Article Text

Download PDFPDF

Concise report
Biologic discontinuation studies: a systematic review of methods
  1. Kazuki Yoshida1,2,
  2. Yoon-Kyoung Sung1,3,
  3. Arthur Kavanaugh4,
  4. Sang-Cheol Bae3,
  5. Michael E Weinblatt1,
  6. Mitsumasa Kishimoto5,
  7. Kazuo Matsui2,
  8. Shigeto Tohma6,
  9. Daniel H Solomon1
  1. 1Division of Rheumatology, Immunology and Allergy, Brigham and Women's Hospital, Boston, Massachusetts, USA
  2. 2Department of Rheumatology, Kameda Medical Center, Kamogawa, Chiba Prefecture, Japan
  3. 3Department of Rheumatology, Hanyang University Hospital for Rheumatic Diseases, Seoul, South Korea
  4. 4Division of Rheumatology, Allergy and Immunology, University of California, San Diego, La Jolla, California, USA
  5. 5Division of Allergy and Rheumatology, St. Luke's International Hospital, Chuo-ku, Tokyo, Japan
  6. 6Department of Rheumatology, Clinical Research Center for Allergy and Rheumatology, Sagamihara National Hospital, Sagamihara, Japan
  1. Correspondence to Dr Kazuki Yoshida, Division of Rheumatology, Immunology and Allergy, Brigham and Women's Hospital, 75 Francis Street, Boston, MA 02115, USA; kazukiyoshida{at}


Objectives We conducted a systematic review to assess the design and ‘failure definition’ in studies of biologic discontinuation in rheumatoid arthritis (RA).

Methods We found 403 studies on PubMed, and included nine published papers and five abstracts from scientific meetings. We used a structured extraction form to collect information regarding study design and outcome (failure) definition.

Results Three types of studies were found: randomised controlled trials, long-term extension studies of clinical trials and prospective discontinuation studies. The largest study had 196 subjects in the discontinuation arm. Most studies allowed concomitant use of non-biologic drugs at biologic discontinuation. Heterogeneity was also found in the failure definition. Although all studies used measures of disease activity, the threshold for failure and the time point of assessment differed among studies. Few studies incorporated changing use of non-biologic drugs or glucocorticoids into the failure definition.

Conclusions Although many studies have examined the outcome of biologic discontinuation, they have all been relatively small. Typical practice studies from registries may add important information but will likely need to rely on a broader failure definition.

  • biologic DMARDs
  • remission
  • discontinuation
  • rheumatoid arthritis

Statistics from



Aggressive treatment strategies utilising biologic disease-modifying anti-rheumatic drugs (DMARDs) have revolutionised the treatment of rheumatoid arthritis (RA). Remission or low disease activity is now a realistic goal for most patients. After this disease state is achieved, it remains unclear whether biologic DMARDs should be continued indefinitely or discontinued.

Many studies have evaluated biologic discontinuation. However, comparing results from these studies is difficult due to heterogeneous designs and non-uniform definition for failing biologic discontinuation (‘failure definition’). Also most studies have focused on RA patients enrolled in biologic trials or standardised protocols, which may not reflect typical practice. To address these issues, rather than focusing on the study results, we evaluated the study designs and outcome (failure) definitions used in previous studies of biologic discontinuation.


Literature search

We conducted a systematic review of the literature. PubMed was searched for the following terms: (rheumatoid OR ‘Arthritis, Rheumatoid’ [MeSH]) AND (TNF OR ‘tumor necrosis factor’ OR abatacept OR adalimumab OR anakinra OR certolizumab OR etanercept OR golimumab OR infliximab OR rituximab OR tocilizumab) AND (withdrawal OR discontinuation OR stop). The search period was limited to 1998 (the year when the first biologic DMARD was approved) to 6 April 2013. So that recent studies would be included, abstracts presented at the 2010–2012 meetings of the American College of Rheumatology and the European League Against Rheumatism were also searched. Only material available in English was considered.

The titles of all identified studies were reviewed by one author (KY). The exclusion criteria included (figure 1): studies that did not use biologics, animal or in vitro studies, non-RA studies, reviews without primary data, studies not focusing on the discontinuation of biologic DMARDs, studies that examined biologic discontinuation for side effects, studies that did not report outcomes after biologic discontinuation for good disease control, studies with less than five subjects, paediatric studies, and meeting abstracts without sufficient publicly available details.

Figure 1

Selection of studies. Literature published after approval of the first biologic disease-modifying anti-rheumatic drug and meeting abstracts in the past 3 years were searched. Steps are explained in the Methods section of the text. ACR, American College of Rheumatology; EULAR, the European League Against Rheumatism; RA, rheumatoid arthritis.

Data extraction

Data were extracted for study characteristics We focused specifically on two aspects of the methods: study design and outcome (failure) definition. Each study's study design was assessed for the sample size of the discontinuation arms, study type, the presence of controls who continued biologic DMARDs, focus on early arthritis, requirements before biologic discontinuation, concomitant non-biologic DMARD and glucocorticoid use. We also determined the outcome (failure) definition used to assess failure after biologic discontinuation. This included disease activity thresholds, when and how often the outcome was assessed, and composite failure definition, such as biologic reuse, non-biologic treatment modification and radiographic outcomes. The outcome of the proportion of patients free of failure at 12 months (6 or 7 months’ result if 12 months’ result was not available) was also described.


Literature search results

We initially identified 403 abstracts in PubMed (last searched on 6 April 2013). After exclusion criteria were applied (see figure 1), nine studies were considered relevant. Additionally, after exclusion of three published studies, five abstracts had sufficient details to be included, resulting in a total of 14 studies.1–14

Study design

The studies were from Europe (n=8),1 ,2 ,5 ,6 ,8 ,10 ,12 ,14 Japan (n=4),3 ,7 ,11 ,13 and the USA and Europe (n=1).9 The sample sizes in the biologic discontinuation arms varied from 9 to 196. Three types of studies were identified (table 1): randomised controlled trials (‘RCT’ in table 1),1 ,2 ,5 ,9 in which discontinuation was randomised; long-term extension of trials (‘LTE’ in table 1),3 ,6 ,10 ,14 and single arm prospective studies of discontinuation (‘DC study’ in table 1)7 ,8 ,11–13 in which patients were prospectively recruited for biologic discontinuation.

Table 1

Comparison of study designs of biologic discontinuation studies

Five studies had control arms in which patients with good disease control were kept on biologic treatment.1–3 ,5 ,9 Four studies randomised discontinuation,1 ,2 ,5 ,9 and three used placebo in place of discontinued biologic treatment and provided at least double blinding.1 ,5 ,9 Four studies focused on early RA patients with less than 1 year of disease,6 ,9 ,10 ,14 while the other studies were conducted on more established RA patients.

Biologic DMARDs were discontinued after pre-specified duration of use in two studies.6 ,14 All other studies used disease activity lower than a specific cut-off as the enrolment criterion. However, the disease activity measures and thresholds varied by study. In one of the studies, biologic discontinuation was assessed as part of a biologic dose reduction protocol,8 that is, biologic DMARDs were tapered after 6 months of good disease control until discontinuation or failure occurred, and failure of both tapering and discontinuation were reported. Another study had a similar tapering protocol, but only discontinuation was reported.10 Two studies addressed maintenance treatment with half-dose biologic DMARDs as well as discontinuation.1 ,5

Use of concomitant non-biologic DMARDs and glucocorticoids at the time of biologic discontinuation varied. Some studies required patients to be off glucocorticoids,7 ,9 ,10 ,13 ,14 while other studies allowed small doses of glucocorticoids (prednisolone-equivalent, <10 mg/day1 ,2 ,6 or <5 mg/day11 ,12). In general, few details were provided about concomitant treatment in many of the studies.

Outcome (failure) definition

After discontinuation of biologic DMARDs, all RA studies used disease activity thresholds as failure criteria (table 2). The cut-off for failure was either low disease activity or remission, but three studies used a relative definition either by itself8 or in combination.2 ,5 Follow-up duration and time of assessment also varied. Most non-interventional studies (LTE study type) defined failure as an increase in disease activity above the threshold at any time (time to failure analysis), while most interventional studies (RCT and DC studies) assessed disease activity at a pre-specified time,2 ,3 ,5–7 ,9 ,14 for example, 6 months after discontinuation.6

Table 2

Comparison of outcome (failure) definitions of biologic discontinuation studies

Biologic DMARD reuse was regarded as failure in all studies except for four studies in which reuse was not allowed.1 ,6 ,9 ,14 In most studies, reuse was according to protocol, that is, reuse was allowed only after treatment was labelled as failed by the main definition.2 ,4 ,5 ,7 ,8 ,10 ,12 Treatment changes in non-biologic treatments at the discretion of the treating physician (increased doses or new introduction of non-biologic DMARDs and glucocorticoids) were used as failure definition in some studies,5 ,7 ,11 but only two studies reported changes in these treatments in detail.3 ,8 Two studies used radiographic progression as part of the failure definition,9 ,11 although the cut-offs chosen may be arguable considering they were lower than the general smallest detectable differences between readers.15

The actual outcomes of the studies varied widely (table 2).


We reviewed studies on biologic discontinuation to examine their designs. We found heterogeneity across studies, particularly in study design and outcome (failure) definition. Study enrolment was usually based on disease activity lower than a given threshold, but this threshold differed across studies. The failure definition was consistent in primarily using disease activity measures. However, concomitant non-biologic treatment at discontinuation and treatment changes after discontinuation were not sufficiently reported in most of the studies, and few clearly defined changes in non-biologic treatments as failure. The ‘success rates’ during the first year varied from 0% to 80%, supporting the substantial heterogeneity noted as regards study methods. In fact, this is precisely why we focused this systematic review on study methods and not on results. While we believe the results of these studies to be very important, the methods need critical appraisal before how the results should be appropriately interpreted is considered.

To collect comparable information, the use of standardised enrolment and failure definition is essential. It is only very recently that RCTs which include biologic continuation control arms have been presented. These are very useful for studying the comparative effectiveness of biologic discontinuation in comparison with biologic continuation. However, such trials apply strict inclusion criteria, creating study cohorts that may not be generalisable. Thus, observational data from registries of typical practice could potentially contribute more clinically relevant information.

There are particular obstacles to registry studies. An increase in disease activity may occur between study visits but not be present at the subsequent study visit, making it difficult to define ‘failure’ only by disease activity at study visits. This ‘interval failure’ issue may be partly addressed by including treatment modification as part of composite failure definition (disease activity increase or treatment modification as failure). This approach is in agreement with the preliminary consensus definition of RA flare proposed by OMERACT, which described flare as an increase in disease activity that results in treatment modification.16 However, as there is no consensus on what treatment changes should be considered clinically significant, there is inherent subjectivity in this definition.

In conclusion, the currently available studies show heterogeneity in designs, and suggest need for standardised definitions of biologic discontinuation in good disease control and failure of biologic discontinuation. Standardised failure definition incorporating changes in treatments in addition to the main criteria by disease activity may allow information from currently available clinical practice registries to be utilised, thereby providing information more relevant to typical clinical practice.



  • Handling editor Tore K Kvien

  • Contributors All authors contributed to the design and execution of the study, interpretation of data and preparation of the manuscript. KY is the guarantor.

  • Funding KY's time at Brigham and Women's Hospital is funded by a scholarship from Kameda Medical Center. DHS receives salary support from NIH-K24AR055989. SCB was supported by a grant of the Korea Healthcare Technology R&D Project, Ministry for Health and Welfare, Republic of Korea (A102065)

  • Competing interests YKS and KM declare no competing interests. KY received honoraria and served as an instructor at a musculoskeletal ultrasonography workshop sponsored by Abbott Japan. AK has conducted sponsored clinical research for Abbott, Janssen, Amgen, BMS, Roche and UCB. SCB has received research grants from Abbott, Bristol Myers Squibb Pharmaceutical, Eisai, GlaxoSmithKline, MSD and Pfizer. MEW has received grant support from Bristol-Myers Squibb and serves in consultant roles for Amgen, Abbott, Janssen, Bristol-Myers Squibb, Roche and UCB. MK has received speaking fees and/or honoraria from Santen Pharmaceutical, Mitsubishi Tanabe Pharma, Pfizer and Abbott Japan. ST has received research grants from Pfizer Japan, Eisai and Chugai Pharmaceutical. DHS receives salary support from institutional research grants from Eli Lilly, Amgen and CORRONA, royalties from Uptodate, and serves in unpaid roles in studies funded by Pfizer and Novartis.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.