Improved outcomes in rheumatoid arthritis with obesity after a weight loss intervention: randomized trial

Abstract

Objective

To examine whether a weight loss intervention programme improves RA disease activity and/or musculoskeletal ultrasound synovitis measures in obese RA patients.

Methods

We conducted a proof-of-concept, 12-week, single-blind, randomized controlled trial of obese RA patients (BMI ≥ 30) with 28-joint DAS (DAS28)  ≥ 3.2 and with evidence of power Doppler synovitis. Forty patients were randomized to the diet intervention (n = 20) or control group (n = 20). Diet intervention consisted of a hypocaloric diet of 1000–1500 kcal/day and high protein meal replacements. Co-primary outcomes included change in DAS28 and power Doppler ultrasound (PDUS)-34. Clinical disease activity, imaging, biomarkers, adipokines and patient-reported outcomes were monitored throughout the trial. Recruitment terminated early. All analyses were based on intent-to-treat for a significance level of 0.05.

Results

The diet intervention group lost an average 9.5 kg/patient, while the control group lost 0.5 kg (P < 0.001). Routine Assessment of Patient Index Data 3 (RAPID3) improved, serum leptin decreased and serum adiponectin increased significantly within the diet group and between the groups (all P < 0.03). DAS28 decreased, 5.2 to 4.2, within the diet group (P < 0.001; −0.51 [95% CI −1.01, 0.00], P = 0.056, between groups). HAQ-Disability Index (HAQ-DI) improved significantly within the diet group (P < 0.04; P = 0.065 between group). Ultrasound measures and the multi-biomarker disease activity score did not differ between groups (PDUS-34 −2.0 [95% CI −7.00, 3.1], P = 0.46 between groups).

Conclusion

Obese RA patients on the diet intervention achieved weight loss. There were significant between group improvements for RAPID3, adiponectin and leptin levels, and positive trends for DAS28 and HAQ-DI. Longer-term, larger weight loss studies are needed to validate these findings, and will allow for further investigative work to improve the clinical management of obese RA patients.

Trial registration

ClinicalTrials.gov, https://clinicaltrials.gov, NCT02881307

RA, clinical trial, obesity, weight loss, ultrasound, outcome measures

Rheumatology key messages

• Obese RA patients on a weight loss diet achieved significant improvements in RAPID3 and adipokines.

• DAS28 and HAQ-DI improved significantly within the diet intervention group; this was not seen in controls.

• This study suggests weight loss may improve obese RA outcomes; further research is warranted.

Introduction

RA patients with obesity represent >30% of North American RA patients [1, 2] and are less likely to attain remission [3–5] and optimal response to therapy than non-obese patients [4–6]. They also have evidence of more joint deformity, functional disability, decreased quality of life and increased cardiovascular risks than non-obese patients [3, 7, 8].

Although clinicians advocate for weight loss because of substantive economic and public health concerns [8–10], there is still a lack of definitive evidence about the effects of weight loss on RA disease activity. In this context, there are no standardized interventions for the reduction of obesity rates in RA patients. Exercise is considered safe [11] but a broad implementation of exercise regimens in RA patients is limited by fears of causing injury and/or aggravating disease, in addition to being highly variable in terms of weight loss results [12–14]. Data suggest that hypocaloric diets have been successful in producing weight loss and health benefits, and the World Health Organization recommends a minimum of 5% body weight loss in obese patients [7, 15, 16]. Yet, few studies have investigated the impact of weight loss in obese RA patients through dietary modifications [17, 18], and no studies have examined its impact comprehensively on RA disease outcomes (clinical, imaging and biomarker measures).

Here we report the results of a proof-of-concept, 12-week, randomized controlled trial (RCT) of a hypocaloric diet in RA patients with obesity to evaluate the effects of weight loss on clinical, imaging and biomarker/adipokine outcomes.

Methods

Forty obese RA patients were recruited from multiple outpatient rheumatology clinical sites throughout the greater Los Angeles, CA area within the University of California, Los Angeles (UCLA) during October 2016 to July 2019. Inclusion criteria included obesity (BMI ≥ 30) and meeting the ACR 2010 classification criteria for RA. Patients were required to be on stable DMARDs for at least 12 weeks prior to enrolment and throughout the study, stable prednisone ≤ 10 mg every day, baseline 28-joint DAS (DAS28)/ESR ≥ 3.2 and baseline 34-joint power Doppler ultrasound (PDUS) score ≥10. Patients were excluded if diagnosed with anorexia, binge eating disorder, or if they had undergone prior bariatric surgery.

Enrolled RA patients were randomized 1:1 to a diet intervention group or control group (Fig. 1). All patients were recommended to walk at least 50 min/week. This was a single blind study, where the joint assessor and musculoskeletal ultrasound assessor were blinded to the patients’ randomization. The study was approved by the institutional review board, and all patients signed an informed consent form. The trial was registered with clinicaltrials.gov (NCT02881307).

Fig. 1

Flow diagram of rheumatoid arthritis patients randomized in the diet intervention trial

DAS28: 28-joint DAS; PDUS: power Doppler ultrasound.

Diet intervention group

Patients randomized to the diet intervention group (n = 20) were placed on a hypocaloric diet of 1000–1500 kcal/day [19–21]. Meal plans included 3–4 meal replacements/day and one meal of 500 Calories. Prescribed calories were based on resting metabolic rate estimated from lean body mass by DXA while recommended dietary protein intake was calculated according to the lean body weight at the ratio of 2.2 g of protein/kg of lean body weight. Meal replacements utilized were protein-rich powder formulas containing 140 Cal, 20 g whey protein, 5 g sugar and 2.5 g fat (Nutrasumma, Phoenix, AZ, USA). Suggested meals included lean protein and >3 cups of non-starchy vegetables (not provided). Patients were given a choice to have a foods-only meal plan as well. Patients met with the dietitian every other week to evaluate adherence to the weight loss programme and for counselling for dietary intake (total of seven in-person visits over 12 weeks). Twenty-four-hour food recalls were performed at baseline, 6 weeks and 12 weeks to evaluate adherence with the intervention.

Control group

Patients randomized to the control group (n = 20) were counselled on the American Heart Association Step 1 Diet: (i) eat nutritious foods from all food groups, (ii) eat less of the nutrient-poor foods, and (iii) avoidance of smoking tobacco and second-hand smoke [22]. Control patients were provided with suggestions to avoid processed foods consumption, eating lean meats, fat-free dairy, decrease trans/saturated fats, decrease salt intake, limitations in alcohol and small portion sizes. Patients met with the dietician to review these recommendations at baseline, and 24 h food recall at baseline, 6 weeks and 12 weeks. The research coordinator contacted control patients by phone at weeks 2, 4, 8 and 10, with in-person visits at baseline, 6 weeks and 12 weeks (three in-person visits and four phone call visits).

Clinical and biomarker assessments

Baseline, 6 weeks and 12 weeks assessments included: weight (kg), BMI, DAS28 joint count using ESR (DAS28/ESR), clinical disease activity index (CDAI), tender joint count (TJC28), swollen joint count (SJC28), patient global assessment visual analogue scale (VAS), physician global assessment VAS, patient pain VAS, HAQ-Disability Index (HAQ-DI), Routine Assessment of Patient Index Data 3 (RAPID3), Functional Assessment of Chronic Illness Therapy (FACIT), multi-biomarker disease activity (MBDA) score, leptin, adiponectin, visfatin and ESR. ELISA kits were used to blindly assay the serum concentrations of leptin (R&D Systems, Minneapolis, MN, USA), adiponectin and visfatin (Ray Biotech, Peachtree Corners, GA, USA) according to the manufacturers’ instructions.

Other measures performed at baseline and 12 weeks only were body composition by DXA and lipid panel. The percentage body fat was calculated from the body composition DXA. In addition, calculation of percentage change in body weight was calculated for the achievement of ≥5% weight loss.

Whole blood obtained in serum-separator tubes was processed according to manufacture instructions. The de-identified serum samples were stored at −80°C. They were batched and shipped to Crescendo Bioscience, Inc. (South San Francisco, CA, USA) on dry ice for testing. MBDA scores were calculated by a validated algorithm based on serum concentrations of 12 protein biomarkers (vascular cell adhesion molecule 1, epidermal growth factor, VEGF-α, IL-6, TNF receptor-I, MMP-1, MMP-3, YKL-40, leptin, resistin, serum amyloid A and CRP), and adjusted for age, gender and BMI [23].

Musculoskeletal ultrasound assessment

PDUS and grey scale synovial hypertrophy (GSUS) assessments were performed at screening, baseline, 6 and 12 weeks using the previously used LAJAX (pronounced ‘Lay-Jax’) 34 joint protocol (US-34, 152 images total per patient visit) [24–26]: bilateral PIP joints 2–5 (dorsal long/short, volar long), IP joint (dorsal long/short, volar long), MCP joints 1–5 (dorsal long/short, volar long), radioulnar joint (dorsal long), midline wrist joint (dorsal long), knee (lateral and medial gutters) and MTP joints 2–5 (dorsal long). Images were scored semi-quantitatively on a scale of 0–3, with real-time scoring at the time of image capture [26]. Total US PDUS-34 and GSUS-34 scores were calculated by the summation of all 34 joint scores (range 0–102). Sonographic exams were performed by an ACR musculoskeletal ultrasound (RhMSUS) certified rheumatologist (V.K.R.) using a MyLab70C US machine (Biosound Esaote, Fishers, IN) with linear probe (12–18 MHz, LA-435), Doppler frequency of 8.3 MHz, pulse repetition frequency of 500 Hz and the gain set above the noise. The inter-reader reliability between sonographers was 0.77 (weighted kappa [G.S.K., V.K.R.]) as seen previously [27], and the intra-reader reliability was 0.80 (weighted kappa [V.K.R.]) based on re-scoring 10% of the scans.

Sample size

This proof-of-concept randomized clinical diet intervention trial had co-primary outcomes: change in DAS28/ESR and PDUS over 12 weeks. The pilot was designed to provide 80% power to detect an effect size of 0.8 in terms of differential change from baseline in each of the co-primary outcomes, assuming a two-sample Student’s t-test and an 0.05 significance level [28]. Di Minno et al. reported a 10% attrition rate [28]. We used a conservative attrition rate of 13%, so the calculated sample size was 60 patients. However due to the stringent inclusion criteria (PDUS-34 ≥ 10, in particular), study enrolment ended after screening of 89 patients and enrolment of 40 RA patients (Fig. 1). Forty-nine patients’ screen failed due to either not meeting PDUS-34 (n = 45) or DAS28/ESR (n = 2), or for other reasons (n = 4).

Statistical methods

Patients were randomized using a permuted block design with variable block sizes of 4 and 6. Patient characteristics were summarized for the full sample, as well as stratified by study arm. Age and RA disease duration were summarized using means and standard deviations, while sex, race, seropositivity and medication history were summarized using frequencies and percentages. Quantitative endpoints were summarized at each assessment point (baseline, 6 weeks and 12 weeks) by study arm using means and standard deviations. Line plots were used to visualize outcome trends in each group. Linear mixed models were used to evaluate effects of the diet intervention on the change from baseline of the various endpoints. The model specification included fixed effects for follow-up time point, study group and their interaction. Random patient effects accounted for multiple follow-up assessments per patient. Inferential testing of changes within each group over 12 weeks were evaluated using linear contrasts, while differential changes between groups were evaluated by testing the group-by-time interaction effects. Natural log-transformations were applied to adiponectin, visfatin, HDL and triglycerides prior to analysis due to substantial right-skewness. Fisher’s exact test was used to compare groups in terms of the proportion of patients experiencing at least 5% weight loss at 12 weeks. All analyses were based on intention-to-treat, and a significance level of 0.05 was used throughout. All available observations were included in the analyses and no data were imputed. Statistical analysis was performed using R v. 4.1.0 (http://www.r-project.org/).

Results

Forty RA patients were randomized to either the diet intervention or the control group (1:1) (Table 1). The average patient age in the study was 55.0 years (s.d. 13.2), 90% were female, and 75% were Caucasian. Eighty percent of patients were seropositive (ACPA or RF), with mean disease duration of 12.8 years. At baseline 17.5% patients were taking prednisone (average dose 2.6 mg/day), and 55% were taking biologic DMARDs or targeted synthetic DMARDs. Two patients in the diet intervention group and one patient in the control group, did not complete the trial due to loss to follow-up.

Table 1

Baseline characteristics

Characteristic	All participants (n = 40)	Diet intervention (n = 20)	Control (n = 20)
Age, mean (s.d.), years	55.0 (13.2)	57.7 (11.9)	52.2 (14.2)
Female, n (%)	36 (90.0)	19 (95.0)	17 (85.0)
Caucasian, n (%)	30 (75.0)	16 (80.0)	14 (70.0)
Seropositive, n (%)	32 (80.0)	16 (80.0)	16 (80.0)
RA disease duration, mean (s.d.), years	12.8 (11.8)	11.6 (10.3)	14.0 (13.3)
BMI, mean (s.d.), kg/m2	34.8 (5.7)	33.4 (5.7)	36.2 (5.4)
Baseline prednisone use, n (%)	7 (17.5)	2 (10.0)	5 (25.0)
Baseline prednisone dose, mean (s.d.), mg	2.6 (1.8)	3.8 (1.8)	2.1 (1.7)
csDMARD, n (%)	20 (50.0)	10 (50.0)	10 (50.0)
bDMARD/tsDMARD, n (%)	22 (55.0)	9 (45.0)	13 (65.0)
DAS28/ESR, mean (s.d.)	5.0 (1.1)	5.2 (1.2)	4.7 (1.0)
CDAI, mean (s.d.)	24.6 (10.1)	25.5 (9.7)	23.7 (10.6)
PDUS-34, mean (s.d.)	18.7 (11.3)	20.8 (10.8)	16.7 (11.8)
GSUS-34, mean (s.d.)	30.7 (11.8)	31.7 (12.4)	29.7 (11.3)
MBDA, mean (s.d.)	40.1 (12.6)	42.4 (14.5)	37.7 (10.2)
HAQ-DI, mean (s.d.)	0.90 (0.59)	0.8 (0.5)	1.0 (0.7)
RAPID3, mean (s.d.)	11.3 (4.7)	12.0 (5.0)	10.7 (4.5)
Patient global, mean (s.d.)	4.5 (2.0)	5.0 (2.1)	4.0 (1.7)
Patient pain, mean (s.d.)	4.6 (2.3)	5.3 (2.4)	3.8 (2.0)

bDMARD: biologic DMARD; CDAI: Clinical Disease Activity Index; csDMARD: conventional synthetic DMARD; DAS28/ESR: 28-joint DAS using ESR; GSUS: grey scale synovial hypertrophy; HAQ-DI: HAQ-Disability Index; MBDA: multi-biomarker disease activity; PDUS: power Doppler ultrasound; RAPID3: Routine Assessment of Patient Index Data 3; tsDMARD: targeted synthetic DMARD.

Weight, BMI and body composition by DXA

Patients in the diet intervention group lost a significant amount of weight, with mean weight loss of 9.5 kg (P < 0.001), and 72.2% of these patients lost ≥5% of their baseline body weight. The control group only lost 0.5 kg (P = 0.68) and 10.5% of these patients lost ≥5% of their baseline body weight. Within the diet intervention group, there was a significant reduction in BMI and percentage DXA body fat over 12 weeks (both P < 0.001). Between the two groups, there was a significant differential change in weight (kg), proportion of patients with ≥5% weight loss, change in BMI and percentage change in DXA body fat over 12 weeks of the study (all P < 0.001) (Table 2).

Table 2

Metrics of weight loss

	Diet intervention	Control	Between group P-value
Weight, mean (s.d.), kg			<0.001
Baseline	89.6 (15.7)	96.8 (16.4)
12 weeks	80.1 (5.5)	96.3 (16.7)
Change	−5.6 (3.5)	−0.3 (2.7)
Within group P-value	<0.001	0.684
BMI, mean (s.d.), kg/m2			<0.001
Baseline	33.4 (5.7)	36.2 (5.4)
12 weeks	29.9 (2.0)	36.3 (5.5)
Change	−2.0 (1.1)	−0.1 (1.1)
Within group P-value	<0.001	0.740
% body fat, mean (s.d.)			<0.001
Baseline	47.4 (4.8)	47.1 (5.9)
12 weeks	44.0 (4.7)	47.2 (5.6)
Change	−2.4 (1.9)	0.1 (1.7)
Within group P-value	<0.001	0.758
%Weight loss at 12 weeks, n (%)
>5%	13 (72.2)	2 (10.5)	<0.001
>10%	2 (11.1)	0	0.230

Clinical disease activity measures and musculoskeletal ultrasound

There was a within group reduction in mean DAS28 of 1.0 in the diet group (5.2 [1.2] to 4.2 [1.2], P < 0.001), 0.3 in the control group (4.7 [1.0] to 4.4 [1.2], P = 0.16) (Fig. 2). In addition, CDAI and TJC28 were significantly reduced in the diet intervention group (both P < 0.01), although SJC28 and ESR were not (Table 3).

Fig. 2

Changes in DAS28, imaging, function and patient-reported outcomes

*Statistical significance within the arm. Bracketed parenthesis indicates P-value for between group comparisons. DAS28/ESR: 28-joint DAS using ESR; HAQ-DI: HAQ-Disability Index; PDUS-34: power Doppler ultrasound 34 joint score; RAPID3: Routine Assessment of Patient Index Data.

Table 3

Disease activity measures and patient reported measures

	Diet intervention	Control	Between group P-value
ACR20 at 12 weeks, n (%)	5 (27.8%)	1 (5.3%)	0.090
CDAI, mean (SD)			0.292
Baseline	25.5 (9.7)	23.7 (10.6)
12 weeks	16.5 (7.5)	18.9 (10.0)
Change	−7.2 (3.9)	−4.4 (10.8)
Within group P-value	<0.001	0.025
FACIT, mean (s.d.)			0.759
Baseline	36.8 (9.0)	37.1 (10.7)
12 weeks	40.7 (9.0)	38.1 (10.2)
Change	4.0 (7.1)	1.0 (7.1)
Within group P-value	0.447	0.753
GSUS-34, mean (s.d.)			0.792
Baseline	31.7 (12.4)	29.7 (11.3)
12 weeks	24.1 (9.3)	23.1 (12.3)
Change	−6.5 (6.9)	−5.8 (9.9)
Within group P-value	0.004	0.013
28 TJC, mean (s.d.)
Baseline	10.1 (5.3)	7.9 (6.5)	0.099
12 weeks	5.3 (4.3)	6.1 (4.7)
Change	−4.2 (3.3)	−1.3 (6.6)
Within group P-value	0.002	0.130
28 SJC, mean (s.d.)			0.831
Baseline	6.2 (3.3)	6.7 (4.6)
12 weeks	4.7 (2.4)	5.6 (4.0)
Change	−0.9 (2.4)	−1.2 (4.9)
Within group P-value	0.329	0.194
ESR			0.317
Baseline	29.0 (26.5)	24.0 (18.0)
12 weeks	19.9 (14.9)	24.6 (19.0)
Change	−3.1 (11.5)	0.7 (9.8)
Within group P-value	0.261	0.779

CDAI: Clinical Disease Activity Index; FACIT: Functional Assessment of Chronic Illness Therapy; GSUS: Grey Scale Synovial Hypertrophy; SJC: swollen joint count; TJC: tender joint count.

No significant differences between groups were seen in the change of the clinical disease activity measures (DAS28, CDAI, TJC28/SJC28) and ESR over 12 weeks (Table 3 and Fig. 2). The DAS28 demonstrated a mean change difference of −0.51 (95% CI −1.01, 0.00, P = 0.056). Interestingly, ∼28% of patients in the diet intervention met ACR20 and only 5% in the control, but this was not significant between the groups (P = 0.09).

PDUS had a numerically greater reduction in the diet group (20.8 [10.8] to 13.1 [9.1], P < 0.001) than in the control group (16.7 [11.8] to 12.3 [12.1], P = 0.01) (Fig. 2), and this was also similarly seen for GSUS (Table 3). However, changes in PDUS and GSUS were not significantly different between groups (mean difference of the PDUS changes −2.0 [95% CI −7.00, 3.1, P = 0.46 and P = 0.79, respectively).

Patient-reported outcomes

Patient global, patient pain, HAQ-DI and RAPID3 significantly improved in the diet group (P < 0.045 for all comparisons). This improvement was not seen in the control group (Fig. 2). Patient global, patient pain and RAPID3 between group assessments were significant as well (P < 0.02). Improvements in patient pain, as a component of the RAPID3, had similar results. HAQ-DI was reduced from 0.8 to 0.6 in the diet group and was unchanged in the control group (1.0–1.0), but the between group analysis for HAQ-DI was not statistically significant (P = 0.065). For the FACIT, there were no within group nor between groups differences.

Biomarkers, adipokines and lipids

Serum leptin decreased over the 12 weeks in the intervention group (P = 0.009), but not in control patients (P = 0.612). The MBDA score, visfatin and adiponectin did not change significantly over time in either group (P > 0.05) (Table 4). Between groups, however, there were significant differences in serum levels of leptin and adiponectin after 12 weeks (P < 0.024 and P < 0.037, respectively) (Table 4). Cholesterol, low density lipoprotein, high density lipoprotein and triglycerides did not change significantly within groups or between groups (data not shown).

Table 4

Biomarkers and adipokine measures

	Diet intervention	Control	Between group P-value
MBDA score, median (Q1, Q3)			0.303
Baseline	37.5 (34, 52.5)	39 (36, 46)
12 weeks	38 (28, 46)	39 (27, 50)
Within group P-value	0.361	0.587
Adiponectina, median (Q1, Q3)			0.037
Baseline	10 (8, 17)	9.5 (4.7, 12)
12 weeks	11 (5, 18)	9 (6, 23)
Within group P-value	0.198	0.087
Leptin, median (Q1, Q3)			0.024
Baseline	39.2 (21.9, 60)	42.2 (26.8, 105.7)
12 weeks	24.6 (17.4, 32.1)	57.2 (23.7, 110.9)
Within group P-value	0.009	0.612
Visfatina, median (Q1, Q3)			0.473
Baseline	0.65 (0.6, 0.9)	0.65 (0.6, 0.9)
12 weeks	0.9 (0.6, 1.2)	0.95 (0.6, 1.2)
Within group P-value	0.452	0.797

a

Adiponectin and visfatin values were analysed on the log scale to account for strong right-skewness. MBDA: multi-biomarker disease activity.

Outcomes with reported clinically important differences

The percentage of patients achieving minimally clinically important differences (MCIDs) for RA outcomes was also examined. For RAPID3 and HAQ-DI, the 12-week MCID achievement was 54% and 44% in the diet intervention group, and 21% and 20% for the control group, respectively. At 12 weeks, the diet intervention group achieved good or moderate EULAR response for DAS28 in 56% of patients, while the control group met EULAR response in 26%. Fifty percent of the diet intervention group met CDAI MCID, while only 26% achieved it in the control group. Twenty-eight percent of patients in the intervention group achieved the MBDA score MCID, while 20% achieved it in the control group.

Safety

Twenty-five subjects (75.8%) reported adverse events (AEs) over the course of the trial. Infection was the most commonly reported AE overall, occurring in two patients (6.1%) in the intervention group and six patients (18.2%) in the control group. Infections included mild upper respiratory infections, urinary tract infections and one patient who developed shingles. Constipation was the second most frequently reported AE overall, occurring in six patients (18.2%) of the diet intervention group and one patient in the control. One subject in the intervention group developed transaminitis, which later resolved. There were no serious AEs during the clinical trial.

Discussion

The Centers for Disease Control and Prevention indicates that the prevalence of obesity and severe obesity in the USA is 42.4% (BMI ≥ 30) and 9.2% (BMI ≥ 40), respectively [29]. Obesity in itself increases the risk for development of RA [30, 31]. Additionally, obese RA patients pose challenges for the rheumatologist in terms of their therapeutic management, due to their reduced therapeutic response and low remission rates [32–34].

It is recommended that obese RA patients lose weight. Yet, there is a paucity of studies evaluating the effects of dietary interventions in obese RA patients, which led us to conduct this proof-of-concept pilot RCT. This 12-week diet intervention RCT was successful in achieving weight loss (the average patient lost 9.5 kg and 72.2% of patients lost ≥5% of their body weight in the intervention arm), with a high retention rate (93%), and overall good safety and tolerability. The patients randomized to the intervention arm had notably significant beneficial differences in RAPID3, patient global, patient pain, and leptin and adiponectin levels. Additionally, DAS28 and HAQ-DI improved significantly within the diet intervention group, with trends favouring the interventional group (P = 0.056–0.065) for the between group comparisons.

Few RCT diet intervention studies have reported DAS28 outcomes [35, 36], and no studies have examined effects on ultrasound and MBDA scores. A cross-over study of 50 RA patients with DAS28 > 2.6 evaluated the effects of an anti-inflammatory diet (n = 26) vs control (n = 24) [35, 36], and did not demonstrate a difference in DAS28 between groups (P = 0.116). However, the diet intervention group showed a significant improvement of DAS28 from baseline (P = 0.012) and no significant difference for the control group. Our study showed similar results. The modest sample size for these trials may have contributed to the lack of significance for the between arm comparisons.

Our pilot study demonstrated that RA patient-reported outcomes (PROs) improved in the diet intervention group compared with the control group for RAPID3, patient global assessment and patient pain. PROs measure different aspects of disease including function, disease activity, fatigue and pain, and these metrics are critically important for the individual RA patient. The majority of published randomized diet intervention studies predominantly examined the Mediterranean diet. These studies evaluated the overall diets’ impact on PROs in all RA patients, although not specifically within obese RA patients. Skoldstam et al. pooled data from three small RA clinical trials published between the 1970s and 2000s, where the diet interventions included a lacto-vegetarian diet, vegan diet and Mediterranean diet [37–40]. Changes in acute phase reactants, pain and physical function were significantly associated with the diet interventions. A 2018 systematic review examined the effects of the Mediterranean diet on RA outcomes, and concluded that unbiased studies are lacking to determine whether the diet impacts PROs [41].

Importantly, we found that adipokine levels were influenced by the dietary intervention in obese RA patients. Adipokines are released by the adipose tissue (whose mass is higher in obese individuals) and influence the pathogenesis of many chronic inflammatory disorders including RA and other autoimmune conditions [42]. Elevated leptin levels promote and sustain chronic inflammation [43], and meta-analyses have shown that serum leptin was significantly higher in RA patients than in healthy controls [44], positively correlating with RA parameters of disease activity (DAS28 and CRP) [45]. Regarding the role of adiponectin in RA, there is some controversy in the literature, possibly related to the fact that this adipokine can have both anti- and pro-inflammatory activities [46]. Future studies will need to address this point.

Obese RA patients suffer from comorbidities like fibromyalgia and osteoarthritis that may impact validated RA clinical outcome measures [47, 48]. For this reason, we also examined more objective measures of disease activity including ultrasound synovitis measures and biomarkers (PDUS-34, GSUS-34, ESR, MBDA scores). While PDUS and GSUS demonstrated statistically significant improvement in both the diet and control groups over 12 weeks, there were no between group differences. MBDA scores and ESR showed essentially no change from baseline in either group. Aside from leptin and adiponectin, improvements in this small trial were predominantly seen in the more subjective disease activity measures (between group RAPID3, pain VAS, patient global VAS; and diet intervention arm TJC, HAQ-DI), while the more objective measures (SJC, ESR, MBDA and PDUS/GSUS) were not significantly different between the two groups. The exact mechanism behind why RA PROs significantly improved between groups is uncertain. Did weight loss improve the underlying RA disease activity that resulted in PRO improvement, did weight loss directly improve PROs, or was there a combination of both? In addition, it is still unclear if a larger or longer trial is required to demonstrate improved objective/semi-objective RA outcomes. This partitioning of outcomes for the subjective vs objective measures was unexpected, but it should be interpreted with caution, given our limited sample size. Further study is warranted.

This study has several strengths. It was a RCT and was single-blind for both clinical and ultrasound assessments. This study included PROs, semi-objective measures (PDUS/GSUS) and objective measures (DXA, MBDA scores, ESR and adipokines).

This study also has limitations. We only recruited 40 of the 60 patients. This may have contributed to its trending but not statistically significant differences in DAS28 and HAQ-DI; thus, a larger longer trial would be desirable. This study was conducted at a single centre examining moderate to severe RA patients with obesity, established disease, predominantly women and evidence of synovitis on ultrasound. Diversity in race/ethnicity, disease duration and gender would have made this a more generalizable study. In addition, we were unable to blind the patients to the study arm allocation or the amount of weight loss, and the patient’s knowledge may have affected the results of the trial. Potential bias may have been introduced since the sonographer and the clinical joint assessor were not blinded to the appearance of the patient. Currently, there is no ‘cure’ for obesity and weight loss is difficult. However, the hypocaloric diet has been shown to be effective in reducing weight with a self-pay model [49]; this approach may not be available to those patients with financial constraints. Attempts were made to avoid the Hawthorne effect (the effect of paying more attention to the diet intervention patients) by using phone calls in lieu of in-person visits, and cost considerations prevented full compensation for this effect.

In conclusion, we report the results of a randomized, single-blind, controlled, pilot diet intervention study tailored to weight loss in obese RA patients. PROs improved significantly between the treatment groups for RAPID3, pain and patient global. Also, DAS28 and HAQ-DI trended towards between-group differences. These findings are important to the RA patient’s overall well-being. Lastly, changes in adiponectin and leptin were also significantly affected by the dietary intervention. Further study is required to validate these findings, but the results are encouraging that a dietary intervention may serve as a helpful adjunctive treatment for RA.

Acknowledgements

We would like to acknowledge Nicolette Morris and Himakar Nigam for their assistance.

Funding: This work was supported by the Rheumatology Research Foundation.

Disclosure statement: The authors have declared no conflicts of interest.

Data availability statement

The data underlying this article will be shared on reasonable request to the corresponding author.