Effect of Remotely Supervised Weight Loss and Exercise Training Versus Lifestyle Counseling on Cardiovascular Risk and Clinical Outcomes in Older Adults With Rheumatoid Arthritis: A Randomized Controlled Trial

Abstract

Objective

To compare a remotely supervised weight loss and exercise intervention to lifestyle counseling for effects on cardiovascular disease risk, disease activity, and patient‐reported outcomes in older patients with rheumatoid arthritis (RA) and overweight/obesity.

Methods

Twenty older (60–80 years), previously sedentary participants with seropositive RA and overweight/obesity were randomized to 16 weeks of either Supervised Weight loss and Exercise Training (SWET) or Counseling Health As Treatment (CHAT). The SWET group completed aerobic training (150 minutes/week moderate‐to‐vigorous intensity), resistance training (two days/week), and a hypocaloric diet (7% weight loss goal). The CHAT control group completed two lifestyle counseling sessions followed by monthly check‐ins. The primary outcome was a composite metabolic syndrome z‐score (MSSc) derived from fasting glucose, triglycerides, high density lipoprotein–cholesterol, minimal waist circumference, and mean arterial pressure. Secondary outcomes included RA disease activity and patient‐reported outcomes.

Results

Both groups improved MSSc (absolute change −1.67 ± 0.64 in SWET; −1.34 ± 1.30 in CHAT; P < 0.01 for both groups) with no between‐group difference. Compared with CHAT, SWET significantly improved body weight, fat mass, Disease Activity Score‐28 C‐reactive protein, and patient‐reported physical health, physical function, mental health, and fatigue (P < 0.04 for all between‐group comparisons). Based on canonical correlations for fat mass, cardiorespiratory fitness, and leg strength, component‐specific effects were strongest for (1) weight loss improving MSSc, physical health, and mental health; (2) aerobic training improving physical function and fatigue; and (3) resistance training improving Disease Activity Score‐28 C‐reactive protein.

Conclusion

In older patients with RA and overweight/obesity, 16 weeks of remotely supervised weight loss, aerobic training, and resistance training improve cardiometabolic health, patient‐reported outcomes, and disease activity. Less intensive lifestyle counseling similarly improves cardiovascular disease risk profiles, suggesting an important role for integrative interventions in the routine clinical care of this at‐risk RA population.

INTRODUCTION

Persons with rheumatoid arthritis (RA) have an increased risk for cardiovascular disease (CVD) with an associated reduction in life expectancy. ¹ , ² , ³ , ⁴ , ⁵ , ⁶ , ⁷ , ⁸ RA is also associated with an increased incidence of multiple age‐related comorbidities, including sarcopenic obesity, osteoporosis, physical disability, and exercise intolerance. ⁹ , ¹⁰ , ¹¹ , ¹² , ¹³ , ¹⁴ As these comorbidities accumulate in older adults with RA, they contribute to drastic reductions in overall health and well‐being. ¹⁵ , ¹⁶ Further, despite improvements in pharmacologic management—such as biologic disease‐modifying antirheumatic drug therapy—of RA inflammatory disease activity, the increased prevalence of RA CVD persists. ² , ³ , ⁴ , ⁵ , ⁶ , ⁷ , ⁸ Given risks for polypharmacy, infection, kidney and liver damage, and other adverse drug effects in older adults with RA, ¹⁵ there is a critical need for nonpharmacologic interventions to improve clinical care in this at‐risk population.

SIGNIFICANCE & INNOVATIONS.

• Lifestyle interventions impart multiple health benefits for patients with rheumatoid arthritis (RA); however, the combined effects of diet and exercise are understudied in older adults with RA who are at high risk for poor health outcomes.

• The combination of a hypocaloric diet, aerobic training, and resistance training benefits RA cardiometabolic disease risk, disease activity, and multiple patient‐reported outcomes; less intensive diet and exercise lifestyle counseling also improves RA cardiometabolic disease risk.

• This study is the first to show that a remotely delivered, multicomponent lifestyle intervention improves overall health in older patients with RA and overweight/obesity.

• Results of this study highlight the importance of lifestyle medicine in the routine clinical care of older adults with rheumatoid arthritis.

As both overweight/obesity and physical inactivity significantly contribute to poor clinical outcomes in persons with RA, ¹⁷ , ¹⁸ , ¹⁹ lifestyle interventions, including weight loss and increased physical activity, have great potential for improving RA clinical care. ¹⁷ For example, intentional weight loss can improve RA disease activity and immune function, ²⁰ , ²¹ , ²² whereas exercise training can also improve RA disease activity, as well as physical function, pain, sleep, fatigue, and overall health. ²³ Indeed, the recent American College of Rheumatology (ACR) guidelines strongly recommend consistent engagement in exercise as part of an integrative approach to manage RA (24). However, specific modes of exercise (eg, aerobic and resistance) and types of diet (eg, Mediterranean style) only received conditional recommendations because of a low certainty of evidence. ²⁴ In particular, studies showing the use of lifestyle interventions for older adults with RA are lacking. Thus, high‐quality studies are essential to advance and optimize the integrative management of RA.

Our group previously showed that a high‐intensity interval training treadmill walking exercise intervention in persons with RA age older than 55 improves cardiorespiratory fitness, disease activity, and innate immune cell function. ²⁵ In fact, improvements in RA disease activity were greatest in older age participants with high erythrocyte sedimentation rate (ESR), low cardiorespiratory fitness, and altered baseline skeletal muscle‐specific metabolism. ²⁶ Overall, cardiorespiratory fitness improvements were associated with positive changes in RA peripheral CD4+ T cell oxidative metabolism. ²⁷ In sum, our previous findings highlight the potential for exercise training to impact clinical outcomes by targeting systemic and tissue‐specific metabolism, especially in older patients with RA. In the present study, we sought to further evaluate the effects of exercise, in combination with weight loss through diet modification, on cardiovascular health, disease activity, and patient‐reported outcomes in a population of patients with RA at high risk for cardiometabolic dysfunction. In previously sedentary, older patients with RA and overweight/obesity, we hypothesized that, as compared with traditional lifestyle counseling, a 16‐week remotely supervised hypocaloric diet and exercise training intervention would result in greater improvements in CVD risk, RA disease activity, and patient‐reported outcomes.

PATIENTS AND METHODS

Participants

After informed consent and baseline assessments, 24 adults aged 60 to 80 years with overweight/obesity (body mass index: 28–40), with seropositive or erosive RA meeting 2010 ACR/EULAR criteria for RA, ²⁸ not currently meeting the 2018 Physical Activity Guidelines for Americans, ²⁹ and without absolute contraindications to exercise were enrolled.

Design

This 16‐week randomized controlled trial compared Supervised Weight loss and Exercise Training (SWET) and a standard‐of‐care control, Counseling Health As Treatment (CHAT). The study started July 2021 and ended February 2023. Of the 24 persons randomized, 21 completed baseline and postintervention assessments and 20 were included in final analyses (Figure 1). Trial design details, including eligibility, randomization, blinding, equipment, and remote interventions, were reported in the published protocol. ¹⁷ Participants were recruited from outpatient rheumatology clinics serving Duke Health in North Carolina. After screening for eligibility and informed consent, participants were examined by the study physician and assessed for readiness to complete cardiopulmonary exercise testing and the study exercise training intervention according to American College of Sports Medicine guidelines. ³⁰ Following baseline assessments, participants were randomized to SWET versus CHAT interventions in a 1:1 allocation ratio blocked by age decile with the variable block size known only to the study statistician. The study was approved by the Duke University Health System Institutional Review Board.

Figure 1.

Study flow diagram. BMI, body mass index; CHAT, Counseling Health As Treatment; PI, principal investigator; RA, rheumatoid arthritis; SWET, Supervised Weight loss and Exercise Training; TNFi, tumor necrosis factor inhibitor.

SWET

Using video conferencing and the study YouTube channel, SWET participants remotely completed three intervention components: hypocaloric diet, aerobic training, and resistance training. Details regarding each intervention component were described in detail in our previously published study rationale and design manuscript. ¹⁷ For the diet component, participants were supervised by a registered dietitian who provided an individualized hypocaloric diet prescription for a goal weight loss of 1 to 2 pounds per week, a loss of 7% body mass in total. Participants attended weekly live virtual group nutrition classes, completed weekly weigh‐ins (A&D Medical PLUSCONNECT wireless weight scale), and reported their weekly food intakes on the MyFitnessPal app, using their study‐provided tablet. For the aerobic training component, participants were supervised by an exercise physiologist, attended a weekly live virtual group aerobic exercise class, and were instructed to complete additional aerobic exercises (eg, RA‐specific aerobic exercise videos on the study YouTube channel) to meet weekly goals of 150 minutes of moderate‐to‐vigorous intensity exercise (ie, 45%–65% VO₂ reserve, determined via cardiopulmonary exercise testing with gas exchange and monitored by both corresponding heart rate via wrist‐worn Garmin device and rating of perceived exertion) and an average of 6,000 steps per day (recorded via Garmin device). For the resistance training component, participants were supervised by an exercise physiologist, attended a weekly live virtual group resistance exercise class, and were instructed to complete an additional resistance exercise session (eg, RA‐specific exercise videos with resistance bands on the study YouTube channel) for a total of twice weekly, nonconsecutive sessions of 10 to 11 exercises (1–3 sets of 8–15 repetitions) targeting all major muscle groups. Prescription for aerobic and resistance training components were based on US physical activity guidelines and in accordance with American College of Sports Medicine guidelines. ²⁹ , ³⁰

CHAT

Considering standard of care to include referrals for nutrition and physical/occupational therapy, CHAT participants remotely completed two 60‐minute lifestyle counseling sessions followed by usual care. ¹⁷ For the first counseling session with a registered dietitian, participants received general dietary recommendations to improve overall health. ³¹ For the second counseling session with an exercise physiologist, participants received physical activity recommendations to improve overall health. ²⁹ CHAT participants were contacted monthly by study staff via phone or email to minimize potential attention bias. Similar to SWET participants, CHAT participants also received food diaries, electronic scales, wrist‐worn Garmin devices, and resistance bands.

Outcome measures

Outcomes were assessed at two timepoints: baseline (preintervention) and 16‐weeks after randomization (postintervention). Clinical assessments, including fasted blood draws, were completed in‐person; Research Electronic Data Capture questionnaires for patient‐reported outcomes were completed online. ³²

The primary outcome was a composite metabolic syndrome z‐score (MSSc), which is a continuous weighted score of five metabolic syndrome components: waist circumference, mean arterial blood pressure, fasting glucose, triglycerides, and high density lipoprotein–cholesterol (HDLc). ³³ , ³⁴ A modified z‐score was calculated for each individual using continuous differences between the Adult Treatment Panel III guideline values and participant values with normalization to SD from study graduates included in these analyses (n = 20). ³⁵ To reduce the impact of outlier values and maintain a conservative estimate of intervention effects, individual MSSc variable values greater than 2 SD from the total cohort mean were excluded from SD and MSSc calculations. To account for sex‐specific Adult Treatment Panel III criteria, we used sex‐specific MSSc equations. For female participants, MSSc was calculated as z‐score:

$$\left[\left(50–\text{HDLc}\right)/16.5\right]+\left[\left(\text{triglycerides}–150\right)/38.5\right]+\left[\left(\text{fasting plasma glucose}–100\right)/11.8\right]+\left[\left(\text{waist circumference}–88\right)/8.0\right]+\left[\left(\text{mean arterial pressure}–100\right)/9.1\right].$$

For male participants, MSSc was calculated as z‐score:

$$\left[\left(40–\text{HDLc}\right)/16.5\right]+\left[\left(\text{triglycerides}–150\right)/38.5\right]+\left[\left(\text{fasting plasma glucose}–100\right)/11.8\right]+\left[\left(\text{waist circumference}–102\right)/8.0\right]+\left[\left(\text{mean arterial pressure}–100\right)/9.1\right].$$

MSSc is reported as a continuous value without maximum or cutoff point.

Anthropometrics and vital signs

With the participant wearing lightweight clothing and no shoes, height and body weight were measured with a stadiometer and digital scale, respectively. Waist circumference was measured with a flexible tape measure at the minimal waist (ie, smallest horizontal circumference above the umbilicus and below the xiphoid process). ³⁶ Resting blood pressure and heart rate were measured after the participant relaxed in a seated position for approximately 5 minutes.

Body composition, cardiorespiratory fitness, muscle strength, and step counts

Fat mass and fat‐free mass were assessed via air displacement plethysmography (BOD POD GS‐X). Cardiorespiratory fitness (peak VO₂) was determined by cardiopulmonary exercise testing with 12‐lead electrocardiography and continuous expired gas analysis (ParvoMedics TrueOne 2400) using a graded treadmill protocol until the participant reached volitional exhaustion (30). Muscle strength was assessed via quadriceps isometric knee extension (HUMAC NORM) as peak and average peak torque across three trials and bilateral hand‐grip strength (Jamar hydraulic hand dynamometer) testing. Step counts during the intervention period were measured and recorded for participants in both SWET and CHAT groups via the wrist‐worn Garmin device (Garmin Forerunner 45).

Blood‐based biomarkers

Phlebotomy was performed after a 12‐hour, overnight fast. ESR and high‐sensitivity C‐reactive protein (CRP) concentration were determined via commercial clinical analysis (Labcorp). Plasma was immediately isolated via centrifugation and stored at −80°C. Stored plasma samples were analyzed for the following: (1) glucose and lipid/lipoprotein profiles using nuclear magnetic resonance LipoProfile testing (LP4 algorithm; Labcorp) ³⁷ , ³⁸ ; and (2) leptin and total adiponectin concentrations using human multiplex immunoassays assessed in duplicate (Meso Scale Discovery). Concentrations were obtained for all samples measured and intra‐assay coefficients of variation for leptin and adiponectin were 5.4 and 3.0%, respectively.

Disease activity

RA disease activity was assessed with the disease activity score‐28 (DAS‐28), which is a composite measure including a self‐reported overall health assessment on a 100 mm visual analog scale; the number of tender and swollen joints determined from a 28‐joint examination by the blinded study physician; and either ESR or high‐sensitivity CRP. ³⁹

Patient‐reported outcomes

Patient‐reported outcomes included medical history and measures of physical health, physical function, mental health, pain, and fatigue derived from the Patient‐Reported Outcomes Measurement Information System (PROMIS) bank (https://commonfund.nih.gov/promis/index). ⁴⁰ Questionnaires were scored using “response pattern scoring” according to PROMIS scoring manuals to calculate individual T‐scores for each outcome measure.

Statistical analysis

Based on data from non‐RA participants completing similar lifestyle interventions, sample sizes were calculated to detect a clinically significant difference in absolute MSSc change between the SWET and CHAT groups of 2.5 (SD = 2.1). ¹⁷ , ³³ , ³⁴ With 26 RA participants enrolled and randomized with an expected attrition rate of 20% to 25% during the intervention period, 10 participants per group provided 80% power at a one‐tailed α of 0.05 to detect a group difference of 2.5, in which each unit difference in MSSc corresponds to a hazard ratio for CVD of 1.49 (CI 1.37–1.62). ⁴¹

Comparative analyses were completed with the intention‐to‐treat principle using SAS statistical software (v.9.4). Missing outcomes post‐intervention were imputed with pre‐intervention values. To assess the impact of the intervention on the primary outcome (i.e., absolute change in MSSc; post‐intervention minus pre‐intervention), difference between groups was assessed with regression modeling, controlling for between‐group differences at baseline, in which β₂ is the group effect for the primary outcome with significance set at P < 0.05:

$${\text{ΔMSSc}}_{\mathrm{i}}={\mathrm{\beta }}_{0+}{\mathrm{\beta }}_1{\text{baselineMSSc}}_{\mathrm{I}+}{\mathrm{\beta }}_2{\text{Group}}_{\mathrm{j}\left(\mathrm{j}=1\text{if SWET},\mathrm{j}=0\text{if CHAT}\right)}+{\mathrm{\epsilon }}_{\mathrm{i}}$$

Between‐group differences in secondary outcomes were assessed similarly with regression modeling without control for type‐I error because of the exploratory nature of these analyses. To inform between‐group analyses, within‐group differences were analyzed with Wilcoxon signed rank tests. The strength of relationships between relative outcome variable changes [(postintervention variable minus preintervention variable)/preintervention variable × 100] were determined with Spearman's rank correlations.

Canonical correlations were performed among all study participants to understand the strength of relationships between intervention component‐specific outcomes with the primary and key secondary outcomes. Set 1 canonical correlation test variables included absolute change in fat mass (ie, diet‐specific component outcome), cardiorespiratory fitness (ie, aerobic training‐specific outcome), and knee extension muscle strength (ie, resistance training‐specific component outcome); set 2 canonical correlation test variables included either the primary outcome (ie, absolute change in MSSc) or key secondary outcomes (ie, absolute change in RA disease activity and patient‐reported outcomes). Standardized canonical coefficients were interpreted as follows: a one SD change in a set 1 variable leads to a coefficient value SD change in the score for the set 2 outcome when the other model variables are held constant.

RESULTS

Results are presented as mean ± SD unless otherwise specified. Participant demographics and flow through the trial are shown in Table 1 and Figure 1.

Table 1.

Participants with rheumatoid arthritis and overweight/obesity baseline clinical characteristics^*

Variables	All participants (n = 20)	CHAT control group (n = 10)	SWET intervention group (n = 10)
Age, mean (SD), y	66.7 (5.4)	65.6 (5.4)	67.7 (5.4)
Sex, female n (%)	16 (80)	9 (90)	7 (70)
Race, n (%)
Black or African American	7 (35)	6 (60)	1 (10)
White	13 (65)	4 (40)	9 (90)
Weight, kg (SD)	84.7 (9.4)	86.3 (11.3)	83.0 (7.1)
RF positive, n (%)	18/19 (94.7)	8/9 (88.9)	10/10 (100)
Anti‐CCP antibody positive, n (%)	17/17 (100)	9/9 (100)	8/8 (100)
Erosions on radiograph present, n (%)	4/14 (35.7)	3/9 (33.3)	2/5 (40.0)
Disease duration, mean (SD), y	15.2 (10.6)	12.6 (11.0)	17.7 (10.1)
DAS‐28‐ESR (SD)	3.5 (1.1)	3.6 (1.0)	3.3 (1.2)
DAS‐28‐CRP (SD)	3.0 (1.1)	3.1 (1.0)	2.9 (1.2)
Disease remission (<2.6), n (%)	8 (40)	3 (30)	5 (50)
Low disease activity (2.6–3.2), n (%)	5 (25)	4 (40)	1 (10)
Moderate disease activity (>3.2–5.1), n (%)	7 (35)	3 (30)	4 (40)
Medication use, n (%)
Antihypertensive	14 (70)	8 (80)	6 (60)
Statin	8 (40)	4 (40)	4 (40)
Aspirin	3 (15)	2 (20)	1 (10)
NSAID	8 (40)	5 (50)	3 (30)
Prednisone	5 (25)	2 (20)	3 (30)
Hydroxychloroquine	6 (30)	4 (40)	2 (20)
Methotrexate	11 (55)	7 (70)	4 (40)
Leflunomide	2 (10)	1 (10)	1 (10)
TNFi	11 (55)	4 (40)	7 (70)

^*CCP, cyclic citrullinated peptide; CHAT, counseling health as treatment; CRP, C‐reactive protein; DAS‐28, disease activity score in 28 joints; ESR, erythrocyte sedimentation rate; NSAID, nonsteroidal anti‐inflammatory drug; RF, rheumatoid factor; SWET, supervised weight loss and exercise training; TNFi tumor necrosis factor inhibitor.

Intervention adherence

For the diet component, SWET participants completed an average of 96.4 ± 7.5% of target weekly weigh‐ins via home scale and participated in 93.1 ± 10.4% of weekly dietitian‐led nutrition classes. For the aerobic training component, SWET participants completed an average of 83.1 ± 46.8% of the goal 150 minutes/week of moderate‐to‐vigorous aerobic exercise within target heart rate range or target rating of perceived exertion. For the resistance training component, SWET participants completed an average 84.2 ± 17.1% of the target two resistance training sessions/week. Throughout the study period, the average daily step count was 7,066 ± 2,117 steps per day for SWET group participants and 7,093 ± 2,965 steps per day for CHAT group participants.

Participant safety

Throughout the study, safety events were monitored, documented, and classified according to the National Institute of Health guidelines. Although no serious adverse events occurred, among study completers, there were four confirmed cases of COVID‐19 (1 case in CHAT; 3 cases in SWET) and 10 reports of musculoskeletal symptoms (6 related to preexisting conditions and 4 related to falls and/or injuries unrelated to study‐related activities), which temporarily limited study participation and/or intervention adherence.

CVD risk

The primary outcome, change in MSSc, did not differ significantly between groups (mean difference in MSSc absolute change (SWET – CHAT): 0.33; 95% CI −0.58 to 1.24); however, MSSc improved in both the CHAT control group (absolute z‐score change: −1.34 ± 1.30; P = 0.01) and the SWET intervention group (absolute z‐score change: ‐1.67 ± 0.64; P = 0.002) (Table 2; Figure 2A). Of the five MSSc components, as compared with CHAT, SWET participants had reductions in waist circumference (mean difference in change: 5.1 cm; 95% CI 2.3–7.9) and HDLc concentration (mean difference in change: −13.8 mg/dl; 95% CI −19.4 to −8.2) (Figure 2B).

Table 2.

Pre‐ and post‐intervention clinical outcomes^*

Variables	CHAT control group		Within group pre‐post P value	SWET intervention group		Within group pre‐post P value	Between group absolute change P value
	Pre (0 weeks) n = 10	Post (16 weeks) n = 10		Pre (0 weeks) n = 10	Post (16 weeks) n = 10
Anthropomorphic and body composition assessments
Weight (kg)	86.3 (11.3)	84.1 (11.3)	0.03	83.0 (7.1)	78.2 (8.1)	0.002	0.04
BMI (kg/m2)	33.6 (3.5)	32.7 (3.3)	0.03	31.3 (1.6)	29.5 (1.8)	0.002	0.04
Fat mass (kg)	42.0 (7.3)	39.8 (7.6)	0.01	37.0 (4.9)	32.3 (4.2)	0.002	0.04
Lean mass (kg)	44.5 (7.8)	44.9 (7.5)	0.43	46.6 (6.8)	46.6 (6.9)	1.0	0.68
Minimal waist circumference (cm)	100.1 (8.5)	98.7 (8.6)	0.20	97.6 (6.4)	91.2 (6.4)	0.003	0.002
Muscle strength and cardiorespiratory fitness assessments
Grip strength, right hand (kg)	19.1 (9.9)	18.6 (9.6)	0.97	27.0 (9.7)	26.6 (7.4)	0.72	0.45
Grip strength, left hand (kg)	17.4 (10.2)	19.6 (10.2)	0.19	24.7 (9.3)	26.4 (7.7)	0.27	0.79
Isometric knee extension peak torque (Nm)	105.9 (48.7)	115.6 (48.7)	0.21	124.0 (39.3)	131.9 (33.4)	0.31	0.94
Isometric knee extension average torque (Nm)	102.6 (48.1)	111.8 (48.0)	0.23	117.8 (37.8)	127.2 (31.5)	0.16	0.83
Peak aVO2, (L/min)	1.61 (0.46)	1.69 (0.47)	0.25	1.62 (0.31)	1.68 (0.43)	0.38	0.67
Peak rVO2, (ml/kg/min)	18.4 (3.9)	19.8 (3.5)	0.07	19.2 (2.6)	21.1 (3.6)	0.04	0.28
Cardiovascular disease risk assessment
MSSc (z‐score)	−1.89 (2.39)	−3.23 (2.80)	0.01	−1.82 (2.01)	−3.49 (2.28)	0.002	0.48
Total cholesterol (mg/dl)	179.3 (21.7)	180.5 (25.3)	0.87	214 (40.2)	184.5 (32.5)	0.08	0.59
LDL cholesterol (mg/dl)	90.5 (24.1)	87.7 (24.3)	0.45	117.9 (33.7)	104.6 (32.5)	0.38	0.78
HDL cholesterol (mg/dl)	72.5 (17.4)	76.7 (19.2)	0.13	68.0 (24.5)	59.6 (15.6)	0.01	0.002
Triglycerides (mg/dl)	90.2 (35.0)	88.3 (41.2)	0.51	188.9 (270.4)	111.6 (81.5)	0.01	0.63
Fasting plasma glucose (mg/dl)	102.2 (10.4)	98.2 (8.2)	0.39	113.6 (46.6)	101.3 (14.7)	0.27	0.95
Mean arterial pressure (mmHg)	96.0 (11.2)	91.2 (13.0)	0.10	94.8 (7.2)	88.9 (9.2)	0.08	0.72
Plasma leptin (pg/ml)	37,110 (22,391)	34,783 (16,642)	0.70	28,548 (8,013)	18,778 (9,908)	0.02	0.02
Plasma adiponectin (μg/ml)	23.0 (12.8)	23.8 (12.4)	0.13	24.1 (18.9)	24.0 (18.3)	0.70	0.47
Plasma adiponectin: leptin ratio	0.00069 (0.00038)	0.00071 (0.00036)	0.63	0.00087 (0.00067)	0.00184 (0.00247)	0.01	0.24
Rheumatoid arthritis disease activity assessment
DAS‐28‐ESR	3.6 (1.0)	3.3 (0.9)	0.37	3.3 (1.2)	2.8 (1.0)	0.18	0.31
DAS‐28‐CRP	3.1 (1.0)	2.9 (0.8)	0.26	2.9 (1.2)	2.1 (0.9)	0.02	0.04
Tender joints (#)	2.7 (4.8)	1.7 (2.1)	0.75	2.4 (2.9)	0.7 (0.9)	0.03	0.04
Swollen joints (#)	4.3 (3.4)	2.3 (1.7)	0.03	3.2 (3.7)	0.9 (1.2)	0.06	0.05
Patient global assessment VAS (mm)	31.1 (20.3)	22.2 (22.8)	0.16	26.4 (26.5)	13.3 (22.9)	0.03	0.47
hs‐CRP (mg/L)	8.2 (13.0)	9.3 (11.5)	0.91	4.7 (9.5)	3.6 (4.4)	0.04	0.24
ESR (mm/hour)	26.0 (24.0)	29.5 (31.1)	0.27	22.3 (18.0)	26.5 (17.6)	0.57	0.89
Patient reported outcome assessment
PROMIS‐physical health (T‐score)	47.3 (7.3)	45.1 (5.8)	0.31	47.7 (5.7)	52.3 (6.6)	0.08	0.01
PROMIS‐physical function (T‐score)	41.7 (5.0)	41.7 (6.4)	0.90	42.3 (6.3)	48.1 (6.5)	0.01	0.01
PROMIS‐mental health (T‐score)	51.6 (6.6)	51.9 (6.6)	0.52	54.1 (8.8)	57.6 (8.4)	0.01	0.02
PROMIS‐cognitive function (T‐score)	53.6 (7.7)	51.1 (8.7)	0.25	49.3 (9.6)	52.3 (8.0)	0.12	0.16
PROMIS‐pain intensity (T‐score)	46.5 (6.1)	45.7 (4.0)	0.50	46.3 (5.8)	41.7 (7.9)	0.04	0.11
PROMIS‐fatigue (T‐score)	49.0 (9.1)	52.0 (6.0)	0.16	49.0 (6.3)	44.0 (9.9)	0.06	0.02

^* Source: Values are shown as mean (SD).

aVO₂, absolute cardiorespiratory fitness; BMI, body mass index; CHAT, counseling health as treatment; DAS‐28, disease activity score in 28 joints; ESR, erythrocyte sedimentation rate; HDL, high density lipoprotein; hs‐CRP, high sensitivity C‐reactive protein; LDL, low density lipoprotein; MSSc, metabolic syndrome z‐score; PROMIS, Patient‐Reported Outcomes Measurement Information System; rVO₂, relative cardiorespiratory fitness, SWET, supervised weight loss and exercise training; VAS visual analog scale.

Figure 2.

Graphs show percent (%) change in cardiovascular disease risk in older patients with RA and overweight/obesity following CHAT control versus SWET lifestyle interventions. (A) Graphs show changes from pre‐ (closed triangle) to post‐intervention (open triangle) in individual RA participants (n = 10 for CHAT; n = 10 for SWET) metabolic syndrome z‐score. Graphs show percent (%) change following interventions in CHAT (closed circle) versus SWET (open circle) group participants; (B) metabolic syndrome components: Waist, MAP, Glucose, Tri, and HDLc; and C) additional lipoprotein parameters: total Chol:HDLc, total HDLp, HDLp size, ApoB, ApoA1, and ApoB:ApoA1 ratio. *P < 0.05 (without multiple testing correction) for between group differences assessed via linear regression modeling. ApoA1, apolipoprotein A1; ApoB, apolipoprotein B; CHAT, counseling health as treatment; CholHDLc cholesterol:HDLc ratio; Glucose, fasting plasma glucose; HDLc, high‐density lipoprotein cholesterol; HDLp, high‐density lipoprotein particles; MAP, mean arterial pressure; SWET, supervised weight loss and exercise training; Tri, triglycerides; RA, rheumatoid arthritis; Waist, waist circumference.

To explore the unanticipated SWET‐specific reduction in plasma HDLc, we analyzed nuclear magnetic resonance–based lipoprotein parameters, including total cholesterol:HDLc ratio, high‐density lipoprotein particle number (HDLp) and particle size, apolipoprotein B (ApoB), apolipoprotein A1 (ApoA1), and ApoB:ApoA1 ratio (Figure 2C). Compared with the CHAT group, the SWET group experienced significant decreases in plasma total HDLp, HDLp size, large HDLp, and ApoA1 following the intervention (P < 0.03 for all) (Supplementary Table 1).

Given the strong association between obesity‐associated hyperleptinemia and hypoadiponectinemia with CVD, ⁴² we analyzed changes in plasma leptin and adiponectin concentrations. As compared with the CHAT control group, the SWET participants had reduced leptin (mean difference in change: 10,697 pg/ml; 95% CI 2,404–18,990) and unchanged adiponectin (mean difference in change: 0.79 μg/ml; 95% CI −1.32 to 2.90). The change in ratio of adiponectin:leptin did not significantly differ between groups (mean difference in change: −0.00095; 95% CI −0.00211 to 0.00021) but increased in the SWET group (absolute adiponectin:leptin ratio change: 0.00097 ± 0.002; P = 0.01).

Body composition, cardiorespiratory fitness, and muscle strength

Body weight and fat mass significantly improved in both the CHAT group (weight change: −2.2 ± 2.6 kg, P = 0.03; fat‐mass change: −2.3 ± 3.2 kg, P = 0.01) and the SWET group (weight change: −4.8 ± 2.4 kg, P = 0.002; fat‐mass change: −4.7 ± 2.6 kg, P = 0.002); however, magnitudes of change were significantly greater in SWET (body weight change mean difference: 2.6 kg, 95% CI 0.3–4.9; fat mass change mean difference: 3.1 kg, 95% CI 0.3–5.9). Lean mass did not change in CHAT or SWET (P > 0.05 for both groups) (Table 2).

The SWET group experienced an increase in peak VO₂ relative to body weight (rVO₂; ml O₂/kg body weight/min) of 10.2 ± 13.9% (P = 0.04), whereas the CHAT group did not change significantly (5.0 ± 6.6%, P > 0.05). Absolute peak VO₂ (aVO₂; L O₂/min) did not change in CHAT or SWET (P > 0.05 for both groups). Neither group experienced significant changes in unilateral isometric knee extension or bilateral grip strength.

Disease activity

Following the intervention, SWET participants improved DAS‐28‐CRP by 22%, which was significantly greater than the 6% improvement in CHAT (mean difference in change: 0.64; 95% CI 0.09–1.19) (Table 2; Figure 3A). DAS‐28‐ESR improved by only 13% in SWET, which was not different from CHAT (mean difference in change: 0.35; 95% CI −0.32 to 1.02). Among all participants, relative improvements in DAS‐28‐CRP were most strongly correlated with decreases in waist circumference (rho = 0.48; P = 0.03) (Figure 3B) and increases in isometric knee extension average peak torque (rho = 0.57; P = 0.01) (Figure 3C) (Supplementary Table 2).

Figure 3.

Changes in disease activity and patient reported outcomes in older patients with RA and overweight/obesity following CHAT control versus SWET lifestyle interventions. (A) Graphs show changes from pre‐ (closed triangle) to post‐intervention (open triangle) in individual RA participant (n = 10 for CHAT; n = 10 for SWET) DAS28‐CRP. Scatter plots depict relationships between percent (%) change in DAS28‐CRP (x‐axis) with % change in (B) waist circumference and (C) average isometric knee extension torque (strength) following lifestyle intervention among all study participants (n = 20). (D) Graphs show % change following interventions in CHAT (closed circle) versus SWET (open circle) group participant PROMIS patient reported outcomes. *P < 0.05 (without multiple testing correction) for between group differences assessed via linear regression modeling. CHAT, counseling health as treatment; DAS‐28‐CRP, disease activity score in 28 joints PROMIS, Patient‐Reported Outcomes Measurement Information System; RA, rheumatoid arthritis; SWET, supervised weight loss and exercise training.

Patient‐reported outcomes

Compared with the CHAT group, the SWET group improved multiple PROMIS metrics, including physical health, physical function, mental health, and fatigue (P ≤ 0.02 for each metric) (Table 2; Figure 3D). Improvements in physical health and physical function were most strongly correlated with increases in peak aVO₂ (rho = 0.60 and rho = 0.56, respectively; P = 0.01 for each) and peak rVO₂ (rho = 0.78 and rho = 0.67, respectively; P < 0.01 for each); physical health improvements also correlated with a decrease in body weight (rho = −0.55; P = 0.01) (Supplementary Table 2). Improvements in mental health were most strongly correlated with decreases in body weight (rho = −0.51; P = 0.02) and mean arterial pressure (rho = −0.53; P = 0.02). Improvements in fatigue were most strongly correlated with decreases in fat mass (rho = 0.49; P = 0.03) and plasma leptin concentrations (rho = 0.47; P = 0.04) and increases in peak rVO₂ (rho = −0.54; P = 0.02) and isometric knee extension average peak torque (rho = −0.47; P = 0.04).

Relative intervention component effects

Relationships between changes in primary and key secondary outcomes were evaluated and shown in Supplementary Table 2. Canonical correlations for changes in fat mass, leg strength, and cardiorespiratory fitness revealed the strongest relative impact of (1) the diet component on changes in MSSc, physical health, and mental health; (2) the aerobic training component on changes in physical function and fatigue; and (3) the resistance training component on change in DAS‐28‐CRP (Table 3).

Table 3.

Canonical correlations between absolute change in intervention component‐specific outcomes with absolute change in primary outcome and key secondary outcomes^*

Variable (absolute change; post‐pre)	MSSc	DAS‐28‐CRP	PROMIS‐physical health	PROMIS‐physical function	PROMIS‐mental health	PROMIS‐fatigue
Fat mass (kg)	0.81	0.68	−0.81	−0.74	−0.79	0.58
Peak aVO2 (L/min)	−0.68	0.02	0.69	0.76	0.47	−0.79
Isometric knee extension average torque (Nm)	0.12	−0.78	0.11	0.16	‐0.47	−0.40

^* Source: Values are shown as standardized canonical coefficients. Diet component represented by absolute change in fat mass; aerobic training component represented by absolute change in peak aVO₂; resistance training component represented by absolute change in isometric knee extension average torque.

aVO₂, absolute cardiorespiratory fitness; DAS‐28‐CRP, disease activity score in 28 joints with C‐reactive protein; MSSc, metabolic syndrome z‐score; PROMIS, Patient‐Reported Outcomes Measurement Information System.

DISCUSSION

Following the 16‐week interventions, older adults with RA and overweight/obesity in both SWET and CHAT groups significantly improved their CVD risk profiles, as measured by the composite MSSc. When compared between groups, the magnitude of MSSc improvement in SWET was not significantly greater than that of CHAT—a finding contrary to our original hypothesis for the primary outcome. Although not to as great of an extent as the SWET group, participants in the CHAT control group also experienced significant improvements in several markers of body composition, including body weight, fat mass, and minimal waist circumference. These beneficial changes in CVD risk profiles exhibited by the CHAT group highlight the importance of lifestyle counseling to improve health in older patients with RA who are motivated to make behavioral modifications. Further study is needed to better identify which patients with RA at risk for CVD would benefit from a less intensive lifestyle counseling program alone (eg, CHAT) versus those who would need a more intensive and supervised lifestyle modification intervention (eg, SWET).

Despite improvements in other aspects of cardiometabolic health, SWET participants had reductions in plasma HDLc, HDL particles, and ApoA1 (ie, a major component of HDLc). HDLc reductions are generally considered to be an adverse health outcome, as therapies that increase HDLc lead to a reduction in CVD risk. ⁴³ As HDLc is one of only five MSSc components, the observed HDLc reductions in the SWET group likely contributed to the lack of group difference in MSSc. One explanation for these findings is that HLDc reductions occur as a direct effect of active weight loss with dietary fat restriction leading to less total chylomicron‐derived lipoprotein production. ⁴⁴ , ⁴⁵ , ⁴⁶ Indeed, in addition to HDLc and triglyceride reductions, participants in the SWET group nonsignificantly reduced total cholesterol and LDLc while maintaining cholesterol:HDLc and ApoB:ApoA1 ratios. Intriguingly, across all participants, decreases in HDLc were associated with improvements in patient‐reported physical health. Thus, the potential for lifestyle interventions to influence HDL as a means to improve health and quality of life deserves further evaluation in older adults with RA. ⁴⁷

The comprehensive SWET program overall demonstrated the powerful ability of a remotely supervised weight loss and exercise intervention to substantially impact a multitude of health markers in older adults with RA and overweight/obesity. In addition to improvements in CVD risk profiles, SWET participants also had beneficial changes in body composition and self‐reported measures of physical health, physical function, mental health, and fatigue. The SWET intervention also elicited improvements in RA disease activity; not only did DAS‐28‐CRP improve by 22% across all SWET participants, but those with low or moderate disease activity at baseline on average experienced an even greater improvement of 35%. These disease activity improvements point toward the potential for more intensive, supervised lifestyle interventions to serve as nonpharmacologic, disease‐modifying therapies for older patients with RA.

Two recent studies also explored the effects of lifestyle interventions on RA disease activity. The “Plants for Joints” trial randomized 83 adults (≥18 years) with RA to 16 weeks of lifestyle counseling (including 10 group sessions focused on whole‐food plant‐based diet, physical activity, and stress management) or usual care control. ⁴⁸ Similar to our findings, participants in the Plants for Joints intervention group comparatively improved disease activity by approximately 26% with concomitant improvements in body composition and metabolic status. In another study, 49 older adults (≥65 years) with RA were randomized to either 20 weeks of moderate‐to‐high intensity aerobic and resistance exercise or active control home‐based light intensity exercise. ⁴⁹ In contrast to our findings, disease activity did not significantly improve following the supervised exercise intervention; these inconsistent findings are likely due in part to a greater proportion (ie, 73% versus 60%) of participants in remission or with low disease activity at baseline and key differences in study design (eg, exercise alone versus exercise plus hypocaloric diet). ¹⁷ , ⁵⁰ Thus, further investigation is critical to delineate the effects of specific lifestyle interventions on improving RA disease activity, including patients with higher disease activity, difficult‐to‐treat disease, and those at‐risk for polypharmacy and medication side effects. ¹⁵

The SWET intervention significantly improved patient‐reported outcomes across multiple domains (as measured by PROMIS). The SWET group not only reported beneficial changes in physical health and physical function, but also in mental health and fatigue. Surprisingly, our patient‐reported outcome findings contrast with those from the Plants for Joints trial, in which the intervention group did not significantly change compared with the control arm for PROMIS measures of depression, fatigue, pain, and physical function. ⁴⁸ These contrasting findings may be due to differences in study design (eg, group counseling physical activity sessions versus individualized exercise training). ¹⁷ , ⁵¹ Notably, current evidence still supports the use of lifestyle interventions for improving RA patient‐reported outcomes; however, further study is needed to optimize lifestyle modification programs specific to patient goals.

Each component of the SWET intervention likely contributed, at least partially, to the beneficial effects of the intervention. Based on results from canonical correlations, the diet component was linked to improvements in MSSc, physical health, and mental health, aerobic training was linked to improvements in physical function and fatigue, and resistance training was linked to improvements in RA disease activity. These differential relationships for each component suggest that a hypocaloric diet, aerobic training, and resistance training exert unique effects, and thus, are each beneficial for improving health in older patients with RA and overweight/obesity. Indeed, as described in persons without RA, combinations of these lifestyle intervention components can have additive, or even synergistic, benefits. ¹⁷ , ⁵² However, given the amount of time and resource allocation needed for multicomponent lifestyle program implementation, further study of intervention components both alone and in combination is needed to optimize clinical care for older patients with RA.

Limitations to consider in this study include the small sample size, a highly motivated group of CHAT participants, lack of blinding for intervention participants, and the potential impact of COVID‐19. Although only 20 participants were included in the final analyses, our study was adequately powered to detect clinically important differences in RA‐related CVD risk (ie, MSSc) based on findings observed in similar studies that included combined aerobic training, resistance training, and weight loss diet in non‐RA populations. ¹⁷ , ³³ , ³⁴ , ⁵³ Although we found no difference between groups in MSSc change, these findings need to be considered in the context of unexpected HDLc decreases in the SWET group and potential contamination from the highly motivated CHAT group participants, who also significantly decreased their body weight and maintained higher levels of physical activity than expected. Nevertheless, compared with CHAT, the SWET group had greater beneficial changes in multiple key secondary measures, including RA disease activity and patient‐reported outcomes. However, significant between‐group differences in patient‐reported outcomes could possibly be related to an enhanced placebo effect, as participants were not blinded to the lifestyle intervention protocol. Further, as these findings only show efficacy for the SWET program, larger trials with greater generalizability are needed to assess effectiveness. Contrary to our expectations for this older, sedentary group of participants with RA, neither intervention group significantly improved muscle strength nor absolute cardiorespiratory fitness. This lack of objective improvements in physical fitness may be due to interference from the weight loss component of the intervention, factors related to study participants, response heterogeneity, and/or a lack of statistical power for secondary analyses; these and other potential contributing factors need to be explored in future, larger lifestyle intervention clinical trials for older adults with RA.

Despite the entirety of the study occurring during the COVID‐19 pandemic, we were able to reach our target sample sizes for intervention completion and final analyses. We also acknowledge the potential physiologic impact of COVID‐19, in which the presence of SARS‐CoV‐2 can further complicate the complex relationship among RA disease activity, comorbidities, and functional capacity. For participants reporting positive COVID‐19 testing, we temporarily adjusted the SWET intervention (eg, by reducing exercise volume and intensity) and, if needed, delayed post‐intervention assessments until after self‐reported recovery or up to a maximum of two weeks; nonetheless, we posit the downstream effects of SARS‐CoV‐2 infection could have impacted several outcome measures, such as cardiorespiratory fitness, disease activity, and patient‐reported outcomes.

In summary, results from this trial support the use of a remotely delivered, supervised weight loss and exercise training program for older adults with RA and overweight/obesity to improve overall health. Following the 16‐week intervention, SWET participants improved CVD risk profiles, RA disease activity, and various patient‐reported outcomes. Furthermore, this trial highlights the importance of providing general diet and physical activity counseling with accompanying self‐monitoring tools, as the CHAT control participants also experienced beneficial changes in weight, fat mass, and CVD risk. Unfortunately for patients with RA, lifestyle counseling for healthy diet and physical activity behaviors is not routinely implemented in clinical practice. ⁵⁴ Findings from our study indicate, at a minimum, integrating even two hours of healthy lifestyle counseling may improve RA management, let alone demonstrate the substantial impact that can be provided by a comprehensive, remotely supervised lifestyle intervention. Further study is needed to assess the durability of these observed beneficial health changes, distinguish the specific health effects of individual lifestyle components (eg, diet versus exercise), and guide implementation strategies for integrating lifestyle medicine to optimize the routine clinical care for older adults with RA.

AUTHOR CONTRIBUTIONS

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 published. Dr. Andonian 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. “Analysis and interpretation of data” section as follows: Andonian, Ross, Sudnick, Pieper, Huebner, Huffman.

Study conception and design

Andonian, Ross, Sudnick, Johnson, Pieper, Belski, Counts, Huebner, Connelly, Siegler, Kraus, Bales, Porter Starr, Huffman.

Acquisition of data

Andonian, Ross, Sudnick, Belski, King, Wallis, Bennett, Gillespie, Moertl, Richard.

Analysis and interpretation of data

Andonian, Ross, Sudnick, Huebner, Huffman.

Supporting information

Disclosure form:

Supplementary Table 1: Pre‐ and post‐intervention lipoprotein outcomes.

Supplementary Table 2: Relationships between relative (%) change in primary and key secondary outcomes following 16‐week intervention in the entire study cohort (n=20).