IL-6 Inhibitors and TNF Inhibitors

Visual Abstract

Highlights

• •The complex duality of IL-6 presents a potential clinically relevant challenge for patients with RA, as treatment with IL-6i may alleviate RA symptoms but may also suppress beneficial cardiometabolic adaptations to exercise.
• •TNFi-treated patients with RA showed a significant exercise-induced cardiac adaptations after 12 weeks, whereas no difference was found in the IL-6i–treated groups.
• •The interaction effect between treatment modalities on beneficial cardiac changes caused by exercise was insignificant.

Summary

Interleukin-6 inhibitors (IL-6i) are commonly used in patients with rheumatoid arthritis to reduce inflammation from chronically increased IL-6. IL-6 levels increase transiently following exercise, exerting numerous positive effects. This study examined if beneficial exercise-induced cardiac adaptations were attenuated in patients with rheumatoid arthritis in concomitant IL-6i treatment compared with tumor necrosis factor inhibitors. Compared with control, we found that the tumor necrosis factor inhibitor–treated group, but not the IL-6i group, had a significant increase in left ventricular mass following 12 weeks of supervised exercise. However, the interaction effect of treatment modalities on exercise-induced cardiac adaptations was insignificant. (Exercise-induced Cardiac Adaptions in Rheumatoid Arthritis Patients During IL-6 vs TNF Antibody Therapy; NCT05215509)

Patients with rheumatoid arthritis (RA) have an increased cardiovascular disease risk compared with the general population,^1,2 that is, independent of traditional cardiovascular disease risk factors, but rather involves inflammatory disease activity, mediated primarily through a plethora of cytokines, including tumor necrosis factor (TNF) and interleukin (IL)-6.^3,4 The predominantly used biological disease-modifying antirheumatic drugs (bDMARDs) for patients with RA are tumor necrosis factor inhibitors (TNFi) and interleukin-6 inhibitors (IL-6i), which are considered equal in efficacy^5,6 and cardiovascular safety.⁷

Although TNF and IL-6 are both proinflammatory cytokines, IL-6, in the context of exercise, is anti-inflammatory and has metabolic effects.⁸ During exercise, skeletal muscle is the source of circulating IL-6 and plasma concentrations thus increase independently of TNF activity.^9,10 Although IL-6, during a bout of exercise, promotes energy for the working muscle, it reduces visceral adipose tissue mass and increases left ventricular mass (LVM) in response to long-term exercise training.11, 12, 13, 14 Given that exercise training has cardiometabolic protective effects¹⁴ and is considered a cornerstone treatment for patients with RA,¹⁵ it is important to clarify whether cardiac adaptations are affected by bDMARD treatment.

The primary objective was to evaluate whether concomitant use of 1 of the 2 different bDMARDs impact the effect of 12 weeks of regular supervised exercise, relative to treatment as usual, on changes in LVM (g) measured by cardiac magnetic resonance imaging (cMRI) from baseline to week 12, in patients with RA in stable treatment with either a TNF or IL-6 inhibitor. The key secondary objectives explored whether, in patients with RA in stable treatment with either a TNF or IL-6 inhibitor, the concomitant use of 1 of the 2 different bDMARDs impact the effect of 12 weeks of regular supervised exercise, relative to treatment as usual, after 12 weeks on changes in: 1) left ventricular stroke volume (LVSV); 2) left ventricular end-diastolic volume (LVEDV); 3) relative peak oxygen uptake (V̇O_2peak); and 4) Disease Activity Score-28 for RA based on the erythrocyte sedimentation rate (DAS-28-ESR).

Methods

Settings and recruitment

The trial design and methods have been published.¹⁶ The study was designed as a 12-week, parallel-group, open-label, randomized single-center clinical exercise trial with blinded assessors, conducted at the Centre for Physical Activity Research (CFAS), Rigshospitalet, University of Copenhagen, Copenhagen, Denmark, from January 2022 to August 2023. We enrolled adult patients with RA treated with either TNFi or IL-6i from outpatient clinics in the Zealand and the Capital Region of Denmark. The trial was pre-registered on ClinicalTrials.gov (NCT05215509) and approved by the Scientific Ethical Committee of the Capital Region of Denmark (approval number H-21010559 with amendments 86424, 87463, 88044, and 100650) and the Danish Medicines Agency (Eudra-CT 2021-005287-21). The study was monitored by the Danish Good Clinical Practice unit and adhered to the principles outlined in the Helsinki Declaration. The data are reported in accordance with the CONSORT (Consolidated Standards of Reporting Trials) guideline.¹⁷ A prespecified statistical analysis plan (SAP version 1.1) was made available at the trial registry on Clinicaltrials.gov before any statistical analyses. All participants provided written informed consent before enrollment.

To be considered eligible for trial inclusion, patients had to be between 18 and 69 years, diagnosed with RA, and treated with either a TNFi or IL-6i for a minimum of 4 months before enrollment. The RA disease had to be well-controlled with a maximum DAS28-ESR score of 3.2 and an oral corticosteroid use of no more than 10 mg/day within 7 days of enrollment, and no intramuscular corticosteroid use within 3 weeks of enrollment. A full list of inclusion and exclusion criteria along with the recruitment process has been published.¹⁶

Randomization

Patients were randomized to 1 of 2 groups (exercise vs control), stratified by sex, and their concomitant use of bDMARDs (TNFi or IL-6i). A researcher otherwise not involved in the study performed a computerized randomization that used blocks of 10 and randomly assigned patients 1:1 from each stratum to an exercise or control group. An unblinded study investigator notified the patient. Because of the nature of the study, blinding of the patients was not possible. Outcome assessors (Simon Jønck, Malte Lund Adamsen, Morten Asp Vonsild Lund) were blinded to group allocation until the database was finalized, and all analyses outlined in the SAP were completed.

Interventions

Standard care consisted of routine visits to RA outpatient clinics independent of this study. At weeks 4 and 8, all patients were contacted by telephone by an unblinded study investigator to document adverse events and compliance with bDMARD treatment. We did not discriminate between different drugs within TNFi or IL6i regimens. All distinct bDMARDs treatment before enrollment was documented. At baseline, all patients were advised to contact an unblinded study investigator in case of adverse events. Patients were instructed to maintain their current level of physical activity (except exercise as part of the intervention) and dietary habits during the study.

Exercise was performed solely on an ergometer bicycle and each session lasted 45 minutes. All sessions were supervised by an educated trainer, and heart rate (HR) was monitored at all times (Polar RS400, Polar). Each session included a 10-minute warm-up at 50% of maximal heart rate (HR_max) followed by a high-intensity interval training (HIIT) regimen that consisted of 4 bouts of 4 minutes at a target of ≥85% HR_max with 3 minutes recovery at 40% to 60% HR_max between each bout. An 8-minute cool-down at 50% HR_max concluded the session. The exercise groups completed 3 weekly sessions for 12 weeks.

Procedures

Two test days, 1 at CFAS and 1 at the Department of Cardiology, Rigshospitalet, were scheduled at baseline and repeated at 12-week follow-up. Before the visit at CFAS, patients were advised to avoid any exercise for 48 hours and were furthermore instructed to arrive fasted for at least 10 hours. The Department of Cardiology, Rigshospitalet, conducted a cMRI and no specific instructions were given before that visit.

Clinical assessment, questionnaires, and blood samples

Physicians Simon Jønck and Malte Lund Adamsen conducted the clinical assessments at baseline and follow-up at CFAS. At baseline, all current medication, previous use of bDMARD, comorbidities, and duration of RA were documented. At both visits, physicians assessed blood pressure, height, and weight as well as tender and swollen joint count along with a patient global health score (0–10 cm) to account for DAS-28-ESR. DAS-28-ESR was preferred from the more frequently used DAS-28-CRP due to IL-6’s influence on C-reactive protein secretion from the liver. A higher DAS-28-ESR score indicates increased RA severity.¹⁸

Blood samples were analyzed using standard procedures at the Department of Clinical Biochemistry, Rigshospitalet, and plasma-IL-6 was analyzed in-house (V-PLEX Custom Human Biomarkers, Meso Scale Discovery Inc.).

Body composition and cardiorespiratory fitness

Body composition was determined by dual x-ray absorptiometry (Lunar Prodigy GE Healthcare; enCORE software version 14, 10, 022). An unblinded researcher supervised the cardiorespiratory exercise test to obtain peak oxygen uptake (V̇O_2peak [mL/min/kg]). The cardiorespiratory exercise test was conducted as a standardized graded exercise test on a bicycle ergometer (Monark LC4, Monark Exercise AB); following a 5-minute warm-up at a fixed load (50 W), workload increased by 20 W every minute until exhaustion. Exhaustion was defined as either voluntary termination or a cadence below 60 rounds/minute. The quality of the test was deemed appropriate if the respiratory exchange ratio exceeded 1.1 and oxygen uptake reached a plateau despite increased workload. Minute ventilation, as well as expired CO₂ and O₂ fractions were measured using indirect calorimetric methods (Quark b2, Cosmed). HR_max was determined as the maximal heart rate achieved during the final minute of the test.

cMRI

The cMRI used a 1.5-T whole-body scanner to obtain steady-state free precession cine images. Images were acquired during repeated breath-holding in 3 long axes and a stack of short axes, covering the entire heart. This process aimed to identify any wall motion abnormalities and enable quantification of cardiac chamber size and mass. For post-processing, semi-automatic evaluation was carried out using commercially available software (cvi42, Circle Cardiovascular Imaging). In certain cases, manual correction of the automatic contour detection for the left ventricular endo- and epicardium were performed in accordance with established guidelines.¹⁹ Based on these procedures, various parameters, including LVM, as well as left ventricular ejection fraction, LVSV, LVEDV, left ventricular end-systolic volume (LVESV), and global longitudinal strain measures were derived. All cMRI analyses were conducted in a blinded manner by experienced personnel at the Department of Cardiology, Rigshospitalet. Body surface area was calculated according to the Mosteller formula.²⁰

Dietary and physical activity measurements

At mid-point of the study (week 6), all patients were informed to fill out a self-reported 3-day dietary (weekends excluded) journal. Dietary data were subsequently processed through an online kcal calculator. Upon completion of the 12-week follow-up visit at CFAS, axial accelerometer-based physical activity monitors (Axivity AX3) were placed on the right hip and right thigh for 5 consecutive days.

Outcomes

The primary outcome measure was LVM measured by cMRI, with the primary endpoint being the difference in change of LVM (g) from baseline to 12-week follow-up in the 2 exercise groups in treatment with either a TNFi or IL6i, relative to standard care in patients with RA.

The key secondary endpoints included differences in change from baseline to 12-week follow-up in LVSV (mL), LVEDV (mL), relative V̇O_2peak (mL/min/kg), and DAS-28-ESR in the exercise groups (TNFi or IL-6i treatment) relative to standard care.

Other (exploratory) outcomes (termed “other secondary outcomes” in the SAP) included LVM/body surface area (g/m²), left ventricular global longitudinal strain (%), left ventricular ejection fraction (%), LVESV (mL), absolute V̇O_2peak (mL/min), total lean mass (g), total fat mass (g), and resting HR (beats/min), which were all analyzed in the same manner as the primary and secondary outcome.

Sample size

The sample size was determined based on a similar study in healthy individuals with obesity.¹³ To detect a difference in mean change of LVM of 12.0 ± 12.3 g assuming a 2-sided unpaired t-test with a 5% alpha level, homoscedasticity, and a power of 80% (1-β), a sample size of 16 patients in each group was necessary. To accommodate potential dropout, we planned to enroll 20 patients in each exercise/bDMARD group (N = 80).

Statistical analysis

We used the intention-to-treat (ITT) patient population to perform our main analyses. Continuous outcomes were analyzed by analysis of covariance, with change from baseline to week 12 as the dependent outcome. The model included the baseline level as a covariate to reduce the random variation, with fixed effect factors for group (exercise vs control), sex (male or female), bDMARD (IL-6i or TNFi), and the interaction between group and bDMARD (ie, Group×bDMARD [an interaction term with 4 levels]). For the main analyses, missing data were conservatively imputed by the baseline carried forward method, effectively corresponding to a non-responder imputation. Data are presented as estimated least square means (LSM), calculated using the emmeans package in R (version 4.3.0)²¹ to derive both estimated marginal means and contrasts between groups with 95% CIs. The model specification underwent visual assessment through normal probability plots and plots of residuals vs fitted values. In case of non-normal distribution, we planned a log transformation and if not suitable, a median with lower and upper quartile are reported. All results from statistical analyses on the primary and key secondary endpoints were accompanied by 2-sided 95% CIs and corresponding P values (superiority defined as P < 0.05). We did not adjust for multiplicity, but instead secondary outcomes were assessed and interpretated through a predefined precise order (gatekeeping). Consequently, 95% CIs not adjusted for multiplicity should not be used in place of hypothesis testing. The predefined per-protocol (PP) assessment included participants who attended at least 80% of exercise sessions (29 of 36), maintained a bDMARD compliance of no <80%, and completed the follow-up. In addition, we performed a predefined sensitivity analysis that included previous bDMARD usage (the count of bDMARDs used before enrollment) and length of RA diagnoses (years). Baseline characteristics are presented in mean ± SD, median with 25th and 75th percentiles (Q1-Q3), or count (percentage). All analyses were performed in R (version 4.3.0).²²

Results

Patient eligibility and baseline characteristics

From January 2022 to April 2023, 299 patients with RA were assessed for eligibility. Of these, 148 (49%) declined participation and 40 (13%) were deemed ineligible. Of the 111 individuals undergoing a second assessment, 35 (32%) were not eligible. Subsequently, 76 patients were invited to baseline visits before randomization. In the IL-6i group (n = 33), 4 individuals (12%) did not complete baseline: 1 was erroneously invited to baseline visits without fulfilling inclusion criteria, 2 declined participation following written consent, and 1 was not able to complete the cMRI due to anxiety. In the TNFi group (n = 43), 3 individuals (7%) did not complete baseline: 1 declined participation following written consent and 2 were unable to complete the cMRI due to anxiety. In total, 69 participants were included in the study. Eight patients (5 in IL-6i + exercise and 1 and 2 in IL-6i + control and TNFi + control) were lost to follow-up. Adherence to the prespecified exercise training criteria of a minimum 80% attendance was not met by 4 and 2 patients in the TNFi and IL-6i groups, respectively. Two participants, 1 from IL-6i + control and 1 from TNFi + exercise did not meet the prespecified bDMARD adherence criterion of minimum 80% (Figure 1). In accordance with the SAP, the recruitment process was terminated early as the inclusion rate for the IL-6i group fell below 2 patients per month. The mean age was 53.5 ± 9.5 years, 56 (81%) were female, and the mean RA duration was 15.9 ± 10.1 years. In general, baseline characteristics appeared comparable among groups; however, 20 (69%) IL-6i patients had received treatment with 2 or more bDMARDs before their current therapy, compared with 2 (5%) TNFi patients (Table 1). In addition, the use of concomitant oral steroid treatment was exclusive to the IL-6i treatment patients (n = 9 [33.3%]); however, only 4 were in daily continuous oral treatment in a mean dosage of 5 ± 2 mg, which were kept stable during the trial, and 5 had oral steroid treatment prescribed as needed (Supplemental Table 1). The median daily calorie intake was 1,856 (Q1-Q3: 1,557-2,028) and 1,563 (Q1-Q3: 1,408-1,928) for the TNFi and IL-6i + exercise, respectively, and 2,022 (Q1-Q3: 1,634-2,236) and 1,691 (Q1-Q3: 1,504-2,006) for the TNFi and IL-6i + control with no differences between groups or strata. Furthermore, there was no difference in sedentary and moderate to vigorous daily physical activity among the 4 groups (Supplemental Table 2).

Figure 1.

Flowchart of Study

BL = baseline; cMRI = cardiac magnetic resonance imaging; DAS28-ESR = Disease Activity Score-28 for rheumatoid arthritis with erythrocyte sedimentation rate; FU = follow-up; ITT = intention-to-treat; PP = per-protocol; RA = rheumatoid arthritis.

Table 1.

Baseline Characteristics

	Exercise (n = 37)	Control (n = 32)	Overall (n = 69)
Demographics
Female/male
Total	30/7	26/6	56/13
TNFi	16/4	16/4	32/8
IL-6i	14/3	10/2	24/5
Age, y
Total	52.5 ± 9.3	54.6 ± 9.8	53.5 ± 9.5
TNFi	51.3 ± 10.9	54.7 ± 9.5	53.5 ± 10.3
IL-6i	53.9 ± 7.1	52.8 ± 10.4	53.4 ± 8.5
BMI, kg/m2
Total	26.9 ± 5.9	26.5 ± 5.4	26.7 ± 5.6
TNFi	26.5 ± 5.2	25.9 ± 5.2	26.2 ± 5.1
IL-6i	27.4 ± 6.7	27.6 ± 5.7	27.5 ± 6.2
RA characteristics
Duration of RA, y
Total	14.8 ± 8.6	17.2 ± 11.5	15.9 ± 10.1
TNFi	15.0 ± 9.1	18.6 ± 12.9	17.3 ± 11.3
IL-6i	14.6 ± 8.7	13.3 ± 7.7	14.0 ± 8.2
DAS-28-ESR
Total	1.8 ± 0.8	2.0 ± 0.8	1.9 ± 0.9
TNFi	2.1 ± 0.7	2.0 ± 0.9	2.1 ± 0.8
IL-6i	1.5 ± 1.1	2.0 ± 0.9	1.7 ± 1.0
Sero-positive RA
Total	27 (73.0)	20 (62.5)	47 (68.1)
TNFi	14 (70.0)	15 (75.0)	29 (72.5)
IL-6i	13 (76.5)	5 (41.7)	18 (62.1)
SR, mm
Total	5 (2.0, 9.0)	4.5 (2.0, 10.5)	5 (2.0, 10.0)
TNFi	7 (5.75, 13.0)	7.5 (4.5, 14.5)	7 (5.0, 13.3)
IL-6i	2 (2.0, 4.0)	2 (1.75, 4.0)	2 (2.0, 4.0)
IL-6, pg/mL
Total	2.7 (0.6, 13.4)	1.6 (0.7, 11.9)	2.1 (0.6, 12.6)
TNFi	0.7 (0.74, 1.1)	0.8 (0.5, 1.1)	0.8 (0.5, 1.1)
IL-6i	13.8 (9.6, 23.0)	12.7 (11.3, 14.9)	13.4 (9.7, 17.5)
Cardiac measures
LVM, g
Total	103.0 (23.0)	102.5 (26.4)	102.8 (24.5)
TNFi	102.9 (26.8)	108.3 (28.6)	105.6 (27.5)
IL-6i	103.1 (18.4)	92.9 (19.8)	98.9 (19.3)
LVSV, mL
Total	85.4 (15.3)	88.7 (19.0)	86.9 (17.0)
TNFi	85.5 (17.4)	93.2 (20.0)	89.3 (18.9)
IL-6i	85.4 (12.8)	81.1 (15.0)	83.6 (13.7)
LVEDV, mL
Total	130.4 (23.8)	134.6 (33.5)	132.4 (28.6)
TNFi	131.9 (26.8)	142.9 (35.3)	137.4 (31.3)
IL-6i	128.7 (20.3)	120.8 (26.1)	125.4 (22.7)
LVM/BSA, g/m2
Total	53.7 (9.2)	54.0 (9.1)	53.9 (9.1)
TNFi	53.0 (11.2)	56.5 (10.0)	54.7 (10.6)
IL-6i	54.6 (6.2)	50.0 (5.6)	52.6 (6.3)
LVEF, %
Total	65.7 (4.7)	66.4 (6.1)	66.0 (5.4)
TNFi	64.8 (4.9)	65.6 (5.6)	65.2 (5.2)
IL-6i	66.6 (4.5)	67.8 (6.8)	67.1 (5.5)
LVESV, mL
Total	45.2 (11.3)	46.0 (17.9)	45.6 (14.6)
TNFi	46.8 (12.4)	49.7 (18.6)	48.3 (15.7)
IL-6i	43.3 (9.9)	39.7 (15.5)	41.8 (12.4)
Resting HR, beats/min
Total	65 (11)	70 (12)	68 (12)
TNFi	65 (9)	67 (12)	66 (10)
IL-6i	66 (12)	75 (12)	69 (13)
SBP, mm Hg
Total	127 (15)	131 (11)	129 (13)
TNFi	128 (17)	131 (11)	129 (14)
IL-6i	126 (12)	130 (13)	128 (13)
DBP, mm Hg
Total	79 (8)	80 (7)	79 (7)
TNFi	78 (9)	80 (7)	79 (8)
IL-6i	80 (7)	77 (6)	79 (7)
Metabolic measures
Total lean mass, kg
Total	45.0 (7.9)	44.7 (9.3)	44.8 (8.5)
TNFi	45.2 (7.3)	46.0 (9.4)	45.6 (8.3)
IL-6i	44.6 (8.7)	42.6 (9.2)	43.8 (8.8)
Total fat mass, kg
Total	29.7 (12.2)	27.8 (12.1)	28.8 (12.1)
TNFi	29.2 (10.8)	27.0 (11.8)	28.1 (11.1)
IL-6i	30.1 (14.1)	29.0 (13.2)	29.7 (13.5)
Fitness measures
Absolute V̇O2peak, mL/min
Total	1,908 (471)	1,985 (602)	1,943 (532)
TNFi	1,949 (430)	2,047 (668)	1,998 (556)
IL-6i	1,859 (524)	1,882 (481)	1,869 (498)
Relative V̇O2peak, mL/min/kg
Total	25.0 (6.4)	26.7 (7.5)	25.8 (6.9)
TNFi	25.5 (5.6)	27.2 (7.5)	26.3 (6.6)
IL-6i	24.4 (7.3)	25.9 (7.7)	25.0 (7.4)
Cardiometabolic comorbidities
Hypertension
Total	6 (16.2)	5 (15.6)	11 (15.9)
TNFi	4 (20.0)	3 (15.0)	7 (17.5)
IL-6i	2 (11.8)	2 (16.7)	4 (13.8)
Diabetes mellitus type 2
Total	0 (0)	2 (6.3)	2 (2.9)
TNFi	0 (0)	1 (5.0)	1 (2.5)
IL-6i	0 (0)	1 (8.3)	1 (3.4)
Cardiovascular medications
Beta-blocker
Total	0 (0)	2 (6.3)	2 (2.9)
TNFi	0 (0)	1 (5.0)	1 (2.5)
IL-6i	0 (0)	1 (8.3)	1 (3.4)
Other antihypertensive
Total	3 (8.1)	5 (15.6)	8 (11.6)
TNFi	3 (15.0)	2 (10.0)	5 (12.5)
IL-6i	0 (0)	3 (25.0)	3 (10.3)
Statins
Total	5 (13.5)	2 (6.2)	7 (10.1)
TNFi	3 (15.0)	1 (5.0)	4 (10.0)
IL-6i	2 (11.8)	1 (8.3)	3 (10.3)

Values are mean ± SD, median (Q1, Q3), or n (%).

BMI = body mass index; BSA = body surface area; DAS-28-ESR = Disease Activity Score-28 for rheumatoid arthritis with erythrocyte sedimentation rate; DBP = diastolic blood pressure; HR = heart rate; IL = interleukin; IL-6i = interleukin-6 inhibitor; LVEDV; left ventricular end-diastolic volume; LVEF = left ventricular ejection fraction; LVESV = left ventricular end-systolic volume; LVM = left ventricular mass; LVSV; left ventricular stroke volume; RA = rheumatoid arthritis; SBP = systolic blood pressure; SR = sedimentation rate; TNFi = tumor necrosis factor inhibitor; V̇O_2peak = peak oxygen uptake.

Adherence did not differ regarding exercise amount and intensity. For TNFi vs IL-6i, patients completed a median of 32 (Q1-Q3: 30-34) vs 33.5 (Q1-Q3: 30-35.5) of the planned 36 sessions and spent a median of 13.3 (Q1-Q3: 11.6-15.6) vs 13.9 (Q1-Q3: 10.8-15.7) minutes at an HR ≥85% of HR_max. Adherence to bDMARD treatment was high (>95%) and did not differ between groups or strata (Supplemental Table 3).

Primary outcome

As illustrated in Figure 2A, compared with patients randomized to the control group, there was a statistically significant improvement in LVM in patients prescribed TNFi (LSM 3.8 g, 95% CI: 0.2-7.4 g) but not in IL-6i-treated patients (LSM 1.3 g, 95% CI: −3.0 to 5.6 g) following exercise. However, comparing these strata, we found no evidence to support a difference in the exercise benefit between bDMARD strata (LSM −2.5 g, 95% CI: −8.2 to 3.2 g; P_interaction = 0.39) (Table 2).

Figure 2.

Structural and Functional Cardiac Outcomes

(A) Change in LVM from baseline to follow-up. Individual data points are presented for each group in gray circles. The estimated least square means change with 95% CIs represented in bold circles. The interaction effect of concomitant IL-6i treatment on exercise-induced changes to LVM was found insignificant. (B) Change in LVSV from baseline to follow-up. Individual data points are presented for each group in gray circles. The estimated least square means change with 95% CIs represented in bold circles. (C) Change in LVEDV from baseline to follow-up. Individual data points are presented for each group in gray circles. The estimated least square means change with 95% CIs represented in bold circles. Ex = exercise; IL-6i = interleukin-6 inhibitor; LVEDV = left ventricular end-diastolic volume; LVM = left ventricular mass; LVSV = left ventricular stroke volume; TNFi = tumor necrosis factor inhibitor.

Table 2.

Outcomes After 12 Weeks in the ITT Population

	Exercise (n = 37)	Control (n = 32)	Difference in Exercise vs Control Groups (95% CI)	Interactiona (95% CI)
ΔLVM, g
Total	3.4 (1.2-5.7)	0.7 (−1.7 to 3.1)	2.7 (0.1-5.5)	—
TNFi	3.9 (1.1-6.8)	0.1 (−2.9 to 3.1)	3.8 (0.2-7.4)	−2.5 (−8.2 to 3.2) (P = 0.39)
IL-6i	2.7 (−0.4 to 5.8)	1.4 (−2.0 to 4.9)	1.3 (−3.0 to 5.6)
Key secondary outcomes
ΔLVSV, mL
Total	2.8 (−0.7 to 6.2)	−1.3 (−5.0 to 2.5)	4.1 (−0.3 to 8.4)	—
TNFi	5.6 (1.4-9.9)	−2.4 (−6.9 to 2.1)	8.0 (2.4-13.7)	−9.0 (−17.8 to −0.3)
IL-6i	−0.8 (−5.4 to 3.8)	0.2 (−5.1 to 5.6)	−1.0 (−7.6 to 5.6)
ΔLVEDV, mL
Total	3.2 (−1.5 to 7.9)	−1.3 (−6.2 to 3.7)	4.5 (−1.1 to 10.1)	—
TNFi	7.1 (2.0-12.2)	−0.5 (−5.8 to 4.9)	7.6 (1.0-14.2)	−8.9 (−19.3 to 1.4)
IL-6i	−0.1 (−5.5 to 5.4)	1.3 (−5.0 to 7.6)	−1.4 (−9.2 to 6.4)
ΔRelative V̇O2peak, mL/min/kg
Total	1.6 (0.4-2.8)	0.2 (−1.1 to 1.5)	1.4 (−0.1 to 2.9)	−
TNFi	1.3 (−0.1 to 2.8)	1.2 (−0.2 to 2.8)	0.1 (−1.9 to 2.0)	3.3 (0.3-6.3)
IL-6i	1.9 (0.3-3.5)	−1.5 (−3.4 to 0.4)	3.4 (1.1-5.7)
ΔDAS28-ESR
Total	−0.2 (−0.5 to 0.1)	0.01 (−0.3 to 0.3)	−0.2 (−0.6 to 0.1)	—
TNFi	−0.2 (−0.6 to 0.2)	0.1 (−0.3 to 0.5)	−0.3 (−0.8 to 0.2)	0.1 (−0.7 to 0.9)
IL-6i	−0.3 (−0.7 to 0.2)	−0.1 (−0.6 to 0.4)	−0.2 (−0.8 to 0.4)

Values are least square means (95% CI).

ITT = intention-to-treat; other abbreviations as in Table 1.

^aInteraction of IL-6i and exercise across strata.

Secondary outcomes

As illustrated in Figure 2B and detailed in Table 2, compared with control, we found a significant improvement in LVSV in patients in TNFi treatment (LSM 8.0 mL, 95% CI: 2.4-13.7 mL) following exercise, whereas no difference was seen for IL-6i-treated patients (LSM −1.0 mL, 95% CI: −7.6 to 5.6 mL). Similarly, as shown in Figure 2C and Table 2, following exercise, LVEDV improved significantly in patients in TNFi treatment (LSM 7.6 mL, 95% CI: 1.0-14.2 mL), but not in IL-6i treatment (LSM −1.4 mL, 95% CI: −9.2 to 6.4 mL) compared with their respective control. Comparison of strata revealed an LSM difference in exercise benefits between bDMARDs of −9.0 mL (95% CI: −17.8 to −0.3 mL) for LVSV and −8.9 mL (95% CI: −19.3 to 1.4 mL) for LVEDV (Table 2).

Compared with control, exercise did not improve V̇O_2peak in patients in TNFi treatment (LSM 0.1 mL/min/kg, 95% CI: −1.9 to 2.0 mL/min/kg). In contrast, IL-6i patients improved V̇O_2peak by LSM 3.4 mL/min/kg (95% CI: 1.1-5.7 mL/min/kg) following exercise compared with control, resulting in an LSM 3.3 mL/min/kg (95% CI: 0.3-6.3 mL/min/kg) difference in V̇O_2peak change across strata in favor of IL-6i (Table 2). For the patient-reported outcome DAS-28-ESR, no changes across bDMARDs and interventions were found (Table 2). The PP and sensitivity analysis (Table 3, Supplemental Table 4) aligned with the ITT analysis.

Table 3.

Outcomes After 12 Weeks in the PP Population

	Exercise (n = 25)	Control (n = 27)	Difference in Exercise vs Control Groups (95% CI)	Interactiona (95% CI)
ΔLVM, g
Total	4.5 (1.7-7.3)	0.4 (−2.4 to 3.2)	4.1 (0.7-7.4)	—
TNFi	4.7 (1.2-8.2)	−0.3 (−3.8 to 3.1)	5.0 (0.7-9.4)	−2.6 (−9.6 to 4.5)
IL-6i	3.0 (−0.3 to 6.2)	1.5 (−2.6 to 5.6)	1.5 (−3.1 to 8.1)
Key secondary outcomes
ΔLVSV, mL
Total	3.6 (−0.7 to 8.0)	−1.1 (−5.4 to 3.2)	4.7 (−0.7 to 10.1)
TNFi	5.8 (0.5-11.1)	−2.6 (−7.8 to 2.6)	8.4 (1.6-15.2)
IL-6i	0.1 (−6.4 to 6.7)	1.5 (−4.8 to 7.8)	−1.4 (−10.1 to 7.3)	−9.8 (−20.7 to 1.2)
ΔLVEDV, mL
Total	5.0 (−0.9 to 10.9)	−1.5 (−7.3 to 4.4)	6.4 (−0.7 to 13.5)
TNFi	7.9 (1.6-14.2)	−0.9 (−7.1 to 5.3)	8.8 (0.7-16.8)	−9.6 (−22.6 to 3.4)
IL-6i	1.4 (−6.3 to 9.2)	2.2 (−5.2 to 9.7)	−0.8 (−11.1 to 9.5)
Δ/Relative V̇O2peak, mL/min/kg
Total	2.1 (0.7-3.4)	−0.1 (−1.4 to 1.3)	2.2 (0.5-3.9)	—
TNFi	1.4 (−0.2 to 3.0)	0.7 (−0.9 to 2.2)	0.8 (−1.3 to 2.8)	3.3 (0.3-7.3)
IL-6i	3.1 (1.1-5.1)	−1.4 (−3.5 to 0.6)	4.5 (1.7-7.4)
ΔDAS-28-ESR
Total	−0.3 (−0.6 to 0.1)	−0.05 (−0.4 to 0.3)	−0.2 (−0.6 to 0.2)	—
TNFi	−0.3 (−0.7 to 0.2)	0.1 (−0.3 to 0.5)	−0.3 (−0.8 to 0.2)
IL-6i	−0.2 (−0.7 to 0.2)	−0.1 (−0.6 to 0.4)	−0.2 (−0.8 to 0.4)	0.07 (−0.7 to 0.8)

Values are least square means (95% CI).

P = per protocol; other abbreviations as in Table 1.

^aInteraction of IL-6i and exercise across strata.

Exploratory outcomes

None of the remaining cMRI measures nor other exploratory outcomes showed any difference with exercise compared with the control group in either of the bDMARD groups. The PP and sensitivity analysis aligned with the ITT (Supplemental Tables 4 to 6).

Safety

No serious adverse events were noted.²³ Within the TNFi groups, 12 participants in the exercise group and 7 in the control group reported adverse events. In the IL-6i groups, 10 participants in the exercise group and 8 in the control group reported adverse events. Except for musculoskeletal pain in the lower extremities in the exercise groups, the characteristics and frequency of adverse events appeared comparable across groups (Supplemental Table 7).

Discussion

In this assessor-blinded, randomized trial, with RA patients in stable TNFi or IL-6i treatment, we investigated the effects of different bDMARD treatments on exercise-induced structural cardiac adaptations. Following a 12-week exercise intervention, we observed an increase in LVM in patients receiving TNFi but not IL-6i compared with the control groups; however, an interaction effect of IL-6i and exercise on LVM was not significant.

Structural cardiac adaptations

To our knowledge, no previous studies have examined the effect of bDMARD treatment on exercise-induced cardiac adaptations in patients with RA. As exercise is a cornerstone treatment for patients with RA, insights into the influence of bDMARDs on exercise training outcomes may be important to enable successful, personalized treatment strategies. The TNFi + exercise group demonstrated an ∼4 g greater increase in LVM compared with the TNFi + control group. In contrast, no significant increase in LVM was observed in the IL-6i + exercise group compared with the IL-6i + control group. Even though the estimated difference of 2.5 g between these 2 outcomes could not be shown to be significant (P = 0.39), the higher estimate in the TNF group may reflect that signaling pathways stimulating physiological LVM expansion are independent of TNFi but may be affected by IL-6i. The physiological increase in LVM to exercise training involves the activation of intracellular protein rapamycin (mTOR) through the exercise-induced stimulation of insulin growth factor-1.²⁴ Studies in skeletal muscles in mice have shown that IL-6 activity was essential to obtain compensatory muscle hyperthropy,²⁵ and others point to an IL-6-driven increase in mTOR signaling²⁶; however, the exact link between IL-6 and mTOR in cardiomyocytes is poorly described. Despite the well-established role of muscle-derived IL-6 as an important myokine for interorgan crosstalk and adaptations to exercise, its role in a cardiac context has previously only been scarcely explored and available results are conflicting. Hence, a higher activation of proteins associated with cardiac physiological hypertrophy has previously been reported in exercising IL-6 knockout vs wild-type mice,²⁷ whereas another study reported an equivalent increase in LVM relative to body weight in IL-6 knockout and wild-type rodents following 6 weeks of exercise.²⁸ In contrast, an exercise study in mice demonstrated an increased myocardial expression of IL-6 receptors and significantly smaller infarct size following ischemic injury, which was not evident in the IL-6 knockout group, indicating a more favorable classic IL-6 signaling and cardioprotective role of exercise-induced IL-6.²⁹ Also, a recent study in healthy abdominally obese individuals found the inhibition of IL-6 activity using the IL-6 receptor inhibitor, tocilizumab, abolished the increase in LVM induced by exercise training,¹³ which is further supported by the current results.

The change in LVM from exercise and concomitant TNFi treatment was relatively small compared with other cMRI-based exercise studies. In healthy populations, post-Covid-19 individuals participating in an identical intervention as this study, increased LVM by 9.1 g,³⁰ obese individuals in a similar 12-week exercise intervention had an increase of 12.0 g,³¹ and young men had a 9-g increase in LVM following 10 weeks of military training consisting of a mixture of strength and endurance exercise.³² Despite difficulty in comparing changes to LVM across studies due to the heterogeneity of participants and interventions, it is intriguing to contemplate whether RA in itself, TNFi treatment, or a combination of both attenuate structural cardiac adaptations to exercise, especially considering that the relative intensity of the exercise intervention in the current study was greater than studies mentioned previously.^30,31 RA induces unfavorable cardiac remodeling by increased LV wall thickness³³ as well as structural cardiac abnormalities and pathological left ventricular hypertrophy, as reviewed by Corrao et al.^34,35 TNFi may also be implicated in cardiac remodeling, as a significant decrease in indexed LVM was observed in patients with RA following a 6-month treatment.³⁶ Taken together, it is clear that RA affects cardiac structure and remodeling; however, whether RA with or without TNFi modifies cardiac adaptations with exercise remains speculative.

Secondary cardiac outcomes

As indicated previously, the lack of an interaction effect of IL-6i and exercise might be explained by a sample size that was too small for the given effect of exercise on changes to LVM in the TNFi group. This is further underlined by the changes seen in the secondary outcomes LVSV and LVEDV, where exercise induced a greater increase in the TNFi group compared with the IL-6i group, relative to control. The 8.0-mL mean increase in LVSV in the TNFi + exercise compared with control is consistent with a cMRI-based study in healthy men undergoing 12 weeks of light and 12 weeks of high-intensity endurance training.³⁷ We found no change in LVSV in IL-6i + exercise. Overall, LVSV relies on the contractile force of the myocardium and the filling of the left ventricle. Both factors can be influenced by exercise through compensatory mechanisms.²⁴ The lack of change in LVSV in IL-6i + exercise is supported by a study that found IL-6 to regulate the increase of human cardiomyocyte contractility through paracrine effects.³⁸ However, previous exercise trials have suggested that up to 50% of the change in LVSV can be explained by a change in LVEDV.³⁹ Therefore, the increase in LVSV in TNFi + exercise might be attributed to the corresponding increase in LVEDV, which may reflect an exercise-induced increase in blood volume and venous return.⁴⁰

Cardiorespiratory fitness

Somewhat surprisingly, exercise allocation was associated with an increase in V̇O_2peak specifically in the IL-6i group, despite the apparent absence of structural cardiac changes. Indeed, this improvement in aerobic fitness of ∼3 mL/min/kg aligns with that of previous exercise trials on patients with RA.⁴¹ Meanwhile, no change in V̇O_2peak occurred in TNFi + exercise despite the documented structural cardiac changes. Given that cardiac adaptations are classically considered the main contributors to changes in aerobic fitness on exercise training, particularly HIIT, we cannot readily explain these seemingly paradoxical findings based on the data at hand. However, the V̇O_2peak increase in the IL-6i group in response to HIIT is substantial and may be clinically relevant, particularly in the context of RA and its potential implications for ongoing trials in other patient groups even though it is unlikely to be caused by changes in LVM. As such, this may suggest that IL-6i could affect other steps in the oxygen transport chain, which were not investigated in this study, but are critical and thus warrant further investigation.

Patient-reported outcome

In terms of symptom severity, no change to DAS-28-ESR was observed. There were, on the other hand, no flares in disease activity, highlighting low-impact HIIT as a safe activity for patients with RA. The lack of improvement may be explained by the enrollees having low RA disease activity at inclusion (DAS-28-ERS ≤3.2). These results are in line with a systematic review that found no significant effects of exercise on DAS-28 as reviewed by Ye et al,⁴² as well as previous findings from a non-randomized study on RA patients with moderate disease activity undergoing 3 months of exercise training with no changes to DAS-28.⁴³ Thus, 3 months of exercise, regardless of whether it is continuous moderate intensity or HIIT, appears to be insufficient for eliciting an effect on symptom severity captured by the DAS-28-ESR.

Study limitations

First, the sample size estimation may have been too small for an interaction analysis because it was based on an assumed exercise-induced 12 ± 12.3 g difference in change between the TNFi and IL-6i groups, whereas we only observed an ∼4 g difference, increasing the risk of a type II error. Second, as patients were stratified to bDMARD treatment as part of routine care before randomization, potential inherent differences between patients in TNFi and IL-6i treatment groups introduce a risk of allocation bias and residual confounding. However, by including patients in stable treatment with low disease activity score, we ensured that patients could adhere to both the bDMARD treatment and exercise intervention. To account for any possible inherent differences, a sensitivity analysis was conducted including duration of RA (years) and number of bDMARDs previously used, with results that aligned with the ITT analyses. Third, because of the nature of the study, blinding of patients was not possible, potentially creating a bias in the patient-reported outcome; however, this is unlikely to be different across bDMARD groups. Nonetheless, we did not find any difference in the patient-reported outcome between the exercise and control groups. Fourth, we were not able to recruit the desired number of IL-6i patients, causing a potentially larger variation in our data. Fifth, the randomization of the IL-6i population was skewed. This was likely caused by a fixed block size (10) that was too large considering the overall size of the study and the stratification of biological sex before randomization. Last, we did not measure blood volume. Consequently, we cannot evaluate whether the effect of LVSV and LVEDV are induced by differences in blood volume changes.

Conclusions

Patients with RA in TNFi treatment demonstrated increased cMRI-based LVM indicative of cardiac remodeling after a 12-week HIIT intervention, whereas no difference was evident with IL-6i treatment. However, concomitant IL-6i compared with TNFi treatment, did not significantly affect the exercise-induced changes to LVM from baseline to week 12.

Perspectives.

COMPETENCY IN MEDICAL KNOWLEDGE: Many studies have reported on the detrimental cardiovascular effects of chronic increased levels of IL-6. Less attention has been given to the beneficial cardiometabolic effects of the transient increase in IL-6 following exercise. The dual role of IL-6 poses a clinically relevant challenge for patients with RA, as IL-6 inhibitors may relieve RA symptoms but could also hinder beneficial cardiometabolic adaptations from exercise. In the present randomized controlled trial, we compared the effect of high-intensity exercise training on changes to LVM, a surrogate marker for beneficial physiological cardiac adaptations, between patients with RA in IL-6i or TNFi treatment. Our findings raise clinically interesting questions regarding physiological adaptations to exercise in patients with RA in concomitant biological treatment, a patient group known for an increased sedentary lifestyle, with increased risk of cardiovascular disease. Greater understanding of the mechanism behind exercise-induced adaptation in these patients may guide future therapies.

TRANSLATIONAL OUTLOOK: These findings call for further studies on the regulatory effect of TNFi and IL-6i on cardiac and cardiorespiratory fitness adaptations in patients with RA. Aside from central adaptations, several different peripheral adaptations may be affected, contributing to a differentiated response to exercise.

Funding Support and Author Disclosures

The Centre for Physical Activity Research is supported by TrygFonden (grants ID 101390, ID 20045, and ID 125132). Dr Jønck was supported by The Danish Rheumatism Association (grants ID R224-A8298-B1963 and ID R212-A7782-B1963), The Rigshospitalet Research Grant, Gangstedfonden (grant ID A40059), King Christian the 10th Foundation, and Snedkermester Sophus Jacobsen and wife Astrid Jacobsens Foundation. Dr Adamsen was supported by the Novo Nordisk Foundation (grant ID: NNF20OC0065929), The Danish Rheumatism Association (grant ID: R202-A7503), and a Holbæk Hospitals research grant. Dr Christensen was supported by a grant from the Danish Heart Foundation (16-R107-A6704-22970) and Danish Cardiovascular Academy (CPD5Y-2021001-DCA), which is funded by the Novo Nordisk Foundation and The Danish Heart Foundation. Section for Biostatistics and Evidence-Based Research, the Parker Institute, Bispebjerg and Frederiksberg Hospital are supported by a core grant from the Oak Foundation (OCAY-18-774-OFIL). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Dr Jønck is currently employed at Novo Nordisk. Dr Adamsen has received a speaking fee from Novartis. Dr Dreyer has received research grants (paid to institution) from BMS and Abbvie outside the present work; and travel expenses from Janssen, UCB, and Boehringer Ingelheim. Dr Køber has received speaker honoraria from AstraZeneca, Boehringer, Novartis, and Novo Nordisk. Dr Pedersen has received speaker honorarium from Novo Nordisk. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.