Effects of high-intensity resistance training in patients with rheumatoid arthritis: A randomized controlled trial




To confirm, in a randomized controlled trial (RCT), the efficacy of high-intensity progressive resistance training (PRT) in restoring muscle mass and function in patients with rheumatoid arthritis (RA). Additionally, to investigate the role of the insulin-like growth factor (IGF) system in exercise-induced muscle hypertrophy in the context of RA.


Twenty-eight patients with established, controlled RA were randomized to either 24 weeks of twice-weekly PRT (n = 13) or a range of movement home exercise control group (n = 15). Dual x-ray absorptiometry–assessed body composition (including lean body mass [LBM], appendicular lean mass [ALM], and fat mass); objective physical function; disease activity; and muscle IGFs were assessed at weeks 0 and 24.


Analyses of variance revealed that PRT increased LBM and ALM (P < 0.01); reduced trunk fat mass by 2.5 kg (not significant); and improved training-specific strength by 119%, chair stands by 30%, knee extensor strength by 25%, arm curls by 23%, and walk time by 17% (for objective function tests, P values ranged from 0.027 to 0.001 versus controls). In contrast, body composition and physical function remained unchanged in control patients. Changes in LBM and regional lean mass were associated with changes in objective function (P values ranged from 0.126 to <0.0001). Coinciding with muscle hypertrophy, previously diminished muscle levels of IGF-1 and IGF binding protein 3 both increased following PRT (P < 0.05).


In an RCT, 24 weeks of PRT proved safe and effective in restoring lean mass and function in patients with RA. Muscle hypertrophy coincided with significant elevations of attenuated muscle IGF levels, revealing a possible contributory mechanism for rheumatoid cachexia. PRT should feature in disease management.


Rheumatoid arthritis (RA) is characterized by, and a major cause of disability. Recently, Giles et al (1) showed that disability in patients with RA is linked to adverse changes in body composition, with mean Health Assessment Questionnaire (HAQ) scores being inversely related to appendicular lean mass (ALM; a surrogate measure of skeletal muscle mass), and directly related to appendicular fat mass. Unfortunately, both reduced muscle mass, termed rheumatoid cachexia (2), and increased adiposity are associated with RA (3). In fact, according to the definitions proposed by Baumgartner et al (4), 67% of our whole-body dual x-ray absorptiometry (DXA)–assessed patients with RA are muscle wasted, and 80% are obese (5–7).

Relative to age- and sex-matched healthy sedentary controls, the loss in lean body mass (LBM) that we observe averages ∼15% (Lemmey et al: unpublished observations); a magnitude in accordance with the 14–16% reductions in body cell mass identified by Roubenoff et al for patients with RA (8, 9). This degree and prevalence of rheumatoid cachexia is alarming because, in addition to diminished function and increased disability, muscle loss is associated with impaired immune and pulmonary function, osteoporosis, glucose intolerance, and increased mortality (10). Consequently, interventions that can increase muscle mass in cachectic individuals have the potential to improve physical performance and decrease morbidity and mortality (11).

In a nonrandomized pilot study (5), we showed that 12 weeks of high-intensity progressive resistance training (PRT; 3 days/week) significantly increased LBM and ALM, decreased percent body fat, and substantially improved objective functional capacity in cachectic patients with RA. Similarly, Hakkinen et al (12) subsequently found that combined strength and aerobic training increased thigh muscle mass and decreased thigh fat mass in female patients with RA. However, these body composition results are at odds with the earlier investigation of Rall et al (13), in which PRT failed to augment LBM in patients with RA. Therefore, the efficacy of PRT in restoring muscle mass in patients with RA requires confirmation.

Also awaiting clarification are the mechanisms by which exercise-induced improvements are made, and further insights into the pathology of rheumatoid cachexia are also needed. It is likely that the insulin-like growth factor (IGF) system plays a key role in these processes, because IGF-1 produced locally in the muscle (mIGF-1) is thought to regulate adult skeletal muscle maintenance and its hypertrophic adaptation to increased loading (14). In support of this, significantly reduced mIGF-1 levels have been identified in conditions characterized by muscle wasting (15–18). Furthermore, the success of exercise training in restoring muscle mass in these conditions appears to be dependent on whether mIGF-1 levels respond to the exercise stimulus (15, 18, 19).

Consequently, in the current randomized controlled study we aimed to confirm our preliminary observations (i.e., that PRT reverses debilitating cachexia and improves function in patients with RA), and to investigate the role of the local IGF system in exercise-induced hypertrophy of skeletal muscle in patients with RA.


Study design and ethics.

This was a randomized, controlled intervention trial conducted July 2004–January 2007. Approval was received from the North West Wales National Health Service Trust Research Ethics Committee.

Subject recruitment and eligibility.

Sample size was determined by power calculations for the primary outcome variable: ALM. This calculation, based on the mean ALM change (0.93 kg) for PRT subjects in our pilot study (4), assumed normal distribution, 2 groups, equal variance, no change in the control group, a common SD for each group (i.e., the SD for our pilot study PRT group, 0.46 kg) (4), α = 0.05, and power = 0.80, resulted in a requirement of 5 subjects per group. To account for potential dropouts, we aimed to recruit 18 subjects per group (i.e., a total of 36 patients).

The progress of patients through the study is depicted in Figure 1. Thirty-six eligible volunteer patients (Table 1) from the Gwynedd Rheumatology Department were randomized to either a high-intensity PRT group or to a home-based, low-intensity range of movement (ROM) exercise (control) group by stratified random allocation, with age, sex, and estrogen status serving as the stratified variables. Of these, 28 attended baseline assessment and commenced training.

Figure 1.

Flow of participants through each stage of the trial. GP = general practitioner.

Table 1. Eligibility criteria for inclusion in the study*
  • *

    ACR = American College of Rheumatology (formerly the American Rheumatism Association); HIV = human immunodeficiency virus.

Fulfills the ACR 1987 revised criteria for the diagnosis of rheumatoid arthritis (44)
Age ≥18 years
Functional class I or II
Not cognitively impaired
Antiinflammatory and/or antirheumatic drug therapy has been stable for the previous 3 months
If on corticosteroids, maintained on a dosage <10 mg/day
Free of other cachectic diseases (e.g., cancer, HIV infection)
Free of medical conditions contraindicating regular high-intensity exercise
Not taking drugs or nutritional supplements known to be anabolic
Not currently undertaking regular, intense physical training
Not pregnant

Outcome measures.

Assessments were conducted at baseline (pretest) and immediately following (posttest) the 24-week training period. For each assessment, subjects presented at approximately the same time of day, having fasted and refrained from strenuous exercise for 24 hours.

Body composition.

Total and regional lean and fat masses were measured with a whole-body pencil-beam DXA scanner (QDR1500, software version V5.72; Hologic, Waltham, MA). Subsequently, ALM (i.e., total arms + legs soft–lean mass), a proxy measure of total body skeletal muscle mass (20), was determined (21); relative skeletal muscle index was calculated (ALM [kg]/height [m2]) (7); and percent body fat was estimated. Relative skeletal muscle index and percent body fat were then used to determine whether patients were cachectic, obese, or, if both, cachectic-obese, according to the definitions of Baumgartner et al (4).

Immediately following DXA scanning, bioelectrical impedance spectroscopy (Hydra 4200; Xitron Technologies, San Diego, CA) was performed to estimate extracellular water and total body water (TBW). Checking these is necessary because DXA measures of body composition are affected by edema. By combining DXA and bioelectrical impedance spectroscopy data, we estimated total body protein using the model of Fuller et al (22), i.e., LBM in grams minus (0.2302 × total bone mineral content in grams) minus TBW in grams.

Muscle strength, physical function, and habitual physical activity.

PRT-specific strength was determined by the total weight lifted per session (i.e., Σ [resistance in kg × number of sets × repetitions per set] for each exercise). To account for the initial neural adaptations (i.e., increased strength due to enhanced muscle fiber recruitment), baseline measures for training-specific strength were taken after 2 weeks of PRT. Maximal voluntary isometric knee extensor strength (KES; at a fixed joint angle of 90°) was measured by an isokinetic dynamometer (Kin-com, Chattanooga, TN). Objective physical function was additionally measured by tests from the Senior Fitness Test (23): a 30-second arm curl, a 30-second chair stand, and a 50-foot walk.

Subjective, patient-reported physical disability was assessed with the Multidimensional HAQ (MDHAQ) (24). Habitual physical activity was estimated by electronic pedometers (Digiwalker DW 200; Yamax, Tokyo, Japan) worn during all waking hours of the first (pretest) and last (posttest) weeks of training. For statistical analyses, the average number of daily steps for each of the 2 assessment weeks was used.

Muscle biopsies and blood sampling.

From patients who consented (PRT group n = 9, control group n = 5), muscle biopsy specimens from vastus lateralis were obtained, as previously described (17), prior to and 3–7 days after the intervention period. Clotted venous blood was collected from all subjects at 0 and 24 weeks, following an overnight fast. Muscle and serum samples were stored at −70°C.

IGF system assays.

Muscle IGF-1 and IGF binding protein 3(IGFBP-3), and serum IGF-1, IGF-2, IGFBP-1, and IGFBP-3 levels were measured in triplicate by specific, in-house radioimmunoassays using established techniques (25–27). The intraassay and interassay coefficient of variations were 3% and 15%, respectively, for IGF-1, 5% and 14%, respectively, for IGF-2, 4% and 13%, respectively, for IGFBP-1, and were 4% and 14%, respectively, for IGFBP-3. To assess IGFBP-3 protease activity, serum was analyzed by Western immunoblotting, and the percentage of fragmented IGFBP-3 was estimated from the pixel densities of the 3 bands (43–46-kd doublet and 30-kd fragment) (28). Muscle IGF-1 and IGFBP-3 values were expressed per total muscle protein.

Disease activity and inflammation.

Disease activity was assessed by the Disease Activity Score in 28 joints (DAS28) (29), and systemic inflammation by the erythrocyte sedimentation rate (ESR).

Exercise intervention.

After baseline assessment, subjects in the PRT group trained twice a week for 24 weeks at a community gym under the supervision of exercise physiologists (KC, SW, and FC). The PRT program consisted of 3 sets of 8 repetitions with a load corresponding to 80% of the 1-repetition maximum (1-RM; i.e., the maximum load lifted for each of the prescribed exercises), with 1–2 minutes of rest between sets, for each of the following multi-stack machine exercises: leg press, chest press, leg extension, seated rowing, leg curl, triceps extension, standing calf raises, and bicep curl.

Training volumes and intensities such as this are considered optimal for inducing muscle hypertrophy (30). To reduce muscle soreness, only 1 set was performed for each exercise in the first week and only 2 sets were performed in the second week. To further facilitate adaptation, 15 repetitions/set at 60% of 1-RM were performed in weeks 1–4 and 12 repetitions/set at 70% of 1-RM were performed in weeks 5 and 6, before progressing to 8 repetitions/set at 80% of 1-RM in weeks 7–24. To ensure maintenance of relative intensities, the 1-RMs were reassessed every 4 weeks. PRT sessions were preceded by a warmup and ended by a cool down, each comprising 10 minutes of low-intensity ROM exercises. Following instruction, control subjects were asked to perform these ROM exercises twice weekly at home. ROM exercise was chosen as a control condition because this form of exercise, while being the type most commonly prescribed for patients with RA, features insufficient intensity to elicit muscle hypertrophy (30). A training diary was maintained by all patients so that compliance and any adverse effects could be evaluated; additionally, control patients were phoned every two weeks. Other than their directed training, subjects were instructed to maintain their habitual physical activity levels and diet during the experimental period.

Statistical analysis.

Differences between groups at baseline were examined using multiple paired t-tests for continuous variables and Fisher's exact probability test for categorical variables. Where no difference was confirmed, treatment effects were assessed by multiple 2 × 2 (time × group) factor analyses of variance with repeated measures for time. The assumptions of sphericity and normality of distribution were verified by Mauchly's test and the Kolmogorov-Smirnov test, respectively. The effect size for the time × group interaction was calculated as eta squared (η2), with thresholds for small, moderate, large, and very large effects set at 0.01, 0.08, 0.26, and 0.50, respectively. After pooling both groups' data, Pearson's correlation coefficient (r) was employed to assess the significance of hypothesized relationships (e.g., changes in arm and leg lean mass and changes in objective measures of upper- and lower-body function, respectively). P values less than 0.05 were considered statistically significant, and P values ranging from 0.05 to 0.10 were considered a trend. Data were analyzed using the Statistical Package for the Social Sciences, version 14 (SPSS, Chicago, IL), and are presented as the mean ± SD.


Twenty-eight patients with established, stable RA completed baseline assessment and commenced either high-intensity PRT (n = 13) or home-based ROM exercise (control; n = 15) (Figure 1). At baseline, the groups were not significantly different for age, sex, disease duration, current disease activity and medication, or estrogen status (Table 2). The subjects who provided muscle biopsy specimens for investigation of local IGF responses to training were not different from their respective groups (either demographically or in training response; data not shown).

Table 2. Baseline characteristics of patients with RA in the PRT and control groups*
VariablePRT (n = 13)Control (n = 15)P
  • *

    Values are the mean ± SD unless otherwise stated. RA = rheumatoid arthritis; PRT = progressive resistance training; DAS28 = Disease Activity Score in 28 joints; ERT = estrogen replacement therapy.

  • For continuous variables, differences between groups were examined using multiple paired t-tests.

Age, years55.6 ± 8.360.6 ± 11.20.201
No. women/men11/212/30.686
Disease duration, months74 ± 76125 ± 1010.146
DAS28 score3.29 ± 1.273.28 ± 1.070.989
Postmenopausal, no. patients99 
ERT, no. patients10 

Compliance, training effect, and safety.

Compliance to training was good for the PRT group. Of the 48 scheduled sessions, subjects completed on average 34.6 sessions (range 11–45 sessions), i.e., 73% of the sessions. The effectiveness of the PRT program was apparent from the mean total session load lifted by each patient, which increased from 4,931 kg at week 2 (training baseline) to 10,803 kg by week 24 (P < 0.0001); an increase in training-specific strength of 119%.

Subjects in the ROM exercise control group reported relatively good compliance for a home-based program, with an average of 25.9 sessions (range 11–48 sessions) completed, i.e., 54%. No training-related injuries or other adverse events were reported by subjects in either group.

Despite its high intensity and volume, the PRT program did not exacerbate disease activity (mean ± SD DAS28 score pretest versus posttest: 3.29 ± 1.27 versus 3.12 ± 1.34 for the PRT group and 3.28 ± 1.27 versus 3.56 ± 0.71 for the control group; P = 0.471 for the interaction), or systemic inflammation (mean ± SD ESR pretest versus posttest: 18.0 ± 12.5 mm/hour versus 17.5 ± 10.5 mm/hour for the PRT group and 26.2 ± 15.1 mm/hour versus 31.1 ± 16.4 mm/hour for the control group; P = 0.285 for the interaction).

Body composition.

The effects of 24 weeks of exercise training on DXA-assessed body composition are presented in Table 3. PRT increased both total LBM (mean ± SD 1,536 ± 1,471 gm) and ALM (1,211 ± 1,235 gm), which is a more precise measure of muscle mass. Conversely, over the same period the control subjects lost an average of ∼400 gm and ∼160 gm in LBM and ALM, respectively. Analyses of the bioelectrical impedance spectroscopy data confirmed that the PRT-induced increases in lean mass were not attributable to fluid retention, because no changes in extracellular water were observed (mean ± SD pretest versus posttest: 13.94 ± 1.75 liters versus 13.97 ± 1.51 liters for the PRT group and 14.61 ± 1.98 liters versus 14.77 ± 2.06 liters for the control group; P = 0.934 for the interaction). Consistent with this data is an increase in mean estimated total body protein (+1,796 gm) following PRT, contrasting with a numerical loss for the controls (−413 gm). Seemingly large losses of total fat mass (−2,298 gm and −1,326 gm for the PRT and control groups, respectively) and trunk fat mass (−2,500 gm and −1,318 gm, respectively) were observed over the intervention period, although none of these changes approached significance. Additionally, PRT reduced the number of patients classified as cachectic, obese, and cachectic-obese, whereas low-intensity ROM training failed to modify this classification in control patients (Table 3).

Table 3. Effects of 24 weeks of high-intensity PRT on body composition in patients with RA*
VariablePRT group (n = 13)Control group (n = 15)Pη2
  • *

    Values are the mean ± SD in kilograms unless otherwise stated. PRT = progressive resistance training; RA = rheumatoid arthritis.

  • For group × time interaction.

  • Effect size, with thresholds for small, moderate, large, and very large effects set at 0.01, 0.08, 0.26, and 0.50, respectively.

  • §


  • Relative skeletal muscle index ≤5.45 kg/m2 and ≤7.26 kg/m2 for women and men, respectively (4).

  • #

    Percent body fat ≥38% and ≥27% for women and men, respectively (4).

  • **

    Coincidence of both cachexia and obesity.

Lean body mass    
 Pretest37.25 ± 3.9540.42 ± 8.92  
 Posttest38.79 ± 4.1740.02 ± 8.680.006§0.26
Appendicular lean mass    
 Pretest14.29 ± 1.7815.65 ± 4.07  
 Posttest15.48 ± 2.2415.49 ± 3.980.002§0.33
Total body protein    
 Pretest6.40 ± 2.027.66 ± 3.56  
 Posttest8.20 ± 1.847.25 ± 3.930.004§0.28
Total fat mass    
 Pretest27.83 ± 11.9531.26 ± 8.74  
 Posttest25.53 ± 10.7529.93 ± 10.430.657 
Trunk fat mass    
 Pretest14.00 ± 6.4616.11 ± 5.70  
 Posttest11.50 ± 5.2314.79 ± 6.090.489 
Cachectic, no.    
Obese, no.#    
Cachectic-obese, no.**    

Physical function.

As anticipated, the anabolic effects of our PRT program concurred with improvements in objectively assessed physical function (Table 4). Relative to baseline, mean performance improved by 30% for the 30-second chair stand test, 25% for KES, 23% for the 30-second arm curl test, and 17% for the 50-foot walk following PRT. The η2 for these improvements indicate large treatment effects. These PRT-induced improvements saw our patients with established RA attain or surpass the respective 50th percentile performance score (sex-weighted) for healthy, age-matched individuals for the chair stand, arm curl, and walk tests (23) (Table 4). In contrast, no changes in performance were observed in control patients.

Table 4. Effects of 24 weeks of high-intensity PRT on objectively assessed physical function in patients with RA*
VariablePRT group (n = 13)Control group (n = 15)Pη2Healthy controls§
  • *

    Values are the mean ± SD unless otherwise stated. PRT = progressive resistance training; RA = rheumatoid arthritis; KES = knee extensor strength.

  • For group × time interaction (2 × 2).

  • Effect size, with thresholds for small, moderate, large, and very large effects set at 0.01, 0.08, 0.26, and 0.50, respectively.

  • §

    50th percentile performance level (weighted for sex, i.e., 23 women, 5 men) for healthy 60–64-year-olds for tests selected from the Senior Fitness Test (26).

  • Significant.

30-second arm curl test, reps     
 Pretest16.1 ± 4.914.5 ± 4.1   
 Posttest19.8 ± 4.615.5 ± 4.20.0270.2716.5
30-second chair test, reps     
 Pretest12.5 ± 4.011.8 ± 3.1   
 Posttest16.3 ± 4.011.9 ± 4.20.0010.4515.2
50-foot walk test, seconds     
 Pretest9.33 ± 2.4010.03 ± 3.78   
 Posttest7.77 ± 1.409.89 ± 4.280.0130.288.0
KES, newtons     
 Pretest323 ± 79308 ± 131   
 Posttest404 ± 122329 ± 1230.0060.34 

The link between muscle mass and physical function is underlined by the correlations between increases in lean mass and improvements in performance. Therefore, change in LBM was consistent with changes in the scores of the 50-foot walk test (r = −0.613, P < 0.0001), the chair rise test (r = 0.383, P < 0.05), the arm curl test (r = 0.292, P = 0.064), and KES (r = 0.251, P = 0.09). More specifically, change in arm lean mass correlated with change in the scores of the arm curl test (r = 0.321, P < 0.05), whereas change in leg lean mass correlated with change in the scores of the 50-foot walk (r = −0.634, P < 0.0001) and chair rise (r = 0.364, P < 0.05) tests, and was weakly associated with KES change (r = 0.224, P = 0.126).

The marked improvements in objectively measured physical function following PRT are not reflected in patients' perceived function, as MDHAQ scores were unchanged (mean ± SD pretest versus posttest: 0.914 ± 0.680 versus 0.817 ± 0.691 for the PRT group and 0.575 ± 0.619 versus 0.575 ± 0.590 for the control group; P = 0.513 for the interaction). Habitual physical activity was unchanged in both groups (mean ± SD pretest versus posttest: 7,975 ± 2,827 steps/day versus 8,207 ± 2,762 steps/day for the PRT group and 6,057 ± 2,605 steps/day versus 5,601 ± 2,320 steps/day for the control group; P = 0.537 for the interaction). Similarly, no changes in diet were reported by subjects in either group.

IGF system.

The IGF data are presented in Table 5. Increases in muscle IGF-1 and IGFBP-3 levels (41% and 73%, respectively) were observed following 24 weeks of PRT. In contrast, mIGF-1 levels decreased (by 28%) and mIGFBP-3 levels remained stable for the control group. For both muscle IGF proteins, the interactions were significant and the η2 indicated large treatment effects. Overall (n = 14), changes in mIGF-1 were significantly correlated to changes in mIGFBP-3 (r = 0.584, P < 0.05). No effects on systemic IGF-1, IGF-2, IGFBP-1, IGFBP-3, or IGFBP-3 fragmentation levels were detected.

Table 5. Effects of 24 weeks of high-intensity PRT on mIGF-1, mIGFBP-3, sIGF-1, sIGF-2, sIGFBP-1, sIGFBP-3, and sIGFBP-3 fragmentation levels in patients with RA*
VariablePRT groupControl groupPη2
  • *

    Values are the mean ± SD unless otherwise stated. PRT = progressive resistance training; mIGF-1 = muscle insulin-like growth factor 1; mIGFBP-3 = mIGF binding protein 3; sIGF-1 = serum IGF 1; sIGFBP-1 = sIGF binding protein 1; RA = rheumatoid arthritis.

  • For group × time interaction.

  • Effect size, with thresholds for small, moderate, large, and very large effects set at 0.01, 0.08, 0.26, and 0.50, respectively.

  • §

    For the PRT group n = 9; for the control group n = 5.

  • Significant.

  • #

    For the PRT group n = 13; for the control group n = 15.

mIGF-1, ng/mg total protein§    
 Pretest1.048 ± 0.2741.306 ± 0.524  
 Posttest1.474 ± 0.3980.938 ± 0.3790.0160.39
mIGFBP-3, ng/mg total protein§    
 Pretest22.74 ± 7.5034.18 ± 13.49  
 Posttest39.32 ± 18.2730.88 ± 12.120.0390.31
sIGF-1, ng/ml#    
 Pretest127 ± 50130 ± 43  
 Posttest130 ± 30126 ± 390.819 
sIGF-2, ng/ml#    
 Pretest1,263 ± 4301,167 ± 332  
 Posttest1,097 ± 2051,097 ± 2680.708 
sIGFBP-1, ng/ml#    
 Pretest22.57 ± 12.3136.98 ± 21.62  
 Posttest21.48 ± 10.4931.64 ± 24.620.498 
sIGFBP-3, ng/ml#    
 Pretest4,275 ± 8643,992 ± 874  
 Posttest4,366 ± 5324,055 ± 8630.321 
sIGFBP-3 fragmentation, %#    
 Pretest36.42 ± 9.2340.15 ± 18.03  
 Posttest38.23 ± 12.6636.96 ± 13.560.411 


In this randomized controlled trial, we have confirmed our pilot study findings (5) that PRT significantly increases muscle mass and restores physical function in patients with RA. In addition, we have identified a probable mechanism for muscle anabolism in RA.

The results of Rall et al (13) suggested that patients with RA, perhaps due to their hypermetabolic state, were resistant to the anabolic effects of exercise. However, subsequent studies (5, 12), and especially the current investigation, which features a more robust methodologic design, refute this. Our pilot study (5) and the current trial have both observed marked increases in LBM, ALM, and total body protein following high-intensity PRT, and Hakkinen et al (12) demonstrated increases in quadriceps femoris thickness (P < 0.001) in female patients with RA following 21 weeks of PRT combined with aerobic training. Similarly, these studies consistently show reductions in fat following training. In the study by Hakkinen et al (12), quadriceps femoris subcutaneous fat thickness decreased (P < 0.001); in our pilot study (4), percent body fat was reduced (P < 0.05) and there was a trend for trunk fat mass to decrease (−752 gm; P = 0.084); and in the current study, physiologically significant reductions in fat mass (2.3 kg) and trunk fat mass (2.5 kg, i.e., 18%) were observed. This repeated suggestion that PRT reduces trunk fat mass is of clinical interest because RA predisposes to central obesity (31) and, consequently, insulin resistance, dyslipidemia, hypertension, and the metabolic syndrome (32).

One difference between the studies that achieved exercise training–induced improvements in body composition in patients with RA (refs.5, 12, and the current study) and that of Rall et al (13), which did not, is volume of training. Whereas in the earlier investigation (13) subjects on average completed only 21 training sessions (2 sessions/week for 12 weeks with 87% compliance), at least 30 sessions were completed, on average, in the other trials. Additionally, the training implemented by Rall and colleagues involved performing only 5 exercises per session (13), whereas our program featured 8 different exercises (ref.5, and the current study).

In terms of the magnitude of training effect, the body composition changes we observed correspond to those typically seen for healthy middle-aged or older subjects following 3–12 months of high-intensity PRT (33, 34). A more direct comparison is provided by Hakkinen et al (12), who found almost identical increases in thigh muscle and comparable reductions in thigh fat in female patients with RA and age-matched, healthy women following completion of the same training program. Therefore, it is now clear that patients with RA are not resistant to the anabolic effects of exercise, and that for patients with RA, as for healthy individuals, an appropriate combination of PRT intensity and volume is required to produce beneficial changes in body composition (35).

Predictably, the increases in muscle mass elicited by our PRT intervention were associated with improvements in strength and objectively assessed function. Although these correlations are mostly moderate, indicating that other factors also contribute to the improvements in physical function induced by PRT in patients with RA (36, 37), this association between muscle hypertrophy and enhanced physical performance replicates the findings from our pilot study (5). Interestingly, given that muscle mass is thought to decline in the general population at approximately 6% per decade after the age of 50 years and strength by approximately 12–14% per decade over the same period (38), there is reasonable agreement from the mean gains we observed in ALM (8.4%) and KES (25%) following 24 weeks of PRT that these training effects are equivalent to the reversal of 14–20 years of sarcopenia. That PRT should have such a positive effect on function and disability in patients with RA is to be expected given the recent finding that both ALM and appendicular fat mass exert significant effects on HAQ score (1).

In common with most exercise intervention studies requiring volunteer subjects, the current study suffers from having a relatively biased sample, i.e., our patient cohort is less disabled than would be anticipated if outpatients were randomly selected. Despite this restriction, the incidence of significant muscle loss (cachexia) among subjects at baseline approached 60%, the incidence of obesity was 79%, and the incidence of cachectic-obesity (i.e., the coincidence of both conditions) was ∼36%. In the general elderly population, classification as either muscle wasted (sarcopenic) or obese significantly increases the likelihood of disability (39), and the coincidence of both (sarcopenic-obesity) independently increases the risk of disability in women 12 fold. It is notable that in the current study, PRT removed most of the subjects previously classified as cachectic and cachectic-obese from these high-risk categories.

The inability of PRT to change MDHAQ scores in the current study is consistent with results from other exercise intervention studies (12, 36, 40), and is likely to be due to the low disability of our patients and the relative insensitivity of this instrument to detect improvements in mildly disabled patients. Although the level of disability for our patients with RA was comparatively mild, it should be noted that, consistent with their reduced muscle mass and increased adiposity, performance of objective function tests by the PRT patients at baseline and the controls throughout was inferior to that of age- and sex-matched, sedentary, healthy individuals. Notably, performance of these tests, which were developed to assess the physical capacity of elderly people to perform activities of daily living (23), was normalized in patients with established RA by 24 weeks of PRT. Because in the current study habitual physical activity, diet, and disease activity remained unchanged, none of these factors can account for the improvements in body composition or function following PRT.

Despite the devastating consequences that impaired physical function has on patients with RA, society, and the economy, little research has been devoted to the metabolic changes that cause the muscle loss underlying much of this decrement (2, 8). This is regrettable because insights into the mechanisms of rheumatoid cachexia and its reversal should lead to improvements in patient treatment and outcome. To further this understanding, in the current study we tested our hypothesis that PRT exerts its anabolic effects in patients with RA, as it does in healthy individuals, primarily through an increase in mIGF-1 activity (14). Our data appear to support this proposed mechanism. The mIGF-1 levels in the 14 patients with RA from whom biopsy samples were taken were reduced relative to levels we observed previously in healthy, sedentary controls of a similar age (mean ± SD 2.78 ± 1.76 ng/mg total protein) (17). This is consistent with the diminished mIGF-1 levels identified in subjects with chronic renal failure (17, 19), chronic heart failure (16), chronic obstructive pulmonary disease (COPD) (18), and advanced aging (15); all of which are conditions characterized by muscle wasting. In turn, mIGF-1 and mIGFBP-3 levels in our patients with RA increased significantly in response to PRT. Although these results should be interpreted conservatively due to the low and uneven biopsy sample sizes, the effect size is large and this finding of augmented mIGF-1 levels with accompanying muscle hypertrophy has also been seen in frail, elderly subjects following PRT (15) and in patients with COPD after high-intensity training (18). Similarly, an increase in mIGFBP-3 accompanying training-induced hypertrophy has been described in patients with hemodialysis (41). In contrast, 12 weeks of high-intensity interval cycling that failed to increase muscle mass in patients with hemodialysis also failed to raise mIGF-1 or IGFBP-3 levels (19). Unlike muscle IGFs, and consistent with previous results in older adults (42, 43), including patients with RA (12), there was no effect of PRT on any of the measured components of the systemic IGF system in our patients.

To summarize, our results confirm that PRT is a safe and effective means of restoring muscle mass and functional capacity in patients with established, stable RA. Consequently, we advocate that, for appropriate patients, PRT programs similar to ours be included in disease management. We also provided a mechanistic insight into muscle anabolism in RA that could affect clinical and pharmaceutical approaches to treatment.


All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be submitted for publication. Dr. Lemmey had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. Lemmey, Marcora, Maddison.

Acquisition of data. Chester, Wilson, Casanova.

Analysis and interpretation of data. Lemmey, Marcora, Maddison, Chester, Wilson, Casanova.