Filgotinib in Moderate to Severe Rheumatoid Arthritis in a Real-World Setting: Up to 6-Month Effectiveness, Patient-Reported Outcomes, and Persistence, and up to 24-Month Safety from More than 1400 Patients in the FILOSOPHY and PARROTFISH Studies

Abstract

Introduction

FILOSOPHY (NCT04871919) and PARROTFISH (NCT05323591) are ongoing, prospective observational European phase 4 studies of filgotinib in patients with rheumatoid arthritis (RA) in a real-world setting. We report the study design, baseline characteristics, and interim results for disease activity measures, patient-reported outcomes (PROs), and treatment persistence up to 6 months and safety up to 24 months.

Methods

Eligible patients had moderate to severe RA and were prescribed filgotinib for the first time in daily practice. In this interim analysis, measures include the proportion of patients achieving low disease activity (LDA), according to Disease Activity Score for 28 joint count using C-reactive protein (DAS28-CRP) or Clinical Disease Activity Index (CDAI), and the proportion achieving a clinically meaningful change from baseline in visual analog scale (VAS) pain (≥ 10 mm reduction) and in Functional Assessment of Chronic Illness Therapy (FACIT)-Fatigue score (≥ 4.0 increase). Treatment persistence was estimated using the Kaplan–Meier method. Treatment-emergent adverse events (TEAEs) are reported.

Results

From May 2021 to April 2025, 1431 patients initiated filgotinib treatment; 93.0% had completed at least 6 months of follow-up (median follow-up 706 days). Improvements in DAS28-CRP and CDAI occurred from month 1. By month 6, DAS28-CRP ≤ 2.6 was reported for 58.1% (494/850) of patients and CDAI remission for 25.7% (221/859). The proportion with LDA was 71.4% (607/850) based on DAS28-CRP and 69.8% (600/859) based on CDAI. The proportion with a clinically meaningful change from baseline in VAS pain and FACIT-Fatigue score was 44.1% (178/404) and 43.8% (176/402), respectively, at week 1, increasing to 69.6% (227/326) and 62.5% (202/323), respectively, at month 6. Treatment persistence (95% CI) at month 6 was 84.0% (82.0, 85.8). Up to month 24, the exposure-adjusted incidence rates per 100 patient-years of exposure (95% CI) of any TEAEs and serious TEAEs were 89.8 (84.1, 95.8) and 9.4 (8.0, 10.8), respectively.

Conclusion

Interim data from FILOSOPHY/PARROTFISH showed improvement of disease activity from month 1 and PROs from week 1, high treatment persistence at 6 months, and no new safety signals. These findings continue to support the safety and effectiveness of filgotinib in daily clinical practice.

Trial Registration

ClinicalTrials.gov: NCT04871919 (2021-04-29), NCT05323591(2022-04-05).

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1007/s40744-026-00874-x.

Key Summary Points

Why carry out this study?

Filgotinib is a Janus kinase 1 preferential inhibitor approved for the treatment of moderate to severe active rheumatoid arthritis (RA), having demonstrated positive results in clinical trials.

FILOSOPHY and PARROTFISH are ongoing, prospective, observational phase 4 studies of filgotinib in patients with RA in Europe, designed to provide valuable real-world data beyond those captured in previous clinical trials.

What was learned from the study?

Interim data from the studies indicate that measures of disease activity (DAS28-CRP and CDAI) and physical function (HAQ-DI) improved as early as month 1, and pain and fatigue as early as week 1, with improvements maintained to month 6.

Treatment persistence at 6 months was high, and no new safety signals were observed.

Overall, these findings support the safety and effectiveness of filgotinib for the treatment of patients with moderate to severe active RA in daily clinical practice, outside of a clinical trial setting.

Introduction

A key goal of rheumatoid arthritis (RA) treatment is to achieve sustained remission or low disease activity (LDA) due to their association with reduced joint damage progression, and improved health-related quality of life, functioning, and patient-reported outcomes (PROs) [1–3]. According to current European Alliance of Associations for Rheumatology (EULAR) recommendations, a treat-to-target strategy should be used, which involves adjusting RA therapies until this goal is achieved [2]. Many patients, however, report being dissatisfied with pain control despite achieving remission or LDA with treatment [4] and continue to experience high levels of fatigue [5].

Current treatment options for patients with RA who have an insufficient response to conventional synthetic (cs) disease-modifying antirheumatic drugs (DMARDs) include biologic (b) DMARDs and targeted synthetic (ts) DMARDS, comprising Janus kinase (JAK) inhibitors [2, 6]. Filgotinib is a JAK1 preferential inhibitor, which was assessed in the phase 2b DARWIN 1 and 2 clinical trials, and the phase 3 FINCH 1–3 clinical trials [7–11]. However, clinical trials are conducted in highly selected patient populations (and under strictly controlled conditions) that may not be reflective of patients seen in routine clinical practice. Therefore, observational, post-marketing studies are increasingly being used to assess the effectiveness and safety of the newer DMARDs in a real-world setting, and to provide answers to questions that are not addressed in randomized clinical trials [12–15]. Evidence from such real-world studies, performed in a more heterogeneous patient population, is valuable for healthcare providers, payors, and organizations developing treatment recommendations [16]. In addition to evaluating disease activity and long-term safety, real-world data are particularly valuable in assessing treatment persistence, which is reflective of both effectiveness and safety, and PROs, such as pain and fatigue, which may not be fully assessable within the timeframe of a clinical trial [4, 17, 18]. Furthermore, use of treatments in specific subgroups of patients with RA (e.g., based on prior or current RA treatments) can be evaluated in larger cohorts.

To complement clinical trial data, the phase 4 FILOSOPHY (NCT04871919) and PARROTFISH (NCT05323591) non-interventional studies were initiated in patients with moderate to severe RA receiving filgotinib in a real-world setting. In this manuscript, we report the study design of FILOSOPHY and PARROTFISH and baseline patient characteristics, along with interim, descriptive data on disease activity, PROs, and persistence (up to 6 months after starting treatment) and safety (up to 24 months) from all 1431 patients who initiated filgotinib treatment.

Methods

Study Design

FILOSOPHY and PARROTFISH are ongoing identically designed, prospective, non-interventional, multicenter cohort studies conducted in patients who are treated with filgotinib for the first time, and in accordance with the local standard of care and product label. As predefined in the study protocol, data from both studies will be pooled. Patients are followed for up to 24 months and managed per routine practice. In FILOSOPHY, patients will have the option of taking part in an additional 24-month extension. Patients who discontinue filgotinib treatment may remain in the study and will continue to be followed up until they reach the end of the study or prematurely discontinue the study (e.g., due to being lost to follow-up, withdrawal of consent, or death).

Study Population

To be eligible for inclusion in the studies, patients had to be aged ≥ 18 years with moderate to severe active RA, being prescribed filgotinib for the first time, and able and willing to use an electronic device for PRO collection. FILOSOPHY enrolled patients in Germany, Belgium, the UK, the Netherlands, Spain, and Italy. PARROTFISH was conducted as a separate study only in France, due to an initial requirement to include only female patients, pending results of ongoing clinical studies on the effect of filgotinib on spermatogenesis. These studies concluded that there was no evidence that filgotinib affected testicular function [19], and the PARROTFISH protocol was subsequently updated in June 2023 to allow for male patients to be enrolled. Exclusion criteria for FILOSOPHY and PARROTFISH included participation in any other interventional or non-interventional study, with the exception of registries that contribute data to other studies, and female participants being pregnant or intending to become pregnant before or while taking filgotinib (as per the product label).

Primary Objective

The primary objective of FILOSOPHY and PARROTFISH is to evaluate filgotinib treatment persistence per routine practice at 24 months, defined as the probability of remaining on treatment at 24 months after treatment initiation.

Secondary Objectives

The studies capture patient information on baseline characteristics, sociodemographic characteristics, RA disease characteristics, previous RA treatments, reasons for treatment changes, comorbidities, and concomitant medications. Baseline characteristics include the reason for starting filgotinib, which is selected by the investigator from the following options: primary inadequate response to previous treatment; loss of response to previous treatment; intolerance (non-allergic) to previous treatment; and “other.” The treatment target is also reported by the investigator upon study entry for each patient, and is selected from the following options: remission by Disease Activity Score for 28 joint count (DAS28); remission by Clinical Disease Activity Index (CDAI); remission by other score; LDA by DAS28; LDA by CDAI; LDA by other score; improvement in patient-defined pain; improvement in patient-defined fatigue; improvement in patient-defined functionality; other treatment target; and treatment target not defined.

Filgotinib persistence (reported as the primary objective at month 24) is also assessed at months 1, 3, 6, 12, and 18, and reasons for discontinuation throughout the study will be described. DAS28 using C-reactive protein (DAS28-CRP), CDAI score, and the Health Assessment Questionnaire Disability Index (HAQ-DI) score are assessed at months 1, 3, 6, 12, 18, and 24 (HAQ-DI was added under a protocol amendment in May 2022). Patient and disease characteristics associated with treatment persistence and/or disease activity will be identified.

The following PROs are assessed weekly from week 1 to month 1, at month 3 and then every 3 months up to month 24: pain assessed on a visual analog scale (VAS); fatigue assessed using the Functional Assessment of Chronic Illness Therapy (FACIT)-Fatigue score (reused with permission); work productivity assessed using the Work Productivity and Activity Impairment: Questionnaire for Rheumatoid Arthritis (WPAI-RA) and disease impact measured using Rheumatoid Arthritis Impact of Disease (RAID) score. RAID was added under a protocol amendment in May 2022. At that time, the frequency of WPAI-RA assessments was also reduced to exclude visits at week 1 and 3 and at months 9, 15, and 21, the aim being to reduce patient burden following the addition of RAID.

Safety assessments include the rates of treatment-emergent adverse events (TEAEs), serious TEAEs and TEAEs of interest, including herpes zoster, opportunistic infections, serious infections, major adverse cardiac events (MACE; which is non-adjudicated and includes acute myocardial infarction, stroke, and cardiovascular (CV) death [MedDRA v26.0] as per the 3-point MACE definition), pulmonary embolism, deep vein thrombosis, fractures, malignant or unspecified tumors (excluding non-melanoma skin cancer (NMSC]), NMSC (assessed by the occurrence of basal cell carcinoma, squamous cell carcinoma of the skin, and neuroendocrine carcinoma of the skin), hyperlipidemia, and gastrointestinal perforation.

Exploratory Objectives

Exploratory objectives include describing filgotinib treatment patterns such as concomitant medication, use of glucocorticoids, dose tapering, and reasons for treatment changes. The study will also evaluate the effect of changes in filgotinib treatment patterns on disease activity and PROs. Treatment adherence (the extent to which patients adhere to the posology as per the label) is assessed using the Compliance Questionnaire Rheumatology (CQR) at months 1 and 3, and then every 3 months up to month 24. Originally the CQR19, a 19-item self-administered questionnaire, was to be used. However, following the protocol amendment in May 2022, this was changed to the CQR5, a shortened 5-item version of the questionnaire, which has been shown to maintain a good level of reliability but decreases patient burden compared with the CQR19 [20]. Other exploratory objectives are to identify patient and disease characteristics associated with the development of TEAEs of interest, and to evaluate changes from baseline in the following parameters at months 1, 3, 6, 12, 18, and 24: CRP; erythrocyte sedimentation rate; total cholesterol; low-density lipoprotein cholesterol; high-density lipoprotein cholesterol and triglycerides; aspartate transaminase; alanine transaminase; bilirubin; serum creatinine; estimated glomerular filtration rate; hemoglobin (Hb); HbA1C; neutrophils; lymphocytes; and platelets.

Data Collection

Data are collected, via electronic case report forms, from the date of filgotinib initiation until study completion at 24 months after enrollment, loss to follow-up, withdrawal of consent, or death, whichever occurs first. In FILOSOPHY, patients have the option of taking part in a 24-month extension, with additional follow-up visits at 30, 36, 42, and 48 months. PROs are collected remotely using electronic devices (patients can use their own device or can be provided with a device). The index date (serving as the anchor for all analyses and for assessing persistence) is defined as the date of first administration of filgotinib.

Ethical Approval

The studies are conducted in accordance with Good Pharmacovigilance Practice guidelines and local regulatory requirements, and were approved by the responsible independent research ethics committee/institutional review board, as applicable (Supplementary materials). All patients provided written informed consent.

Sample Size and Statistical Analyses

The studies aimed to recruit approximately 1500 patients in total. This sample size was primarily determined based on feasibility and consistency with previous non-interventional studies in patients with moderate to severe RA treated with JAK inhibitors, which reported 2-year persistence of approximately 50% [21]. Assuming a true persistence of 50%, a sample size of 1500 allows estimation of this proportion with approximately ± 2.5% precision.

Analysis of Endpoints

Treatment persistence is a time-to-event endpoint analyzed using the Kaplan–Meier method (to derive persistence at 24 months and other timepoints). Treatment persistence is defined as the probability of remaining on filgotinib treatment without permanent discontinuation or death. Time to discontinuation or death was censored at the date of last available evidence of filgotinib use.

As study visits might not have occurred on protocol-specified days, observations for longitudinal endpoints were assigned to mutually exclusive analysis windows. Disease activity and PROs were analyzed descriptively at these fixed time windows; these included longitudinal continuous endpoints (e.g., DAS28-CRP at month 1), longitudinal categorical endpoints (e.g., DAS28-CRP within a certain range at month 1), and longitudinal binary endpoints (e.g., clinically meaningful change from baseline in DAS28-CRP at month 1). Safety outcomes were summarized descriptively and analyzed using exact event dates, with exposure-adjusted incidence rates.

Assessments and Subgroups in this Interim Analysis

In this interim analysis, as-observed data are reported from May 2021 up to April 2025. This data cut was selected as it was the point at which the last enrolled patient was expected to have completed 6 months of follow-up after treatment initiation. Baseline characteristics and disease activity measures, PROs, and treatment persistence (up to month 6) are reported for all enrolled patients with a measurement at the specified analysis window. Reported measures include median DAS28-CRP and CDAI, the proportion of patients with LDA according to DAS28-CRP (a score of ≤ 3.2) or CDAI (a score of ≤ 10), and the proportion with a DAS28-CRP of ≤ 2.6 or in CDAI remission (a score of ≤ 2.8). PROs include median HAQ-DI (measured on a scale of 0–3, with higher values indicating greater disability), median VAS pain score (measured of a scale of 0–100 mm, with higher values indicating more pain), median FACIT-Fatigue score (measured on a scale of 0–52, with higher values indicating less fatigue), and the proportion of patients with a clinically meaningful change from baseline in VAS pain score (defined as a reduction of ≥ 10 mm [22]) or in FACIT-Fatigue score (defined as an increase of ≥ 4.0 [23]).

Measures of disease activity and PROs are reported for all treated patients and for the following subgroups: patients receiving filgotinib monotherapy and patients receiving filgotinib in combination with csDMARDs for RA as reported on study day 8 (day 8 was used rather than baseline to avoid inclusion of patients with overlapping treatment start and end dates rather than genuine combinations); and patients who were advanced therapy (AT)-naïve (no prior history of tsDMARD or bDMARD for RA) and AT-experienced (prior history of tsDMARD or bDMARD for RA).

Safety assessments include TEAEs of interest and TEAEs occurring in ≥ 2% of patients, taking all events reported up to the time of the analysis into account (up to 24 months). TEAEs of interest are reported for all patients and in the following subgroups: patients aged ≥ 65 years and/or with ≥ 1 CV risk factor and patients aged < 65 years with no CV risk factors. The following were considered CV risk factors: being a current or former smoker; having a family history of CV disease, or having a medical history of diabetes; peripheral vascular disease; hypertension; dyslipidemia; CV disease; or ischemic central nervous system vascular disorders.

In this interim analysis, measures of disease activity and PROs were analyzed “on treatment” meaning that observations after the permanent discontinuation of filgotinib were excluded and missing data were not imputed. Similarly, TEAEs and deaths were analyzed for the on-treatment period, which included events up to 30 days after permanent discontinuation. Data for the filgotinib 200 mg and filgotinib 100 mg doses were pooled.

Results

Patient Disposition and Baseline Characteristics

Enrollment began in May 2021 and was completed in March 2024, with a total of 1459 patients enrolled in both studies combined (553 in Germany, 205 in Belgium, 166 in the UK, 165 in the Netherlands, 155 in France, 110 in Spain, and 105 in Italy). Of the 1459 enrolled patients, 1431 initiated treatment with filgotinib (28 patients did not initiate filgotinib treatment). As of April 2025 (current data cut), the median follow-up was 706 days and median duration of filgotinib treatment was 566 days. At that time, 93.0% of patients had completed at least 6 months and 30.0% of patients had completed 2 years of follow-up; 38.9% of patients had prematurely discontinued treatment. Of all treated patients, 13.9% (199/1431) discontinued treatment due to lack of efficacy and 10.6% (152/1431) due to adverse events (Fig. S1).

At baseline, median age of all treated patients was 56.0 years, 76.8% were female, and the median duration of RA was 8.1 years (Table 1). A medical history of hypertension and dyslipidemia were reported in 28.7% and 13.8% of patients, respectively.

Table 1.

Baseline characteristics and CV risk factors

	All filgotinib-treated patients (N = 1431)
Baseline demographics
Age, years	56.0 (49.0–64.0)
Female sex, n (%)	1099 (76.8)
BMI, kg/m2	26.4 (23.1–30.5); n = 1286
TJC28	6.0 (3.0–12.0); n = 1374
SJC28	4.0 (2.0–7.0); n = 1372
DAS28-CRP	4.5 (3.7–5.3); n = 1134
CDAI score	23.0 (16.0–32.0); n = 1123
VAS pain score	63 (48–76); n = 474
FACIT-Fatigue score	26 (18–35); n = 468
RA duration, years	8.1 (3.3–15.7); n = 1421
RF positive and/or ACPA positive, n (%)a	839 (58.6)
Smoking status, n (%)b
Former or current smoker	496 (34.7)
Non-smoker	684 (47.8)
Family history of myocardial infarction, n (%)c	96 (6.7)
Comorbidities of interest/medical history, n (%)
Hypertension	411 (28.7)
Dyslipidemia	197 (13.8)
Diabetes	105 (7.3)
CV disease	82 (5.7)
Peripheral vascular disease	35 (2.4)
Ischemic CNS vascular disorders	33 (2.3)

Values are median (interquartile range), except where indicated

ACPA anti-citrullinated protein autoantibodies, BMI body mass index, CDAI Clinical Disease Activity Index, CNS central nervous system, CV cardiovascular, DAS28-CRP Disease Activity Score 28 with C-reactive protein, FACIT-Fatigue Functional Assessment of Chronic Illness Therapy–Fatigue, RA rheumatoid arthritis, RF rheumatoid factor, SJC swollen joint count, TJC tender joint count, VAS visual analog scale

^a165 patients (11.5%) were RF negative and ACPA negative; the remaining 427 patients (29.8%) cannot be classified due to missing values for RF, ACPA, or both

^cData are missing for 251 patients (17.5%)

^cData are missing for 315 patients (22.0%)

In total, 94.1% of patients had received prior treatment with DMARDs for RA: 90.1% had received ≥ 1 prior csDMARD, 61.4% had received ≥ 1 prior bDMARD, and 17.8% had received ≥ 1 prior tsDMARD (Table 2). A total of 64.5% of patients had received prior treatment with bDMARDs or tsDMARDs (i.e., were AT-experienced), while 35.5% of patients were AT-naïve.

Table 2.

Prior and current treatments

	All filgotinib-treated patients (N = 1431)
Prior exposure to
At least one DMARD for RA	1346 (94.1)
At least one csDMARD for RA	1290 (90.1)
At least one bDMARD for RA	879 (61.4)
At least one tsDMARD for RA	255 (17.8)
Reason to start filgotinib
Loss of response to previous treatment	768 (53.7)
Primary inadequate response to previous treatment	459 (32.1)
Non-allergic intolerance to previous treatment	128 (8.9)
Othera	67 (4.7)
Not reported	9 (0.6)
Main target for filgotinib treatment
Remission by DAS28	710 (49.6)
Target not defined	262 (18.3)
LDA by DAS28	166 (11.6)
Remission by CDAI	111 (7.8)
Improvement in patient-defined pain	96 (6.7)
Improvement in patient-defined functionality	34 (2.4)
LDA by CDAI	17 (1.2)
Remission by other score	7 (0.5)
LDA by other score	3 (0.2)
Improvement in patient-defined fatigue	1 (< 0.1)
Other	12 (0.8)
Not reported	12 (0.8)
Study treatment
Filgotinib monotherapy	741 (51.8)
Filgotinib + csDMARD for RA	674 (47.1)
Filgotinib + methotrexate	546 (38.2)
Filgotinib + other csDMARDs	128 (8.9)
Filgotinib dose
200 mg	1306 (91.3)
100 mg	125 (8.7)

Values are n (%)

(b/cs/ts)DMARD (biological/conventional-synthetic/targeted synthetic) disease-modifying antirheumatic drug, CDAI Clinical Disease Activity Index, CV cardiovascular, DAS28 Disease Activity Score 28, LDA low disease activity, RA rheumatoid arthritis

^aOther reasons, which were collected via a free-text field, included (but were not limited to): better CV risk profile, clarification of thromboembolic risk due to prior therapy with tofacitinib, disease activity/active RA, radiological progression, physician or patient decision, refusal of subcutaneous medication, request for oral treatment, and resuming treatment after pregnancy

The most common reasons for starting filgotinib were loss of response to previous treatment (53.7%) and primary inadequate response to previous treatment (32.1%) (Table 2). For nearly half of patients (49.6%), the predefined treatment target at study entry was DAS28 remission (Table 2).

In total, 51.8% of patients received filgotinib monotherapy and 47.1% received filgotinib in combination with csDMARDs for RA (Table 2). At baseline, 35.4% of patients in the monotherapy group, and 38.4% of those in the combination therapy group received concomitant glucocorticoids. The majority of patients (91.3%) received filgotinib 200 mg; 8.7% received filgotinib 100 mg.

Clinical Outcomes

Disease Activity

Among patients with available assessments, median (interquartile range [IQR]) DAS28-CRP decreased from 4.5 (3.7–5.3) at baseline to 2.9 (2.0–3.7) at month 1 and 2.3 (1.7–3.3) at month 6. Similarly, median (IQR) CDAI score decreased from 23.0 (16.0–32.0) at baseline to 10.0 (5.0–17.0) at month 1 and 6.0 (2.0–12.0) at month 6 (Table 3).

Table 3.

DAS28-CRP, CDAI, and HAQ-DI scores from baseline to month 6, including subgroup findings

	All filgotinib-treated patients (N = 1431)	Study treatment			Prior RA treatment
		Filgotinib monotherapy (n = 741)	Filgotinib plus csDMARD (n = 674)	AT-naïve (no prior bDMARD or tsDMARD) (n = 508)		AT-experienced (prior bDMARD or tsDMARD) (n = 923)
DAS28-CRP
Baseline	4.5 (3.7–5.3) n = 1134/1431 (79.2%)	4.5 (3.7–5.3) n = 587/741 (79.2%)	4.5 (3.8–5.3) n = 532/674 (78.9%)	4.4 (3.7–5.3) n = 382/508 (75.2%)		4.5 (3.8–5.3) n = 752/923 (81.5%)
Month 1	2.9 (2.0–3.7) n = 742/1428 (52.0%)	2.8 (2.0–3.8) n = 387/741 (52.2%)	2.9 (2.0–3.7) n = 352/674 (52.2%)	2.7 (1.9–3.5) n = 276/505 (54.7%)		3.0 (2.1–3.8) n = 466/923 (50.5%)
Month 3	2.5 (1.8–3.4) n = 876/1337 (65.5%)	2.5 (1.7–3.3) n = 473/687 (68.9%)	2.4 (1.8–3.5) n = 402/648 (62.0%)	2.2 (1.7–3.0) n = 295/475 (62.1%)		2.7 (1.8–3.6) n = 581/862 (67.4%)
Month 6	2.3 (1.7–3.3) n = 850/1227 (69.3%)	2.3 (1.7–3.4) n = 443/627 (70.7%)	2.3 (1.7–3.3) n = 406/598 (67.9%)	2.0 (1.6–3.1) n = 301/450 (66.9%)		2.4 (1.8–3.5) n = 549/777 (70.7%)
CDAI score
Baseline	23.0 (16.0–32.0) n = 1123/1431 (78.5%)	23.0 (16.0–32.0) n = 585/741 (78.9%)	24.0 (17.0–32.0) n = 527/674 (78.2%)	22.0 (16.0–31.0) n = 388/508 (76.4%)		24.0 (17.0–33.0) n = 735/923 (79.6%)
Month 1	10.0 (5.0–17.0) n = 776/1428 (54.3%)	10.0 (5.0–17.0) n = 418/741 (56.4%)	10.0 (5.0–16.0) n = 354/674 (52.5%)	8.0 (4.0–14.0) n = 292/505 (57.8%)		11.0 (6.0–18.0)n = 484/923 (52.4%)
Month 3	7.0 (3.0–13.0) n = 909/1337 (68.0%)	7.0 (3.0–13.0) n = 493/687 (71.8%)	7.0 (3.0–14.0) n = 415/648 (64.0%)	5.0 (2.0–10.0) n = 306/475 (64.4%)		8.0 (4.0–15.0)n = 603/862 (70.0%)
Month 6	6.0 (2.0–12.0) n = 859/1227 (70.0%)	6.0 (2.0–12.0) n = 465/627 (74.2%)	6.0 (3.0–12.0) n = 393/598 (65.7%)	4.0 (2.0–10.0) n = 308/450 (68.4%)		7.0 (3.0–14.0) n = 551/777 (70.9%)
HAQ-DI score
Baseline	1.3 (0.8–1.8) n = 340/1431 (23.8%)	1.3 (0.6–1.7) n = 168/741 (22.7%)	1.3 (0.8–1.9) n = 169/674 (25.1%)	1.4 (0.8–1.9) n = 129/508 (25.4%)		1.3 (0.8–1.8) n = 211/923 (22.9%)
Month 1	0.8 (0.3–1.3) n = 232/1428 (16.2%)	0.9 (0.2–1.4) n = 116/741 (15.7%)	0.8 (0.3–1.3) n = 116/674 (17.2%)	0.8 (0.1–1.3) n = 93/505 (18.4%)		0.9 (0.4–1.4) n = 139/923 (15.1%)
Month 3	0.8 (0.3–1.4) n = 276/1337 (20.6%)	0.8 (0.1–1.4) n = 137/687 (19.9%)	0.8 (0.3–1.4) n = 139/648 (21.5%)	0.5 (0.0–1.4) n = 103/475 (21.7%)		0.9 (0.3–1.4) n = 173/862 (20.1%)
Month 6	0.8 (0.1–1.4) n = 241/1227 (19.6%)	0.7 (0.1–1.4) n = 118/627 (18.8%)	0.8 (0.3–1.4) n = 123/598 (20.6%)	0.4 (0.0–1.2) n = 96/450 (21.3%)		0.9 (0.3–1.5) n = 145/777 (18.7%)

Values are median (interquartile range) and the number of patients with a measurement/the number of patients on treatment (proportion of patients with available data) at each timepoint. Data are from the on-treatment analysis as observed at the time of the data cut

AT advanced therapy, CDAI Clinical Disease Activity Index, (b/cs/ts)DMARD (biologic/conventional-synthetic/targeted synthetic) disease-modifying antirheumatic drug, DAS28-CRP Disease Activity Score 28 with C-reactive protein, HAQ-DI Health Assessment Questionnaire Disability Index

Median DAS28-CRP and CDAI score improved from baseline to month 6 among patients with available data. Similar patterns were observed in patients receiving filgotinib monotherapy, those receiving filgotinib in combination with csDMARDs, AT-naïve and AT-experienced patients (Table 3).

Among patients with available assessments, the proportion with DAS28-CRP ≤ 2.6 was 6.1% (69/1134) at baseline increasing to 58.1% (494/850) at month 6. The proportion with LDA based on DAS28-CRP (including those with DAS28-CRP ≤ 2.6) was 13.4% (152/1134) at baseline, increasing to 71.4% (607/850) at month 6 (Fig. 1a). The proportion with CDAI remission was 0.9% (10/1123) at baseline, increasing to 25.7% (221/859) at month 6. The proportion of patients with low disease activity based on CDAI (including those in remission) was 8.9% (100/1123) at baseline increasing to 69.8% (600/859) at month 6 (Fig. 1b). Again, similar patterns were observed across the subgroups included in this interim analysis (Fig. S2a, b).

Fig. 1.

Patients with DAS28-CRP ≤ 2.6 or LDA (a) and CDAI remission or LDA (b). Data are from the on-treatment analysis as observed at the time of the data cut. No imputation of missing data was performed. Total percentages above each column are based on patient numbers so may not equal the sum of each percentage due to rounding. BL baseline, CDAI Clinical Disease Activity Index, DAS28-CRP Disease Activity Index 28 with C-reactive protein, LDA low disease activity, M month

PROs

Among the relatively small number of patients with available HAQ-DI data, median (IQR) scores improved from 1.3 (0.8–1.8) at baseline to 0.8 (0.3–1.3) at month 1 and remained stable at month 6 (0.8 [0.1–1.4]). Similar improvements were observed in the aforementioned subgroups (Table 3). Among patients with available VAS pain assessments, the median (IQR) score improved from 63 mm (48–76) at baseline to 52 mm (34–68) at week 1 and 29 mm (11.5–53) at month 6 (Fig. 2a). Similarly, median (IQR) FACIT-Fatigue score improved from 26 (18–35) at baseline to 29 (21–38) at week 1 and 35 (25–43) at month 6 (Fig. 2b). A similar pattern of early, and thereafter sustained, improvement in VAS pain and FACIT-Fatigue score was observed in patients who received filgotinib monotherapy and in those who received filgotinib in combination with csDMARDs (Fig. S3a, b). Improvements for both PROs were also seen in AT-naïve and AT-experienced patients (Fig. S3c, d).

Fig. 2.

VAS pain (a) and FACIT-Fatigue scores (b) over time in all filgotinib-treated patients. Data are from the on-treatment analysis as observed at the time of the data cut. No imputation of missing data was performed. Dotted line indicates score for patient acceptable symptom state for VAS pain score [37, 38]. BL baseline, FACIT-Fatigue Functional Assessment of Chronic Illness Therapy–Fatigue, IQR interquartile range, M month, VAS visual analog scale

Among patients with available assessments, the proportion with a clinically meaningful change from baseline in VAS pain score was 44.1% (178/404) at week 1 and 69.6% (227/326) at month 6 (Fig. 3a); for FACIT-Fatigue score the proportions were 43.8% (176/402) and 62.5% (202/323), respectively (Fig. 3b). In each of the subgroups assessed in this interim analysis, the proportion of patients achieving clinically meaningful changes from baseline in VAS pain score and in FACIT-Fatigue score increased from week 1 up to month 6 (Fig. S4a–d).

Fig. 3.

Patients with clinically meaningful changes from baseline* in VAS pain (a) and FACIT-Fatigue score (b). *A reduction of ≥ 10 mm in VAS pain score and an increase of ≥ 4.0 in FACIT-Fatigue score were considered clinically meaningful changes from baseline. Data are from the on-treatment analysis as observed at the time of the data cut. No imputation of missing data was performed. BL baseline, FACIT-Fatigue Functional Assessment of Chronic Illness Therapy–Fatigue, M month, VAS visual analog scale, W week

Treatment Persistence

At the time of the current interim analysis, treatment persistence (95% confidence interval [CI]) was 84.0% (82.0, 85.8) at month 6 (Fig. S5).

Safety

Based on pooled data for the filgotinib 200 mg and 100 mg doses, the exposure-adjusted incidence rate (EAIR) (95% CI) per 100 patient-years of exposure (PYE) was 89.8 (84.1, 95.8) for any TEAEs on treatment, 9.4 (8.0, 10.8) for serious TEAEs, and 8.5 (7.3, 9.9) for TEAEs leading to premature discontinuation of filgotinib. The most frequent TEAEs, occurring in ≥ 2% of patients, are listed in Table S1.

The EAIR per 100 PYE (95% CI) was 0.0 (0.0, 0.3) for serious infections, 8.4 (7.1, 9.8) for COVID-19, 1.3 (0.9, 1.9) for herpes zoster (1.4 [0.4, 3.5] in vaccinated patients; 1.3 [0.8, 2.0] in unvaccinated patients; 77.0% of vaccinated patients received Shingrix), and 0.2 (0.1, 0.5) for opportunistic infections. For CV events, the EAIR (95% CI) per 100 PYE was 0.6 (0.3, 1.0) for MACE, 0.2 (0.1, 0.6) for pulmonary embolism, 0.2 (0.1, 0.6) for transient ischemic attack, 0.1 (0.0, 0.4) for deep vein thrombosis, and 0.1 (0.0, 0.4) for cardiac failure. Other events of interest included fractures and malignant or unspecified tumors (excluding NMSC), with EAIRs per 100 PYE of 1.8 (1.3, 2.5) and 0.6 (0.3, 1.0), respectively. In terms of NMSC, the EAIR per 100 PYE for basal cell carcinoma was 0.2 (0.1, 0.5), and it was 0.0 (0.0, 0.3) for both squamous cell carcinoma of the skin and neuroendocrine carcinoma of the skin. The EAIR for hyperlipidemia was 0.1 (0.0, 0.4). No cases of gastrointestinal perforation were reported.

Of patients aged ≥ 65 years and/or with ≥ 1 CV risk factor, 68.1% (656/963) had TEAEs, as did 57.5% (269/468) of those aged < 65 years with no CV risk factors. EAIRs for TEAEs of interest according to age and/or CV risk are shown in Table 4.

Table 4.

EAIRs for TEAEs of interest according to age and/or CV risk (on-treatment analysis)

Adverse event	Aged ≥ 65 years and/or with CV risk factor (including ever smoked) EAIR per 100 PYE (95% CI) (n = 963)	Aged < 65 years with no CV risk factors (including ever smoked) EAIR per 100 PYE (95% CI) (n = 468)
Herpes zoster	1.4 (0.8, 2.2)	1.1 (0.4, 2.3)
Opportunistic infections	0.2 (0.0, 0.6)	0.2 (0.0, 0.9)
Serious infections	0.1 (0.0, 0.4)	0.0 (0.0, 0.6)
MACE	0.7 (0.4, 1.4)	0.3 (0.0, 1.1)
Pulmonary embolism	0.2 (0.0, 0.6)	0.3 (0.0, 1.1)
Transient ischemic attack	0.3 (0.1, 0.8)	0.2 (0.0, 0.9)
Deep vein thrombosis	0.1 (0.0, 0.4)	0.2 (0.0, 0.9)
Cardiac failure	0.1 (0.0, 0.4)	0.2 (0.0, 0.9)
Fractures	1.7 (1.1, 2.6)	2.1 (1.1, 3.5)
Malignant or unspecified tumors (excluding NMSC)	0.7 (0.4, 1.4)	0.3 (0.0, 1.1)

MACE is non-adjudicated and includes acute myocardial infarction, stroke, and CV death (MedDRA v26.0)

CI confidence interval, CV cardiovascular, EAIR exposure-adjusted incidence rate, MACE major adverse cardiovascular events, MedDRA Medical Dictionary for Regulatory Activities, NMSC non-melanoma skin cancer, PYE patient-years of exposure, TEAE treatment-emergent adverse event

There were 10 deaths on treatment. Two deaths, one caused by pneumococcal meningitis and one caused by pneumonia and pulmonary embolism, were considered by the investigator to be possibly related to study treatment (relationship to treatment was not adjudicated). The remaining eight deaths were considered unrelated to treatment (primary cause of death is shown in Table S2).

Discussion

In this interim analysis, patients treated with filgotinib demonstrated improvements in DAS28-CRP, CDAI, and HAQ-DI scores as early as month 1, the first timepoint at which these outcomes were assessed, and in patient-reported pain and fatigue as early as week 1. Improvements in each of these measures were observed in the filgotinib monotherapy and combination therapy subgroups, and in patients who were AT-naïve and those who were AT-experienced. Treatment persistence at month 6 was high (84.0%). No new safety findings were observed with a median treatment duration of 566 days, and safety results were in line with those observed in the clinical trial program [9–11].

As in this interim analysis, improvements in disease activity were consistently reported across the filgotinib clinical trial program. For example, in FINCH 1, of patients treated with filgotinib 200 mg (who had a mean baseline DAS28-CRP of 5.8), 50% achieved DAS28-CRP ≤ 3.2 at week 12, as did 61% at week 24 [9]; in FINCH 2, the proportions were 40.8% at week 12 and 48.3% at week 24 (with a mean baseline DAS28-CRP of 5.9) [10]. At these timepoints, the proportion of patients reaching DAS28-CRP ≤ 3.2 in FILOSOPHY and PARROTFISH was 69.7% and 71.4% (with a median baseline score of 4.5). Findings from FILOSOPHY and PARROTFISH demonstrated rapid improvements in disease activity, with assessments taking place from an earlier timepoint than in the clinical trials (month 1 vs month 3). A small proportion of patients had DAS28-CRP ≤ 2.6 or CDAI remission at baseline (6.1% and 0.9%, respectively). These patients may have entered the FILOSOPHY and PARROTFISH studies having discontinued prior treatment due to adverse events or other reasons unrelated to effectiveness.

At month 6 of FILOSOPHY and PARROTFISH, 69.8% of patients achieved CDAI LDA and 25.7% achieved CDAI remission. Improvements in disease activity have been reported in other observational studies assessing RA treatments, including JAK inhibitors, in a real-world setting. A 6-month interim analysis of the RA-BE-REAL observational study showed that in patients treated with baricitinib, 62.4% achieved CDAI LDA, including 25.6% in CDAI remission, while in patients treated with any other tsDMARD or any bDMARD, 55.8% achieved CDAI LDA, including 18.5% in CDAI remission [15]. In the UPHOLD study of patients treated with upadacitinib, 61.1% achieved CDAI LDA (including 20.2% in remission) at 6 months, increasing to 68.6% (including 28.0% in remission) at 12 months [24]. Of those initiating treatment with a JAK inhibitor in the JAK-pot registry, adjusted response rates were 56% and 15% for CDAI LDA and remission, respectively, at 1 year [25]. As longer-term data become available from FILOSOPHY and PARROTFISH, the improvement in disease activity—evident from the present interim analysis—can be assessed in context of those observed in other registries and real-world studies, over similar time periods.

It has been shown that the effectiveness of tumor necrosis factor (TNF) inhibitors improves when administered in combination with csDMARDs [26]. In contrast, in FILOSOPHY and PARROTFISH, a similar improvement in disease activity was seen in patients treated with filgotinib monotherapy and in those treated with filgotinib in combination with csDMARDs. In terms of the AT subgroups, numerically greater improvements in disease activity were generally seen in AT-naïve patients than in AT-experienced patients. However, no formal statistical testing was performed to compare these subgroups, since the study was not designed for such comparisons. Therefore, findings regarding the subgroups should be interpreted with caution.

Although PROs data were not available for all patients, improvements in pain and fatigue were observed in FILOSOPHY and PARROTFISH from week 1, which is a new finding in the context of JAK inhibitor therapy for RA in real-world practice. Improvements in pain and fatigue were previously observed in each of the FINCH clinical trials [9–11]. In the FINCH trials, mean reductions from baseline in VAS pain score of between 10 and 20 mm were reported at week 2, which was the first timepoint at which pain was assessed [27]. With improvements already observed after 1 week in FILOSOPHY and PARROTFISH, interim data are consistent with a rapid reduction of pain levels in patients treated with filgotinib. This is an important finding given that prompt pain relief is considered by patients to be a top priority of their RA treatment, and remains so over time [4, 28]. In FILOSOPHY and PARROTFISH, median VAS pain score improved from 63 mm at baseline to 29 mm at month 6, and HAQ-DI score improved from 1.3 at baseline to 0.8 at month 6. These PROs have been assessed in other real-world RA studies: in the RA-BE-REAL study, mean VAS pain score in patients treated with baricitinib improved from 58.9 mm at baseline to 34.5 mm at month 6, while mean HAQ-DI improved from 1.4 at baseline to 1.0 at month 6 [15]; in a real-world, observational study of patients treated with baricitinib in Spain (the ORBIT-RA study), VAS pain score improved from 66 mm at baseline to 41 mm at month 6 [29].

PROs are being evaluated with the use of electronic devices in the FILOSOPHY and PARROTFISH studies. This allows outcomes to be conveniently reported by patients at regular intervals and without the need for clinic attendance. The extent to which PROs are completed will provide insight into patients’ willingness and ability to regularly use electronic devices to capture health-related information. Notably, following baseline assessment, such evaluations are being completed from as early as week 1, which is earlier than in previous filgotinib clinical trials and allows for capturing such information in a patient-relevant timeframe.

Safety findings were generally in line with those previously reported for filgotinib. In an integrated safety analysis of filgotinib clinical trials, the EAIR for any TEAE (defined as those that occurred during treatment or up to 30 days after treatment) was 101.5 per 100 PYE. The EAIR per 100 PYE was 6.4 for serious TEAEs, 4.7 for TEAEs leading to premature treatment discontinuation, 0.4 for MACE (positively adjudicated), and 2.0 for serious infections [30]; the EAIRs per 100 PYE seen in FILOSOPHY/PARROTFISH were 89.8 for any TEAE, 9.4 for serious TEAEs, 8.5 for treatment discontinuation, 0.6 for MACE (unadjudicated), and 0.0 for serious infections. In the 12-month interim analysis of the UPHOLD study, it was reported that 11.7% of patients had TEAEs resulting in drug discontinuation [24]. In FILOSOPHY and PARROTFISH, 10.6% discontinued treatment due to TEAEs up to 2 years.

Treatment persistence is an important factor in achieving the goal of sustained remission/LDA, as discontinuation of therapies may result in worse clinical outcomes in patients with RA and, in turn, increased costs to healthcare systems [31, 32]. Treatment persistence in a real-world setting is influenced by multiple factors and should not be interpreted as a direct measure of effectiveness. Nevertheless, it may provide an indication of treatment benefit and tolerability, and the feasibility of continued treatment in routine practice [25, 33, 34]. Long-term treatment persistence will be an important focus of future analyses of FILOSOPHY and PARROTFISH and will be interpreted alongside other reported outcomes, including measures of disease activity and PROs. In this interim analysis, treatment persistence was estimated to be 84.0% at month 6, which is in line with the 6-month filgotinib persistence rate reported in an Italian registry (84.5%) [35]. For context, a real-world study of baricitinib in Spain reported the 6-month persistence rate to be 79.7% [29]. In the RA-BE-REAL study, the cumulative discontinuation rate at 6 months was 23.3% for the cohort who received any bDMARD or tsDMARD other than baricitinib (60% of whom received a TNF inhibitor) [15].

Real-world studies such as FILOSOPHY and PARROTFISH allow a broader, more diverse patient population to be assessed compared with clinical trials, which have strict inclusion and exclusion criteria and follow stringent protocols. Consequently, real-world studies allow findings from clinical trials to be confirmed in a routine clinical setting, and allow valuable additional data to be collected that may not necessarily be ascertained through clinical trials [16, 36]. Further to confirming findings from the FINCH program, the FILOSOPHY and PARROTFISH studies will capture data regarding aspects such as work productivity and absenteeism, which have not been reported in the previous clinical trials. Gathering such information will provide valuable insights into how day-to-day living is impacted by RA and potentially improved with filgotinib treatment.

In terms of limitations, as FILOSOPHY and PARROTFISH are non-interventional studies, they may be affected by selection or information bias and attrition problems that can sometimes occur with such studies over the longer term. In addition, the studies lack a comparator group. The use of electronic devices to capture PROs may also introduce bias, due to attrition, lack of complete data, or insufficient digital health literacy. However, use of standard measurements and provision of appropriate training aimed to minimize this issue. Data are reported as observed with no imputation for missing values, and results are descriptive, with no statistical testing performed to enable comparisons between subgroups. Outcome data are not available for all enrolled patients. Data may be missing because patients did not reach 6 months of follow-up (e.g., if the data cut occurred before the planned month 6 visit); patients discontinued filgotinib treatment; patients did not provide outcome data within the target window; or baseline data were unavailable (meaning change from baseline measures could not be determined). In the case of HAQ-DI, data are available for a limited proportion of patients because HAQ-DI was added under a protocol amendment. The present findings summarize available “on-treatment data,” which are descriptive in nature, and are therefore not intended to support formal inference on effectiveness. Given that no imputation strategies were implemented, the findings may be prone to attrition bias leading to a potential overestimation of treatment effectiveness. Because of the interim nature of the analysis, current findings must be interpreted with caution. However, as the studies progress the longer follow-up will allow outcomes to be comprehensively evaluated.

Conclusions

The FILOSOPHY and PARROTFISH studies allow data on filgotinib to be captured from patients with RA in a real-world setting, complementing those already obtained from clinical trials. Interim data show early improvements in disease activity, physical function, pain, and fatigue that were maintained to month 6. Improvements were seen when filgotinib was used in different therapeutic strategies. Given the descriptive nature of the analyses presented, observed differences between subgroups may be subject to bias and are not intended to provide evidence of comparative inferences. Treatment persistence remained high at month 6 and early safety findings are in line with available evidence from randomized clinical trials of filgotinib. Future results from the FILOSOPHY and PARROTFISH studies will allow longer-term assessment of the effectiveness and safety of filgotinib and enable additional outcomes to be evaluated that have not been reported from clinical trials.

Supplementary Information

Below is the link to the electronic supplementary material.

Author Contribution

Gerd R. Burmester, James Galloway, Karen Bevers, Jérôme Avouac, Susana Romero-Yuste, Roberto F. Caporali, Neil Betteridge, and Patrick Verschueren contributed to the concept and design of the study, and the acquisition of the data; Carole Van der Donckt, Thomas P.A. Debray, and Ouafia Tandjaoui Bouzid contributed to the analysis of data. All authors contributed to interpretation of data, critically reviewed the manuscript, and approved the final version for submission.

Funding

The study was sponsored by Alfasigma S.p.A. (Bologna, Italy). Alfasigma was involved in the conceptualization and design of the study, collection and analysis of data, the decision to publish, and in the preparation of the manuscript. The journal’s Rapid Service Fee was funded by Alfasigma.

Data Availability

Anonymized individual patient data will be shared upon request (beginning 6 months and ending 5 years following publication of 2-year follow-up data) for research purposes dependent upon the nature of the request, the merit of the proposed research and the availability of the data and its intended use. Scientifically sound proposals should be directed to https://vivli.org/ourmember/alfasigma/. Further details concerning Alfasigma’s commitment to clinical trial data sharing can be found at https://www.alfasigma.com/our-company/ethics-and-transparency/scientific_transparency/.

Declarations

Conflict of Interest

Gerd R. Burmester reports consultancy and speaker fees from AbbVie, Alfasigma, BMS, Eli Lilly & Co., Galapagos, Janssen, Novartis, Pfizer, Sanofi, and UCB. James Galloway reports speaker fees from AbbVie, Eli Lilly & Co., Galapagos, Janssen, Pfizer, and UCB; consultancy fees from AbbVie, Alfasigma S.p.A., Eli Lilly & Co., Galapagos, Janssen, and Pfizer; and grant/research support from AstraZeneca, GSK, Janssen, and Pfizer. Karen Bevers reports consultancy fees from Alfasigma S.p.A. and Galapagos. Jérôme Avouac reports speaker fees from AbbVie, BMS, Eli Lilly & Co., Galapagos, Novartis, and Pfizer; consultancy fees from AbbVie and Galapagos; and grant/research support from BMS, Fresenius Kabi, Novartis, and Pfizer. Susana Romero-Yuste reports consultancy fees from AbbVie, AstraZeneca, Eli Lilly & Co., and UCB; and speaker fees from AbbVie, AstraZeneca, Eli Lilly & Co., and Galapagos. Roberto F. Caporali reports consultancy fees from AbbVie, Alfasigma, BMS, Celltrion, Eli Lilly & Co., Fresenius Kabi, Galapagos, Janssen, Johnson & Johnson, Novartis, Pfizer, and UCB; and speaker fees from AbbVie, Alfasigma, BMS, Celltrion, Eli Lilly & Co., Galapagos, Janssen, Johnson & Johnson, Novartis, Pfizer, and UCB. Neil Betteridge reports consultancy fees from Alfasigma, Galapagos, Grünenthal, Ipsen, and Sanofi; and speaker fees from Galapagos, Grünenthal, and Sanofi. Carole Van der Donckt is an employee of Alfasigma Belgium BV. Thomas P.A. Debray received consulting fees from Alfasigma for this work and has provided paid methodological and statistical consulting to several pharmaceutical and biotech organizations during the past decade. Specific companies are disclosed where directly relevant to the submitted work. Ouafia Tandjaoui Bouzid is a former employee of Alfasigma Belgium BV. Patrick Verschueren reports speaker fees from AbbVie, Alfasigma, Eli Lilly & Co., and Galapagos; consultancy fees from Alfasigma, Boehringer Ingelheim, Citryll, Eli Lilly & Co., and Galapagos; support for attending meetings from Fresenius Kabi and AbbVie; and grant/research support from Galapagos and Pfizer.

Ethical Approval

The studies are conducted in accordance with Good Pharmacovigilance Practice guidelines and local regulatory requirements, and were approved by the responsible independent research ethics committee/institutional review board, as applicable (Supplementary materials). All patients provided written informed consent.