees were stretched from 90° to full extension in cycles of 6 s. This unit was targeted at the knee extensors. 2) Heel raises: with the knees in almost complete extension, the heels were raised as long as the subjects could sustain this. After briefly resting back on the foot platform, the heels were raised anew. This unit was targeted at the foot plantar flexors. 3) Toe raises: with knees almost in complete extension, the toes were raised as long as possible. After briefly relieving, the toes were raised anew. This unit was targeted at the foot dorsal flexors. 4) “Kicks”: the knees were extended from 90° flexion to extension as quickly and forcefully as possible. Between each of these 10 kicks, subjects relaxed completely for 10 s.
 

For each training session, units 1–4 were done once and in ascending order. There was one morning session and one in the afternoon. In the morning sessions, exercise units 1–3 were performed for at least 60 s. If subjects managed to perform them for 100 s, the vibration frequency was increased. Initially, the vibration frequency was set to 19 Hz. In the afternoon, subjects were asked to exercise with a lower resting platform reaction force (60–80% of the value achieved in the morning) and to run through units 1–3 for 60 s each, with as many iterations as possible. Trained staff members supervised all training sessions.
 

Measurements
 

To reduce circadian and dietary effects, measurements were performed at the same time of day in each individual subject, and meals were identical for each subject on the days of the measurements. All subjects refrained from caffeine and alcohol from midnight, and from vitamin C supplements for 24 h. On the testing days, subjects did not perform exercise before testing. Because of the scheduling of the measurements it was impossible for the subjects to be tested in a fasting state. However, after midnight the diet was carefully controlled. Subjects received low-fat meals, and meals were identical for each measurement. Measurements were performed after an acclimatization period of at least 20 min after instrumentation. Blood pressure and heart rate were measured at the onset of the measurement. Blood pressure was measured manually by the standard auscultatory method. Heart rate was derived from the electrocardiogram.
 

Ultrasound measurements Resting blood cell velocity and diameter of the common femoral artery (CFA) and superficial femoral artery (SFA) were measured in the left leg, using an echo Doppler device (Megas, ESAOTE, Firenze, Italy) with a 5- to 7.5-MHz broadband linear array transducer (16, 17). The angle of inclination for the velocity measurements was consistently below 60°, and the vessel area was adjusted parallel to the transducer. In addition, resting blood cell velocity and diameter were measured in the right brachial artery and in the left common carotid artery. The brachial artery represented a conduit artery supplying a limb that is subject to less intense deconditioning during bed rest. The common carotid artery was included as a reference vessel.
 

For reactive hyperemia and flow-mediated dilatation (FMD) of the SFA, a cuff was placed around the left upper thigh 3–4 cm below the bifurcation of the CFA. The cuff was inflated to a suprasystolic pressure of 220 mmHg for 5 min. After cuff deflation, hyperemic flow velocity in the SFA was recorded on videotape for the first 25 s, followed by a continuous registration of the vessel diameter for 5 min to determine FMD. Studies in the radial and brachial conduit arteries have proven that the vasodilatation response to hyperemic response after 5 min of distal arterial occlusion is endothelium and nitric oxide dependent (21, 35). Therefore, our FMD response most likely reflects endothelium-dependent dilatation. Endothelium-independent vasodilatation of the SFA was determined in 12 subjects. After a resting period of at least 20 min to reestablish baseline conditions, a systemic dose of nitroglycerin (0.4 mg) was administered sublingually to determine the endothelium-independent vasodilatation of the SFA, which is indicative for smooth muscle function and nitric oxide responsiveness. Vessel diameter of the SFA was continuously recorded between 2 and 6 min after nitroglycerin administration. We have reported the reproducibility for the measurements in the SFA previously as 1.5% for diameter, 14% for blood flow, and 15% for relative FMD changes (17).
 

Data Analysis
 

Ultrasound. For resting diameter measurements, two consecutive longitudinal vessel images were frozen at the peak systolic and end-diastolic phase and analyzed off-line. Three measurements were performed per diameter image. Mean diameter was calculated as (1/3 . systolic diameter) + (2/3 . diastolic diameter). The average of 10–12 Doppler spectra waveforms was used to calculate peak velocity and mean velocity. Mean blood flow (ml/min) was calculated as 1/4 .3.14(pie)(mean diameter)² .mean velocity (cm/s) . 60; peak blood flow (ml/min) was calculated as 1/4 .3.14(pie).(systolic diameter)²
peak velocity (cm/s) .60; regional peak wall shear rate (PWSR, s^-1) was calculated as 4.(peak velocity/systolic diameter), and mean wall shear rate (MWSR, s ^-1) was calculated as 4.(mean velocity/mean diameter). Reactive hyperemic blood flow was calculated from blood velocity 15–25 s after cuff release and the baseline vessel diameter. Although maximal reactive hyperemia may occur slightly earlier, we used this time frame to obtain data from all measurements in all subjects. In addition, we made the assumption that the diameter 15–25 s after cuff release is similar to baseline diameter. Delta PWSR and delta MWSR were defined as the differences between rest and hyperemic responses. Vessel diameters after reactive hyperemia were measured off-line from videotape at 50, 60, 70, 90, 120, 180, and 240 s after cuff release and at 2, 3, 3.5, 4, and 5 min after nitroglycerin administration. FMD and endothelium-independent vasodilatation were expressed as relative (%) diameter change from baseline of the end-diastolic diameter. Because the FMD response is directly proportional to the magnitude of the stimulus (30), the FMD response was also expressed relative to the delta shear rate. Ratios were calculated for the FMD/delta PWSR and FMD/delta MWSR. The ratio between the maximal FMD and endothelium-independent vasodilatation was expressed as FMD/nitroglycerin- mediated dilatation. Ultrasound analysis has been described in more detail previously (17).

Statistical Analysis
 

Data are presented as means ± SE. Differences in the response to bed rest between the BR-RVE group and the BR-Ctrl group were tested with repeated-measures ANOVA with time as within-subject factor and group as between-subject factor (Statistical Package for Social Sciences, SPSS 12). The time factor represents the overall effect of bed rest. The time-by-group factor was used to test the effect of the RVE countermeasure. Statistically significant differences between the groups were further analyzed with unpaired t-tests at bed rest day 25 and bed rest day 52. Differences were considered to be statistically significant at P < 0.05.
 

RESULTS
 

Subjects
 

There were no significant differences between the groups for any of the baseline characteristics (Table 1). All subjects completed the study. During the bed rest period the subjects in the BR-RVE group were exposed to 89 exercise sessions of ˜30 min (8 min pure exercise time).
 

Heart Rate and Blood Pressure
 

During the bed rest period, resting heart rate increased significantly in the BR-Ctrl group (P< 0.05, Table 2). Changes in heart rate during bed rest were significantly different between the BR-Ctrl and BR-RVE groups, and heart rate was significantly lower in the BR-RVE group compared with the BR-Ctrl group at BR25 and BR52 (P < 0.05, Table 2).Blood pressure did not change significantly during bed rest and was not different between the groups.
 

Diameter and Blood Flow of the CFA and SFA
 

The data for the CFA in the exercise group are based on seven subjects, because the diameter of the CFA could not be assessed in one subject because of vessel wall irregularities. The diameter of the CFA and SFA decreased significantly during bed rest (P < 0.001 for time). This decrease was significantly attenuated in the exercise group compared with the BR-Ctrl group in both the CFA and SFA (Fig. 1, A and C, P =0.001 and P <0.001 for group
time, respectively). The blood flow in the CFA and SFA did not change during bed rest and did not differ between the groups (Fig. 1, B and D).
 

Diameter and Blood Flow of the Brachial and Carotid Artery
 

The diameter of the brachial artery decreased significantly during bed rest (P =0.016 for time) but did not differ between the exercise and control group (Fig. 2C). The diameter of the carotid artery, and the blood flow in the brachial artery and in the carotid artery did not change during bed rest (Fig. 2, A, B, and D).
 

Reactive Hyperemia, FMD, and Nitroglycerin-Mediated Dilatation of the SFA
 

Reactive hyperemic blood flow did not significantly decrease after bed rest in both groups (BR-Ctrl: from 989 ±98 to 716 ±58 ml/min, BR-RVE: from 1,119 ±104 to 1,076 ± 107 ml/min).
 

At bed rest day 25, one FMD measurement in the control group and one in the exercise group failed; therefore the ANOVA is based on 14 subjects. FMD increased significantly during bed rest (P =0.007, Fig. 3A, n =14). This increase tended to be less in the exercise group than in the BR-Ctrl group (P =0.07). FMD was lower in the exercise group than in the BR-Ctrl group on bed rest day 25 (P =0.008), but not on bed rest day 52 (P =0.55). Nitroglycerin-mediated dilatation increased significantly over time (P =0.002, n =12) with no difference between the groups (Fig. 3C). These findings for FMD and nitroglycerin-mediated dilatation were similar if absolute instead of relative changes in diameter were analyzed. FMD corrected for MWSR did not change significantly over time nor between groups over time (Fig. 3B, n =14). Changes in FMD corrected for nitroglycerin-mediated dilatation were significantly different between groups (Fig. 3D, n =10). At bed rest day 25, corrected FMD tended to be lower in the exercise group (P =0.05).

 

DISCUSSION
 

This study is the first to characterize the adaptation of diameter and endothelial function of the leg conduit arteries to bed rest deconditioning. The diameter of the CFA and SFA decreased after bed rest, whereas baseline blood flow did not change. Both FMD and endothelium-independent dilatation of the superficial femoral artery increased significantly after bed rest, indicating increased reactivity to nitric oxide after bed rest, possibly by increased nitric oxide sensitivity or increased smooth muscle sensitivity to vasodilators. In addition, this is the first study to demonstrate that RVE can effectively attenuate the diameter decrease of conduit arteries of the leg.
 

Bed Rest Deconditioning and Vascular Dimension
 

After bed rest without exercise, the diameter of the CFA decreases by 13 and 17%, at BR25 and BR52, respectively. This suggests that most of the adaptation in arterial diameter occurs in the first 4 wk of bed rest deconditioning. After 4 wk of hindlimb unloading in rats, an animal model for physical inactivity and microgravity, the lumen diameter of the femoral artery also decreased significantly by ˜8% (54). Furthermore, our results are in agreement with a 12% decrease in diameter of the CFA after 4 wk of deconditioning by unilateral lower limb suspension in humans (4), suggesting the same degree of deconditioning in several models of physical inactivity. However, in the paralyzed legs of spinal cord-injured individuals the diameter of the CFA is 30% smaller than in healthy control subjects (16). This adaptation is completed within 6 wk after the occurrence of a spinal cord injury (16). Hence, the decrease in arterial diameter appears to be larger in spinal cord-injured individuals than in able-bodied, immobilized individuals. This can be attributed to the presence of some physical activity of the legs during bed rest as opposed to no activity because of paralysis. Notably, the time course of arterial diameter adaptation is very similar in bed rest and spinal cord injury. The adaptation of conduit artery diameter to bed rest deconditioning may reflect structural and/or functional changes. Nitroglycerin 0.4 mg sublingually has been shown to produce a maximal vasodilatation in both coronary and brachial arteries (10, 36). In addition, maximal dilatation of the femoral artery to nitroglycerin closely resembles maximal vasodilatation in response to another strong vasodilator stimulus, 12 min of ischemia combined with ischemic exercise (4). Therefore, the response to nitroglycerin can be used as a measure of near maximal arterial diameter in the SFA (Fig. 4). Overall, maximal diameter decreased with bed rest (P < 0.01), suggesting that structural changes occur in conduit arteries in response to bed rest.

 

The decrease in brachial diameter of 5% was small compared with the effect of bed rest deconditioning in the legs. This may be attributed to the specific antigravity and locomotion functions of the legs. After 7 days of bed rest, baseline diameter of the brachial artery did not change (7); this is probably due to the shorter duration of bed rest in that study. The lack of effect of bed rest on carotid artery diameter can be explained by the minor effect of physical inactivity on the cerebral circulation.
 

Bed Rest Deconditioning and Blood Flow
 

Baseline leg blood flow did not decrease after bed rest. Former studies used plethysmography and reported a decrease in leg blood flow at the arteriolar level (14, 27, 31, 39). All these studies applied 6° head-down-tilt bed rest (14, 27, 31, 39). Louisy et al. (31) demonstrated that a large portion of the decrease in blood flow was already present after 1 day of head-down tilt bed rest. In the first 24–48 h, head-down tilt bed rest causes a pronounced decrease in plasma volume (11), which may be responsible for a large part of the blood flow decrease in these studies. In contrast, Bonde-Petersen et al. (6) also used plethysmography but reported no changes in leg blood flow after 20 days of horizontal bed rest. Interestingly, Takenaka et al. (49) used echo Doppler ultrasound in the same subjects and reported a decrease in leg blood flow. It is not possible to make a detailed comparison with our echo Doppler data because Takenaka et al. did not report on changes in diameter and velocity
 

The present findings are in agreement with a previous study of deconditioning due to unilateral lower limb suspension in human volunteers (4). After limb suspension, diameter of the leg conduit arteries decreased, whereas leg blood flow did not decrease. Furthermore, even in extreme deconditioning due to paralysis after spinal cord injury, with a dramatic decrease in arterial diameter, several studies have reported no differences in resting leg blood flow, as measured with echo Doppler ultrasound (17, 37). Studies using exercise training have provided clues that conduit arteries adapt primarily to peak blood flow and peak oxygen demand during exercise (20, 33). Baseline diameter seems to adapt to maximal blood flow during bouts of exercise rather than to resting blood flow (20, 33). In the present bed rest study the loss of periods of high blood flow and high shear stress in the group without exercise would explain the decrease in arterial diameter, without changes in baseline blood flow. In agreement with the results in the legs, blood flow in the arm did not change during bed rest.
 

Bed Rest Deconditioning and Endothelial Function
 

FMD, indicative for endothelial function, was significantly increased after 52 days of bed rest. This corresponds with an increase in FMD after 28 days of deconditioning by lower limb suspension (4) and with an increase in FMD in the paralyzed legs of spinal cord-injured individuals (17). However, when FMD is corrected for its eliciting stimulus MWSR (30), the increase in FMD is no longer statistically significant in the present and the cited (4, 17) studies. In the present study, the shape of the figure changes very little when this correction for shear rate is applied, with an increase in standard error (Fig. 3, A and B). This suggests that the number of subjects is too low for this type of correction. However, correction of FMD for PWSR instead of MWSR results in a trend toward increased FMD after bed rest (P =0.059) with a significant difference between groups (P =0.018). Likewise, in a previous study virtually all spinal cord-injured individuals had a higher FMD response per delta shear rate (17). Moreover, bed rest deconditioning causes a significant increase of FMD of the brachial artery (7). Combined, these data provide evidence that FMD increases after deconditioning.


 

In hindlimb-unloaded rats, endothelium-dependent vasodilatation of the lower abdominal aorta in response to acetylcholine is reduced. This decrease in vasodilatation is probably due to endothelial dysfunction, but changes in smooth muscle cell nitric oxide sensitivity may also be responsible (18). At the level of the resistance arteries and arterioles, endotheliumdependent vasodilatation and nitric oxide synthase expression are reduced in the soleus muscle after unloading (26, 46, 53), whereas endothelium-independent vasodilatation is enhanced (26). Therefore, animal data largely suggest a reduction in endothelium-dependent dilatation combined with changes in nitric oxide responsiveness at the level of the smooth muscle cells. In contrast, an upregulation of inducible nitric oxide synthase has also been demonstrated after hindlimb unloading (45, 50). The changes in endothelial function in this animal model and our human model are distinctly different. Apart from interspecies differences, hindlimb unloading causes more microgravity effects than horizontal bed rest. In addition, most changes in rats were observed in the soleus muscle with a decrease in baseline blood flow in the absence of changes in endothelial function in the gastrocnemius muscle (53), whereas in our study blood flow did not change. Nevertheless, the animal data do illustrate that deconditioning may also alter smooth muscle responsiveness.

 

In the present study, FMD corrected for endothelium-independent dilatation does not increase after 52 days of bed rest (Fig. 3D). This suggests that mainly nitric oxide responsiveness or general vasodilator responsiveness of the smooth muscle cell is enhanced after bed rest and not endothelial function and nitric oxide availability. In contrast, exercise training in animals and humans with endothelial dysfunction and vigorous exercise in healthy subjects specifically increase endotheliumdependent dilatation (34). Therefore, the physiological mechanism of the increase in vascular function as a result of exercise or deconditioning appears to be fundamentally different and seems to be located in the endothelium for exercise and mainly in the smooth muscle cell for bed rest deconditioning.
 

It has been suggested that increase in arterial diameter after exercise training is due to expansive remodeling in response to peak shear stress during exercise (20, 33). Parallel to this reasoning, the observed decrease in diameter after bed rest may represent inward remodeling as an adaptation to diminished exposure to periods of high shear stress. The 16% decrease in maximal diameter of the SFA in the BR-Ctrl group was attenuated to 5% in the RVE-group (Fig. 4, P < 0.01), suggesting that RVE significantly attenuated the effect of bed rest on blood vessel structure. RVE has been shown to increase heart rate (41), oxygen uptake (41), and leg blood flow (28). Therefore, periods of high shear stress are not absent in the BR-RVE group, which explains the observed attenuation of the decrease in baseline and maximal arterial diameter in the BR-RVE group. Nevertheless, the stimulus of RVE on the conduit arteries is probably too low to completely prevent vascular adaptations to bed rest. Moreover, the lack of increase in heart rate after 52 days of bed rest in the BR-RVE group as opposed to the BR-Ctrl group suggests that RVE is an effective countermeasure for some aspects of bed rest deconditioning. In accordance, resistive exercise has been shown to be an effective countermeasure against other detrimental effects of bed rest, such as loss of muscle size and function (1). Whether the effect of RVE is due to the vibration exercise component, the resistive exercise component or the combination of both cannot be determined in the present study design.
 

The FMD and nitroglycerin-mediated response did not differ between the BR-Ctrl and BR-RVE group before and after 52 days of bed rest. Nevertheless, the BR-RVE group appears to follow a different time course of adaptation, with significant differences between groups after 25 days of bed rest. Possibly, RVE only delays the adaptation of endothelial function to bed rest, whereas the reactivity to nitric oxide increases similarly in both groups. One might argue that in the BR-RVE group exercise should have caused an increase in FMD. However, 52 days of bed rest represents an immense deconditioning stimulus, specifically in the legs. In addition, vigorous systemic exercise is needed to improve endothelium-dependent dilatation in healthy subjects without endothelial dysfunction (9). Therefore, it is well conceivable that in our bed rest study the deconditioning stimulus on the endothelium overruled the exercise stimulus.
 

Limitations
 

Endothelium dependency of FMD has been established more extensively in the conduit arteries of the arm than of the leg. However, both Rubanyi et al. (44) and Pohl et al. (40) have demonstrated that an intact endothelium is required for FMD of the femoral artery. Studies of FMD in the arm have shown that both ischemia at the measurement site and prolonged ischemia (15 min) decrease the contribution of nitric oxide to FMD (21, 35). Because we measured FMD proximal of the arterial occlusion cuff and in response to 5 min of ischemia, our results likely reflect endothelium-dependent dilatation.
 

In the setting of the study it was not possible to perform the measurements in the fasting state. Because FMD is decreased after high-fat meals (22, 52), we carefully controlled the subjects’ diets. Subjects received identical, low-fat meals before each measurement. Baseline arterial diameter and FMD are not affected by low-fat meals (22, 52). Therefore, we are confident that we minimized the confounding effects of food intake.
 

Some of the subjects smoked until the start of the study. Smokers were equally distributed among the BR-Ctrl and BR-RVE groups. Although there have been reports that smoking may not affect endothelium-dependent dilatation (25, 32), most evidence suggests that smoking decreases FMD (8). In a hallmark study by Celermajer et al. (8), former smokers with an average time since cessation of 6 yr tended to have better FMD than current smokers. To our knowledge, data on the effect of short-term cessation of smoking of maximal 8 wk on endothelium-dependent dilatation are lacking. Our FMD results were similar if smokers were excluded from the analysis. Therefore, smoking does not appear to have an important influence on our results.
 

In conclusion, the diameter of the leg conduit arteries decreases after bed rest, whereas baseline blood flow remains unchanged. Both FMD and endothelium-independent dilatation of leg arteries increase significantly after bed rest, indicating increased reactivity to nitric oxide after bed rest, possibly by increased nitric oxide sensitivity or increased smooth muscle vasodilator capacity. In addition, RVE can effectively attenuate the diameter decrease of conduit arteries of the leg but seems only to delay the effect of bed rest on endothelial function.

The detection threshold of the study design failed to show any changes in bone density using an intention-to-treat analysis for either the placebo or treatment group. Regression analysis on the a priori study group demonstrated a significant effect of compliance on efficacy of the intervention, particularly at the lumbar spine (p=0.004). Posthoc testing was used to assist in identifying various subgroups that may have benefited from this treatment modality. Evaluating those in the highest quartile of compliance (86% compliant), placebo subjects lost 2.13% in the femoral neck over 1 year, whereas treatment was associated with a gain of 0.04%, reflecting a 2.17% relative benefit of treatment (p = 0.06). In the spine, the 1.6% decrease observed over 1 year in the placebo group was reduced to a 0.10% loss in the active group, indicating a 1.5% relative benefit of treatment (p = 0.09). Considering the interdependence of weight, the spine of lighter women (<65 kg), who were in the highest quartile of compliance, exhibited a relative benefit of active treatment of 3.35% greater BMD over 1 year (p = 0.009); for the mean compliance group, a 2.73% relative benefit in BMD was found (p = 0.02). These preliminary results indicate the potential for a noninvasive, mechanically mediated intervention for osteoporosis. This non-pharmacologic approach represents a physiologically based means of inhibiting the decline in BMD that follows menopause, perhaps most effectively in the spine of lighter women who are in the greatest need of intervention.

Key words: osteoporosis, anabolic, mechanical loading, vibration, low-level, frequency, osteogenic, muscle, skeleton, aging, menopause, bone, antiresorptive
 

INTRODUCTION
 

OSTEOPOROSIS, A DISEASE CHARACTERIZED by the progressive loss of bone tissue, is one of the most common complications of aging.(1) After menopause, BMD can continue to decline at a rate as high as 3%/year in some women,(2–5) resulting in 70% of women over the age of 80 having BMD measurements more than 2.5 SDs below young normal values.(6) Intervention strategies that slow the loss of bone soon after menopause may result in a significant reduction of fractures in those individuals at greatest risk.(7)
 

To date, prevention of bone loss has been approached principally through pharmacologic intervention, the longterm safety of which remains uncertain.(8) These pharmacologic approaches inherently ignore that a significant portion of the skeleton’s structural success can be attributed to bone’s sensitivity to alterations in its mechanical environment, with its “form follow function” characteristics ensuring that sufficient mass is placed to withstand the rigors of functional activity.(9) In essence, physical stimuli represent both an endogenous anabolic stimulus to bone tissue(10) and an antiresorptive factor that can actively inhibit osteoclastogenesis.( 11)
 

The skeleton’s sensitivity to its physical environment infers that such non-pharmacologic signals could provide an exogenous treatment regimen for the inhibition of bone loss. Whereas long-term exercise has been shown to increase BMD in young people,(12) this sensitivity seems to be greatly reduced in the elderly.(13) Moreover, exercise, and the predilection to falls that it may invite, could promote the very fractures that the intervention is prescribed to prevent. In contrast to the relatively well-accepted anabolic influence of high mechanical forces, recent work has led to the hypothesis that extremely small physical stimuli, at sufficiently high, but physiologically relevant, frequencies, can be critical determinants of bone morphology(14) and thus represent a unique means of mediating bone quantity and quality.
 

Using a surgically invasive model on the ulnae of aged (4 year old) turkeys, high-frequency (30 Hz), low-magnitude (200 microstrain) signals were successful in stimulating an increase in cortical bone, whereas high-amplitude (3000 microstrain), low-frequency (1 Hz) signals failed to be anabolic.(15) Delivering these signals noninvasively for 10 minutes/day, a floor plate vibrating vertically at 90 Hz, inducing strain in the bone of less than 10 microstrain, successfully inhibited disuse osteopenia caused by 23 h and 50 minutes of tail suspension in the rat, whereas 10 minutes/ day of normal weight-bearing activity failed to curb this loss.(16)
 

Using a surgically invasive model on the ulnae of aged (4 year old) turkeys, high-frequency (30 Hz), low-magnitude (200 microstrain) signals were successful in stimulating an increase in cortical bone, whereas high-amplitude (3000 microstrain), low-frequency (1 Hz) signals failed to be anabolic.(15) Delivering these signals noninvasively for 10 minutes/day, a floor plate vibrating vertically at 90 Hz, inducing strain in the bone of less than 10 microstrain, successfully inhibited disuse osteopenia caused by 23 h and 50 minutes of tail suspension in the rat, whereas 10 minutes/ day of normal weight-bearing activity failed to curb this loss.(16)
 

In longer-term animal studies, 1 year of daily, 20-minute sessions of low-level (0.3g, where g = earth’s gravitational field, or 9.8 m/s2), high-frequency (30 Hz) mechanical stimulation to the hind limbs of adult female sheep stimulated a 43% increase in bone density in the proximal femur, measured by CT.(17) This increase was achieved through a 36% increase in the thickness of individual trabeculae and a 45% increase in their number,(18) contributing to a 12% increase in stiffness and 27% increase in strength of the cancellous bone from the femur.(19)
 

The work reported here evaluates, in humans, whether such a noninvasive, low-level mechanical signal, induced noninvasively into the musculoskeletal system, is able to inhibit the bone loss that follows menopause. Considering the fiber type–specific sarcopenia that parallels aging,(20) we believe the bone wasting that occurs in older adults results not only from the diminished levels of activity, but from the attenuated 20- to 50-Hz muscle dynamics that normally arise during long-duration activities such as quiet standing. Thus, we hypothesize that “reintroducing” the lowmagnitude, high-frequency dynamics back into the musculoskeletal system will re-establish a key regulatory stimulus to the bone tissue and thus inhibit the reduction of BMD that follows menopause.
 

MATERIALS AND METHODS
 

Study subjects
 

The protocol and study design were reviewed and approved by Creighton University’s Human Use Committee, and all clinical work was completed at the Creighton University School of Medicine’s Osteoporosis Center. Women meeting the 3 to 8-year postmenopausal criteria were recruited from the greater Omaha area by newspaper, radio, and television advertising and from existing subjects within Creighton’s Osteoporosis Center. Informed consent was obtained from qualified volunteers who agreed to participate in the study. Inclusion criteria included normal nutritional status (as determined by questionnaire), stable weight maintenance (i.e., no elective weight loss or diet), estimated daily calcium intake of >=500 mg/day, and the capability of following the protocol for daily use of the device as well as understanding and providing informed consent. Because of design constraints of the oscillating device, the body mass of included subjects had to be greater than 45 kg and less than 84 kg.
 

Exclusion criteria consisted of any pharmacologic intervention for osteopenia within the previous 6 months, any use of steroids, current smoking status, consumption of excessive alcohol (>2 drinks/day), evidence of osteomalacia, Paget’s disease, osteogenesis imperfecta, gastrointestinal disease, or history of malignancy, and/or any prolonged immobilization of the axial or appendicular skeleton within the last 3 years. Subjects were also excluded if they had evidence of spondyloarthrosis, thyrotoxicosis, psychomotor disturbances, hyperparathyroidism, renal or hepatic disease, and chronic diseases known to affect the musculoskeletal system (e.g., muscular dystrophy), and/or were engaged in high-impact activity at least three times per week (including but not limited to tennis, aerobics, running, weight-bearing activity or exercise more intense than fast walking).
 

Subjects not excluded by medical history and who met the inclusion criteria of 3–8 years past menopause underwent a battery of standard laboratory tests (e.g., Health Screen 20, urinalysis, hematology, and bone-specific markers; Metra, Sausalito, CA, USA), as well as lateral X-ray views of the thoracic and lumbar spine. In this second tier examination, subjects were excluded with physical or radiographic evidence of fractures or osteophytes. No patient exclusion was based on BMD status (T or Z scores). If the inclusion/exclusion criteria were satisfied by the medical history, laboratory data, and X-ray data, the subject was enrolled in the study. Over the course of 2 years, a total of 70 women were enrolled in the study.
 

Active and placebo devices were manufactured and assigned a device number to coincide with a randomization code. Each woman successfully recruited into the study was provided with a mechanical device (see below), which was delivered to her home and set up by a technician. Throughout the course of the study, subjects and investigators were blinded as to which device was an active or placebo unit, and all information regarding the randomization scheme was kept confidential and secure.
 

Design of the vibration platform
 

To induce low-level physical stimulation in a controlled manner, an apparatus was designed that used a small, lowforce (18N), but highly linear, moving coil actuator (model LA18–18; BEI San Marcos, CA, USA) to impose peak to peak vertical accelerations of 0.2g at a frequency of 30 Hz on a body mass of up to 85 kg. The device was designed such that a very small driving force would produce vertical accelerations of the subject’s body mass and the supporting spring loaded plate (Fig. 1). With incorporation of appropriate accelerometer feedback from the plate surface, control circuitry was sufficient to reduce non-translational modes of vibration caused by motion or positional changes of the subject.(21) As demonstrated in human volunteers, foot-based, whole-body vibrations above 25 Hz (cycles per second) and below 1g can safely be transmitted into the lower appendicular and axial skeleton without producing any detrimental skeletal resonances. The measured transmissibilities in the skeleton are all significantly below 1.0 at frequencies above 25 Hz, with ˜70% of the ground-based signal reaching the trochanter of the femur and L₃ in the spine.(22)

Experimental design
 

Sample size projections (discussed below) determined that 64 women would be required to address the principal hypothesis, that is, women who used an active device at least 80% of the prescribed time would show a significant inhibition of the bone loss that follows menopause. The study was also designed such that subjects who dropped out within the first months of participation would be replaced. The initial cohort of 64 women was randomly distributed into one of two groups, and individual treatment began as soon as each subject was enrolled in the study. Each subject was randomly assigned to the active or placebo group according to a confidential, randomized number sequence generated by an independent statistical consultant and without regard to baseline BMD or matching between groups.
 

In the initial recruitment group, active devices, which vibrated at 30 Hz, 0.2g peak to peak, were provided to 32 women, whereas 32 women received a placebo device. At this intensity level, with a total displacement of 55 µm, the motion of the active platform is slightly discernible because the intensity is just above the perception level for vibration.( 23) To help obscure the active/placebo status of the devices, each device emitted a low-frequency audible sound to suggest that every plate was “active.” Throughout the course of the study, neither the investigators nor the subjects were informed whether the device was active or placebo, reinforcing the blinded nature of the study.
 

Each coded device was delivered to the subject’s home, and the subject was instructed how to stand on it for two 10-minute treatments/day, separated by a minimum of 10 h, for 7 days/week. By delivering the devices to the subject’s home, each person was insulated from other participants in the study and intersubject device comparison was avoided, which also aided in the blinded study design. The subjects were advised to use the device in any location in their home that was convenient for them. Subject compliance was recorded by an electronic monitor integrated within the device, which tabulated time, date, and duration of each treatment, throughout the 1-year period. After the 10-minute treatment period, both active and placebo devices shut off automatically. If the subject interrupted any given 10- minute period (e.g., stepping off to answer the phone), this disruption was detected through a plate surface pressure switch, signaling the device to emit an acoustic warning and the treatment would pause until the subject returned. If the subject did not return within 10 minutes, the device would record the time activated and automatically shut off.
 

No incentive was given for maximizing compliance, the device emitted no visible or audible warnings if daily use was undersubscribed, and the study was designed such that the investigators did not prompt the subjects to use the device. Percent compliance was measured as the total number of treatments in which the subject stood on the device for at least 8 minutes, divided by two times the number of days the devices were in the subject’s home times x 100.
 

Clinical assessment
 

Baseline BMD was determined by DXA (QDR 2000; Hologic, Waltham, MA, USA), with measurements taken at four skeletal locations: proximal right and left femora, lumbar spine, and the distal one-third of the nondominant radius. Subjects were phoned to come in for follow-up scans at approximately 3, 6, and 12 months. Care was taken to position the patient in the same way at each scan, and the same bone density technician performed each scan. A bone phantom was used to calibrate the DXA machine each day.(24) At baseline and completion of the study, to approximate change in bone remodeling status, serum and urine samples were taken, and markers of bone formation and resorption were measured. At completion of the study, a written “exit” questionnaire was requested from each subject, which asked about ease and convenience of use and whether, in the subject’s judgment, they were on a placebo or active device.
 

Statistical analysis
 

After 12 months of treatment, the primary outcome measure was, in subjects with at least 80% compliance, a significant difference between changes in BMD of the spine and femur in the active and placebo groups. Secondary outcome measures were serum indices of bone formation and resorption. The sample size was determined by anticipating a balanced study with a difference in bone density loss between active and placebo groups of 2% over 1 year, assuming a population SD of 2.4%. A final group size of 56 was calculated to be required to attain a power of 0.80 with an œ of 0.05. With a 10% drop-out rate projected (N = 6), a recruitment goal of 64 was set (N = 32 in each group). While the active/placebo status of the devices was not revealed, any subject who withdrew within the first 3 months of treatment was replaced by a subject who received the same device status.
 

The study results were analyzed in collaboration with an independent statistical consultant (Boston Biostatistics, Wellesley, MA, USA), and no data imputation was performed. The data were initially evaluated in an “intentionto- treat” analysis using the 12-month DXA scan or the scan at the last follow-up visit, and included the results of all subjects enrolled in the study, both treatment and placebo. Analyses were performed a priori using all subjects, first by simple population t-test, and second by multiple linear regression, with body mass and compliance as covariates. Posthoc analyses were performed for all subjects with baseline and 12-month DXA data and for whom full electronically recorded compliance data were available. In posthoc testing, the interaction of compliance and treatment was assessed in a linear effects model, with least square means generated at the specified compliance levels reflecting the intercepts of the three compliance quartile boundaries (59.1%, 76.6%, 85.9%). Because of the reported relationship between osteoporotic fracture risk and body build,(25) a three-way interaction of treatment, compliance, and subject weight (bisected at 65 kg; consistent with NHANES II body weights of females in this age range(26)) was investigated both in a linear effect model and by a simple t-test dichotomizing compliance at the 80% and 60% levels. p values less than 0.05 were considered statistically significant; no posthoc corrections were undertaken.
 

RESULTS
 

In total, 70 (33 active and 37 placebo) subjects were randomized into the study and were included in the intention-to-treat analysis. Six (one active and five placebo) subjects withdrew within the first 3 months and therefore had no DXA follow-up. Each of the six people who withdrew was replaced by a new subject who entered into the same treatment type. Of the 64 subjects who had at least two DXA measurements, 8 did not have a 12-month DXA scan; therefore, the remaining 56 subjects (28 active and 28 placebo each with a 12-month DXA scan) formed the a priori analysis group. Complete electronically recorded compliance data on 10 of the remaining 56 subjects were not available, and thus the per-protocol analysis (group used for posthoc analysis purposes) considers only the subset of 46 subjects (26 active and 20 placebo) where a full electronic record of compliance was available. There was one adverse reaction of treatment reported (headache), which came from a woman in the placebo group. All active devices were reassessed at the end of the study and found to be within 5% of the 30 Hz, 0.2g criteria, as per the original dynamic parameters at the initiation of the study. Furthermore, at the end of the 12-month period, the audible acoustic signal, intended to obscure the active/placebo status of the platform, was functioning in all devices.
 

At the completion of the study, the randomization code was broken, and a comparison of the two groups, active and placebo, was determined. Although the study was not powered to detect demographic differences, age, height, femur, and spine BMD at baseline were not significantly different between the groups. However, at baseline, the placebo group’s average weight was 5 kg higher than the active group (p < 0.03), and the body mass index (BMI) of the placebo group was 2 kg/m2 higher (p < 0.04; Table 1).
 

 

An intention-to-treat analysis of all 70 subjects was undertaken using a bootstrap technique to permit estimation of the response in subjects with incomplete data.(27) In neither the active nor placebo group did changes in bone density exceed the detection threshold of the study design. In the femoral neck, the active group lost 0.69% of their BMD versus a 0.27% loss in the placebo group. In the trochanter, the active group lost 0.07% of their BMD versus a 0.19% loss in the placebo group . In the lumbar spine, the active group lost 0.51% versus a loss of 0.65% in the placebo population (p = 0.45).
 

Fifty-six subjects (28 active and 28 placebo) had a 12- month DXA scan, and this group constituted the a priori study group. A wide range in compliance with device use was observed in this population, ranging from 1% to 95%. When the device was used, however, 98.4% of what constituted a complete treatment (>8 minutes) was a full 10- minute treatment. Thirty-seven percent of subjects completing the study were at least 80% compliant (10 active and 7 placebo), whereas 72% of subjects were at least 60% compliant (19 active and 14 placebo). Whereas the placebo population had consistently higher losses of BMD in the lumbar spine, femoral neck, and trochanter regions of the skeleton than that measured in the active treatment groups, no significant differences were observed on population averages.

Because of the large range of compliance, multiple regression analysis was performed on the a priori populations to identify the relationship between compliance and efficacy. Strong positive associations between device usage and changes in BMD were observed at all three sites of interest (Table 2). Using compliance and weight as covariates, BMD of the spine was found to increase 0.071% for each percent increase in compliance of device use (p = 0.0039). Projecting this correlation to an “idealized patient” who was 100% compliant, and assuming the bone remodeling response to be linear, this would correspond to a 7.1% increase in BMD over the course of the year. For the trochanter at 100% compliance, BMD would be projected to increase by 5.1% (p = 0.085), and for the femoral neck, BMD would increase at a projected rate of 1.8% over the course of the year (p = 0.54). Correspondingly, BMD changes in the placebo population demonstrated no association at the trochanter and lumbar spine and a negative association for the femoral neck (p = 0.001).

                                                                
 

Posthoc analysis of the per protocol group, examining efficacy at each intercept of compliance quartiles, used least square means generated at the specified compliance level for those subjects in that quartile and treatment group performed without corrections. Based on the suggestion of a treatment and compliance interaction as seen in Table 2, a linear prediction model was constructed to investigate the general influence of compliance (i.e., percent of total possible treatments completed; Table 3). A significant interaction of treatment and compliance was observed for femoral neck BMD changes (p = 0.06), with the active treatment showing a relative benefit over placebo of 2.17% when the subjects were 86% compliant. Similar observations are seen at the trochanter (relative benefit of 1.23% at 86% compliance; p = 0.21) and at the lumbar spine (1.5% relative benefit; p = 0.09). Factoring in weight improves the efficacy of treatment, with the benefit of treatment ranging from 2% to 3% at all three sites, with p values ranging from 0.19 to 0.009.

 Considering weight as an interacting influence on spine BMD, the subjects were stratified into groups above and below 65 kg (Fig. 2; Table 4). In the lower-weight cohort, in the highest quartile of compliance (86%), there was a 3.17% loss of bone in the spine in the placebo group compared with a 0.18% gain in BMD in the active group, suggesting a 3.35% relative benefit of treatment (p < 0.009; Table 3). Similarly, in this lower-weight, high-compliance group for the femoral neck, there was a 2.23% loss over the course of the year in the placebo group compared with a 0.13% loss in the active group, representing a 2.1% relative benefit of treatment. For the trochanter, the relative benefit was 1.92% over the course of 1 year of treatment.

 

Figure 3 provides a plot of the quartiles derived from the linear modeling with the placebo group providing the mean of the three-quartile values for each treatment site. In the lumbar spine, a 0.1% loss in the highest quartile of compliance was relatively better than the 1.55% loss experienced by the lowest compliance group. This 1.55% loss in the lowest compliance group was similar to the 1.76% loss measured in subjects standing on a placebo device. In the trochanter region, a 0.76 gain was determined for the highest compliance group, whereas a 0.5% loss was experienced by the lowest compliance group, a loss that was similar to the 0.71% loss observed in the placebo group. The femoral neck, as well, demonstrated a dose-dependent response with a 0.04% gain in the highest-compliance group versus a 1.18% loss in the low-compliance group. This 1.18% loss was similar to the 1.24% loss measured in the placebo group. In the distal radius, there were no significant differences between any of the compliance groups and the placebo group.
 

Serum indices of bone formation and resorption were evaluated at baseline and at the end of the study to determine if the mechanical intervention influenced bone remodeling activity. Dietary calcium (self-reported) was the only variable that seemed significantly different at baseline. At 12 months, hydroxyproline levels fell 16% in the placebo group but only 3% in the active group, reflecting a 13% difference (p = 0.07). Phosphorus (baseline value = 3.7) was up 1.3% in the active group but fell 4% in the placebo group, reflecting a 5% difference (p = 0.08). No significant changes were seen in bone-specific alkaline phosphatase (which went up in both groups), total alkaline phosphatase (which went down in both groups), creatinine (which did not change), osteocalcin, or parathyroid hormone (PTH). Every 3 months, either by telephone or visits to the Center, patients were asked if they exercised more or changed any other aspect of their lifestyle. No trends were identified.
 

In their exit interviews, the subjects expressed concern that two 10-minute/day treatments were difficult to schedule but that they may be more encouraged to use the device if efficacy was demonstrated and if a single use per day were possible. Approximately 20% of the active subjects guessed incorrectly in terms of whether they had an active device, and 30% of the placebo subjects guesses were incorrect as to the status of their device.
 

DISCUSSION
 

This study examines the safety and potential efficacy of a very-low-magnitude physical stimulus to inhibit loss of BMD, which is based on the musculoskeletal system’s strong sensitivity to mechanical stimuli. The physical stimulus is imposed noninvasively into the weight-bearing skeleton through ground-based accelerations. The nature of the vibratory stimulus is based on providing a surrogate for the spectra of high-frequency muscle-based signals that attenuate with aging.(20) In addition to large amplitude mechanical forces (and resultant strains) associated with vigorous activity, smaller magnitude strain signals are continually evident in bone,(14,28) and it is these signals that we are trying to mimic. When the 12-month human data are considered in an a priori analysis, the results indicate a potential benefit of treatment strongly dependent on compliance, as standing on the device for close to 20 minutes/day was associated with a greater ability to prevent bone loss. Using linear regression analysis to determine the effect of full 100% compliance indicates that an “idealized” subject who used the device for the full 20 minutes/day could have up to 7% higher lumbar spine BMD and 5% higher BMD in the trochanter than those who did not use the device at all. Compliance, however, is difficult to ensure in any study,(29) and strategies to improve use must be considered.
 

The exit interviews indicated that a “twice per day” regimen made it difficult to fit into a working schedule. Possibly, exposure time could be reduced if the potency of the mechanical signal could be increased, perhaps by increasing the amplitude to above 0.2g, which may take advantage of the interdependence of cycle number and strain magnitude,(30) or to identify alternative frequencies or waveform combinations that may be more effective.(31) Examining subject commitment to a shorter treatment duration, a recent feasibility study has shown that, over 6 months of treatment in an elderly female population (75–90 years old), using a 10-minutes/day, 30 Hz stimulus at 0.3g, a mean compliance of 93% was maintained.(32) Considering the difficulty in fitting in two 10-minute treatment regimens, it is also possible that compliance would have been improved had a single 20-minute session been used.
 

Posthoc analysis indicates that this intervention may be more effective in lighter women than in heavier women, particularly in the spine (Fig. 2). Considering that BMD is positively correlated with body mass,(25) these data in turn also suggest that the mechanical stimulus works best in those women with lower BMD (i.e., effective in women who require it), specific to those skeletal sites that need treatment (no significant differences were observed in the radius between active and placebo subjects). The individualized “sensitivity” to the mechanical signal is consistent with findings in the mouse, where the anabolic potential of the mechanical stimulus is realized in inbred strains with low bone density (e.g., B6), whereas there is only low responsivity to altered mechanical environments in the high-density strains (e.g., C3H).(33)
 

This study indicates that low-level mechanical stimuli may have the potential to prevent bone loss in the postmenopausal population, but failed to stimulate the formation of bone. In contrast, the stimulus used in this study was shown in animal studies to be strongly anabolic,(17–19) an observation supported by recent work addressing the effects of 0.3g vibration on bone density in children with cerebral palsy(34) and adolescent females (10–13 years old) in the lowest quartile of BMD.(35) Whether the anabolic response was observed because the signal was delivered to the skeleton of children rather than adults or because the amplitude was 50% greater (0.3g rather than 0.2g) is not yet clear. Considering that the bone strain resulting from these vibrations are two orders of magnitude below those levels that initiate microdamage,(36) this indicates that anabolism can be achieved without putting the skeleton at structural risk. With this in mind, it is relevant to note that in a recent study reported by Torvinen et al.,(37) vibration 40 times greater than the signals examined here (8g as opposed to 0.2g) failed to stimulate any form of bone response. Whether this was because the study was relatively brief (8 months), used healthy young adults (and therefore there was no “signal” lacking that required replacement), or that the amplitude was so great as to be beyond any form of physiologic relevance (as in light that is too bright, sound that is too loud, or pressure that is too great), is difficult to determine at this point.
 

No adverse reactions were reported in the active group. Nevertheless, vibration of the human body is undeniably a complex issue,(38) and considering the variety of pathologies it may exacerbate, including low back pain,(39) circulation disorders,(40) and/or neurovestibular dysfunction,(41) it must be approached carefully. ISO 2631 gives “provisional guidance as to acceptable human exposure” to whole-body vibration in the 1- to 100-Hz band for a sitting or standing person,(42) defining numerical values of the “fatigue-decreased proficiency boundary” over a 24-h period. Sinusoidal frequencies in the range of 25–32 Hz allow for a 4-h exposure at 0.4g, well exceeding acceleration levels and times under investigation with this device. The safety of signals that exceed 1g, for even a short duration, may be of some concern.(43)
 

There is general perception within the skeletal disciplines that signals must be large to represent a positive influence on bone mass and morphology.(44) These data support the premise that extremely small mechanical signals may also be capable of serving as a regulatory influence on skeletal architecture, the “outcome” of which seems to be a more uniform distribution of stresses in trabecular bone under load.(45) This regulatory influence may be achieved directly, by mechanical strain, or indirectly, through amplification of the signal by intramedullary pressure(46) or fluid flow(47) in the bone tissue. Alternatively, the regulatory response may be regulated through a system such as neuromuscular feedback amplified by the low-level signals exceeding a stochastic threshold(48) or by stimulating skeletal muscle pump activity, resulting in significant effects on circulatory flows and fluid flow through the bone tissue.(49) Even considering the complicated nature of the physical mechanism, there can be little doubt that the biological means of controlling bone adaptation are even more complex.(50)
 

Bone architecture is but one of several critical risk factors associated with long bone fractures. For example, postural stability and muscle strength contribute to fracture risk on a par with BMD.(51) If the physical stimulus investigated here does represent a surrogate for the signals lost by sarcopenia, it is entirely possible that the muscle may benefit from treatment as well, enhancing muscle strength,(52) and coupled with the neurovestibular system, improve postural stability.( 48)
 

This prospective, randomized, double-blind, and placebocontrolled study has provided important preliminary results, and clinical support for the hypothesis that extremely low level physical stimuli may provide an effective means to inhibit bone loss, particularly for those who cannot or will not comply with traditional pharmacologic interventions for osteoporosis.(53)
 

ABSTRACT.
 

Rønnestad, B.R. Comparing the performance-enhancing effects of squats on a vibration platform with conventional squats in recreationally resistance-trained men. J. Strength Cond. Res. 18(4):000-000. 2004.—The purpose of this investigation was to compare the performance-enhancing effects of squats on a vibration platform with conventional squats in recreationally resistance-trained men. The subjects were 14 recreationally resistance-trained men (age, 21–40 years) and the intervention period consisted of 5 weeks. After the initial testing, subjects were randomly assigned to either the ‘‘squat whole body vibration’’ (SWBV) group (n = 7), which performed squats on a vibration platform on a Smith Machine, or the ‘‘squat’’(S) group (n = 7), which performed conventional squats with no vibrations on a Smith Machine. Testing was performed at the beginning and the end of the study and consisted of 1 repetition maximum (1RM) in squat and maximum jump height in countermovement jump (CMJ). A modified daily undulating periodization program was used during the intervention period in both groups. Both groups trained at the same percentage of 1RM in squats (6–10RM). After the intervention, CMJ performance increased significantly only in the SWBV (p < 0.01), but there was no significant difference between groups in relative jump height increase (p =0.088). Both groups showed significant increases in 1RM performance in squats (p < 0.01). Although there was a trend toward a greater relative strength increase in the SWBV group, it did not reach a significant level. In conclusion, the preliminary results of this study point toward a tendency of superiority of squats performed on a vibration platform compared with squats without vibrations regarding maximal strength and explosive power as long as the external load is similar in recreationally resistance-trained men.

KEY WORDS. whole body vibration, resistance training, strength adaptations, squat, CMJ
 

INTRODUCTION
 

Lately, it has been hypothesized that mechanical vibration at a low amplitude and high frequency of the whole body can positively influence muscle performance (8–10, 15, 49, 55, 57–59). Nazarov and Spivak (38) were among the first to highlight the association between strength and power development and whole-body or segment-focused vibration training. They assumed that repetitive, eccentric vibration loads with small amplitudes would effectively enhance strength, because of a better synchronization of motor units. In the last decade, remarkable enhancements in strength and power after vibration training have been presented. A single vibration bout has been shown to result in acute and temporary effects when it comes to muscle power and/or strength of the lower extremities and arm flexors
 

The mechanisms mediating this acute effect of vibration on neuromuscular performance are not entirely understood. The mechanical action of vibration mediates fast and short changes in the length of the muscle-tendon complex. This may induce a nonvoluntary muscular contraction termed the ‘‘tonic vibration reflex’’ (TVR). TVR is believed to depend upon the excitation of the primary muscle spindle (Ia) fibers (11, 18, 35, 47). Thus, potential extra excitatory inflow during vibration stimulation is partly related to the reflex activation of the a-motoneuron. Accordingly, researchers have reported an increase of the root mean square EMG (EMG_rms) of the biceps brachii muscle in boxers exercising with a vibrating dumbbell that was twice as high as a voluntary arm flexion with a load equal to 5% of the subject’s body mass (7). Also, Torvinen et al. (57) found an increase in EMG_rms in the calf muscles during whole body vibration (WBV). In accordance with the latter, studies have demonstrated a facilitation of the excitability of the patellar tendon reflex by vibration applied to the quadriceps muscle (12), vibration induced drive of a-motoneurons via the Ia neuron loop (48), and activation of the muscle spindle receptors after applying vibrations (30). However, if muscle spindles are stimulated for a long period of time by vibration, they will finally fatigue (6). This, in turn, is seen as reduction in EMG activity, motor-unit firing rates, and muscle contraction force. It is possible that the ideal vibration period to achieve acute strength/power gains is individual, thus fatigue may explain why some of the studies find no positive effect after one acute bout of vibration (16, 45, 58). However, a confounding explanation exists. Vibrations also seem to depress some monosynaptic spinal reflexes (e.g., H-reflex) (17, 34). The decrease in the reflex is primarily related to a presynaptic inhibitory mechanism, involving a depolarization of Ia afferents (21). The practical effects of these reflexes regarding resistance training are unclear.
 

Some studies have examined the effect of WBV training on muscle performance over a longer period. Bosco et al. (8) studied the effect of a 10-day training program with daily series (5 x 90 seconds) of vertical sinusoidal vibrations at a frequency of 26 Hz on subjects who had no previous experience with resistance training. They found significant improvement in the height and mechanical power during a 5-second continuous jumping test. However, a period of 10 days is too short to determine the long-term effects of WBV. Runge et al. (49) presented gains of 18% in chair-raising time in fit elderly persons after 8 weeks of WBV training (3 times a week at 27 Hz). Recently, Torvinen et al. (59) presented a study of 8- month WBV (4 minutes per day, 3–5 times per week, with 25–45 Hz). The subjects were young and healthy nonathletic adults. They found a significant 7.8% improvement in vertical jump height in the vibration group. On the isometric extension strength of the lower extremities, grip strength, shuttle run, and postural sway the vibration intervention had no effect. Similar results have been presented after 4 months of WBV training with an identical training protocol (58).
 

Neither of the studies mentioned above compared the performance-enhancing effects of WBV with those of conventional resistance training, so we cannot tell if there is a difference in strength improvement between the two training methods. However, other studies have compared these 2 training methods for a longer period (6–12 weeks), and have concluded with similar and significant improvement in strength regarding WBV and conventional resistance training with moderate intensity (15, 55). Both these studies included only untrained subjects, and untrained people improve their strength dramatically in the beginning of a strength-training period (40). Thus, if there are any differences in strength gain between the WBV training and conventional resistance training, it is difficult to detect it in previously untrained subjects. Regarding conventional resistance training, studies indicate that training at an intensity similar to 80–90% of 1 repetition maximum (1RM) is best for improving strength (4, 64). The studies of Delecluse et al. (15) and Schlumberger et al. (55) did not carry out conventional resistance training in this intensity zone, so it can be claimed that it was not an optimal strength training regime. Issurin et al. (26) took the latter into consideration when they studied the effects of ‘‘vibratory stimulus training’’ on strength, using a ‘‘sitting bench-pull apparatus’’ with 44 Hz vibration frequency 3 times per week for 3 weeks with men who had not previously trained on resistance exercises. A control group performed exactly the same training protocol except from the vibration stimulus (6 sets of sitting bench-pulls with the load gradually increasing from 80 to 100% of 1RM). The group using vibration showed an increase in maximum strength of 49.8%, whereas the group using conventional resistance training without vibration showed an improvement of 16.1%.
 

In the latter study, the vibration training induced significant greater strength improvement compared with conventional resistance training. Because WBV training is used by professional athletes (9, 27, 33, 36), it is of great interest to repeat the study of Issurin et al. (26) on resistance-trained subjects. Thus, the purpose of this study was to compare the effects of squats performed on a vibration platform (VP; NEMES-LC, Ergotest, Rome, Italy) with conventional squats without vibrations on 1RM and countermovement jump ([CMJ]; a measurement of explosive strength after stretch shortening of the muscles), in resistance-trained men during a 5-week overreaching period of peaking. Both groups trained with a load equal to 6–10RM. With the results of Issurin et al. (26) in mind, it was hypothesized that squats performed on a VP are superior to conventional squats when the subjects are training with the same external load on the Olympic bar.
 

METHODS
 

Experimental Approach to the Problem
 

To address the question of whether squats performed on a VP are superior to conventional squats without vibrations in resistance-trained men, the effects of 5 weeks with squat training on 1RM and CMJ were compared. Both groups trained at the same intensity (number of RM); the only difference was that 1 group performed the squats on a vibration platform (Figure 1). The subjects carried out all squats (both testing and training), in both groups, on a Smith Machine (Gym Bo, Gelsenkirchen, Germany) to avoid a balance problem on the VP during the squats.
 

Subjects
 

Sixteen men (age, 21–40 years; height, 177.8 ± 6.5 cm; weight, 76.2 ± 8.8 kg) served as subjects. Two subjects withdrew before completion of the study, due to causes unrelated to the study. All subjects had participated regularly in resistance training (minimum 3 times a week during the last year) and completed at least 1 bout of squats each week. To be included in the study, the lifters had to lift at least 2.2 times their body weight in a 1RM squat. To make sure there were no differences in training periodization, the subjects provided written information about their training regimen during the last year. Full advice was given to the subjects regarding the possible risk and discomfort that might be involved, and the subjects gave their written informed consent. The study was approved by the Regional Ethics Committee of the Norwegian Research Council for Science and Humanities.
 

Subjects were randomly divided into 2 different training groups. The ‘‘squat whole body vibration’’ (SWBV) group (n = 7) trained squats on the VP on a Smith Machine. The ‘‘squat’’ (S) group (n = 8) trained conventional squats (without a VP) on a Smith Machine.
 

Testing was administered at the beginning and at the end of the 5-week training intervention. Because all the subjects had completed at least 1 bout of squats per week during the last year, we did not spend time on familiarization with the squat exercise. The order of tests was similar before and after the training intervention. The posttests were accomplished at approximately the same time of the day as the pretests, 3 days after the last workout to avoid acute effects of WBV and to reassure proper recovery after the last workout. All subjects completed at least 91% of the workouts.
 

Training
 

The 5-week training period consisted of 3 workouts during the first, third, and fifth weeks, and 2 workouts during the second and fourth weeks. The subjects completed 13 workouts on nonconsecutive days (Table 1). Each subject performed a standardized 10-minute aerobic warmup before each workout; 2–3 warm-up sets of squat were also performed with gradually increased weight. All subjects were supervised by the investigator at every workout during the first 2 training weeks, and thereafter at least once a week.
 

Training volume (total reps performed) and intensity (RM) were altered similarly for the 2 groups. During the first week, both groups performed 3 sets of 10RM in each bout of exercise, during the second and third training week they completed 4 sets of 8RM, and during the last 2 weeks they trained with 4 sets with 6RM (Table 1). Subjects were encouraged to continuously increase their RM loads during the intervention. Subjects were allowed assistance on the last rep. However, to achieve a modified daily undulating periodization, the subjects were told to reduce their load on the Olympic bar by 10% approximately every third workout (this was coordinated between the 2 training groups). Daily undulating periodization is characterized by frequent alterations in the intensity and volume (43, 44). This program seems to place considerably stress on the neuromuscular system, because of the rapid and continuous change in program variables (44), and thus elicits greater strength gains than a linear periodized program. The subjects in SWBV group performed their squats on a VP with a frequency of 40 Hz. Subjects were prohibited from performing any other strength-building exercises on the legs during the 5-week training intervention.
 

Testing
 

We used 1RM as a measure of pretraining strength in squats. Squat testing and training was performed on a Smith Machine. The pre- and posttesting was done on the same equipment with identical subject-equipment positioning overseen by the same trained investigator.
 

Jumping Measurements
 

The subjects performed a 10-minute warm-up, consisting of cycling at a workload of 60–70 W. Thereafter they performed 4 trials of CMJ. The flight time of each single jump was recorded using an infrared light mat (Muscle Lab, Ergotest Technology A.S, Langesund, Norway), interfaced to a personal computer. To avoid immeasurable work, horizontal and lateral displacements were minimized, and the hands were kept on the hips throughout the jumps. During CMJ, the angular displacement of the knees was standardized so that the subjects were required to bend their knees to approximately 908. The obtained flight time (t) was used to estimate the height of the rise of body center of gravity (h) during CMJ (i.e., h = gt²/8, where g = 9.81 m·s^-2). The coefficient of variation regarding test-retest reliability for a similar test has been found to be 4.3 % (63). The best performance was used for statistical analysis.
 

1RM Measurement
 

Before the 1RM squat test, subjects performed a standardized warm-up consisting of 3 sets with a gradually increasing load (40, 75, and 85% of expected 1RM) and decreasing number of reps (12, 7, and 3). The knee-angle during the 1RM squat had to be 908 to be accepted. To assure similar knee angle in the pre- and posttest for all the subjects, the subjects’ squat depth was individually marked at the pretest depth of the buttock on a list. Thus, the subject had to reach his individual depth (touch his list with the buttock) in the posttest to get his lift accepted. The first attempt in the test was performed with a load approximately 5% below the expected 1RM load. After each successful attempt, the load was increased by 2–5% until failure in lifting the same load in 2–3 consecutive attempts. The rest period between each attempt was 3 minutes. The coefficient of variation for test-retest reliability for this test has been found to be <2% (41).

Statistical Analyses
 

All values given in the text, figure, and tables are mean ± SD. Paired t-tests were used for within-groups comparisons, and unpaired t-tests were used to compare the relative changes in strength and jump height between groups. Bonferroni adjustments were made to account for tests of 2 variables. Thus, p values of 0.025 were used for each of the 2 variables (1RM and CMJ).
 

1RM test
 

There was no significant difference between the groups at the pretest in 1RM. In both groups, 1RM squat increased during the training intervention (p < 0.01, Table 2). Although there was a trend toward a greater relative strength increase in the SWBV group compared with the S group (32.4 ± 9.0% vs. 24.2 ± 3.9%, respectively; p = 0.046), it did not reach a significant level when Bonferroni adjustments were made (Table 2).
 

 

CMJ test
 

There was no significant difference between the groups at the pretest. Only the SWBV group significantly improved their jump height (p < 0.01, Table 2), but there was no significant difference between groups in relative jump height increase (p =0.088).
 

DISCUSSION
 

This is the first study on resistance-trained subjects that compares the effects of WBV training and conventional resistance training on 1RM in squats and maximal CMJ, where the external load is similar between 2 groups. The preliminary results of this study point toward a trend in which squats performed on a VP is superior to conventional squats regarding maximal strength and explosive power. It seems that this advantage depends on heavy external loading in addition to WBV. Both groups increased their 1RM in squats during the training intervention, and the relative strength increase was greater in the SWBV group than the S group (32.4 ± 9.0% vs. 24.2 ± 3.9%, respectively; p = 0.046). The jumping performance, CMJ, was significantly improved in the SWBV groups, but there was no significant difference between the groups (p = 0.088). It may be speculated that the lack of significant differences between the groups is related to the fact that this study contains only 7 subjects in each group and the intervention lasted only 5 weeks.
 

Several other studies have found positive effects of WBV on CMJ (8, 15, 56, 58, 59). In contrast to the present study, Delecluse et al. (15) found in untrained subjects that WBV training is superior to conventional resistance training when it comes to improvement in CMJ. However, in the latter study, there was a significant higher CMJ performance recorded in the conventional resistance group compared with the other groups in the pretest condition. Thus, it may be argued that the potential for progression in CMJ was smaller for this group.
 

The first adaptation mechanism of a skeletal muscle to resistance training is believed to be neural change, due to an almost immediate increase in strength at the onset of training and the absence of (measurable) hypertrophy (3, 13, 51). The exact mechanism by which resistance training can improve neuromuscular activation is not known, but there are several possible explanations which could cause this enhancement (e.g., increase in motor unit synchronization, co-contraction of the synergistic muscles or increased inhibition of the antagonist muscles [52]). These explanations have also been used to explain the effects of WBV on jumping performance (7, 15, 58, 59). All these studies were accomplished with subjects who had no previous resistance training, and neural adaptation seems to dominate in the early adaptation phase of resistance training (51, 53). The present study was carried out with resistance-trained men, with whom the neural adaptation phase should have reached a plateau. However, neural adaptation can not be ruled out, because of the specificity principle: a change in the training program, such as different exercises and/or intensity, could trigger a transient burst of neural and muscular adaptations (52). The modified daily undulating periodization training regime and the introduction of vibration training could potentially result in neural adaptations. This is supported by the relatively great improvement in 1RM strength in both the S and SWBV groups (24.2 ± 3.5 and 32.4 ± 8.9%, respectively). In line with this, Ha¨kkinen et al. (24, 25) found increased integrated EMG activity in elite weightlifters, indicating the importance of neural adaptations in experienced strength and power athletes.
 

The trend toward superiority of the SWBV group regarding 1RM strength in this study is in accordance with earlier results with untrained subjects. Issurin et al. (26) found that, with previously untrained subjects, applying vibrations (44 Hz) while training with a load 80–100% of 1RM, is superior to training with the same external load without vibrations. However, Delecluse et al. (15) and Schlumberger et al. (55) compared conventional resistance training with WBV training and found no differences regarding strength improvement. This result may have been caused by the lack of external load in the WBV group. Other studies have not found improvement in maximum strength after WBV interventions (16, 54, 58– 60). The reason is unclear, but the lack of external load in all these studies may indicate that this is important to achieve strength gains after WBV training.
 

The mechanisms mediating the apparently superior effect of performing squats on a VP vs. conventional squats, regarding 1RM strength, are not fully understood. An increase in isometric contraction strength induced by TVR has been well documented after local vibratory stimulation applied to the tendon or muscle (1, 18, 29). Armstrong et al. (2) found similar results when subjects were holding a cylindrical handle vibrating at 40 Hz, resulting in 52% increase in grip strength. The TVR may have contributed to the results of Bosco et al. (7), who found an increase of the EMG_rms of biceps brachii muscle in boxers who were exercising with a vibrating dumbbell twice as high as a voluntary arm flexion, with a load equal to 5% of the subject’s body mass. Also, Torvinen et al. (57) found an increase in EMG_rms in the calf muscles during vibration. In accordance with the latter, studies have demon strated a facilitation of the excitability of the patellar tendon reflex by vibration applied to quadriceps muscle (12), vibration induced drive of a-motoneurons via the Ia loop (47), and vibration activation of the muscle spindle receptors (30). Rittweger et al. (46) also found significantly greater EMG mean frequency over the vastus lateralis after exercise with vibrations than without vibrations. These studies indicate that exercising with vibrations achieves superior excitation of the motoneurons to exercising without vibrations. Sale (50) suggested that full activation of the muscle may lead to motor unit fatigue, and due to this training effect, may increase the strength. The motor units in the SWBV group did perhaps get more fatiguing stimulus because of increased TVR, and thus superior gains in 1RM compared with the S group.
 

The alpha-motoneuron is the final point of summation for all the descending and reflex inputs, and the net membrane current of this motoneuron determines the discharge pattern of the motor unit and thus the muscle activity (37). De Gail (14) states that TVR is able to cause an increase in recruitment of the motor units through activation of muscle spindles and polysynaptic pathways. In addition, the WBV waves propagate from the distal links to muscles located proximally and activate a greater number of muscle spindles. Their discharge activates a larger fraction of the motor pool and recruits many previously inactive motor units into contraction (27). This increased activity of motor units may enable the SWBV group to train with heavier loads than the S group, and thereby optimize the stimulation of higher recruitment threshold motor units and muscle tissue mass with each workout (42).
 

Another possible explanation concerns the difficulty in achieving full muscle activation by voluntary effort during dynamic exercise, when large muscle groups are involved (28). It is likely that the vibrations may cause partial activation of the muscles, and their mobilization at the beginning of the effort will be faster. Thus it is possible that the group which trained with vibrations could train with heavier loads and get a better stimulus for strength increase. Evidence also indicates that voluntary activation is a limiting factor in force production, and that improvements in force generated per unit cross-sectional area are responsible for the initial gain in strength (20). The possibility of enhanced capacity of the muscle to perform work when vibrations are applied simultaneously with external load was demonstrated by Liebermann and Issurin (33). The 1RM in isotonic elbow flexions for Olympic athletes increased significantly (8.3%) while applying vibrations (44 Hz) to the maximum lift, compared with conventional maximum lift without vibrations. Similar results were presented by Issurin and Tenenbaum (27). They found significant increase in mean and maximal power in elite athletes when vibration was applied (44 Hz). This is in accordance with the result of this study, where the SWBV group tended to train with a higher percent of their 1RM, compared with the S group, although this difference was not significant (data not shown). Thus, it seems as the vibrations increases the intensity of the lift rather than reduce it.
 

Although not measured in this study, a certain degree of hypertrophy may be expected after 5 weeks of intensive resistance training (56). In rats, a vibration-induced enlargement of slow- and fast-twitch fibers has been demonstrated (39). Thus it is possible that the vibrations gave an extra hypertrophy stimulus. Another potential explanation is that the vibrations resulted in greater stretch/ tension on the contractile elements (either directly through the TVR itself, or by increased capacity to lift heavier loads via the TVR). Stretch/tension seems to be an essential stimulus for muscle growth
 

Another stimulus for muscle growth is the androgen hormone testosterone. Testosterone is able to affect muscle growth via increased amino acid uptake and protein synthesis in the muscle cells (5, 19, 23, 61). Bosco et al. (10) found that acute exposure to WBV causes increased plasma concentrations of testosterone. The same acute testosterone response is also seen after a single bout of resistance exercise when the workout involves large muscle groups, relative heavy resistance (85–95% of 1RM), moderate to high volume of exercises, and short rest intervals between the sets (31). Whether the addition of vibrations in the SWBV group induced a larger testosterone response than the S group is not known.
 

It may be argued that differential psychological factors due to training on the VP might affect the motivation, and because of that promote greater effort in each single session in the SWBV group compared with the S group. This study did not control for psychological factors, but the results of Delecluse et al. (15) indicate no placebo effect of vibration training.
 

The study design makes it impossible to answer the reasons behind the tendency of difference in 1RM gain between the 2 groups, because no neurogenic enhancement or changes in the morphological structure of the muscles could be demonstrated (neither EMG recordings nor muscle biopsies were performed).
 

In conclusion, this preliminary study on recreationally resistance-trained men indicates that CMJ height was significantly increased only by the squats performed on the VP. Both training interventions led to a significant improvement regarding 1RM in squats. There was a tendency toward superior 1RM improvement in the SWBV group, compared with the S group, but this did not reach a statistically significant level (p = 0.046). Possible explanations for this tendency toward differences in training adaptations may be related to neural adaptation, TVR, or a more favorable hormone milieu regarding muscle growth during the SWBV strength-exercise protocol.
 

The above-noted findings suggest that vibration is a potentially efficient training stimulus. Future studies should include a sufficient number of subjects and focus on comparing the long-term effects of WBV with external loads to conventional resistance training to explore the mechanisms behind these apparent differences.
 

PRACTICAL APPLICATIONS
 

This study indicates that when recreationally resistancetrained men perform squats with the same external load, there is a tendency toward superiority of squats performed on a VP compared with conventional squats without vibrations regarding 1RM in squat and maximal CMJ height. Consequently, it seems as though optimal strength gains in resistance-trained subjects are achieved by adding vibration to the conventional resistance training. This superior effect of vibrations on strength seems to depend on relatively heavy external resistance (6– 10RM). Therefore, instructions from a qualified instructor are advised before adding the relatively heavy external load needed to optimize strength gains.
 

ABSTRACT
 

The short-term effects of whole-body vibration as a novel method of somatosensory stimulation on postural control were investigated in 23 chronic stroke patients. While standing on a commercial platform, patients received 30-Hz oscillations at 3 mm of amplitude in the frontal plane. Balance was assessed four times at 45-min intervals with a dual-plate force platform, while quietly standing with the eyes opened and closed and while performing a voluntary weight-shifting task with visual feedback of center-of-pressure movements. Between the second and third assessments, four repetitions of 45-sec whole-body vibrations were given. The results indicated a stable baseline performance from the first to the second assessment for all tasks. After the whole-body vibration, the third assessment demonstrated a reduction in the root mean square (RMS) center-of-pressure velocity in the anteroposterior direction when standing with the eyes closed (P < 0.01), which persisted during the fourth assessment. Furthermore, patients showed an increase in their weight-shifting speed at the third balance assessment (P < 0.05) while their precision remained constant. No adverse effects of whole-body vibration were observed. It is concluded that whole-body vibration may be a promising candidate to improve proprioceptive control of posture in stroke patients.
 

Key Words: Cerebrovascular Accident, Balance, Vibration, Proprioception
 

There is a general and strong belief that somatosensory stimulation (SSS) promotes brain plasticity, although the underlying mechanisms are not well known.1 Studies regarding the functional recovery in stroke patients have suggested beneficial effects of SSS in terms of motor functions, balance, and activities of daily living. In 1993, a randomized, controlled trial first indicated that electro- acupuncture applied at the paretic body side facilitated recovery of balance, mobility, and activities of daily living in postacute stroke patients, 2 and some of these effects were shown to persist for up to 2 yrs after stroke as assessed by posturography. 3 However, recently, a new randomized, controlled trial from the same research group no longer demonstrated differential effects on motor function, mobility, or activities of daily living of either high-intensity– low-frequency transcutaneous electrical nerve stimulation or electroacupuncture applied at the paretic body side when compared with sham–transcutaneous electrical nerve stimulation of low intensity and high frequency.4 These differences in outcome could not be unambiguously explained. Although others have demonstrated potentially strong immediate effects of transcutaneous electrical nerve stimulation applied at the contralesional side of the neck on postural orientation and stability while sitting in postacute stroke patients with spatial neglect,5,6 the long-term effects of transcutaneous electrical nerve stimulation or (electro-)acupuncture on functional recovery from stroke are, as yet, equivocal. The effects of other forms of SSS are still unknown.
 

One form of SSS that shows considerable promises for rehabilitation is vibration therapy. Priplata et al.9 reported that randomly vibrating insoles could ameliorate age-related impairments in balance control. Apparently, vibration is able to enhance postural stability in persons without specific neurologic diseases. It is, therefore, possible that in stroke patients intensive and deep stimulation of muscle afferents using vibration may induce stronger sensorimotor recovery than more superficial stimulation of (sub)cutaneous afferents by electrical stimulation. In addition, based on functional magnetic resonance imaging and positron emission tomography studies showing plastic changes in both cerebral and cerebellar hemispheres after unilateral stroke, it may be that application of SSS at both sides of the body is more effective than only at the paretic side. Against this background, the purpose of this study was to demonstrate beneficial short-term effects of whole-body vibration (WBV) on postural control in chronic stroke patients and to register any possible adverse effects.
 

METHODS
 

Subjects.
 

A total of 23 ischemic stroke patients (13 men, 10 women; mean age 58.1 ± 11.4 yrs) with a poststroke interval of at least 6 mos (mean interval 23.3 ± 11.4 mos) were recruited from an outpatient population of a rehabilitation center to participate in this research study. Eight patients had a right-hemisphere lesion, and 15 had a left-hemisphere lesion.
To be included, patients must have only one stroke in their lifetime, be able to stand without support for 30 secs, understand the goal and methods of the study, and give their informed consent. Patients with non–stroke related sensory or motor impairments and those with medication that could interfere with postural control were not included. Also patients with contraindications for WBV (pregnancy, recent fractures, gall or kidney stones, malignancies, cardiac pacemaker, recent thromboembolic or infectious disease) and patients already treated with WBV were excluded. To obtain reference values for postural control, 23 healthy, elderly controls (mean age 63.9 ± 9.3 yrs) were recruited, mostly relatives of employees of the rehabilitation center. The Committee on Research Involving Human Subjects, Region Arnhem-Nijmegen, approved the study methods, and all patients gave their written informed consent according to specified guidelines.
 

Intervention.
 

All patients were subjected to one series of four consecutive repetitions of 45-sec WBV with a 1-min pause between administrations. WBV was provided with a commercially available device (Galileo 900, Galileo2000, Enschede, The Netherlands) (Fig. 1). This apparatus consists of a moveable rectangular platform built within a circular ground surface on which a support bar is mounted at the front. The platform makes fast oscillating movements around a sagittal axis in the RMS middle. Subjects were required to stand on the platform with their feet at an equal and standardized distance from the axis of rotation so that the vibration amplitude was approximately 3 mm. The frequency was set at its maximum of 30 Hz. While standing on the vibration platform, subjects were instructed to maintain a “squat” position with slight flexion at the hips, knees and ankle joints, to damp the vibrations approximately at the pelvic level. They were allowed to hold the support bar in front of them. An experienced therapist guided all WBV administrations.
 

Assessments.
 

Postural control and symmetry were assessed in terms of center-of-pressure (COP) movement and position registered with a dualplate force platform (LM-100KA, Kyowa Electronic Instruments) connected to a personal computer, sampling vertical ground reaction forces at a rate of 60 Hz. During all balance registrations, patients stood barefoot on the force platform with their arms (if possible) alongside their trunk and their feet against a fixed foot frame with an interheel distance of 8.4 cm and a toeing-out angle of 9 degrees. Every balance assessment consisted of two consecutive test series. Each series incorporated two 30-sec quietstanding tasks (one with eyes open and one with eyes closed) and one 30-sec voluntary weight-shifting task in a fixed sequence (eyes open, eyes closed, weight-shifting task), which was repeated in the reversed order (weight-shifting task, eyes closed, eyes open). During quiet standing with the eyes closed, patients were wearing a pair of closed, dark goggles. From each quiet-standing registration, the overall COP was calculated for every set of force samples from the two plates. Then, the lateral deviation of the mean position of this COP from the sagittal midline, relative to the base of support width, was determined as a measure of weightbearing asymmetry. The RMS COP velocity in both the anteroposterior and lateral directions was calculated as a measure of postural instability, because it integrates changes in COP amplitude and frequency. Indeed, of various posturographic measures, the COP velocity has been shown to be most reliable, sensitive to task manipulations, and related to measures of functional balance. During the weight-shifting task, realtime visual COP feedback was provided by a color monitor placed in front of the subject at eye level. Two stationary square goals were presented on the monitor at either side of the virtual vertical through the middle of the screen (corresponding to the sagittal midline of the body), such that approximately 15% extra weight had to be borne on each leg to reach the middle of the corresponding goal. During each registration, always one of the two goals was lit by the computer indicating the target. Subjects had to shift their weight toward this target as fast and fluently as possible and hold their COP within it for 1 sec to make a hit. As soon as a hit was made, the contralateral goal was lit and became the target. In this way, subjects made standardized frontal-plane weight shifts at a selfselected speed. Patients were allowed to practice this weight-shifting task until an optimal performance was reached. From each weight-shifting registration, the number of hits was calculated as a measure of weightshifting speed, whereas the lateral COP displacement per weight shift was determined as a measure of imprecision, after adjusting for individual differences in the intergoal distance, according to Geurts et al.17
 

To assess their clinical status, all patients were subjected to a standardized functional evaluation consisting of the Motricity Index of the affected lower limb as a measure of muscular strength, the Berg Balance Scale as a measure of functional balance, and the Functional Ambulation Categories as a measure of gait independence.
 

Procedure.
 

Each patient underwent four balance assessments at 45-min intervals at the same part of the day. The first two assessments served to establish a baseline performance. After the first assessment (A), the functional evaluation was completed in approximately 30 mins. After the second assessment (B), patients were allowed to rest for about half an hour, which was followed by the four WBV administrations. Then, the third (C) and fourth (D) balance assessments were made. Between the latter two assessments, patients were allowed to rest. Figure 2 shows a time schedule for the assessments.
 

For each of the four balance assessments, identical posturographic measures from the two test series per assessment were averaged into one result to reduce intrasubject variability. Because the balance measures yielded rather skewed distributions across patients, possible differences between balance assessments were identified by means of the (nonparametric) Wilcoxon’s matched-pairs signed-ranks test. First, assessment A (mean of test series 1 and 2) was compared with assessment B (mean of test series 3 and 4) to determine a stable baseline. Then, assessment C (mean of test series 5 and 6) and assessment D (mean of test series 7 and 8) were both compared with the average result of assessments A and B (mean of test series 1–4) to determine any immediate or prolonged effect of WBV. The healthy elderly control subjects were only tested once using the same methods (one assessment consisting of two test series).
 

RESULTS
 

The functional measures are presented in Table 1 and indicate that all patients could walk independently with or without aids or supervision but had impaired balance skills and motor functions of the affected leg. All patients were able to tolerate the selected 30-Hz frequency already during the first administration of WBV. No administration of WBV had to be aborted due to immediate adverse effects nor did patients mention any subjective complaints after the vibration.
 

 

None of the selected balance measures showed a significant change between balance assessments A and B, indicating a stable baseline performance (P > 0.22). When quiet standing at assessment C was compared with the average results of RMS assessments A and B, the RMS COP velocity in the AP direction during the eyes-closed condition showed a small but significant decrease in 22 patients (one patient was not able to perform the eyes-closed condition) (P = 0.009). This beneficial effect was still noticeable at assessment D (P = 0.01) (Fig. 3). As for the weight-shifting task, the number of hits showed a small but significant increase at assessment C (P = 0.027), but no longer at assessment D, in 21 patients (two patients were not able to perform the weight-shifting task) (Fig. 4). All other balance measures remained stable across the four balance assessments.
 

To test for possible aspecific learning effects due to repeated testing, we also analyzed the four test series 1–4 within the first two assessments (A and B) individually. No systematic improvement was found between any pair of consecutive test series nor between the first and fourth test for any selected balance measure (P > 0.09). The influence of functional level (Motricity Index, Berg Balance Scale, Functional Ambulation Categories) on the effects of WBV was not tested because of the relatively small number of severely affected patients.
 

DISCUSSION
 

This within-subject study investigated the occurrence of any shortterm effects of WBV on postural control in stroke patients as a novel method of SSS. Indeed, it has been reported that vibration is one of the strongest methods for stimulating proprioception, capable of long-lasting postural effects in healthy subjects. Only chronic patients were included who had their stroke at least 6 mos previously because they were assumed to be relatively stable in their balance performance compared with postacute stroke patients. Although all patients could walk independently to some extent, most subjects had a suboptimal, moderate, or poor score on the Berg Balance Scale, indicating substantial balance problems. This result is corroborated by the posturographic balance measures indicating substantially greater COP velocities compared with healthy elderly.
 

Preliminary evidence was found of short-term beneficial effects of WBV on postural control in stroke patients. As for quiet standing, the COP velocity in the sagittal plane decreased slightly but systematically in most of the subjects, indicating a tendency toward improved postural stability after WBV. The finding that this effect was only significant while standing with the eyes closed may be explained by a relatively great reliance on proprioceptive information during visual deprivation in stroke patients. If WBV specifically promotes proprioceptive control of standing balance in stroke patients, one would indeed expect greater functional effects while standing with the eyes closed than with the eyes opened. This reasoning, however, does not explain why no such effect was found for quiet-standing control in the frontal plane. As for the weight-shifting task, the number of hits slightly but systematically increased in most subjects, yielding a tendency toward improved weightshifting capacity after WBV. The fact that the level of weight-shifting precision did not change precludes a possible “speed–accuracy tradeoff.” This positive effect of WBV also on self-paced frontal plane weight shifts may again be related to proprioceptive stimulation because loading and unloading the legs is highly dependent on proprioceptive feedback.23 Indeed, it is assumed that WBV primarily increases proprioceptive input (mainly through Ia-afferents),24,25 thus stimulating a sensory system that is of vital importance to postural control. Based on recent insights in brain plasticity, it is possible that bilateral proprioceptive stimulation may induce spinal and cortical reorganization both through the affected and nonaffected body sides.
 
 

 

This study used a within-subject design and not a parallel group design because it was anticipated that the presumably small short-term effects of WBV would be hard to demonstrate in a group comparison due to relatively large within-group variability of balance performance. As a consequence, aspecific learning effects related to repeated testing may have influenced the balance improvements that were found. There are, however, several arguments against this possibility. Most importantly, comparing the first four balance test series did not provide any evidence of aspecific learning effects. Because such effects are usually strongest between the first two or three repetitions, it seems unlikely that they would have played a significant role in this study after the fourth test. Second, with regard to the weightshifting results, the observed data pattern, in which significance was lost at the last assessment (D), does not match with a learning process in which one would expect further improvement or at least stabilization. We therefore conclude that this study provides preliminary evidence of specific short-term beneficial effects of WBV on postural control in chronic stroke patients.
 
 

 

The finding that no adverse effects occurred and that nearly all patients reported pleasant subjective sensations both during and after the vibration therapy suggests that WBV may also be a safe application of SSS in (chronic) stroke patients. This latter conclusion is further supported by our experiences with postacute stroke patients included in an ongoing randomized, controlled trial investigating the effects of prolonged vibration therapy (daily during 6 wks) on postural control. Nevertheless, further research is needed to determine both the safety, the short-term effectiveness, and the long-term effectiveness of WBV in different groups of stroke patients.