Study design and patients
Data were included from 4 phase 3 randomized clinical studies. The study design and patient inclusion/exclusion criteria for each study have been described previously [4,5,6,7]. Briefly, 684 patients in RA-BUILD (NCT01721057) and 527 patients in RA-BEACON (NCT01721044) with active RA were randomized 1:1:1 to receive placebo or 2- or 4-mg baricitinib once daily (QD); 1305 patients in RA-BEAM (NCT01710358) were randomized 3:3:2 to receive placebo QD or 4-mg baricitinib QD or a subcutaneous injection of adalimumab every 2 weeks; and 588 patients in RA-BEGIN (NCT01711359) were randomized 4:3:4 to receive oral MTX every week (QW) or 4-mg baricitinib QD or 4-mg baricitinib QD plus MTX QW. Beginning at week 16 (week 24 in RA-BEGIN) in each of the studies, non-responders, defined as patients with a lack of improvement of at least 20% in both tender joint count and swollen joint count compared to baseline, received rescue treatment as outlined in Additional file 1: Table S1. All patients completing these 4 phase 3 studies were eligible to enter the long-term extension study, RA-BEYOND.
Patients who were not receiving glucocorticoids prior to randomization were not permitted to initiate glucocorticoid therapy during the study, including intra-muscular or intra-articular glucocorticoids. After rescue, new glucocorticoids or increases in doses of ongoing concomitant glucocorticoids were permitted. Topical, intranasal, intra-ocular, and inhaled glucocorticoids were permitted.
The primary endpoint in the studies was the American College of Rheumatology 20% (ACR20) response rate at week 12 (week 24 in RA-BEGIN). In RA-BEAM and RA-BUILD (the two studies in which patients had an inadequate response to csDMARDs), patients recorded daily symptoms in an electronic diary from the first day of treatment through week 12, referred to as diary patient-reported outcomes (PROs). Diary entries included duration of morning joint stiffness (MJS) and numeric rating scales (NRS) for MJS severity, worst tiredness, and worst joint pain. Scores for the NRS ranged from 0 to 10, with 10 being the worst level [8]. Additional details regarding the study designs can be found in Additional file 1: Table S1.
Each study was conducted in accordance with the principles of the Declaration of Helsinki and Good Clinical Practice Guidelines and approved by each center’s institutional review board or ethics committee. All patients provided written informed consent. The studies were designed by the sponsors, Eli Lilly and Company and Incyte Corporation, with input from an academic advisory board in which non-Lilly authors of this manuscript participated. All authors participated in the preparation and review of this manuscript and approved the final version.
Characteristics of interruptions
During each study, temporary interruptions of study drug initiated by the investigators were required to be documented in an electronic case report form, providing the time of the last dose, the reason for the interruption, and the duration of the interruption. On the form, investigators could choose one of 4 reasons for the interruption: “adverse event (AE),” “abnormal laboratory values,” “suspected pregnancy,” or “investigator decision.” If investigators selected “AE” as the reason for the interruption, they were immediately prompted to link a specific AE to the interruption in the electronic case report form. Investigators might have listed an abnormal laboratory value as an “AE,” rather than selecting the separate option of “abnormal laboratory value” as the reason for the interruption. In these instances, we reviewed the abnormal laboratory values individually to gain a more complete understanding of laboratory abnormalities that led to interruptions. If “abnormal laboratory value” was selected as the reason for interruption, a query was sent to the investigator for specific information regarding the type of laboratory test and abnormal finding.
Temporary interruption was defined as a temporary withholding of study drug that was followed by the resumption of study drug at a later time point. Interruptions were based on the daily tablet baricitinib study drug, including non-baricitinib groups, which represented interruptions of the matching placebo for baricitinib. In some instances, the study drug was interrupted with the intent of restarting, but instead led to a permanent discontinuation of treatment, which will be referred to as “initiated interruptions.” Patients who had their daily tablet of baricitinib or matching baricitinib placebo temporarily interrupted were compared across their originally assigned treatment groups for trends in the frequency, duration, and reason for baricitinib interruptions within each study. For interruptions among all baricitinib-treated patients, including those who were originally randomized to placebo or active-control, but were rescued or switched to baricitinib during the studies, we provide a detailed summary of the reasons for the interruptions.
Impact of interruptions on efficacy
Post hoc analyses to assess the impact of interruptions on efficacy outcomes in the short- and long-term were conducted in patients who were csDMARD-inadequate responders (IR) and MTX-IR using pooled data of patients randomized to placebo and baricitinib in RA-BEAM and RA-BUILD. First, the effect on longer-term efficacy outcomes was assessed by the percentage of patients with ACR20 and ACR50 response and with Disease Activity Score using 28 joints based on C-reactive protein (DAS28-CRP) ≤ 3.2 at week 24. Responses were compared between patients with and without interruptions of any duration during the first 24 weeks or up to rescue.
Second, although standard RA assessments of efficacy response and disease activity measures, with their associated acute phase reactants, were collected only at scheduled visits approximately every 4 weeks, the effect on short-term efficacy outcomes was assessed through the symptoms recorded by daily diary during the first 12 weeks of RA-BEAM and RA-BUILD. These diary scores were examined at time landmarks in patients who were randomized at least 7 days prior to interruption, had an interruption that lasted at least 3 days, and were then retreated with study drug. Symptoms were evaluated at the following milestones:
-
1.
Study drug initiation (average of values obtained within the first 3 days following randomization)
-
2.
Pre-interruption (average of up to 3 most recent values obtained in the 7 days prior to interruption)
-
3.
Last scores during the interruption (average of 3 most recent values during the last 7 days of the interruption)
-
4.
Post-interruption (average of last 3 available values obtained following interruption and prior to any subsequent interruption or week 12 study visit).
These 4-point profiles were compared between patients treated with and interrupting baricitinib (2 mg and 4 mg doses combined) and patients treated with and interrupting matching placebo.
To further explore the short-term loss of response and increase of symptoms during interruptions, an additional analysis was conducted of the 4 daily diary measures for the first temporary interruption per patient during the first 12 weeks when these measures were collected. For each measure, patients were classified into 1 of the following categories:
-
1.
Patients who reported no worsening (possibly even improvement) of symptoms during the interruption.
-
2.
Patients who reported only minimal worsening of symptoms during the interruption, defined as an increase of < 3 units of the NRS or ≤ 30 min of MJS duration. An increase of < 3 units in the NRS was chosen as representative of a change within severity category based on prior use of 3- to 4-point spans between severity categories of mild (0–3), moderate (4–6), and severe (7–10) for MJS [9]. An increase of 30 min of MJS duration was chosen as representative of a change within severity category based on prior use of 30-min spans between severity categories of mild (1–30 min), moderate (31–60 min), and severe (> 60 min) [10].
-
3.
Patients who temporarily reported greater than minimal worsening of symptoms during the interruption, defined as an increase of 3+ units of the NRS or 31+ minutes of MJS duration, but whose last reported symptom scores during the interruption were improvements, not worsening, or minimal worsening.
-
4.
Patients who reported greater than minimal worsening of symptoms during the interruption that did not resolve prior to resumption of treatment. For patients in this last category, the time for their symptoms to increase beyond minimal worsening was categorized as occurring during the first week, during the second week, or beyond the second week of the interruption. For MJS duration, these timings were considered for the first increase of symptoms by ≥ 61 min.
Because the numbers of patients for each of these efficacy assessments varied depending on efficacy and daily diary data available, the flow chart in Additional file 2: Table S2 outlines and explains the sample sizes for each assessment.
Impact of interruptions on safety
Patient-level data were assessed to provide details on AEs and specific laboratory values that led to temporary interruptions. Reoccurrences of AEs and laboratory findings that led to a subsequent interruption after study drug was resumed were assessed via examination of patient-level data. For reoccurrence of AEs, matching preferred terms using the Medical Dictionary for Regulatory Activities (MedDRA) with a date subsequent to the date of drug resumption were identified in the database. For abnormal laboratory findings, a manual review of investigator-specified laboratory values was conducted to determine if abnormalities with these same analytes reoccurred once baricitinib was resumed. For a reoccurrence of the abnormal laboratory value to be considered, it had to be at least as high (or low) as the value that led to the initial interruption.
To assess whether initial tolerability effects persisted upon reinitiation of baricitinib, AEs reported during the first 4 weeks of treatment with baricitinib, from pooled data across all 4 studies, were compared to AEs reported in the 4 weeks after resuming baricitinib following a temporary interruption for the purpose of exploring whether the frequency and nature of the AEs were similar.
Statistical analyses
Descriptive statistics are presented for the overall summary of interruptions, including reasons for and duration of interruptions and for efficacy and safety effects related to interruptions. No formal hypothesis tests were conducted to compare randomized treatment groups in terms of the number or nature of interruptions nor to compare specified groups with respect to the exploratory post hoc short-term and long-term efficacy analyses.