Instability Resistance Training Improves Neuromuscular Outcome in Parkinson's Disease : Medicine & Science in Sports & Exercise

Journal Logo


Instability Resistance Training Improves Neuromuscular Outcome in Parkinson's Disease


Author Information
Medicine & Science in Sports & Exercise 49(4):p 652-660, April 2017. | DOI: 10.1249/MSS.0000000000001159
  • Free



This study compared the effects of resistance training (RT) and RT with instability (RTI) on neuromuscular and total training volume (TTV) outcomes obtained as part of the Instability Resistance Training Trial in Parkinson's disease. It also used a linear multiple regression (forward stepwise method) to identify the contribution of neuromuscular outcomes to previously published improvements in the timed-up-and-go test and the Unified Parkinson's Disease Rating Scale, motor subscale score.


Thirty-nine patients with moderate to severe Parkinson's disease were randomly assigned to three groups: control (C), RT, and RTI. RT and RTI groups performed resistance exercises twice a week for 12 wk, and only the RTI group used unstable devices to perform resistance exercises. The following neuromuscular outcomes were assessed: quadriceps muscle cross-sectional area, root mean square and mean spike frequency of electromyographic signal, peak torque, rate of torque development, and half relaxation time of the knee extensors and plantarflexors during maximum ballistic voluntary isometric contractions. TTV was calculated for lower limb exercises.


From pre- to posttraining, RTI improved all of the neuromuscular outcomes (P < 0.05) except half relaxation time of the knee extensors (P = 0.068), despite the lower TTV than RT (P < 0.05). RTI was more effective than RT in increasing the root mean square values of vastus medialis, mean spike frequency of gastrocnemius medialis, and rate of torque development of plantarflexors (P < 0.05). Stepwise regression identified the changes in mean spike frequency of gastrocnemius medialis as the best predictor of improvements in timed-up-and-go test (R2 = 0.58, P = 0.002) and on-medication Unified Parkinson's Disease Rating Scale, motor subscale scores (R2 = 0.40, P = 0.020).


RTI optimizes neuromuscular adaptations, which partially explains mobility and motor sign improvements in patients with Parkinson's disease.

Parkinson's disease (PD) is a neurodegenerative and highly debilitating disorder that causes motor signs (tremor, rigidity, bradykinesia, and postural instability) and mobility impairment, which are predictors of reduced survival in patients (28). Motor signs are associated with deficits in neuromuscular outcomes, such as voluntary muscle strength (40), motor unit activation (40), rate of torque development (RTD) (6), and relaxation time (6). Mobility impairment is associated with deficits in maximum strength (36). Thus, improving neuromuscular adaptations could contribute to mitigate the motor signs and mobility impairments of patients with PD.

Progressive resistance training (RT) is an effective nonpharmacological therapy to enhance muscle strength of patients with PD (8,38). However, to the best of our knowledge, no randomized controlled trial (RCT) has investigated the effects of RT on several neuromuscular outcomes that are associated with the severity of motor signs and mobility impairment in patients with PD. In addition, RT does not seem effective in causing significant and minimal detectable changes in mobility (timed-up-and-go test [TUG]) (32,37,38) and significant and clinically important changes in on-medication motor signs (Unified Parkinson's Disease Rating Scale, motor subscale score [UPDRS-III]) of patients with PD (7,8,11,21,38). Recently, we published the results of our Instability Resistance Training Trial in PD (IRTT-PD) (38) comparing the effects of RT and RT with instability (RTI) on the TUG values and on-medication UPDRS-III scores of patients with PD. Our findings demonstrated that only RTI produced significant and positive changes in both outcomes (38), suggesting that RTI may produce different neuromuscular adaptations than traditional RT.

The unstable devices used in RTI (e.g., BOSU® and balance disc) are likely to require not only muscle strength, but also time and amplitude varying changes in muscle activation to maintain body balance while performing the exercises (2–4) due to high motor complexity (38). As a consequence, RTI should induce greater neuromuscular adaptations (e.g., altered motor unit activation and increased RTD) than RT for patients with PD. However, the training-induced neuromuscular advantages of RTI over RT are still unknown. Total training volume (TTV = sets [number] × repetitions [number] × external load [kg]) is usually lower in RTI when compared with RT (4), which may decrease exercise load. As there is a positive relationship between TTV and increases in muscle cross-sectional area (CSA) (35,39), it is possible that the low TTV in RTI attenuates CSA increases. Given that our recently published IRTT-PD demonstrated that RTI was effective in improving TUG values and on-medication UPDRS-III scores, whereas RT did not improve these clinical outcomes (38), it is possible to suggest that neuromuscular changes observed in the current study accompanying RTI may be associated with improvements described previously in the TUG and on-medication UPDRS-III scores (38).

Therefore, the purposes of this study were to compare the effects of 12 wk of RT and RTI on neuromuscular adaptations and on TTV in patients with PD and to use a linear multiple regression (forward stepwise method) to identify neuromuscular variables that may be associated with previously published improvements in TUG and on-medication UPDRS-III scores (38). We hypothesized that RTI will induce greater neuromuscular adaptations than RT except CSA, which will be greater after RT than RTI due to the high TTV for RT.



All of the patients were recruited from an institution devoted to the care of PD (Brazil Parkinson Association). The diagnosis of idiopathic PD was confirmed by a specialist in movement disorders in accordance with UK Parkinson's Disease Society Brain Bank diagnostic criteria (24). The criteria for eligibility were as follows: (a) Hoehn and Yahr stage between 2 and 3; (b) stable use of medication; (c) 50 to 80 yr of age; (d) not participating in structured physical training in the last 3 yr; (e) not presenting neurological disorders other than PD, significant arthritis, and cardiovascular disease; and (f) not having a Mini-Mental State Examination score <23 (15). All of the patients were fully informed of the risks and benefits associated with this study and signed an informed consent form. This study was approved by the University's Ethics Committee (approval number 2011/12), and it was registered at the National Clinical Trial (; RBR-53S3RK).

Experimental Design

The IRTT-PD was a prospective, parallel-group, single-center RCT that compared the effects of 12 wk of RT and RTI on neuromuscular and TTV outcomes in patients with moderate to severe PD. Patients were assessed in the clinically “on” state (fully medicated) within 1.5 to 2 h of taking their morning dose of the dopaminergic drugs. Importantly, the neuromuscular and TTV are secondary outcomes as they represent a subset of the outcomes of this IRTT-PD study previously published (38). The clinical outcomes (TUG and on-medication UPDRS-III) were assessed on the first and second visits as previously described (38). Afterward, neuromuscular outcomes were assessed on the fourth, fifth, seventh, and ninth visits and were conducted at baseline and after 12 wk of intervention, in the same order and at the same time of day (in the morning). Quadriceps CSA (CSAQ) of the patient's most affected leg was assessed on the fourth visit. On the fifth and seventh visits, with a 48-h interval between them, patients performed two familiarization sessions with the maximum ballistic voluntary isometric contraction (MBVIC) test of the knee extensor and plantarflexor muscles of their most affected leg. On the last testing day (ninth visit), patients performed the actual MBVIC tests. Surface EMG from the vastus lateralis (VL), vastus medialis (VM), and gastrocnemius medialis (GM) muscles was synchronized with torque data and recorded for offline analysis. Root mean square (RMS) and mean spike frequency (MSF) of VL, VM, and GM muscles were obtained using customized software (Visual Basic, Microsoft). In addition, peak torque (PT), RTD, and half relaxation time (HRT) of the knee extensors and plantarflexors were obtained from torque data. The TTV for each lower limb exercise was determined over the experimental period. After baseline assessments, patients were classified into quartiles based on their TUG scores (38). Patients from each quartile were randomly assigned to the nonexercising control group (C), RT group, or RTI group. Details about power analysis have been previously published (38).

To estimate test–retest reliability (i.e., second familiarization session and the pretest with the MBVIC of the knee extensor and plantarflexor muscles), we calculated the typical error: TE = SD of the difference between the second familiarization session and the pretest/√2 and the TE expressed as the coefficient of variation (CV) of the measurement (CV% = [TE of the difference between the second familiarization session and the pretest/means of the second familiarization session and the pretest values] × 100) (22). We also calculated the minimal detectable change (MDC95 = 1.96 × √2 × standard error of measurement), which is the minimal amount of change between two points in time that indicates a true statistical change beyond random measurement error (20).

Assessments and Procedures

Timed-up-and-go test (TUG)

The patient was timed while he or she rose from an armchair (seat height, 46 cm), walked as quickly as possible at a comfortable and safe pace to a line on the floor 3 m away from the chair, turned and walked back to the chair, and sat down again. Two test trials were performed with a 1-min interval between trials (30). The shortest time was used for analysis.


The UPDRS-III includes 14 items scored from 0 to 4 (0 = no motor signs and 4 = severe motor signs). Most of these 14 items have right and left scores, resulting in a maximum possible score of 108, which indicates great motor severity (12). In cases of missing values, prorated imputation was implemented following the procedures described previously (19).


CSAQ was obtained through magnetic resonance imaging (Signa LX 9.1; GE Healthcare, Milwaukee, WI). Patients lay in the device in a supine position with straight legs. A bandage was used to restrain leg movements during image acquisition. An initial image was acquired to determine the perpendicular distance from the greater trochanter to the inferior border of the lateral epicondyle of the femur, which was defined as thigh length. CSAQ images were obtained at 50% of the segment length with 0.8-cm slices for 3 s. The pulse sequence was performed with a view field between 400 and 420 mm, a time repetition of 350 ms, an echo time of 9 to 11 ms, two signal acquisitions, and a reconstruction matrix of 256 × 256. The images were transferred to a workstation (Advantage Workstation 4.3, GE Healthcare) to determine CSAQ. The segment slice was divided into the following components: subcutaneous fat tissue, skeletal muscle, bone, and residual tissue. Then the CSAQ of the patient's most affected leg was assessed by computerized planimetry. A single observer, blinded to experimental design, conducted all of the analyses (9). Previous work from our group has observed a CV between 1% and 1.68% for CSAQ assessments (26,27).


Initially, each patient was seated in the isokinetic dynamometer chair (Biodex System 4, Biomedical Systems®, Newark, CA) in an upright position. Body stabilization was provided by two shoulder straps that crossed the participant's chest, a waist strap, and a thigh strap. To assess the knee extensor's and plantarflexor's torque, the estimated center of rotation of the knee and ankle joints was aligned with the dynamometer center of rotation. The knee and ankle joints were fixed at 60° and 90° angles, respectively, from the horizontal. The knee extensor's and plantarflexor's torque was assumed to be equal to the torque produced by the Biodex motor, corrected by gravity torque on the shank. Patients then performed a specific warm-up on the dynamometer, which consisted of three 4-s submaximal isometric contractions separated by 3-min intervals. Three minutes after the warm-up period, participants performed two 3-s MBVIC separated by a 3-min interval. Individuals were instructed to produce torque as fast and hard as possible, hold at maximal torque for 2 s, and then relax. Strong verbal encouragement and visual feedback, via the computer screen, were provided during the tests. The trials with the highest knee extensor and plantarflexor PT were used to assess the neuromuscular variables described in the next section (e.g., RMS, MSF, RTD, and HRT) (33,45). A customized script (Visual Basic, Microsoft) was used to calculate PT (45). The TE, CV, and MDC of PT assessments were 1.98 N·m, 2.64%, and 5.48 N·m for knee extensors and 1.65 N·m, 3.37%, and 4.56 N·m for plantarflexors, respectively.

Surface EMG signal

Ag/AgCl electrode bars (model TSD150TM; Biopac Systems, Goleta, CA) were placed over the belly of the VL, VM, and GM muscles with an interelectrode distance of 2 cm and aligned in parallel with the expected muscle fiber orientation of patient's most affected leg. Before electrode placement, the skin area was shaved, abraded, and cleaned with an isopropyl alcohol pad to reduce skin impedance. In addition, a ground electrode was positioned at the tibia on its medial side. Electrode placement was marked with semipermanent ink in the pretest assessment. Then electrode locations were remarked throughout the experimental protocol, assuring the same electrode placement in the pre- and posttests. EMG signal was differentially amplified with a gain of 1000 and filtered with a band-pass filter with low and high cutoff frequencies of 20–500 Hz, respectively, at −3 dB. The EMG amplifiers had an input-to-noise ratio less than 1 μV RMS and an effective common rejection mode of 85 dB. The EMG signal (Miosystem; Miosystem Co., São Paulo, Brazil) was synchronized with torque data obtained during the MBVIC through a 12-bit A/D convertor at a sampling rate of 1000 Hz and analyzed offline. The RMS of the EMG signal was calculated during a 1000-ms window around PT using a customized script (Visual Basic, Microsoft) (45). The TE, CV, and MDC of the RMS were 4.04 μV, 8.95%, and 11.19 μV for VL muscle; 1.73 μV, 3.55%, and 4.81 μV for VM muscle; and 1.70 μV, 3.63%, and 4.72 μV for GM muscle, respectively.


Spike analysis of the surface EMG signal was used to make inferences about muscle recruitment during isometric contractions (16,17,29). A spike is defined by an upward and downward deflection in the interference pattern of the surface EMG where both deflections cross zero and are at least 100 μV in amplitude. Any deflections that did not constitute discrete spikes as previously defined were assumed to be background noise and were not analyzed (16). MSF was obtained by number of spikes and divided by the total duration of the EMG sample (16). MSF of the VL, VM, and GM muscles were analyzed during a 1000-ms window around PT using a customized script (Visual Basic, Microsoft). The TE, CV, and MDC of the MSF were 8.92 Hz, 4.50%, and 24.73 Hz for VL muscle; 4.12 Hz, 2.24%, and 11.41 Hz for VM muscle; and 2.17 Hz, 3.30%, and 6.01 Hz for GM muscle, respectively.


RTD was calculated as ΔT·Δt−1 (where ΔT represents the change in torque and Δt represents the time lag, in this case 200 ms) from the torque onset (i.e., mean baseline torque plus 2 SD of the baseline torque) to 200 ms after torque initiation for the knee extensor and plantarflexors muscles. A customized script (Visual Basic, Microsoft) was used to calculate RTD (45). The TE, CV, and MDC of the RTD were 9.62 N·m·s−1, 3.36%, and 26.67 N·m·s−1 for knee extensors and 3.35 N·m·s−1, 3.33%, and 9.27 N·m·s−1 for plantarflexors, respectively.


The HRT of the knee extensor and plantarflexor muscles was calculated based on the time necessary to reach the mean torque value between the onset of torque decay and complete relaxation (i.e., zero torque) (42), calculated as follows: the same experienced researcher clicked on a point close to the beginning of torque decay. The onset of torque decay was defined recursively using a moving average algorithm (10 ms) up to the point in which the next average torque value was less than or equal to the previous one. Then the researcher clicked in a point in which torque values were zeroed. Similarly, a recursive algorithm searched for a point in which torque values were greater than baseline values plus 2 SD. The TE, CV, and MDC of the HRT were 3.63 ms, 1.74%, and 10.06 ms for knee extensors and 2.84 ms, 1.79%, and 7.86 ms for plantarflexors, respectively.


The TTV of the 12 wk of RT and RTI for each lower limb exercise (half-squat, plantarflexion, and leg press) was calculated as follows: number of sets × number of repetitions × external load (kg).

Training Programs

Details of the RT and RTI protocols used in the IRTT-PD have been previously published (38). Briefly, the C group did not perform any exercise training. They were provided with bingo games and education about PD through lectures and everyday activities once a week for 60 min by the Brazil Parkinson Association for 12 wk. Both RT and RTI groups trained twice a week (nonconsecutive days) for 12 wk (24 training sessions). A linear periodization in which training load progressed from high-volume low-intensity to low-volume high-intensity loads was implemented in an attempt to maximize training adaptations during the 12-wk training period (14). In the first month, patients performed 2–3 sets of 10–12 repetitions maximum; in the second month, they performed 3–4 sets of 8–10 repetitions maximum; and in the third month, they performed 4 sets of 6–8 repetitions maximum. A rest interval of 2 min was allowed between exercises and sets. RT and RTI consisted of conventional lower limb (i.e., half-squat, plantarflexion, and leg-press) and upper limb (latissimus dorsi pulldown and chest press) resistance exercises. Exercises were performed in the following order: half squat, latissimus dorsi pulldown, plantarflexion, chest press, and leg press. Initial training load was adjusted throughout the sets to allow patients to perform between 10–12 repetitions maximum. Afterward, training load was systematically increased (5%–10%) whenever patients were able to perform the predefined repetitions maximum for two consecutive training sessions (e.g., 10–12 repetitions maximum in the first month, 8–10 repetitions maximum in the second month, and 6–8 repetitions maximum in the third month). For the RTI group, there was a progressive and concomitant increase in load/resistance and degree of instability of the exercises during the 3-month period. Unstable devices were changed throughout the experimental period from the least to the most unstable devices. All of the patients of the RTI group progressed from one unstable device to another throughout the 3 months (38). An unstable device was changed to a more unstable device whenever patients decreased body sway considerably and force production increased abruptly when performing exercises. Because the emphasis of the RTI group was to progressively increase the degree of instability, if patients were not able to perform an exercise with a higher training load because of the greater instability of the new unstable device, patients maintained the load from the last session. Unstable devices (i.e., balance pad, dyna discs, balance discs, BOSU, and Swiss ball) were placed between the bases of support of patient (i.e., the body area responsible for sustaining most of his body weight and/or on the point of force application) and each resistance exercise (38). All of the training sessions were monitored by two investigators.

Statistical Analyses

Initially, the normality and the presence of extreme observations were assessed by Shapiro–Wilk test and box plots, respectively. If normality was not met, data were log-transformed. Then a mixed model was performed for each outcome, having group (C, RT, and RTI) and time (before and after) as fixed factors and patients as a random factor (41). Whenever a significant F value was obtained, a Tukey test adjustment was used for multicomparison purposes. Within-group (pre- to postchanges) and between-group (postchanges) effect sizes (ES) were calculated using Cohen's d (5) for each outcome. ES values were classified as small (ES ≤ 0.49), medium (ES = 0.50–0.79), and large (ES ≥ 0.80). TTV for each lower limb exercise was compared between training groups (RT vs RTI) using independent t-tests. A linear multiple regression (forward stepwise method) was used, and only the changes in independent variables (neuromuscular outcomes) that presented a significant relationship with the changes in dependent variables (TUG values and on-medication UPDRS-III scores) previously published (38) and a low collinearity with previous selected independent variables entered in the model (P < 0.05). Two variables entered the TUG regression model (i.e., MSF of GM muscle and HRT of plantarflexors) and one variable entered the UPDRS-III regression model (i.e., MSF of GM muscle). Results are presented as mean ± SD. The statistical procedures were performed using the software SAS 9.2® (Institute Inc., Cary, NC), and the level of significance was set at P < 0.05.



Ninety-one patients volunteered to participate in this study and signed the written consent. Thirty did not fulfill inclusion criteria (significant arthritis and cardiovascular disease), and 15 had family problems that prevented their participation in the study. Thus, 46 patients performed baseline testing, but 1 had back pain, 1 died, and 5 did not want to continue in the study. Thus, 39 patients, 13 in each group, comprised the final sample (Fig. 1).

The trial profile with schematic representation of participant recruitment and allocation.

The characteristics of the patients are presented in Table 1. There were no between-group differences in demographic, anthropometric, and clinical characteristics at baseline (P > 0.05). No adverse effects were reported during the trial, and adherence to the protocol was high for both training groups (23.6 ± 0.5 sessions [98%] for RT and 23.3 ± 0.7 sessions [97%] for RTI).

Characteristics of the patients with Parkinson's disease (n = 39) by group at baseline (mean ± SD).

At pretraining, there were no between-group differences in any of the neuromuscular outcomes (P > 0.05).

Neuromuscular Outcomes


There was a significant group–time interaction for CSAQ (F2,36 = 49.59, P < 0.0001). CSAQ values increased significantly in the RT (mean difference [MD] = 402.0 mm2, 95% confidence interval [CI] = 289.5–514.6, P < 0.001, ES = 0.38) and RTI groups (MD = 220.2 mm2, 95% CI = 107.6–332.8, P < 0.001, ES = 0.17) but decreased significantly in the C group (MD = −117.1 mm2, 95% CI = −229.7 to −4.5, P = 0.045, ES = −0.07) at posttraining. There were no between-group differences at posttraining (Table 2).

Neuromuscular outcomes in the pre- and posttraining assessments for each group of patients with Parkinson's disease (mean ± SD).


There was a significant group–time interaction for RMS of VL (F2,36 = 33.63, P < 0.0001), VM (F2,36 = 46.10, P < 0.0001), and GM muscles (F2,36 = 57.64, P < 0.0001). RMS values of VL, VM, and GM muscles increased significantly in the RT group (MD = 26.7 μV, 95% CI = 13.5–39.8, P < 0.001, ES = 1.21; MD = 17.2 μV, 95% CI = 7.1–27.2, P < 0.001, ES = 0.83; and MD = 34.7 μV, 95% CI = 22.9–46.4, P < 0.0001, ES = 2.67, respectively) and RTI group (MD = 42.5 μV, 95% CI = 29.3–55.3, P < 0.0001, ES = 3.23; MD = 40.6 μV, 95% CI = 30.5–50.6, P < 0.0001, ES = 2.45; and MD = 50.4 μV, 95% CI = 38.6–62.1, P < 0.0001, ES = 2.91, respectively) at posttraining. In addition, between-group differences were observed at posttraining. RMS values of VL muscle were larger in the RT (MD = 38.1 μV, 95% CI = 13.0–63.2, P < 0.0007, ES = 2.28) and RTI groups (MD = 51.1 μV, 95% CI = 26.1–76.2, P < 0.0007, ES = 3.06) than C group. RMS values of VM muscle were greater in the RTI group than C group (MD = 47.9 μV, IC = 22.9–72.9, P < 0.0001, ES = 2.20) and RT group (MD = 26.1 μV, 95% CI = 1.1–51.1, P = 0.036, ES = 0.97). RMS values of GM muscle were greater in the RT (MD = 50.0 μV, 95% CI = 27.2–72.8, P < 0.011, ES = 2.77) and RTI groups (MD = 70.7 μV, 95% CI = 47.9–46.5, P < 0.0012, ES = 3.92) than C group (Table 2).


There was a significant group–time interaction for MSF of VL (F2,36 = 23.13, P = 0.002), VM (F2,36 = 24.80, P = 0.012), and GM muscles (F2,36 = 51.54, P < 0.0001). MSF values of VL and VM muscles increased significantly only in the RTI group (MD = 102.0 Hz, 95% CI = 69.1–134.8, P < 0.001, ES = 0.78; and MD = 90.8 Hz, 95% CI = 52.3–129.3, P < 0.001, ES = 0.71, respectively). MSF values of GM muscles increased in the RT (MD = 20.7 Hz, 95% CI = 5.9–35.5, P = 0.022, ES = 0.67) and RTI groups (MD = 96.4 Hz, 95% CI = 81.6–111.2, P < 0.0001, ES = 3.22) at posttraining. In addition, between-group differences were observed at posttraining only for MSF values of the GM muscle. MSF values of GM were greater in the RTI group than C group (MD = 110.3 Hz, 95% CI = 70.3–150.4, P < 0.0001, ES = 3.93) and RT group (MD = 75.1 Hz, 95% CI = 35.5–115.1, P < 0.0001, ES = 3.03) (Table 2).


There was a significant group–time interaction for PT of knee extensors (F2,36 = 14.92, P = 0.001) and plantarflexors muscles (F2,36 = 47.91, P < 0.0001). PT values of knee extensor and plantarflexor muscles increased significantly in the RT (MD = 34.7 N·m, 95% CI = 17.8–51.7, P < 0.0001, ES = 1.39; and MD = 28.8 N·m, 95% CI = 16.5–41.1, P < 0.0001, ES = 1.16, respectively) and RTI groups (MD = 34.5 N·m, 95% CI = 17.5–51.4, P < 0.0001, ES = 1.14; and MD = 50.4 N·m, 95% CI = 32.2–78.7, P < 0.0001, ES = 2.05, respectively) at posttraining. In addition, between-group differences at posttraining were observed. PT values of knee extensors muscles were greater in the RT group (MD = 41.2 N·m, 95% CI = 3.9–78.5, P < 0.0001, ES = 1.27) and RTI groups (MD = 39.0 N·m, 95% CI = 1.7–76.3, P < 0.0001, ES = 1.21) than the C group. Similarly, PT values of plantarflexors muscles were also greater in the RT (MD = 32.9 N·m, 95% CI = 3.8–61.9, P < 0.0001, ES = 1.43) and RTI groups (MD = 25.1 N·m, 95% CI = 29.0–87.0, P < 0.0001, ES = 2.53) than the C group (Table 2).


There was a significant group–time interaction for RTD of knee extensor (F2,36 = 50.67, P < 0.0001) and plantarflexor muscles (F2,36 = 79.54, P < 0.0001). RTD values of knee extensor muscles decreased in the C group (MD = −34.4 N·m·s−1, 95% CI = −67.6 to −0.6, P < 0.001, ES = −1.23) and increased significantly only in the RTI group (MD = 123.9 N·m·s−1, 95% CI = 90.3–157.4, P < 0.0001, ES = 2.07) at posttraining. RTD values of plantarflexors muscles increased significantly in the RT (MD = 37.7 N·m·s−1, 95% CI = 14.0–61.4, P < 0.001, ES = 0.80) and RTI groups (MD = 129.4 N·m·s−1, 95% CI = 102.5–150.0, P < 0.0001, ES = 2.77) at posttraining. In addition, between-group differences at posttraining were observed. RTD values of plantarflexors muscles were greater in the RTI group than the C (MD = 138.5 N·m·s−1, 95% CI = 75.5–201.9, P < 0.0001, ES = 3.14) and RT groups (MD = 89.5 N·m·s−1, 95% CI = 26.3–152.6, P < 0.0001, ES = 1.67) (Table 2).


There were no significant changes for HRT of the knee extensors muscles (F2,36 = 1.66, P = 0.204). There was a significant group–time interaction for the HRT of plantarflexors muscles (F2,36 = 14.79, P = 0.001). The HRT of plantarflexors muscles decreased significantly only in the RTI group (MD = −36.4 ms, 95% CI = −53.4 to −19.4, P < 0.001, ES = −0.62). There were no between-group differences at posttraining (Table 2).


TTV was significantly lower in the RTI group compared with the RT group for the half-squat (MD = −879.5 kg, 95% CI = −1725.1 to −34.2, P = 0.048, ES = −0.69) and plantarflexion (MD = −4023.2 kg, 95% CI = −7425.4 to −621.9, P = 0.017, ES = −0.87) exercises. There was also a trend toward lower TTV in RTI group than RT group for the leg-press exercise (MD = −3901.3 kg, 95% CI = −8249.2 to 447.2, P = 0.059, ES = −0.67) (Fig. 2).

Mean ± SD for the TTV for the RT and RTI groups in the half-squat, plantarflexion, and leg-press exercises. *Difference between groups (P ≤ 0.05).

Linear multiple regression approach

Linear multiple regression analysis, using a forward stepwise method, was performed only for RTI group because it presented with the best results for neuromuscular outcomes, as well as for the TUG values and UPDRS-III scores previously published (38). The regression analysis identified the changes in MSF of GM muscle (R2 = 0.58, P = 0.002) and changes in the HRT of plantarflexors (R2 = 0.19, P = 0.016) as the best predictors of improvement in TUG. It also identified the changes in MSF of GM muscle (R2 = 0.40, P = 0.020) as the best predictor of improvement in on-medication UPDRS-III score.


To the best of our knowledge, this is the first RCT to test the effects of RT and RTI on several neuromuscular outcomes and TTV of patients with PD. Our hypotheses were that RTI would induce greater neuromuscular adaptations than RT except CSAQ increase, which would be greater after RT than RTI because of higher TTV in RT. We confirmed greater neuromuscular adaptations (RMS of VM, MSF of GM, and RTD of plantarflexors muscles) for RTI than RT but showed similar CSAQ increase for both training protocols (Table 2) despite the higher TTV for RT than RTI (Fig. 2). In addition, our findings demonstrated that MSF of GM explained most of the variance related to the improvements in TUG (R2 = 0.58, P = 0.002) and on-medication UPDRS-III scores (R2 = 0.40, P = 0.020).

The neural drive to the muscles is reduced in patients with PD (18,44). This manifests as slowness (6) and weakness (40) in these patients. This is due to dopaminergic deficits that result in impaired activity in basal ganglia nuclei (1). It is possible that RTI engages those impaired neural areas, possibly causing use-driven neural plasticity that improves the neural drive to the muscles. Compared with RT, RTI demands higher muscle activation (2–4) to simultaneously keep balance and to produce torque while performing the exercises (38). It is conceivable that greater demands imposed on central neural systems during RTI resulted in use-dependent neural plasticity that resulted in improving the descending neural drive, which in turn is related to the improved neuromuscular functions observed in our study. There is some evidence in our results to support the idea of improved descending drive after RTI. First, RTI resulted in greater EMG amplitude (i.e., RMS) of VL muscle and motor unit discharge frequency (i.e., MSF) of GM muscles during isometric ballistic movements. Second, RTI produced greater increments in RTD than RT at both the knee and ankle joints. These findings are interesting because areas of the basal ganglia (i.e., internal globus pallidus and subthalamic nucleus) are directly related to RTD (43). In addition, the improvements in RTD were accompanied by increases in MSF, which is critical to improve the RTD (25,33). Third, the MSF of GM muscle changes explained 40% and 58% of the improvement in TUG values and on-medication UPDRS-III scores after RTI, respectively. Fourth, previous study demonstrated a significant negative correlation between quadriceps activation and PD motor signs; the lower the quadriceps activation, the greater the motor signs (r = −0.65, P = 0.005) (40). Thus, we hypothesize that RTI may have resulted in use-dependent neural plasticity engaging those deficient neural areas in PD and, therefore, increased descending drive. However, more evidence has to be gathered to substantiate this speculation.

It is important highlight that both training groups presented with similar increases in PT at the knee and ankle joints. Our results are in accordance with previous studies demonstrating PT increases in the lower limb joints of patients with PD after RT (13,34). These results are interesting as previous studies have demonstrated that isometric force appears to be reduced when using unstable devices in training protocols, whereas similar muscle activation are observed when performing exercises on unstable surfaces (e.g., a Swiss ball) and stable surfaces (e.g., a flat bench) (2). In this sense, high muscle activation during RTI (Table 2), albeit low exercise loads (Fig. 2), may have contributed to induce similar PT increases in the RTI group compared with RT group (Table 2).

This is the first study to demonstrate decrease in HRT in patients with PD. The ability of the muscle to relax after force production is impaired in patients with PD (6) and associated with PD motor signs (6). In fact, bradykinesia seems to affect relaxation time more than force production (6). Long relaxation times might account for slowness in motor tasks in which patients are required to alternately activate and relax a muscle couple (i.e., agonist and antagonist), such as walking. During this motor task, it is necessary to contract and relax dorsiflexor and plantarflexor muscles crossing the ankle joint that are alternately activated during the propulsion at the end of the support phase of the gait cycle (31). Accordingly, the decrease in the HRT of the plantarflexors muscles after RTI contributed to improve mobility (R2 = 0.19, P = 0.016).

Regarding TTV, the RTI group trained with lower TTV for the half-squat and plantarflexion exercises compared with the RT group (Fig. 2). This occurred probably because the RTI group training protocol required increasing not only external load but also the degree of instability, whereas RT required increasing only the external load. Thus, whenever patients were not able to support a larger exercise load on a more unstable device, the load was not changed during subsequent sessions, allowing the optimization of movement control. Dibble et al. (10) reported increase in CSAQ by 6% after 12 wk of high-intensity RT in patients with PD. Surprisingly, despite low TTV for RTI group, CSAQ changes were not different between training groups (RTI, 4.0%; RT, 6.8%; Table 2). These findings challenge the suggestions that TTV is important to induce hypertrophy (35,39) and that unstable devices should not be used in training when hypertrophy is the goal (3,4). One possible explanation is that cocontraction may be high during RTI, which could reduce net joint torque (2) but not agonist muscle contraction intensity. Agonist muscle is then producing similar amount of tension, which seems to be one of the main stimuli for hypertrophy (23). Consequently, both training programs were effective in inducing hypertrophy in patients with PD (Table 2). On the other hand, it is important to highlight that the patients with PD in the present study were sedentary (i.e., no structured physical training in the last 3 yr); thus, although the RTI group trained with lower training load than RT group, this may have been sufficient to produce increasing in CSAQ after 12 wk of RTI.

Although the present study presented robust changes in neuromuscular outcomes after RTI than after RT, the sample size of the present trial was small (13 patients per group); thus, future studies should replicate the positive results of this study with a larger sample size of patients with PD to confirm the advantages of RTI over RT for PD patients.

In summary, our results demonstrated that instability added to RT (i.e., RTI) is more efficacious than RT alone on selected neuromuscular outcomes in patients with PD despite low TTV for RTI group. Some of these positive changes were associated with improvements in mobility and on-medication motor signs in patients with PD (38). Therefore, this RCT describes an innovative intervention that can improve neuromuscular adaptations associated with the improvement in motor function in patients with PD.

The authors thank Associação Brasil Parkinson (ABP), Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP), Coordenacão de Aperfeiçoamento de Pessoal de Nível Superior (CAPES), Conselho Nacional de Desenvolvimento Cientifico e Tecnológico (CNPQ), Prêmio Pemberton Coca-Cola, Diagnósticos das Américas S/A (DASA), and Center for Psychobiology and Exercise Studies. The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article. This study was supported by a grant from the FAPESP (2011/042423, 2012/03056-4, and 2013/04970-4), CAPES (3095/2015-00), and (CNPq-406609/2015-2). The authors declare that the results of this study are presented clearly, honestly, and without fabrication, falsification, or inappropriate data manipulation.

The authors declare that they have no conflict of interests, financial or otherwise. The results of the current study do not constitute endorsement by the American College of Sports Medicine.


1. Albin RL, Young AB, Penney JB. The functional anatomy of basal ganglia disorders. Trends Neurosci. 1989;12(10):366–75.
2. Anderson KG, Behm DG. Maintenance of EMG activity and loss of force output with instability. J Strength Cond Res. 2004;18(3):637–40.
3. Behm DG, Anderson KG. The role of instability with resistance training. J Strength Cond Res. 2006;20(3):716–22.
4. Behm DG, Drinkwater EJ, Willardson JM, Cowley PM. The use of instability to train the core musculature. Appl Physiol Nutr Metab. 2010;35(1):91–108.
5. Cohen J. Statistical Power Analysis for the Behavioral Sciences. Hillsdale (NJ): L. Erlbaum Associates; 1988. pp. 29–35.
6. Corcos DM, Chen CM, Quinn NP, McAuley J, Rothwell JC. Strength in Parkinson's disease: relationship to rate of force generation and clinical status. Ann Neurol. 1996;39(1):79–88.
7. Corcos DM, Comella CL, Goetz CG. Tai chi for patients with Parkinson's disease. N Engl J Med. 2012;366(18):1737–8; author reply 8.
8. Corcos DM, Robichaud JA, David FJ, et al. A two-year randomized controlled trial of progressive resistance exercise for Parkinson's disease. Mov Disord. 2013;28(9):1230–40.
9. de Souza EO, Tricoli V, Paulo AC, et al. Multivariate analysis in the maximum strength performance. Int J Sports Med. 2012;33(12):970–4.
10. Dibble LE, Hale TF, Marcus RL, Droge J, Gerber JP, LaStayo PC. High-intensity resistance training amplifies muscle hypertrophy and functional gains in persons with Parkinson's disease. Mov Disord. 2006;21(9):1444–52.
11. Dibble LE, Hale TF, Marcus RL, Gerber JP, LaStayo PC. High intensity eccentric resistance training decreases bradykinesia and improves quality of life in persons with Parkinson's disease: a preliminary study. Parkinsonism Relat Disord. 2009;15(10):752–7.
12. Fahn S, Elton RL, UPDRS Program Members. Unified Parkinson's Disease Rating Scale. In: Fahn S, Marsden CD, Goldstein DB, Calne DB, editors. Recent Developments in Parkinson's Disease, Vol. 2. Florham Park (NJ): Macmillan Healthcare Information; 1987. pp. 153–163, 293–304.
13. Falvo MJ, Schilling BK, Earhart GM. Parkinson's disease and resistive exercise: rationale, review, and recommendations. Mov Disord. 2008;23(1):1–11.
14. Fleck SJ. Periodized strength training: a critical review. J Strength Cond Res. 1999;13(1):82–9.
15. Folstein MF, Folstein SE, McHugh PR. “Mini-mental state.” A practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res. 1975;12(3):189–98.
16. Gabriel DA, Basford JR, An K. Training-related changes in the maximal rate of torque development and EMG activity. J Electromyogr Kinesiol. 2001;11(2):123–9.
17. Gabriel DA, Lester SM, Lenhardt SA, Cambridge ED. Analysis of surface EMG spike shape across different levels of isometric force. J Neurosci Methods. 2007;159(1):146–52.
18. Glendinning DS, Enoka RM. Motor unit behavior in Parkinson's disease. Phys Ther. 1994;74(1):61–70.
19. Goetz CG, Luo S, Wang L, Tilley BC, LaPelle NR, Stebbins GT. Handling missing values in the MDS-UPDRS. Mov Disord. 2015;30(12):1632–8.
20. Haley SM, Fragala-Pinkham MA. Interpreting change scores of tests and measures used in physical therapy. Phys Ther. 2006;86(5):735–43.
21. Hass CJ, Collins MA, Juncos JL. Resistance training with creatine monohydrate improves upper-body strength in patients with Parkinson disease: a randomized trial. Neurorehabil Neural Repair. 2007;21(2):107–15.
22. Hopkins WG. Measures of reliability in sports medicine and science. Sports Med. 2000;30(1):1–15.
23. Hornberger TA, Chien S. Mechanical stimuli and nutrients regulate rapamycin-sensitive signaling through distinct mechanisms in skeletal muscle. J Cell Biochem. 2006;97(6):1207–16.
24. Hughes AJ, Daniel SE, Kilford L, Lees AJ. Accuracy of clinical diagnosis of idiopathic Parkinson's disease: a clinico-pathological study of 100 cases. J Neurol Neurosurg Psychiatry. 1992;55(3):181–4.
25. Klass M, Baudry S, Duchateau J. Age-related decline in rate of torque development is accompanied by lower maximal motor unit discharge frequency during fast contractions. J Appl Physiol. 2008;104(3):739–46.
26. Laurentino GC, Ugrinowitsch C, Roschel H, et al. Strength training with blood flow restriction diminishes myostatin gene expression. Med Sci Sports Exerc. 2012;44(3):406–12.
27. Lixandrao ME, Ugrinowitsch C, Bottaro M, et al. Vastus lateralis muscle cross-sectional area ultrasonography validity for image fitting in humans. J Strength Cond Res. 2014;28(11):3293–7.
28. Lo RY, Tanner CM, Albers KB, et al. Clinical features in early Parkinson disease and survival. Arch Neurol. 2009;66(11):1353–8.
29. Magora A, Gonen B. Computer analysis of the shape of spike from electromyographic interference pattern. Electromyography. 1970;10(3):261–71.
30. Podsiadlo D, Richardson S. The timed “Up & Go”: a test of basic functional mobility for frail elderly persons. J Am Geriatr Soc. 1991;39(2):142–8.
31. Prince F, Corriveau H, Hebert R, Winter DA. Gait in the elderly. Gait Posture. 1997;5(2):128–35.
32. Prodoehl J, Rafferty MR, David FJ, et al. Two-year exercise program improves physical function in Parkinson's disease: the PRET-PD randomized clinical trial. Neurorehabil Neural Repair. 2015;29(2):112–22.
33. Ricard MD, Ugrinowitsch C, Parcell AC, et al. Effects of rate of force development on EMG amplitude and frequency. Int J Sports Med. 2005;26(1):66–70.
34. Roeder L, Costello JT, Smith SS, Stewart IB, Kerr GK. Effects of resistance training on measures of muscular strength in people with Parkinson's disease: a systematic review and meta-analysis. PLoS One. 2015;10(7): e0132135.
35. Ronnestad BR, Egeland W, Kvamme NH, Refsnes PE, Kadi F, Raastad T. Dissimilar effects of one- and three-set strength training on strength and muscle mass gains in upper and lower body in untrained subjects. J Strength Cond Res. 2007;21(1):157–63.
36. Schilling BK, Karlage RE, LeDoux MS, Pfeiffer RF, Weiss LW, Falvo MJ. Impaired leg extensor strength in individuals with Parkinson disease and relatedness to functional mobility. Parkinsonism Relat Disord. 2009;15(10):776–80.
37. Schilling BK, Pfeiffer RF, Ledoux MS, Karlage RE, Bloomer RJ, Falvo MJ. Effects of moderate-volume, high-load lower-body resistance training on strength and function in persons with Parkinson's disease: a pilot study. Parkinsons Dis. 2010;2010:824734.
38. Silva-Batista C, Corcos DM, Roschel H, et al. Resistance training with instability for patients with Parkinson's disease. Med Sci Sports Exerc. 2016;48(9):1678–87.
39. Sooneste H, Tanimoto M, Kakigi R, Saga N, Katamoto S. Effects of training volume on strength and hypertrophy in young men. J Strength Cond Res. 2013;27(1):8–13.
40. Stevens-Lapsley J, Kluger BM, Schenkman M. Quadriceps muscle weakness, activation deficits, and fatigue with Parkinson disease. Neurorehabil Neural Repair. 2012;26(5):533–41.
41. Ugrinowitsch C, Fellingham GW, Ricard MD. Limitations of ordinary least squares models in analyzing repeated measures data. Med Sci Sports Exerc. 2004;36(12):2144–8.
42. Vahtrik D, Gapeyeva H, Aibast H, et al. Quadriceps femoris muscle function prior and after total knee arthroplasty in women with knee osteoarthritis. Knee Surg Sports Traumatol Arthrosc. 2012;20(10):2017–25.
43. Vaillancourt DE, Mayka MA, Thulborn KR, Corcos DM. Subthalamic nucleus and internal globus pallidus scale with the rate of change of force production in humans. Neuroimage. 2004;23(1):175–86.
44. Valls-Sole J, Pascual-Leone A, Brasil-Neto JP, Cammarota A, McShane L, Hallett M. Abnormal facilitation of the response to transcranial magnetic stimulation in patients with Parkinson's disease. Neurology. 1994;44(4):735–41.
45. Wallerstein LF, Tricoli V, Barroso R, et al. Effects of strength and power training on neuromuscular variables in older adults. J Aging Phys Act. 2012;20(2):171–85.


© 2017 American College of Sports Medicine