Article Text

Concise report
Abatacept or tocilizumab after rituximab in rheumatoid arthritis? An exploratory study suggests non-response to rituximab is associated with persistently high IL-6 and better clinical response to IL-6 blocking therapy
  1. Sudipto Das1,2,
  2. Edward M Vital1,2,
  3. Sarah Horton1,2,
  4. Domini Bryer2,
  5. Yasser El-Sherbiny1,2,3,
  6. Andrew C Rawstron4,
  7. Frederique Ponchel1,2,
  8. Paul Emery1,2,
  9. Maya H Buch1,2
  1. 1Leeds Institute of Rheumatic and Musculoskeletal Medicine, University of Leeds, Leeds, UK
  2. 2NIHR Leeds Musculoskeletal Biomedical Research Unit, Leeds Teaching Hospitals NHS Trust, Leeds, UK
  3. 3Clinical Pathology Department, Faculty of Medicine, Mansoura University, Mansoura, Egypt
  4. 4Haematological Malignancy Diagnostic Service, Leeds Teaching Hospitals NHS Trust, Leeds, UK
  1. Correspondence to Dr Maya H Buch, Leeds Institute of Rheumatic and Musculoskeletal Medicine, Chapel Allerton Hospital, Chapeltown Road, Leeds LS7 4SA, UK; m.buch{at}


Objectives To evaluate the efficacy and safety of two different targeted approaches—abatacept or tocilizumab—after rituximab therapy in rheumatoid arthritis, and to explain observed difference in efficacy using blood and synovial studies of interleukin 6 (IL-6) and B cells in patients receiving rituximab therapy.

Methods Consecutive series of patients who had discontinued rituximab therapy owing to inefficacy or toxicity were treated with abatacept (n=16) or tocilizumab (n=35). Clinical response and reasons for discontinuation were evaluated. Serial blood and synovial samples were obtained from a group of 57 and 25 rituximab-treated patients, respectively, and were analysed for B cells and IL-6 using flow cytometry, immunohistochemistry and quantitative real-time PCR.

Results In the abatacept group, mean (SEM) Disease Activity Score in 28 joints calculated using the erythrocyte sedimentation rate (DAS28-ESR) reduced from 5.69 (0.42) at baseline to 4.94 (0.44) at 6 months (p=0.12). In the tocilizumab group: mean (SEM) DAS28- ESR reduced from 5.75 (0.21) at baseline to 3.28 (0.26) at 6 months (p<0.001). This was paralleled by a significant swollen joint count reduction in the tocilizumab (5.47 (0.70) to 2.70 (0.61), p=0.033), but not abatacept (6.23 (1.3) to 4.15 (1.2), p=0.26), group. In the synovium, despite complete depletion of B cells in 19/22 patients, IL-6 mRNA expression was not significantly reduced after rituximab. Blood B cell numbers remained low 12 months after rituximab. Serum IL-6 was raised at baseline and significantly higher in rituximab clinical non-responders (p=0.035) than responders. A significant reduction in serum IL-6 was seen in rituximab clinical responders (p=0.005) but not in non-responders (p=0.237).

Conclusion In patients with rheumatoid arthritis for whom rituximab therapy failed despite adequate B cell depletion, IL-6-directed therapy might be a more logical and effective treatment choice than T cell costimulation blockade. Further controlled studies investigating other possible mechanisms are needed to validate these initial findings.

Statistics from

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.


Multiple biological disease-modifying antirheumatic drug (bDMARD) therapies are licensed for the management of rheumatoid arthritis (RA), and various guidelines and recommendations on how these may be used have been published.1–3 These have, however, illustrated a knowledge gap on the optimal sequencing of bDMARDs.

Rituximab is commonly used as a second-line bDMARD after initial tumour necrosis factor-inhibitor (TNFi) failure.4–6 If rituximab therapy is not successful, subsequent options include returning to an alternative TNFi, or employing a different bDMARD class such as abatacept—a fusion protein of the Fc region of human IgG1 and CTLA4 extracellular domain that blocks T cell costimulation; or tocilizumab—a humanised monoclonal antibody against the IL-6 receptor. Both agents have demonstrated efficacy in patients with RA for whom methotrexate and a TNFi have failed.7–10

Interleukin 6 (IL-6) and costimulation both have important roles in B cell function in RA pathogenesis.11 This may have implications for the suitability of subsequent targeted treatment, since alternative bDMARDs after rituximab are usually started before complete B cell repopulation. Although abatacept and tocilizumab are often used they have not been formally studied in patients with low or undetectable B cells. One previous report of the use of alternative bDMARDs after rituximab has been published, but most of the patients in that study received a TNFi.12 One other small study evaluated safety but not efficacy.13

In this study we (i) report efficacy and safety of two series of consecutive patients who received either abatacept or tocilizumab after inadequate response or toxicity to rituximab; and (ii) report on synovial and blood B cell and IL-6 levels in a group of patients after rituximab therapy for RA to evaluate whether certain targeted approaches may be more suitable after B cell depletion therapy.


Abatacept or tocilizumab after rituximab: clinical observational study

To assess efficacy of follow-on bDMARDs after rituximab, we analysed clinical response of all patients with active RA who had received abatacept or tocilizumab after two infusions of 1000 mg rituximab between September 2007 and July 2011 in a single-centre, prospective, observational database with follow-up data until December 2011 or end of treatment. Each drug was prescribed according to the licensed dose: tocilizumab 8 mg/kg 4-weekly and abatacept <60 kg: 500 mg; ≥60 kg ≤100kg: 750 mg; >100 kg: 1000 mg on days 1, 15, 29 and then every 28 days.

Consecutive series of patients (whose treatment was mainly allocated according to national algorithm/funding restrictions rather than on clinical grounds only) were recruited for this observational study.

Synovial and blood study of non-response to rituximab

To assess the effect of rituximab on IL-6, we analysed serum and synovial biopsy samples from patients with RA who had received rituximab. Serial serum samples were obtained at 0 (n=57), 3 (n=44) and 6 (n=49) months. Synovial biopsy samples were obtained from 25 patients at baseline and 22 patients at 16 or 26 weeks. Clinical outcomes and B cell data have been reported as part of a previous study.14 Normal range for B cells was evaluated in a group of 24 healthy controls and 95 patients with active RA receiving methotrexate.

These studies were performed under local ethics committee approval and informed consent was obtained accordingly.

Outcome measures

Baseline characteristics and previous treatments were recorded. Because tocilizumab may have a selective effect on C-reactive protein (CRP)15 that is not seen with abatacept we chose to use the Disease Activity Score in 28 joints (DAS28) calculated using the erythrocyte sedimentation rate (ESR) at baseline and 6 months as our main efficacy measure and have presented tender and swollen joint count data separately. EULAR response, rates of remission (DAS28<2.6) or low disease activity (DAS28<3.2), duration of treatment, adverse events were recorded. Details of assays for blood and synovial B cells and IL-6 are provided in an online supplementary file.

Statistical analysis

Serum IL-6 levels and synovial IL-6 gene expression were compared between groups using the Mann–Whitney U test and within groups using the Wilcoxon signed ranks test. Association of serum IL-6 with DAS28 and CRP was tested using Pearson's product moment correlation coefficient. Paired t tests were used to compare change in DAS28 for each group. p Values <0.05 were considered significant. SPSS V.19 was used.


Abatacept or tocilizumab after rituximab: clinical observational study

Baseline characteristics

Fifty-one patients were evaluated, of whom 40 (78.4%) were female and 11 (21.6%) were male; median (range) age was 60.2 (24.4–81.7) and disease duration 12 (4–56) years.

Baseline characteristics of each group are summarised in table 1. These were comparable, although a longer disease duration and higher swollen joint count was seen in the abatacept group and higher CRP in the tocilizumab group.

Table 1

Baseline characteristics at time of initiation of abatacept or tocilizumab

Prior treatment

Ten (19.6%) patients had received one cycle of rituximab, 21 (41.2%) received two cycles and 20 (39.2%) three or more cycles. Of these, 21 (41.2%) were primary non-responders to rituximab, 20 (39.2%) had secondary non-response. The remaining 10 (19.6%) changed treatment owing to toxicity. Before rituximab, 7/51 (13.7%) patients had been primary non-responders to a TNFi, 24/51 (47.1%) had been secondary non-responders and 17/51 (33.3%) had toxicity to a TNFi. Two patients had not received a prior TNFi and one patient’s response to a TNFi was not clearly definable.

Clinical response

Clinical response data are summarised in table 2.

Table 2

6-Month clinical details of abatacept and tocilizumab treatment groups

Of the 16 patients treated with abatacept, three patients discontinued before the 6-month assessment (see ‘Safety’ section); of the remaining 13 patients, 6-month data were available for 11. Twelve (75%) patients discontinued treatment while four (25%) patients continued to receive abatacept. Median (range) duration of response was 8.5 (0–42) months. The most common reason for discontinuation was toxicity in six (37.5%) patients with secondary non-response in three (18.8%) patients.

Of the 35 patients in the tocilizumab group 6-month data were available for 30 patients. Twenty-six (74.3%) patients continued to receive tocilizumab and nine (25.7%) discontinued. The median (range) duration of response was 7 (0–24) months. The most common reason for discontinuation was again toxicity in five (14.3%) patients, followed by primary non-response in three (8.6%) patients.


In the abatacept group, of the six (37.5%) patients who discontinued owing to toxicity, three experienced infusion reactions, two had a lupus-like rash and one had recurrent lower respiratory tract infections.

In the tocilizumab group, of the five (14.3%) patients who discontinued owing to toxicity, two had infusion reactions, one had a lupus-like rash, one developed psoriasis de novo, one had culture-negative endocarditis; one that had been switched to tocilizumab owing to hypogammaglobulinaemia on rituximab had a continued decline in IgG. In this last patient, IgG recovered when tocilizumab was stopped; tocilizumab was restarted at 4 mg/kg without further hypogammaglobulinaemia.

Synovial and blood study of rituximab non-response

Synovial characteristics of rituximab non-response

Results of synovial B cell and IL-6 studies are shown in figure 1.

Figure 1

Blood and synovial B cell and interleukin 6 (IL-6) studies in patients with active rheumatoid arthritis (RA) receiving rituximab. (A) CD19+ cells (B cells) in synovium as measured by immunohistochemistry and digital image analysis. (B) Synovial IL-6 mRNA gene expression as measured by Taqman low-density array of RNA extracted from synovial biopsy samples. (C) Peripheral blood B cells as measured by highly sensitive flow cytometry in EULAR responders and non-responders. The dotted lines marked HC and RA show the median B cell count for rituximab-naïve healthy controls and active RA patients respectively. (D) Serum IL-6 titre as measured by ELISA in EULAR responders and non-responders. *Significant p value.

After 16 or 26 weeks of rituximab, synovial B cells were completely depleted in most patients (figure 1A) with B cells detected in only 3/22 patients, and at only 4 cells/mm2 in two of these. In contrast, there was only a trend towards partial reduction in synovial IL-6 mRNA gene expression (figure 1B).

Blood characteristics of rituximab non-response

Total peripheral blood B cells remained low in EULAR responders and non-responders 6 and 12 months after rituximab compared with healthy controls or controls with active RA (figure 1C).

At baseline, serum IL-6 correlated with CRP as expected (R=0.612, p<0.001), but not with DAS28-ESR (R=0.242, p=0.089). Overall there was a significant but variable reduction in serum IL-6 after rituximab (mean percentage reduction 19%, 95% CI 0% to 54%, p=0.004). However, in EULAR non-responders compared with responders, serum IL-6 level was higher before rituximab (p=0.035) and was not suppressed by treatment, resulting in a significantly higher level at 6 months (p=0.009; figure 1D).


Although there is evidence for the efficacy of several bDMARDs with differing targets and mechanisms of action in RA, data on optimal sequencing of bDMARDs are limited; with a notable paucity of data on bDMARDs after rituximab therapy. Studies evaluating characteristics of non-response, such as our report, provide insights that may guide the most appropriate choice of follow-on treatments.

We previously showed that most cases of rituximab non-response were related to ineffective B cell depletion.14 This led to the hypothesis that such patients would then respond well to a further cycle of rituximab, which we showed to be effective for a large proportion of patients.14 With further experience we now observe a group of patients with continued active disease despite complete B cell depletion and after multiple cycles of rituximab, suggesting truly B cell-independent disease.

In the clinical study we observed that IL-6 targeted therapy was more effective than costimulation blockade in patients who had not responded to, or had lost an earlier response to, rituximab. In light of these response profiles, we measured synovial and blood B cells and IL-6 in a rituximab-treated cohort and showed that this clinical observation may be linked to high IL-6 activity. Despite complete depletion of B cells in the synovium, continued IL-6 expression was seen, presumably from the many well-described non-B cell sources.16 Systemically, baseline serum IL-6 levels were higher in clinical non-responders to rituximab than in responders, and were not suppressed after treatment at 6 months.

Hence, rituximab non-response was characterised at 6 months by persistently low synovial and blood B cells, but persistently raised synovial and blood IL-6. These observations suggest that targeting IL-6 would be a more logical follow-on bDMARD strategy than costimulation blockade. B cells function as antigen-presenting cells in RA; our previous study showed that the most important effect in the synovium after abatacept therapy was reduction of B cell number.17 ,18 Hence B cell function and T cell costimulation may be co-dependent mechanisms, limiting the utility of costimulation blockade in patients with active disease despite B cell depletion.

We directed our investigation of response principally towards the relationship between B cell depletion and IL-6. This study cannot explain any specific mechanism by which B cell depletion leads to reduction in IL-6 (apart from reflecting the generic consequences of suppression of inflammation). However, the dissociation we demonstrate between B cell depletion and sustained high IL-6 production in a subgroup of non-responders suggests that the relationship between B cells and IL-6 varies between different groups of patients. We have not investigated, and cannot exclude, other mechanisms of non-response. Analysis of antigen presentation by B cells in rituximab response and non-response would require functional work that was beyond the scope of this study, but warrants further investigation.

Observational studies such as this have clear limitations, particularly ‘confounding by indication’ and a formal randomised trial is required to compare efficacy of these treatment sequences definitively. Nevertheless, this initial exploratory evaluation suggests that there may be differences in clinical response with a potential mechanistic explanation and emphasises the need to design studies that investigate optimal use of bDMARDs with parallel investigation into their mechanism of action and predictors of response.


EMV is a National Institute for Health Research (NIHR) clinical lecturer. MHB is a NIHR clinician scientist.


Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

    Files in this Data Supplement:


  • Handling editor Tore K Kvien

  • SD and EMV contributed equally.

  • Contributors All authors were involved in either conception and design, or analysis and interpretation of data. SD, EMV and MHB drafted the article and SD, EMV, FP, PE and MHB revised it critically for important intellectual content. All authors gave final approval of the version to be published.

  • Funding Roche provided rituximab for the synovial and blood study. A NIHR doctoral research fellowship (RTF/01/097) funded EMV and the laboratory work. MHB is funded by a NIHR Clinician Scientist Award (CS/08/08/04). The observational abatacept and tocilizumab-treated cohorts is part of the Leeds standard patient care biologics database and does not receive any pharma support.

  • Competing interests EMV has received honoraria from Roche. PE and MHB have received honoraria for presentations and/or consultation from Bristol-Myers Squibb, Roche and Chugai.

  • Ethics approval Leeds West and Glasgow ethics committees (Eudract 2005-006674-43).

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