FormalPara Key Summary Points

Why carry out this study?

Obesity has been shown to negatively affect the effectiveness of some rheumatoid arthritis (RA) treatments; in addition, some RA treatments are associated with weight gain.

This post hoc analysis of the FINCH 1–3 studies was conducted to assess the effect of filgotinib on the body mass index (BMI) of patients with RA and to evaluate the efficacy and safety of filgotinib according to baseline BMI.

What was learned from the study?

Filgotinib did not substantially change the BMI in patients with RA.

Baseline BMI did not appear to affect the efficacy of filgotinib.

Certain adverse events, including cardiovascular-related adverse events, occurred more frequently in high-BMI subgroups.

Introduction

Rheumatoid arthritis (RA) is a chronic autoimmune disease that can lead to substantial pain, disability, and increased mortality [1,2,3]. As in the general population, a significant proportion of patients with RA (approximately two-thirds) are overweight or are obese [4, 5]. A recent study showed that although trends in body mass index (BMI) over time are similar in patients with and without RA, BMI is significantly more variable in those with RA [6]. Furthermore, the relationship between obesity and RA is complex. Due to the pain and joint damage associated with the condition, patients with RA may have inactive lifestyles, increasing the likelihood of obesity [7]. Secretion of pro-inflammatory cytokines from adipose tissue contributes to bone and cartilage damage, and, therefore, excess body weight may exacerbate RA disease activity [5, 8]. Such pro-inflammatory cytokines may also result in changes in body composition, which can extend to reduced muscle mass (cachexia), although this does not necessarily result in weight loss [3, 9]. Patients with RA and obesity have worse clinical and patient-reported outcomes and are less likely to achieve remission, compared with patients with RA without obesity [10,11,12]. The pharmacokinetics and effective dose of any drug is highly dependent on the BMI, and obesity in patients with RA has been shown to negatively impact the effectiveness of some RA treatments, such as tumor necrosis factor (TNF) inhibitors [12, 13]. In addition, weight gain has been reported as an effect of some RA treatments, including TNF inhibitors [14,15,16]. Therefore, for physicians to accurately inform patients and prescribe appropriate therapy, it is important to determine the potential impact of RA treatment on body weight or BMI and to understand if and to what extent therapeutic effects of RA treatments will be affected by a patient’s weight/BMI. Such information has not yet been determined for filgotinib, a Janus kinase (JAK) 1-preferential inhibitor, for the treatment of moderate-to-severe RA.

The primary objective of this post hoc analysis was to assess the effect of filgotinib on the BMI in patients with RA. The secondary objective was to assess the efficacy and safety of filgotinib according to baseline BMI. Analyses were based on data from 3452 patients from the FINCH 1–3 studies [17,18,19]; the safety analysis also included data from the FINCH 4 and DARWIN 1–3 studies (total filgotinib exposure = 8085 patient-years) [20,21,22,23].

Methods

Study Design

The effect of filgotinib on the BMI and the impact of BMI on the efficacy of filgotinib was assessed using data from the FINCH 1–3 studies. FINCH 1, 2, and 3 were randomized, double-blind, active- or placebo-controlled phase 3 trials of filgotinib 100 mg and 200 mg in patients with active RA. FINCH 1 (NCT02889796; N = 1755) was a 52-week study in which patients with an inadequate response to methotrexate were randomized to receive filgotinib (100 or 200 mg), adalimumab or placebo, each with background methotrexate. The primary endpoint was the proportion of patients to achieve American College of Rheumatology (ACR)20 response at week 12 [17]. FINCH 2 (NCT02873936; N = 448) was a 24-week study in which patients with an inadequate response to biologic disease-modifying anti-rheumatic drugs (bDMARDs) were assigned filgotinib (100 mg or 200 mg) or placebo, in addition to one or two protocol-specified conventional synthetic DMARDs. The primary endpoint was the proportion of patients to achieve ACR20 response at week 12 [18]. In FINCH 3 (NCT02886728; N = 1249), a 52-week study, patients who were methotrexate naïve received filgotinib 200 mg, methotrexate, or filgotinib 100 mg or 200 mg with methotrexate. The primary endpoint was the proportion of patients to achieve ACR20 response at week 24 [19]. The studies did not specify an upper BMI limit within their inclusion criteria, although patients were to be excluded if they had a chronic, uncontrolled condition, such as morbid obesity, that would, in the opinion of the investigator, put them at increased risk during the study [17,18,19]. To assess the impact of BMI on the safety of filgotinib, data from the FINCH 4 study and the DARWIN 1–3 studies were also assessed, in addition to safety data from FINCH 1–3. DARWIN 1 (NCT01888874) and DARWIN 2 (NCT01894516) were phase 2b, 24-week studies in which patients received filgotinib 50 mg, 100 mg or 200 mg, or placebo, either as an add-on treatment to methotrexate (DARWIN 1) or as monotherapy (DARWIN 2) [21, 22]. The phase 2 DARWIN 3 study (NCT02065700) and phase 3 FINCH 4 (NCT03025308) study are ongoing, open-label long-term extensions for patients who completed treatment in the DARWIN or FINCH parent studies, respectively [20, 23].

Ethical Approval

All studies were conducted in accordance with the Declaration of Helsinki of 1964, and its later amendments, and approved by the institutional review board or ethics committee; all patients provided written informed consent to participate in the study.

Assessments

Baseline Characteristics

Using pooled data from the three FINCH studies, baseline characteristics were assessed according to baseline BMI subgroup: < 25 (underweight/normal weight), 25 to < 30 (overweight), or ≥ 30 kg/m2 (obese). BMI categories of < 18.5 or ≥ 40 kg/m2 (corresponding to underweight or severe obesity, respectively) were not included as separate subgroups as these each encompassed less than 5% of the total number of patients. BMI (kg/m2) was calculated as follows: weight (kg)/height (m)2.

Changes from Baseline in BMI

Mean changes from baseline in BMI were assessed over time by treatment group for FINCH 1, 2, and 3. For each study, the following were assessed according to treatment group and baseline BMI subgroup: mean change from baseline in BMI at week 12 and week 24, the proportion of patients whose BMI increased by incremental thresholds (1, 2, and 3 kg/m2, etc.) and cumulative probability of change from baseline in BMI at week 12 and week 24. The proportion of patients who moved from one BMI category into the ≥ 25 to < 30 kg/m2 category or the ≥ 30 kg/m2 category, and the proportion of patients whose body weight increased from baseline by > 7%, were also assessed according to treatment group.

Efficacy

The efficacy of filgotinib was assessed by treatment group and baseline BMI subgroup at week 12 for FINCH 1 and 2 and at week 24 for FINCH 3. Efficacy measures included the proportion of patients to achieve: ACR20/50/70 response; Disease Activity Score 28 using C-reactive protein (DAS28-CRP) low disease activity and remission (≤ 3.2 and < 2.6, respectively); and Clinical Disease Activity Index (CDAI) low disease activity and remission (≤ 10.0 and ≤ 2.8, respectively). Mean change from baseline in Health Assessment Questionnaire–Disability Index (HAQ-DI) was also assessed.

Predictive Analysis Regression

Exploratory analyses were performed to further study the association between baseline BMI and treatment response using multivariate linear regression, with the following included as covariates: age, sex, ethnicity, smoking status, anti-cyclic citrullinated peptide (anti-CCP) antibody status, steroid use, and filgotinib dose. The analysis was performed with study included as a fixed effect, then repeated with study included as a random effect.

Safety

The safety of filgotinib was assessed by calculating the exposure-adjusted incident rate (EAIR) of advents events (AEs) per 100 patient-years of exposure, according to the baseline BMI subgroup, using integrated data from seven RA clinical trials (FINCH 1–4 and DARWIN 1–3) [24].

Results

Baseline Characteristics

Mean (SD) baseline BMI was 26.7 (5.83) kg/m2 in FINCH 1, 30.2 (7.59) kg/m2 in FINCH 2 and 27.7 (6.41) kg/m2 in FINCH 3. Supplementary Figure S1a shows the distribution of BMI values at baseline using pooled data from FINCH 1–3. Distributions were broadly similar in each study (Supplementary Figure S1b–d). Baseline characteristics according to baseline BMI, based on pooled data from FINCH 1–3, are shown in Table 1 (median values are provided in Supplementary Table S1). The overall mean duration of RA was 6.4 years. There were 1332 patients in the BMI < 25 kg/m2 subgroup, 1102 patients in the 25 to < 30 kg/m2 subgroup and 1018 patients in the ≥ 30 kg/m2 subgroup. The proportion of patients with rheumatoid factor and with anti-CCP antibodies present at baseline decreased with increasing BMI. The proportion of patients with prior exposure to bDMARDs was greatest in the ≥ 30 kg/m2 subgroup (21.2% vs. 9.9% and 13.2% in the < 25 and 25 to < 30 kg/m2 subgroups, respectively). Disease severity indicators (including HAQ-DI, DAS28-CRP, and CDAI) showed a slight increase with increasing BMI, but values were generally similar across BMI subgroups. In the ≥ 30 kg/m2 subgroup, 64.0% of patients had a history of cardiovascular risk, compared with 30.1% and 47.9% of those in the < 25 and 25 to < 30 kg/m2 subgroups, respectively. Mean creatinine clearance increased with increasing BMI (97.2 ml/min, 111.6 ml/min and 138.6 ml/min in the < 25, 25 to < 30 and ≥ 30 kg/m2 subgroups, respectively) as did the proportion of patients with high-sensitivity C-reactive protein (CRP) level ≥ 6 mg/l (57.8, 59.5, and 67.0% in the < 25, 25 to < 30, and ≥ 30 kg/m2 subgroups, respectively).

Table 1 Baseline characteristics according to baseline BMI (pooled data from FINCH 1–3)

Change from Baseline in BMI

In general, mean changes from baseline in BMI were similar across treatment groups (Fig. 1a–c) and slightly greater with active treatment than with placebo. When assessed by baseline BMI subgroup at week 12 and 24, the greatest increase was 0.8 kg/m2, which was seen at week 24 in the filgotinib 200 mg treatment arm of FINCH 3, in both the < 25 and 25 to < 30 kg/m2 subgroups (Table 2). In all three FINCH studies, the majority of patients had increases in BMI of ≤ 1 or 2 kg/m2 regardless of treatment group or baseline BMI; very few patients had increases of ≥ 4 kg/m2 (Supplementary Tables S2–4). The proportion of patients who changed from one BMI category to another is shown in Supplementary Table S5. At week 24, the proportion of patients who moved into the obese state ranged from 1.7% (adalimumab) to 2.1% (placebo) in FINCH 1; 2.1% (placebo) to 5.1% (filgotinib 100 mg) in FINCH 2; and 1.6% (filgotinib 200 mg + methotrexate) to 5.9% (filgotinib 200 mg) in FINCH 3. The proportion of patients with a change from baseline in body weight of > 7% at week 24 was 6.2, 6.3, 4.5, and 2.9% in the filgotinib 200 mg, filgotinib 100 mg, adalimumab, and placebo arms, respectively (FINCH 1); 6.3, 9.4, and 5.3% in the filgotinib 200 mg, filgotinib 100 mg, and placebo arms, respectively (FINCH 2); and 9.9, 15.3, 15.5, and 8.6%, in the filgotinib 200 mg plus methotrexate, filgotinib 100 mg plus methotrexate, filgotinib 200 mg, and methotrexate arms, respectively (FINCH 3) (Supplementary Table S6). Cumulative probability plots of change from baseline in BMI according to baseline BMI in FINCH 1–3 are presented in Supplementary Figures S2–4.

Fig. 1
figure 1

Mean change from baseline in BMI over time in FINCH 1 (a) FINCH 2 (b), and FINCH 3 (c). Safety analysis set (includes enrolled patients who received ≥ 1 dose of study drug). Baseline value was the last available value collected on or prior to first dose of study drug. In FINCH 1 at week 24, all subjects assigned to PBO who did not discontinue study drug were re-randomised 1:1 to either FIL200 or FIL100 and continued in the study through week 52. For re-randomized PBO patients, baseline value after week 24 was the last available value collected on or prior to the first dose of FIL. Follow-up visit was defined as from (last dose date of any study drug + 8 days [FINCH 1 and 3] or + 6 days [FINCH 2]) to (last dose date of any study drug + 30 days). BMI (kg/m2) = [weight (kg)/height (cm)2] × 10,000. ADA adalimumab, BMI body mass index, FIL(100/200), filgotinib 100/200 mg, MTX methotrexate, PBO placebo, SD standard deviation

Table 2 Mean change (SD) from baseline in BMI (kg/m2) according to baseline BMI and treatment group

Efficacy

Filgotinib 200 mg with or without methotrexate was more efficacious than controls regardless of baseline BMI at each timepoint for most measures, including ACR20, ACR50, and ACR70 responses (Fig. 2a–c), DAS28-CRP low disease activity or remission (Fig. 3a,b), CDAI low disease activity or remission (Supplementary Figure S5a,b) and HAQ-DI (Supplementary Figure S6). Across BMI subgroups, filgotinib 200 mg was generally more efficacious than filgotinib 100 mg for most efficacy outcomes.

Fig. 2
figure 2

Proportion of patients to achieve ACR20 (a), ACR50 (b), and ACR70 (c) responses at week 12 (FINCH 1 and 2) or week 24 (FINCH 3), by baseline BMI. Full analysis set. Values above each column are the number of evaluable patients. ACR American College of Rheumatology, ADA adalimumab, BMI body mass index, FIL100/200 filgotinib 100/200 mg, MTX methotrexate, PBO placebo

Fig. 3
figure 3

Proportion of patients to achieve DAS28-CRP low disease activity (a) or remission (b) at week 12 (FINCH 1 and 2) or week 24 (FINCH 3) by baseline BMI. Full analysis set. Values above each column are the number of evaluable patients. ADA adalimumab, BMI body mass index, DAS28-CRP Disease Activity Score 28 using C-reactive protein, FIL100/200 filgotinib 100/200 mg, MTX methotrexate, PBO placebo

Predictive Regression

After accounting for patient age, sex, ethnicity, smoking status, anti-CCP status, steroid use, and filgotinib dose, a linear regression model with study included as a fixed effect indicated that a higher BMI was associated with a poorer DAS28-CRP response. However, the absolute difference in response was small, with a 10 kg/m2 difference in BMI corresponding to a 0.2-unit difference in DAS28-CRP change (Supplementary Table S7). Similar results were observed when study was included as a random effect (Supplementary Table S8).

Safety

Table 3 summarizes EAIRs of AEs of special interest across baseline BMI subgroups, using integrated safety data from FINCH 1–4 and DARWIN 1–3. In patients with BMI ≥ 30 kg/m2, the EAIR of serious treatment-emergent adverse events (TEAEs) was higher than in those with a BMI of 25 to < 30 or < 25 kg/m2 in both the filgotinib 200 mg and filgotinib 100 mg groups. The EAIR of venous thrombotic and embolic (VTE) events was numerically higher in the ≥ 30 kg/m2 subgroup than in the other subgroups within the filgotinib 200 mg arm, although the absolute number of events was low. Within the filgotinib 200 mg arm, the EAIR (95% CI) of major adverse cardiovascular events (MACE) was 0.5 (0.2, 1.0) in the ≥ 30 kg/m2 subgroup versus 0.3 (0.1, 0.7) and 0.3 (0.2, 0.7) in the 25 to < 30 and < 25 kg/m2 subgroups, respectively. In the filgotinib 200 mg treatment arm, the rate of serious infections was numerically highest in the ≥ 30 kg/m2 subgroup, while in the filgotinib 100 mg treatment arm, it was numerically highest in the < 25 kg/m2 subgroup. The EAIRs of deaths, herpes zoster, and malignancies excluding non-melanoma skin cancer were similar across BMI subgroups for each filgotinib dose. When comparing filgotinib doses, the EAIR of deaths was numerically higher with filgotinib 200 mg vs. 100 mg in the 25 to < 30 and ≥ 30 kg/m2 subgroups but not in the < 25 kg/m2 subgroup. Long-term data are not available for the active comparator or placebo groups.

Table 3 EAIR of adverse events per 100 PYE by baseline BMI (integrated data from FINCH 1–4 and DARWIN 1–3)

Discussion

During the FINCH 1–3 studies, BMI did not substantially change from baseline with filgotinib treatment. Filgotinib 200 mg was generally more efficacious than controls, regardless of baseline BMI. Filgotinib 100 mg was also generally more efficacious than controls across BMI subgroups but to a lesser extent and less consistently than the higher dose. Across BMI subgroups, filgotinib 200 mg was generally more effective than filgotinib 100 mg for most efficacy outcomes. Similar to our findings, a pooled analysis of data from six phase 3 trials showed that compared with placebo, tofacitinib treatment was associated with greater improvements in efficacy measures for each baseline BMI subgroup assessed (< 25, 25 to < 30, and ≥ 30 kg/m2) [25]. Previous studies have demonstrated that obesity has a negative impact on response to TNF inhibitors, particularly infliximab, even when dosed according to body weight [26,27,28]. In contrast, therapeutic response to the interleukin (IL)-6 receptor inhibitor tocilizumab (administered intravenously according to body weight) was found to be unaffected by baseline BMI [29]. A recent study found that the difference in effect of obesity on therapeutic responses to TNF inhibitors and tocilizumab was likely due to mechanisms involved in the production of pro-inflammatory cytokines; the authors concluded that therapies other than TNF inhibitors would be more suitable for patients with obesity [30]. These findings may be relevant to JAK inhibitors, as they also inhibit IL-6 signaling [31] and therefore may be less impacted by BMI than TNF inhibitors. In our study, while the results of the predictive regression analysis suggested that higher BMI may be associated with poorer DAS28-CRP responses to filgotinib, the absolute differences in response were unlikely to be clinically meaningful. In addition, this association may be partially influenced by the higher CRP levels seen in the ≥ 30 kg/m2 subgroup at baseline. Of note, we observed that the proportion of patients who had prior exposure to bDMARDs increased with increasing BMI, which may indicate that prior bDMARDs have been less effective in patients with obesity.

Pooled safety data from FINCH 1–4 and DARWIN 1–3 showed that the rate of TEAEs, including serious TEAEs and VTEs, was numerically higher in the ≥ 30 kg/m2 subgroup than in the 25 to < 30 or < 25 kg/m2 subgroups. These findings are as expected given that high BMI is a known risk for the TEAEs reported [32,33,34,35]. Cardiovascular risk associated with RA treatment is an important consideration, especially in patients with obesity who are already at increased cardiovascular risk. In our analysis, as expected, increased levels of CRP, which are associated with increased cardiovascular risk, were observed in patients with the highest BMI at baseline. In addition, the proportion of patients with cardiovascular risk factors, which included a medical history of ischemic heart disease, was greater in the ≥ 30 kg/m2 subgroup than in the 25 to < 30 or < 25 kg/m2 subgroups, and the former subgroup also had a numerically higher rate of VTEs. This observation is in line with the results of a post hoc analysis of the Oral Rheumatoid Arthritis Trial (ORAL) Surveillance study, which found tofacitinib-treated patients with a history of atherosclerotic cardiovascular disease to be at increased risk of MACE [36]. In our analysis, the rate of serious infections was highest in the ≥ 30 and < 25 kg/m2 subgroups in the filgotinib 200 mg and 100 mg treatment arms, respectively; mean baseline fasting glucose levels also generally increased with increasing BMI, which may be linked to increased risk of infection [37]. Our assessment of baseline characteristics showed that creatinine clearance increased with increasing BMI. This finding is consistent with previously published data which also indicated a positive correlation between these two parameters [38]. Our analysis showed that the proportion of RF-positive patients decreased with increasing BMI, in agreement with an association that has been described previously in patients with RA [39].

Weight gain is reported as an uncommon side effect (may affect up to one in 100 people) of baricitinib and tofacitinib, and as a common side effect (may affect up to one in ten people) of upadacitinib [40,41,42]. Compared with patients taking placebo, a numerically greater proportion of those treated with baricitinib experienced abnormal weight gain, defined as an increase of 7% or more from the highest value recorded at baseline, regardless of baseline body weight [43]. The proportion of patients with abnormal weight gain decreased with increasing baseline body weight or BMI; the difference with placebo was only statistically significant for patients with the lowest body weight at baseline (those who had a body weight < 60 kg and a BMI < 25 kg/m2) [40]. Across the FINCH studies, there was no apparent dose-dependent effect of filgotinib on body weight, as measured by an increase of > 7%, and the proportion of patients whose body weight increased by > 7% from baseline was generally similar between active comparators.

There is no standard definition for abnormal or clinically meaningful changes in BMI. Weight loss of ≥ 5% is often used to indicate a clinically meaningful reduction for patients with obesity [44]. As noted above, an increase of 7% or more from the highest recorded baseline body weight was used to define an abnormal increase in patients in baricitinib trials [43]. Whether or not a change in BMI is clinically relevant is dependent on the patient’s baseline BMI. In patients who are underweight at baseline, weight gain may be a return to normal weight, so it would not be considered an AE. However, if a patient of normal weight at baseline became overweight or obese, weight gain would be considered an AE. Patients who are overweight or obese may lose weight if their RA improves and they become more active.

Limitations of the analysis include its post hoc nature and the possible effect of external weight loss interventions, such as changes to diet or physical activity, which could not be controlled or measured. In addition, weight gain is a reported side effect of corticosteroids [45], which were taken concurrently by 44.2% of patients at baseline. Information regarding corticosteroid use throughout the studies, such as the proportion of patients who discontinued corticosteroids or tapered the dose of corticosteroids, is lacking, so it is not possible to ascertain their potential impact on the findings. Furthermore, while the mean corticosteroid dose at baseline has been reported, we are unable to report the dose level for each corticosteroid administered. Given that specific corticosteroids are dosed differently to reach equivalent levels the extent to which the mean dose can be interpreted is limited. Ethnicity data were also not available for the pooled populations by BMI subgroup, so it is not possible to determine possible impact on weight. Patients with a BMI ≥ 40 kg/m2 were not included as a separate subgroup given the low number of patients comprising this category. Therefore, outcomes in patients with severe obesity, who may present in clinical practice, were not assessed. There are also limitations associated with the use of BMI as a parameter. In RA, changes in body composition, specifically a loss of lean body mass and an increase in adiposity, have been reported, which may be associated with increased cardiovascular risk [7, 46, 47]. Such changes may be predominantly due to the systemic inflammation associated with the condition. However, alterations may also be induced by corticosteroids or bDMARDs [46]. In the current analysis, BMI ≥ 30 kg/m2 was used as a surrogate for obesity; there is an absence of data regarding changes in fat mass/body composition, which can only be determined by dual-energy X-ray absorptiometry or bioelectrical impedance analysis [48]. In addition, cardio-metabolic risk may be better reflected by other measures, such as waist circumference or waist-to-hip ratio [7, 25]. However, BMI is a suitable parameter for physicians to use in clinical practice.

Conclusions

Patient weight/BMI is an important consideration when treating RA, as obesity has been shown to negatively affect clinical outcomes and response to treatment [10,11,12]. Findings from this post hoc analysis of the FINCH 1–3 studies indicate that filgotinib did not result in substantial changes from baseline in BMI, and that it was an effective treatment for RA across all BMI subgroups. While the predictive regression analysis indicated poorer DAS28-CRP response with higher BMI, the difference was small and not considered clinically meaningful. As expected, the incidence of serious TEAEs and VTE events was higher in the ≥ 30 kg/m2 baseline BMI subgroups than in the 25 to < 30 or < 25 kg/m2 baseline BMI subgroups.