Statistics from Altmetric.com
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.
As in previous years, the consensus group to consider the use of biological agents in the treatment of rheumatic diseases met during the 13th Annual Workshop on Advances in Targeted Therapies in March 2012. The group comprised rheumatologists from a number of universities among the continents of Europe, North America, South America, Australia and Asia.
Pharmaceutical industry support was obtained from a number of companies for the annual workshop itself, but these companies had no part in decisions about the specific programme or about the academic participants attending this conference. Representatives of the supporting sponsors participated in the initial working groups to supply factual information. The sponsors did not participate in the drafting of the consensus statement.
This consensus was prepared from the perspective of the treating physician.
In view of the new data for abatacept, B cell-specific agents, interleukin 1 (IL-1) antagonists, pegloticase, tocilizumab (TCZ) and tumour necrosis factor α blocking agents (TNFi), an update of the previous consensus statement is appropriate. To enable ease of updating, the 2011 (data from March 2010 to January 2011 updates are incorporated into the body of the consensus, while 2012 updates (February 2011 to January 2012) are separated and highlighted. The consensus statement is annotated to document the credibility of the data supporting it as much as possible. This annotation is that of Shekelle et al and is described in online supplementary appendix 1.1 We have modified the Shekelle annotation by designating all abstracts as ‘category D evidence’, whether they describe well-controlled trials or not, as details of the study were often not available in the abstracts. Further, the number of possible references has become so large that reviews are sometimes included. If they contain category A references, they will be referred to as category A evidence.
The 192 rheumatologists and bioscientists who attended the consensus conference were from 21 countries and were selected for their expertise in the use of biological agents for the treatment of rheumatic diseases. The number of attendees and participants was limited so that not everyone who might have been interested could be invited. All participants reviewed a draft document developed by the coauthors, based on a review of all relevant clinical published articles relating to abatacept, rituximab (RTX) and belimumab (B cell specific therapy), IL-1 blocking agents (including anakinra, canakinumab and rilonacept), pegloticase, TCZ and TNFi (five agents). The draft was discussed in small working groups. The revisions suggested by each group were discussed by all participants in a final open session, and this led to a final document, representing this updated consensus statement.
It is hoped that this statement, which is based on the best evidence available at this time, and is modified by expert opinion, will facilitate the optimal use of these agents for patients with conditions approved by the Food and Drug Administration (FDA) or European Medicines Agency (EMA) for clinical use.
The formatting of this document is arranged as follows: general introduction and general statements followed, in alphabetical order, by each biological agent arranged by generic name or general mechanism (when appropriate). Within each biological agent, the data are arranged by indication and the information is arranged according to clinical use such as dosing, time to response, etc. Some combinations of indications occur; appropriate safety is included after clinical use, also in alphabetical order.
Individual patients differ in the clinical expression and aggressiveness of their disease, its concomitant structural damage, the effect of the disease on their quality of life (QoL) and the symptoms and signs engendered by their disease. They also differ in their risk for, and expression of, side effects to drugs. All these factors must be examined when considering biological treatment for a patient, as well as the toxicity of previous and/or alternative disease-modifying antirheumatic drug (DMARD) use.
As increasing evidence has accumulated on the efficacy and clinical use of biological agents for the treatment of gout, psoriatic arthritis (PsA) and ankylosing spondylitis (AS), these diseases will be discussed separately from rheumatoid arthritis (RA). Adverse reactions, unless disease-specific, however, will remain combined for all indications.
In general, in RA, when response to treatment is being measured or when patients are followed up over time, validated quantitative measures for clinical trials can be used. These include the Disease Activity Score (DAS), Simplified Disease Activity Index (SDAI), Clinical Disease Activity Index (CDAI), Routine Assessment of Patient Index Data (RAPID), Health Assessment Questionnaire Disability Index (HAQ-DI), visual analogue scales (VAS) or Likert scales of global response or pain by the patient or global response by the doctor. Other validated measures for individual patient care, joint tenderness and/or swelling counts, and laboratory data may all be used and may be appropriate measures for individual patients (category A/B evidence2–8). The doctor should evaluate a patient's response using one of the above instruments to determine the patient's status and change.
For PsA, measures of response such as joint tenderness and swelling, enthesitis and dactylitis, global and pain response measures, functional indices and acute phase reactants, both as single measures and as part of composite measures, have been used.3 ,5 ,9
For AS, measures such as the Bath Ankylosing Spondylitis Disease Activity Index (BASDAI) and the Bath Ankylosing Spondylitis Functional Index (BASFI) are used; they have been used in clinical trials, but have not been validated for routine clinical practice (category C evidence2). In this disease, clinical measures such as joint tenderness and swelling, spinal motion, global and pain response measures, functional indices and acute phase reactants have been used and are validated.
Pregnancy remains a controversial topic when using biological agents for rheumatic diseases. For all but the TNFi and IL-1 blocking agents, there are too few data to draw any conclusions. Since a lack of association is extremely difficult to prove, no biological agents can be assumed to be safe. In the absence of such data, this recommendation depends on the US FDA designation. Abatacept, RTX and TCZ have a category C designation, while TNFi and IL-1 blocking agents are designated as category B (see specific drugs).
The appropriate use of biological agents require doctors experienced in the diagnosis, treatment and assessment of RA, PsA, AS and other rheumatic diseases. The physicians need to be aware of the data regarding long-term observations of efficacy and toxicity, including cohort studies and data from registries. Because biological agents have adverse effects, patients or their representatives should be provided with information about potential risks and benefits so that they can give informed consent for treatment. For ease of references, the biological agents are listed in alphabetical order: abatacept; B cell therapy; belimumab; IL-1 blocking agents; pegloticase; TCZ and TNFi.
Caveats and comments
The details of efficacy measures among the biological agents depend on multiple factors such as previous experience of DMARDs or biological agents, comorbidities, concomitant drugs and disease activity. Lack of space prevents detailed discussion of these factors, so that, drugs will be designated as either ‘effective’ or ‘ineffective’ without further gradations, depending on statistical and other differences. The only place in this consensus statement where words such as ‘mild, moderate, severe’ applied to the disease state are included is in the ‘indication’ sections for each drug or drug class.
Abatacept (blocking costimulation of T cells at the CD28–CD80/86 binding site)
One agent which modulates T cell activation (abatacept) has been approved in the USA, Europe, Japan and other countries.
In the USA, abatacept is approved for treatment of moderate-to-severe active RA as monotherapy or with DMARDs in moderate to severe adult RA. It can be used as first-line treatment or after an adequate trial of methotrexate (MTX) or another effective DMARD. In early RA, abatacept has been approved in North America in MTX-naïve patients in combination with MTX (category A evidence6–8 ,10 ,11). In the EU, abatacept had been approved by the EMA in combination with MTX for moderate-to-severe active RA after an inadequate response to one or more DMARDs, including MTX or a TNFi (category C evidence12).
Abatacept may be given when the next dose of a biological agent such as a TNFi would normally be given (category B evidence13). Abatacept has been used with MTX and other DMARDs (category A/B/C evidence 10 ,11 ,14–17).
Juvenile idiopathic arthritis
In the USA, abatacept is recommended for treatment of active polyarticular juvenile idiopathic arthritis (JIA) in patients aged ≥6 years as monotherapy or with MTX. In Europe, abatacept in combination with MTX is indicated for the treatment of moderate to severe polyarticular JIA in patients aged ≥6 years who have insufficient response to other DMARDs, including at least one TNFi (category A/B evidence17–19).
The adult dosing regimen is 500 mg, 750 mg or 1000 mg given intravenously at 0, 2 and 4 weeks, then every 4 weeks (category C evidence20). Subcutaneous dosing (125 mg weekly) has been approved by the FDA for RA (category A evidence21) with or without intravenous loading.
In an open-label, single-arm trial, clinical efficacy was maintained with no increased safety concerns when patients with RA were switched from more than 4 years of intravenous abatacept to subcutaneous administration for the next 3 months (category C evidence22).
An open trial with abatacept (10 mg/kg) failed to show efficacy in 15 TNFi-naïve patients with AS and also in 15 patients with AS for whom TNFi had failed (category C evidence23).
Overall immunogenicity to subcutaneous abatacept was low and consistent with intravenous abatacept. It was not significantly affected by a 3-month interruption/reintroduction and neither did the stop–start schedule have any major impact on efficacy and safety (category C evidence24).
Time to response
Some patients respond to abatacept within 2–4 weeks, using the American College of Rheumatology (ACR) response criteria. Most adult patients respond within 12–16 weeks of starting treatment. It may take longer in children; see below (category A evidence19 ,25). Patients continue to improve for up to 12 months (category A evidence7 ,8 ,26). Of the children for whom TNFi had previously failed, 11% obtained a clinically meaningful ACR20 response (category A evidence18 ,19).
Pharmacoeconomic and quality of life
Quality of life: QoL and other patient-related outcomes such as sleep, fatigue and/or activity improve with abatacept (category A evidence29).
In some patients response with abatacept was maintained for up to 5 years in long-term open-label extension studies (category C/D evidence27–31) in both TNF-incomplete responders (TNF-IR) and MTX-incomplete responders (MTX-IR) (category D evidence32) and for up to 2 years in MTX-naïve patients with early RA (category C evidence33).
In meta-analyses, abatacept, TCZ, RTX and TNFi generally showed similar efficacy and safety (category A evidence34).
Comparison with TNFi and other biological agents
In a randomised, double-blind, placebo-controlled trial, the clinical efficacy of abatacept (10 mg/kg) was similar to that of infliximab (3 mg/kg); there were fewer serious adverse events and serious infections in the abatacept-treated patients (category A evidence35).
The nationwide Danish DANBIO registry was analysed to compare TCZ with abatacept in patients with RA, most of whom (>90%) were TNFi-IR. Good and equivalent responses were noted to both abatacept and TCZ in routine care (category C evidence36).
Switching to abatacept
Abatacept was effective after TNFi or RTX (category D evidence37 ,38). The efficacy and safety during 6 months of treatment with abatacept in TNF-IR was similar with or without wash-out (category C evidence12). In MTX-IR, TNFi were more effective than abatacept (category A evidence34).
Structural changes: Abatacept in combination with MTX inhibits or reduces radiographic progression in RA in MTX-IR and in those with very early inflammatory arthritis (category A/B/C evidence27 ,33 ,39–41).
Juvenile idiopathic arthritis
Abatacept is given as intravenous infusions of 10 mg/kg for weight <75 kg, 750 mg for weight of 75–100 kg and 1000 mg for weight >100 kg. All regimens are given intravenously at 0, 2 and 4 weeks, then every 4 weeks) (category A evidence19). Subcutaneous abatacept has not yet been studied in children.
Psoriatic arthritis: An open-label trial of patients with PsA demonstrated a low degree of efficacy in the arthritis component and little benefit for the psoriatic skin lesions (category A evidence42).
The incidence of autoimmune disorders remained stable over time in the intravenous and subcutaneous abatacept clinical trial database (category D evidence43).
Antibodies against subcutaneous abatacept, an ELISA assay, occurred in 8% of patients but none were persistent. There was a tendency to an increase in autoantibodies if one to two doses of abatacept were missed. No relationship was found between the antibodies and clinical response or adverse events (category A evidence44).
Combining biological agents
Concurrent treatment with abatacept and a TNFi is not recommended. Four patients with systemic-onset JIA refractory to anakinra alone have been treated effectively with a combination of abatacept and anakinra without serious adverse events during 8–17 months of follow-up (category D evidence45).
All patients in abatacept intravenous and subcutaneous phase 3 trials were screened for tuberculosis (TB) with a tuberculin skin test (TST). They were still included if the screen was positive and they were treated for latent TB. To date, a number of cases of TB have been seen in the clinical trial programme and during extension studies (category C/D evidence46 ,47). It is appropriate to screen patients considered for abatacept treatment for TB according to local practice.
Patients with chronic obstructive pulmonary disease (COPD) treated with abatacept had more serious lower respiratory tract infections than patients treated with placebo; therefore its use in patients with RA and COPD should be undertaken with caution.
In comparison with placebo, the incidence of serious infections with abatacept was increased in clinical trials at 12 months, but not in a meta-analysis pooling of 6- and 12-month safety data (category A evidence46 ,47). In a review of clinical trial data, the incidence of hospitalisations for infections remained stable for up to 5 years, and the incidence was not significantly different in the long-term extension as compared with the blinded phase of clinical trials (3.0 vs 2.1/100 000 patient-years) (category A evidence47).
A recent observational study appeared to show that serious infections were fewer with abatacept than infliximab, etanercept, adalimumab or RTX (category B evidence48).
In the large recent Abatacept in Inadequate responders to Methotrexate (AIM) trial, rates of serious infection were 4.2/100 patient-years within the study's double blind period, and 3.2/100 patient-years within the cumulative, 3-year open-label extension (category B evidence49).
Of 1879 patients from the subcutaneous abatacept clinical trials database, with 3086 patient-years exposure (range 2–56 months), the incidence of serious infections was 1.94/1000 patient-years, with pneumonia, urinary tract infections and gastroenteritis being most common. This did not differ from the intravenous data, where the rate of serious infection events (SIEs) was 2.87/1000 patient-years (category D evidence50).
The CORRONA database documented a serious infection risk ratio of 0.68 (95% CI 0.48 to 0.96) for abatacept compared with infliximab (p<0.05). Other biological agents (adalimumab, etanercept and RTX) were similar to abatacept (category B evidence48).
For abatacept in combination with other biological agents, the rate of serious infections is 4.4% compared with 1.5% in controls (category C evidence16). The use of abatacept with TNFi is not recommended, as an increased incidence of serious infections was noted when the combination was used (category A evidence51 ,52). There are minimal data on the combination of abatacept and RTX.
The safety of abatacept treatment in patients with active hepatitis B virus (HBV) is unknown, although a small case series recently reported eight patients with HBV infection (six had latent disease) who were treated with abatacept. In the four patients receiving background antiviral agents no HBV reactivation occurred, while reactivation did occur in all four not receiving background antiviral agents (category D evidence53).
Injection site reactions: These occurred with an incidence of 2.22/100 patient-years (CI 1.74 to 2.82) with subcutaneous abatacept (category D evidence54).
Malignancies: A lymphoma occurred in a double-blind trial with abatacept versus none in the placebo group; four additional cases occurred in the open-label extension (cumulatively 5/8, 39 person-years), while an epidemiological overview showed no increase (category B/D evidence43 ,55). A comparison of intravenous and subcutaneous abatacept clinical trial data with national registries showed no increased rates of lymphoma, lung, breast, colorectal or total malignancies, although the control populations were not completely comparable (category D evidence30 ,43). Epidemiological experience in six RA cohorts shows no increased rate of solid malignancies compared with the control cohorts, although continued monitoring is necessary (category C evidence55).
Skin reactions: New-onset psoriasis occurred with an incidence of 0.29/100 patient-years (CI 0.15 to 0.56) in the subcutaneous clinical database (category D evidence56).
Vaccinations: There was a decreased response to flu, tetanus and pneumococcal vaccinations when abatacept was used in healthy volunteers (category C evidence57). Flu and pneumococcal vaccinations in patients with RA receiving abatacept were reduced, comparable to previous reports in patients with RA receiving MTX (category D evidence58).
On the basis of theoretical concerns, live vaccines should not be given while a patient is receiving abatacept or within 3 months of using abatacept.
The number of cases of pregnancy in patients using abatacept is too small for any definite conclusion to be drawn (see the general statement). According to the US FDA, this drug is considered category C, meaning ‘No human studies and animal studies either show risk or are lacking. However, potential benefits may justify potential risks’.
B cell therapy
Belimumab is an inhibitor of B-lymphocyte stimulator (BLyss) protein that is indicated for moderately active systemic lupus erythematosus (SLE). Subjects with active lupus nephritis or active central nervous system disease were excluded from the study protocols. A further study examining response of African American subjects is underway.
In two phase III trials, belimumab was compared with placebo, with a background of standard of care that allowed some adjustment in background medication (category A evidence59). Efficacy in the trials was judged by a new SLE responder index that has three components designed to document:
overall improvement in disease activity;
freedom from deterioration in any manifestation of SLE;
agreement with clinical judgement.
Response occurred in 38.8% of the standard of care plus placebo group, compared with 50.6% of the standard of care plus 10 mg/kg belimumab group (p<0.0001). This drug has been approved by the US FDA for use in mild–moderate SLE (category A evidence59).
Dosing and clinical use
Belimumab is given as 10 mg/kg intravenously in 5% dextrose/water or normal saline over 1 h. It should be given every 2 weeks for the first three doses and then every 4 weeks. As infusion reactions and hypersensitivity reactions can occur, histamine antagonist premedication is often given prophylactically. Patients should be observed for a least 1 h after each infusion, particularly early on in the course. Some reactions were seen 3–5 h after infusion (category A evidence59).
Quality of life
Different components of the Short Form-36 (SF-36) improved significantly in the BLyss-52 and BLyss-76 studies versus placebo (category A evidence59).
Data are available from the pooled phase 2/3 data over up to 76 weeks (category A evidence59). The death rate over the whole clinical programme was 0.56/100 patient-years for belimumab and 0.54/100 patient-years for placebo.
Serious adverse events occurred in just over 15% of the patients in both groups. Treatment-related adverse events resulting in discontinuation occurred in 6.5% (belimumab) and 7.1% (placebo) of the patients. The malignancy rate, also, did not differ between patients receiving placebo and belimumab-treated patients: 0.28/100 (0.03 to 1.03) patient-years for placebo versus 0.45/100 (0.27 to 0.58) patient-years for belimumab. Neither bacterial nor opportunistic infections occurred statistically more often in the patients receiving belimumab than the placebo patients (category A evidence60 ,61).
It is advisable to immunise before starting drug treatment, and live vaccines should not be used during treatment with belimumab.
In a summary of the pooled phase 2/3 data, belimumab was used for up to 6 years in some patients with SLE and continuation of treatment was documented for >5 years in 12.8% of the patients (category D evidence62 ,63).
As results in 1000 to 2000 patients during phase 2/3 studies represent relatively short-term exposure, long-term experience with larger numbers of patients using belimumab will be needed to fully define the adverse event profile of this drug.
RTX is a chimeric anti-CD20 monoclonal antibody, which was approved in 1997 for treatment of indolent CD20, B cell non-Hodgkin's lymphoma (NHL) and chronic lymphocytic leukaemia.
A consensus statement on the use of RTX in patients with RA has been published (category D evidence64).
RTX is approved by the FDA in the USA with MTX for the treatment of moderate-to-severe RA in patients who have had an inadequate response to at least one TNFi (category A/D evidence64–66; FDA and EMA label; category C/D evidence67–72). An indirect meta-analytic comparison, using 18 studies and one abstract, showed no differences among abatacept, RTX, TCZ and golimumab in TNF-IR patients with RA (category A evidence34). It may also be used when TNFi are not suitable (category D evidence73 ,74). RTX, in combination with MTX, has also been studied in MTX-naïve patients with RA, where it resulted in significantly improved clinical outcomes and a reduction in the progression of radiographic joint damage (compared with MTX alone). RTX may therefore be considered for use in MTX-naïve patients with RA.
One open-label trial and a case report in patients with PsA demonstrated a low degree of effectiveness in the arthritis component of PsA and possible benefit in psoriatic skin lesions (category D evidence75 ,76).
Antineutrophil cytoplasmic antibody-associated vasculitis: RTX is approved by the FDA for treating antineutrophil cytoplasmic antibody-associated vasculitis (category A evidence77). It is indicated for patients with newly diagnosed and relapsing non-life-threatening granulomatosis with polyangiitis (GPA, formerly known as Wegener's granulomatosis) and microscopic polyangiitis.
In a randomised controlled trial (RCT), including 197 subjects with GPA or microscopic polyangiitis, RTX, 375 mg/m2 on four consecutive weeks for induction of remission was not inferior to oral cyclophosphamide. Both arms included a tapering dose of corticosteroids. Remission was defined as no disease activity without any corticosteroids at 6 months (category A evidence77). RTX was more effective than oral cyclophosphamide for patients with relapsing disease, many of whom had received cyclophosphamide previously. There were no differences in adverse events.
After cyclophosphamide treatment, patients were treated with azathioprine while those who had been given RTX were assigned to placebo. At 18 months no differences were seen in the percentages of patients with persisting remission, nor in renal outcome or adverse events (category D evidence78). Open series showed that re-treatment is effective in patients with relapsing disease (category C evidence79).
A very small open uncontrolled trial with 2×1 g RTX failed to show efficacy in 10 TNFi non-responders and showed an Assessment of Spondylo-Arthritis International Society 20 (ASAS20) response in only five of 10 TNFi-naïve patients with active AS (category C evidence80). Although discouraging, the study was very small with low statistical power so a fully powered study should be done.
In RA, RTX is administered intravenously as two 1 g or two 500 mg infusions (given with 100 mg methylprednisolone or equivalent) separated by an interval of 2 weeks. These doses are relatively equivalent clinically, although higher doses are associated with a numerically earlier response, greater ability to achieve higher degrees of clinical response and retard radiographic progression more than the lower dose (category A/D evidence66 ,67 ,69 ,70 ,72 ,74 ,81–86).
Repeated treatment courses are effective in previously responsive patients with RA, and treatment of patients with a partial response after 6 months can result in additional response at week 48 (category B/C evidence85–89). A second course of RTX may achieve a clinical response even in patients who did not respond to the first course; this may relate to the extent of peripheral B cell depletion (category C evidence90).
Most patients who received subsequent courses did so 24 weeks or more after the previous course, and none received repeated courses earlier than 16 weeks after the previous course. Treatment with RTX every 6 months showed better clinical efficacy than on-demand treatment, with no significantly increased adverse events (category B evidence86 ,89).
Persistence: RTX is effective for up to 9.5 years for up to 17 courses in patients with an inadequate response to MTX for whom conventional DMARDS have failed or who have used one or more TNFi. Adverse events rates did not increase over time (category A/B/D evidence69 ,84 ,88 ,89 ,93–96).
Quality of life: RTX improves fatigue (category D evidence97).
TNFi switchers/degree of response
In a retrospective, non-randomised, open-label study, and in an observational study comprising several thousand subjects, patients who failed to respond to one or more TNFi had better clinical responses when switching to RTX than when switching to another TNFi (category C/D evidence98 100, category D evidence99).
Improvement was demonstrated in patient-related outcomes such as HAQ-DI, patient global VAS, fatigue, disability and QoL (category A/D, evidence101 ,102). RCTs show that the combination of RTX with MTX yields better clinical efficacy for RA than RTX monotherapy (category A evidence69 ,93). Preliminary data from observational studies suggest that combination of RTX with leflunomide yields even higher responses than MTX (category D evidence99 ,100).
A small 23 patient study in patients with RA receiving RTX plus statins suggested that statins might decrease response to RTX, but baseline differences among patients and small study size mandate further study of this issue (category C evidence103).
Structural changes: RTX inhibits radiographic progression in both MTX-naïve patients and in TNF-IR (category B evidence104 ,105). In RA, at 1 year, in combination with MTX, the 1000 mg×2 regimen decreased radiographic progression in comparison with MTX alone. This effect was maintained over 2 years (category B evidence106).
Combining biological agents
One small (N=51) open, randomised study of RTX plus abatacept or etanercept was not associated with more SIEs or increased clinical benefit over 6 months (category C evidence107). This early small open study is of interest, but a fully powered study is needed.
Hepatitis: RTX treatment is normally contraindicated in hepatitis B, since fatal hepatitis B reactivation has been reported in patients with NHL treated with RTX. In the case of occult or of latent HBV, alanine transaminase should be measured regularly. If the transaminases are, or become, raised and HBV DNA is found, the result should be checked with sensitive assays. Hepatitis B status should be assessed before treatment (category D evidence108 ,109).
RTX has been used in hepatitis C virus (HCV)-associated cryoglubulinaemic vasculitis with both positive and negative results (category A/D evidence110).
Infections: Cases of progressive multifocal leucoencephalopathy (PML) have been seen in patients with systemic rheumatic diseases with and without RTX treatment (FDA communication).
The utility of baseline screening for JC virus for patients with rheumatic diseases has not been established.
In general, patients who did not respond to TNFi will also have been prescreened for the presence of active or latent TB. In the RA clinical trials of RTX those with active or latent TB were excluded. Others were screened by chest radiograph examination, but were not screened for latent TB by purified protein derivative testing. There is no evidence of an increased incidence of TB in patients with NHL treated with RTX (category B evidence113). There are insufficient data to determine the need to screen for TB before starting treatment. Thus the clinician should be vigilant for the occurrence of TB during treatment.
RTX should not be given in the presence of serious or opportunistic infections.
Similar to the TNFi and the other biological agents, a small increase in serious bacterial infections was seen in patients receiving RTX. There was no further increase in the incidence of SIEs with up to nine courses of treatment (category A/B/D evidence69 ,113–117). No increase in the rate of serious infections was seen in a cohort of 259 patients who received another biological agent after RTX treatment compared with patients receiving RTX treatment before a biological agent (category D evidence116 ,118).
In an observational cohort study of patients with RA who switched biological agents, a multivariate analysis adjusting for other infectious risk factors suggested that the risk of infection in those switching to RTX was no different than that for patients switching to infliximab, etanercept, adalimumab or abatacept (category C evidence48).
Baseline immunoglobulin levels were generally normal in patients entering clinical studies, and decreased levels of IgM, IgA and IgG have been seen with RTX. After repeated courses of RTX, a proportion of patients develop IgG or IgM levels below the lower limit of normal. These patients have a statistical increase in infection rate in open studies (category C/D evidence117). Also, low pretreatment IgG levels conferred a higher risk of serious infection after RTX was started (category C evidence119). In clinical trials, on the other hand, no increase in serious infections was reported in the patients with normal pretreatment IgM levels who developed low IgM levels during RTX treatment (category B/D evidence70 ,82 ,107 ,115 ,120).
B cell levels have been measured in clinical trials, but their importance in routine practice has not been proved. Depletion of the CD20 B cell subpopulation by routine measures did not predict clinical response in patients with RA (category C/D evidence118 ,121–126). This suggests that the timing of re-treatment should be based on disease activity (category B evidence89).
Infusion reactions: The most widespread adverse events are infusion reactions, which are most common with the first infusion of the first course (up to 35%) and are reduced with the second and subsequent infusions (to about 5–10%). Intravenous corticosteroids reduce the incidence and severity of infusion reactions by about 30% without changing efficacy (category A/C/D evidence67 ,69 ,74 ,82 ,106 ,117 ,126 ,127). Rare anaphylactoid reactions have occurred when RTX was used (category D evidence112).
Malignancies: There is no evidence that RTX is associated with an increased incidence of solid tumours in RA. Nevertheless, vigilance for the occurrence of solid malignancies remains warranted during treatment with RTX (category B/C evidence113 ,117 ,126).
Neurological syndromes (see infections): pregnancy
Although more than 200 pregnancies have been reported among mothers exposed to RTX, the data are too incomplete and also too confounded (eg, by the concomitant use of potentially teratogenic drugs) to allow definitive conclusions (category C evidence128–130). The antibody, as an IgG, may be excreted in mother's milk (category C evidence130).
According to the US FDA, RTX is considered category C, meaning ‘no human studies and animal studies either show risk or are lacking. However, potential benefits may justify potential risks’.
Because of possible B cell depletion in the fetus after RTX, it is recommended that RTX be discontinued 1 year before a planned pregnancy, although these recommendations are not specifically data-driven (category C evidence128–130).
Skin reactions: Rare reports of psoriasis, including severe cases, and rare instances of vasculitis have been reported in patients with RA, SLE and NHL after RTX treatment (category D evidence131–134). The causative role of RTX in these circumstances remains unknown.
RTX significantly decreased the immune response to neoantigen (keyhole limpet haemocyanin), and pneumococcus as well as to flu vaccination, whereas delayed-type hypersensitivity responses and responses to tetanus were unchanged (category B evidence135).
Humoral responses to flu vaccination were modestly restored 6–10 months after RTX administration. Of note, patients with previous flu vaccination were more likely to develop protective titres to vaccination, arguing for yearly vaccination for all patients (category B evidence135). No data are available on the success of vaccination against flu after several courses of RTX.
Since RTX causes B cell depletion, it is recommended that any vaccinations involving B cell responses, such as those to prevent pneumonia and flu, should be given before the start of treatment (category A evidence120). Until further data are available, the use of live attenuated vaccines should only be given before the use of RTX.
IL-1 blocking agents
One IL-1 blocking agent, anakinra (IL-1 receptor antagonist), has been approved for use in RA. Two IL-1 inhibitors, rilonacept (IL-1 Trap) and canakinumab (anti-IL-1β monoclonal antibody) have been approved for use in cryopyrin-associated periodic syndromes (CAPS) (category A evidence134 ,136–139).
Anakinra may be used for the treatment of active RA, alone or in combination with MTX, at a dose of 100 mg/day subcutaneously (category A evidence140–142). In Europe, the anakinra label requires prescription in combination with MTX. Anakinra is recommended for the treatment of active RA after an adequate trial of non-biological DMARDs or with other DMARDs (category A/C evidence137 ,138 ,143 ,144). No trials of anakinra as the first DMARD for patients with early RA have been published. A meta-analysis showed that anakinra is less effective than TNFi in RA (category B evidence145).
Cryopyrin-associated periodic syndromes
Anti-IL-1 agents prompt major and sustained clinical benefit in children and adults with CAPS, including severe familial cold autoinflammatory syndrome, Muckle–Wells syndrome, neonatal-onset multisystem inflammatory disease/chronic infantile neurological cutaneous, articular syndrome and familial cold autoinflammatory syndrome/familial cold urticaria (category A/C evidence146–148). These are all rare conditions due to mutations in the NALP3 gene, in which a major role for IL-1 has been shown. There is no evidence that any single IL-1 inhibitor is more effective than another in CAPS, although dosing regimens and tolerability may differ.
Juvenile idiopathic arthritis and adult-onset Still's disease
IL-1 blockade with anakinra is effective in a proportion of patients with systemic-onset JIA and adult-onset Still's disease (category B evidence149) (see online supplementary table in Appendix for additional references). It may be less effective when initiated in patients with widespread arthritis and sustained disease (category C evidence150).
Time of response: Anakinra can lead to significant improvement in symptoms, signs and/or laboratory parameters of RA within 16 weeks, and can inhibit or induce slowing of radiographic progression (category A evidence137 ,140 ,141 ,151). If improvement is not seen by 16 weeks, discontinuation of anakinra should be considered.
Safety: The safety profiles of these agents have largely been established in patients with RA receiving anakinra. Use of newer drugs (canakinumab or rilonacept) or use in non-approved indications may disclose other safety concerns (category A/C evidence136 ,153).
To date, there is no indication that use of anakinra is associated with an increased incidence of TB (category D evidence154).
Serious bacterial infections are more common in patients receiving anakinra, and their incidence is higher than in patients with RA using non-biological DMARDs. The increased incidence of infection was greatest in patients who were also receiving corticosteroids or >100 mg/day anakinra (category A evidence50 ,142). Patients should not start or continue anakinra if a serious infection is present (category A evidence50 ,142 ,155). Treatment with anakinra in such patients should only be resumed if the infection has been adequately treated.
Combining biological agents
When anakinra was used in combination with etanercept, there was no increase in efficacy. However, the incidence of serious infection was increased in comparison with either drug used alone. Therefore, the combination of anakinra and etanercept should not be prescribed (category A evidence143).
Injection site reactions
A dose-related incidence of injection site reactions, affecting up to 70% of patients, has been reported with the use of anakinra. These reactions often do not require treatment and seem to moderate with continued use in most patients (category A evidence137 ,140 ,142 ,156).
According to the US FDA, this drug is considered category B. (No evidence of risk in humans. If no adequate human studies are done, no animal studies have been done or animal studies show risk but human studies do not.)
In one controlled trial, anakinra did not inhibit anti-tetanus antibody response (category D evidence151).
Canakinumab is effective in placebo-controlled randomised clinical trials of CAPS (category A evidence134 ,136). Canakinumab is indicated in the USA and Europe in adults, adolescents and children with CAPS aged ≥4 years with body weight >15 kg (category A/C evidence134 ,136 ,146 ,147).
Canakinumab is given subcutaneously every 8 weeks at a dose of 150 mg for patients with body weight > 40 kg, and at 2 mg/kg for patients with body weight >15 kg and <40 kg. No dose adjustment is needed in patients with end-stage renal disease (category C evidence153).
Clinical use: In CAPS canakinumab induces rapid clinically significant responses across different disease severity phenotypes, including paediatric age (remission or minimal disease activity by day 8) (category D evidence157), and induces sustained remission in paediatric patients with CAPS.
Nine patients with CAPS remitted, most within 7 days after single canakinumab doses. Time to relapse was 49 days (95% CI 29 to 68) (category D evidence157). Response continued over 2 years, with only 10% relapsing and 24% requiring increasing doses or frequency (category C evidence158). There is no evidence that canakinumab is more effective than rilonacept in CAPS.
Among 109 patients with CAPS over 2 years of treatment, 66% had infections, 10.8% developed a serious adverse event and 8% had mild-to-moderate injection site reactions (category C evidence158).
Vaccination: Vaccination response was normal to influenza and meningococcus (category B evidence159).
Rilonacept is given subcutaneously once a week at a dose of 160 mg for patients > 18 years and at 2.2 mg/kg for patients between 12 and 18 years of age. There is no evidence that rilonacept is more effective than canakinumab in CAPS.
There is no need to change rilonacept dosing in end-stage renal disease, based on a single-dose study (category D evidence153).
Pegloticase (uricase inhibition)
Pegloticase is a recombinant polyethylene-glycol conjugated uricase. It catalyses the oxidation of uric acid to allantoin and thus lowers serum uric acid levels (category A evidence160).
Intravenous pegloticase is approved by the FDA for the treatment of chronic gout in adult patient's refractory to conventional treatment (FDA, #1252930). These patients are generally defined as those with chronic gout in whom serum uric acid has failed to normalise to appropriate levels and where the signs and symptoms are not adequately controlled with standard urate-lowering treatments at maximum medically appropriate doses or for whom these drugs are contraindicated (category A evidence160).
Pegloticase was effective when given intravenously as 8 mg every 2 weeks in two replicate, randomised, double-blind, placebo-controlled studies. Background urate-lowering treatments were not used (category A evidence161). Pegloticase decreased plasma uric acid levels to ≤6 mg/dl for ≥80% of the time during 3 and 6 months in 42% of the pegloticase patients compared with 0% of the placebo-treated patients (p<0.001). Complete tophus resolution occurred in 40% of the patients receiving 8 mg pegloticase every 2 weeks, and 7% of the placebo-treated patients (p<0.05).
Pegloticase is given as an intravenous infusion in 250 ml normal saline or equivalent over >2 h. The usual dose is 8 mg every 2 weeks (category A evidence161). Because adverse events can occur immediately after infusion, it is recommended that patients be observed for about 1 h after completion of the infusion. To decrease the incidence of infusion-related reactions, use of a histamine blocking agent on the evening before and the morning of the infusion is suggested, as are paracetamol 1000 mg on the morning of the infusion and intravenous hydrocortisone 200 mg or equivalent immediately before infusion (category A evidence161).
If infusion reactions occur, the rate of infusion should be decreased or the drug stopped. Infusions can be completed using a slower rate of infusion in some cases (category A evidence161).
Reduced clinical efficacy, as well as a higher rate of infusion reactions and antibody formation (see later), was observed were seen in patients who received the drug every 4 weeks (category A evidence161).
Function and quality of life
Pegloticase 8 mg every 2 weeks improves the HAQ-DI by its minimum clinically-important difference (MCID; −0.2 and −0.22) and also improves the SF-36 physical component summary score by approximately its MCID (4.4 and 4.9) (category A evidence161).
About 63% of the 151 patients from phase 3 studies entering long-term, open-label extensions continued to receive the drug for ≥18 months and there is a small amount of data for follow-up to about 2.4 years (category D evidence161 ,162). When patients had sustained levels of uric acid <6 mg/dl during the phase 3 studies, this continued for at least 18 months in 84% (59 patients). These results were associated with continued improvement in signs, symptoms, function and well-being (category D evidence161).
Anti-pegloticase antibodies were detected in most evaluable patients, but only the highest anti-pegloticase antibody titres (>1/2430) correlated with infusion reactions. Conveniently, there is a good correlation between loss of plasma urate-lowering efficacy and infusion reactions. Plasma urate levels should be monitored before each infusion. When plasma uric acid levels rose to >6 mg/dl despite continuing infusion, 71–91% of the patients developed infusion reactions (category A evidence160).
Anaphylaxis or reduced blood pressure occurred in 5.1% of 273 patients in clinical trials (category A evidence160).
Cardiovascular risk factors are common in this group of patients. Deaths occurred during phase 3 trials (four receiving pegloticase (2.4%) versus three (7%) in the placebo group). Two patients in the pegloticase group who died had cardiovascular disease and an additional patient who developed a myocardial infarction survived. Other serious cardiovascular events such as heart failure, arrhythmia, transient ischaemic attack, unstable angina and coronary revascularisation occurred in 2.3–7.1% of pegloticase-treated patients and in 0% of the placebo group (category A evidence160), demonstrating a preponderance of serious cardiovascular events in the pegloticase groups.
Despite continuing prophylaxis with colchicine and/or non-steroidal anti-inflammatory drugs (NSAIDs), gout flares were the most common side effect during clinical trials, occurring in about 75% of the pegloticase-treated patients during the first 3 months of the phase 3 studies (compared with 53% of those receiving placebo). During the second 3 months they still occurred but at a rate equivalent to that of patients receiving placebo (41–67%) (category A evidence160). During long-term treatment the gout flares subsided (category D evidence161).
Infusion reactions occurred relatively commonly (26–41%) despite pretreatments. They were characterised by urticaria, chest pains, erythema, pruritis and/or dyspnoea. They were serious in 5–8% of patients. Slowing the rate of infusion or stopping and restarting the infusion at a slower rate reduced some of these manifestations (category A evidence160).
Tocilizumab (IL-6 inhibition)
TCZ is approved for treatment of RA in the EU, in the USA and a number of other, including Japan. It can be used in combination with MTX in DMARD-IR and TNFi-IR patients (category A/B/C/D evidence165 ,167 ,169–173).
In the EU, it is approved as monotherapy for the treatment of moderate to severe active RA in patients who are MTX intolerant (category A/D evidence156 ,163–165 ,167 ,173–177) or when continuation of MTX is inappropriate.
The FDA has approved TCZ for use in patients with moderate to severe RA as monotherapy.
Juvenile idiopathic arthritis and other indications
Ankylosing spondylitis: In a small open-label study in AS, TCZ did not appear to be efficacious (category D evidence180).
TCZ reduces signs and symptoms of active RA in DMARD-IR (including MTX-naïve patients) or TNFi-IR (category A/D evidence156 ,164–166 ,181). TCZ monotherapy is effective predominantly in DMARD-IR or TNFi-IR patients (category A/D evidence164 ,177).
The dosing regimens recommended vary by indication and country and are shown in table 1. In adults, TCZ is given intravenously every 4 weeks in a dose of 4 or 8 mg/kg, with a maximum of 800 mg per infusion. In general, 8 mg/kg is more effective than 4 mg/kg (table 2). In the USA TCZ is approved at a starting dose of 4 mg/kg; based on the clinical response, escalation to 8 mg/kg is possible. In the EU and elsewhere, 8 mg/kg is the recommended starting dose (category A/D evidence156 ,163–165 ,174 ,182 ,193).
In a double-blind RCT, MTX+TCZ was not clinically better than TCZ alone for MTX-IR as shown by DAS remission and ACR20, 50 and 70 responses (category B evidence156).
Time to response
Response can occur as early as 2–4 weeks in some patients, but it may take ≥24 weeks in other patients (category A/C/D evidence164 ,182 ,184 ,185). TCZ can be effectively restarted after long-term withdrawal (category D evidence186).
Persistence: Long-term efficacy data up to 5 years showed good sustainability in those patients continuing to receive treatment (category B evidence187).
Several other studies comparing the effectiveness of TCZ with that of other biological agents in MTX–IR, DMARD IR and TNFi-IR suggested that TCZ had no statistically significant advantage for the ACR50 response (category A evidence188 ,189). In one study using a mixed-treatment comparison there was a statistically better response to TCZ than to other biological agents for the ACR70 response (category B evidence190).
An indirect pairwise meta-analysis showed that efficacy after one or multiple TNFi failures did not differ between different agents (category A evidence166).
Cost-effective analysis based on UK standards (which may be different in other contexts) showed that TCZ was cost-effective at the threshold commonly used within the UK (category B evidence191). An analysis from Finland showed that biological DMARDs, including TCZ, were cost-effective in DMARD-IR (category C evidence192).
Quality of life
TCZ decreases fatigue (category C evidence97).
Structural changes: TCZ inhibits or reduces radiographic progression in patients for whom MTX, DMARDs or TNFi has produced an inadequate response (category A/D evidence167 ,193 ,194). It inhibits or reduces radiographic progression as monotherapy in those with high baseline risk for radiographic damage (category A/B evidence175 ,195 ,196).
In an RCT including MTX-IR patients, no difference in radiographic progression was seen whether patients were receiving combination MTX + TCZ or TCZ monotherapy (category D evidence197).
Cardiovascular endpoints and lipid levels
The overall long-term effect of TCZ on cardiovascular outcomes is not known. In a follow-up for up to 5 years (category D evidence181 ,198–202), there was no apparent increase in cardiac event rates. Hypertension and cerebrovascular accidents have been observed (category A/D evidence174 ,181 ,199 ,203–206). In a follow-up with a median of 1.5 years, there is no increase in the rate of cerebrovascular accidents (category D evidence199).
Increases in mean fasting plasma lipid levels were seen in TCZ-treated patients, including total cholesterol, low-density lipoprotein (LDL), triglycerides and high-density lipoprotein (HDL) (category A/D evidence199 ,203 ,204 ,206 ,207).
Changes of subfractions of lipoproteins when using TCZ have been analysed. TCZ increased very low density lipoprotein, HDL and a chylomicron subfraction of LDL. Small dense LDL particles remained unchanged while reducing C-reactive protein (CRP), lipoprotein a and fibrinogen. Changes in lipid levels and apolipoprotein levels correlated with soluble and clinical inflammatory markers in RA. Effects on cardiovascular (CV) risk will require long-term observational studies to assess whether these lipid changes affect CV outcome (category D evidence208).
Lipid levels should be monitored 1–2 months after the start of treatment and then every 6 months. Hyperlipidaemia should be managed according to local recommendations. Initiation of statin treatment when using TCZ effectively reduces lipids (category D evidence209).
In a pooled analysis of five pivotal TCZ trials and their extensions, ‘non-fatal myocardial infarction, stroke or death due to cardiovascular cause (major cardiovascular events)’ was associated with age, a history of coronary artery disease and disease activity rather than TCZ (category A evidence202).
Gastrointestinal: Within the controlled clinical trial programme, generalised peritonitis, lower gastrointestinal perforation, fistulae and intra-abdominal abscesses have been reported (overall rate 0.26/100 patient-years compared with no events in the control arm). The concomitant use of corticosteroids and NSAIDs may increase the risk of these events. TCZ should be used with caution in patients with a history of intestinal ulceration or diverticulitis (category D evidence210 ,211).
Among 7901 Japanese patients receiving 8 mg/kg TCZ, followed up for 28 weeks, 0.16% developed gastrointestinal perforations (0.30/100 patient-years) (category C evidence212). A complete literature review of lower gastrointestinal perforation in patients with RA using DMARDs has been published. The risk of diverticular perforation may be slightly higher but probably not statistically different with TCZ than with conventional DMARDs or TNFi but lower than for corticosteroids (ORs: TCZ 1.6; TNFi 1.3; corticosteroids 2.9) (category B evidence213).
In clinical trials, a decrease in the absolute number of neutrophils was found in a higher proportion of patients treated with TCZ than in those receiving placebo. A few patients showed a decrease of polymorphonuclear cells to <1000 cells/mm3 and, rarely, <500 cells/mm3. In one large clinical trial comparing TCZ with MTX, reversible grade 3 neutropenia associated with TCZ occurred in 3.1% with TCZ compared with 0.4% with MTX (category A evidence165).
In children with systemic JIA (n=112), 15% of the patients developed grade 3 and 2% grade 4 neutropenia; all changes were transient and were not associated with severe infections (category A evidence214).
This change usually occurs early after a dose and is transient. Complete blood counts should be monitored regularly according to local labels (usually every 4–8 weeks). In one study, there was an accompanying increase in infections, but this was not seen in most studies (category A/D evidence165 ,215–218).
Neutropenia occurred in 0.8% and 2.4% of patients receiving 4 then 8 or 8 mg/kg TCZ plus a DMARD, respectively, while it was 5.1% among patients receiving 8 mg/kg TCZ as monotherapy (category D evidence219).
Hepatic aminotranferase and bilirubin
Increases in alanine aminotransaminase and aspartate aminotransferase occurred with similar frequency with TCZ monotherapy and MTX alone (category A/D evidence220–222). In combination with DMARDs including MTX, increases are more common than with TCZ alone. Increases in bilirubin, mostly indirect and sometimes associated with Gilbert's syndrome, occur separately and are not associated with hepatic dysfunction. Liver function should be monitored regularly.
Recommendations for the management of TCZ-related laboratory abnormalities have been included in the EMA and FDA package, which are consistent with those for MTX. Liver test abnormalities have been documented within the TCZ clinical trials (category A/D evidence156 ,163 ,165 ,174 ,176 ,181 ,182 ,199 ,203 ,204 ,220–222). Cases have been reported of TCZ-induced hepatic failure (category D evidence223).
In two 6-month, controlled clinical studies, the rate of serious infections in the 4 and 8 mg/kg arms were numerically higher in the TCZ than placebo+DMARD or placebo+MTX trials (1.47 vs 0.78) (category B evidence187 ,215 ,216). The rates were stable for up to 5 years in open-label extensions of controlled trials (category A/D evidence165 ,187 ,199 ,204 ,215 ,220).
In children with systemic JIA (n=112), the rate of serious infection was 11/100 patient-years (category A evidence214).
The integrated safety summary of all core TCZ RCTs and open-label extensions, the large Japanese postmarketing cohort, and the Rapid Onset and Systemic Efficacy (ROSE) trials documented serious infection rates of 4.9–9.1/100 patient-years in those treated with 8 mg/kg TCZ (category A/B/C/D evidence202 ,224 ,225).
Similar to other biological treatments, rates of serious infection were significantly increased in TCZ-treated patients aged ≥65 years, those using ≥5 mg/day of prednisone and those with underlying respiratory disease (category C evidence225).
The down-regulatory effect of TCZ on the acute phase reactant, CRP, may limit the usefulness of CRP as a diagnostic indicator for infections. TCZ should not be given in the presence of serious or opportunistic infections (category D evidence174). As with other biological agents, careful observation for bacterial infections is necessary (category B/D evidence199 ,203 ,204).
Tuberculosis and opportunistic infections
A recent postmarketing observational study in Japan observed a rate of tuberculosis of 0.22/100 patient-years in TCZ-treated patients (category C evidence225). An integrated safety analysis from the global phase III programme demonstrated an opportunistic infection rate of 0.23/100 patient-years (category A evidence202).
Patients should be screened for (latent) TB before treatment. See TNFi section for details of TB screening.
Cases of localised herpes zoster infection have occurred in clinical trials and in postmarketing studies. In the latter, the rate of herpes zoster was 0.61/100 patient-years (category C/D evidence202 ,225).
Hepatitis B or C
The safety of TCZ in patients with active hepatitis B or C is unknown, as patients with positive serologies were excluded from clinical trials.
Serious infusion reactions during/after treatment with TCZ are uncommon (category A/D229).
Anaphylactic reactions have occurred, including a fatal reaction (category D evidence230). Even a single case, however, means that discontinuation of TCZ treatment is required in patients with a previous severe allergic reaction to TCZ.
There is no evidence that TCZ treatment is associated with an increased incidence of malignancies in patients with RA (category A/D evidence156 ,163–165 181–183 ,199). Systematic safety surveillance should be performed during TCZ treatment as required for other biological agents.
An analysis of pooled data from clinical trials and ongoing long-term extension studies of patients with RA who received one or more doses of TCZ (N=4,009; 10 994patient-years) indicates that the overall malignancy rates remained stable with continued TCZ treatment and did not exceed reported malignancy rates from the SEER database (category B/D evidence7 ,8). Similarly, a systematic review, a meta-analysis of clinical trials and a 2-year follow-up trial reported no increase in solid malignancies (breast, lung, colon, prostate) in patients treated with TCZ compared with those treated with placebo (category A/D evidence9 ,10 ,14 ,229).
There have been too few cases of pregnancy during the use of TCZ for any conclusions to be drawn (category C evidence130). According to the US FDA, this drug is considered category C ‘No human studies and animal studies either show risk or are lacking. However, potential benefits may justify potential risks’.
Erythroderma has been ascribed to TCZ (category D evidence231).
Safety and response to vaccinations were evaluated in patients with RA receiving TCZ. Most patients could be effectively immunised with flu and pneumococcal vaccine, although titres were lowered (category D evidence227 ,232). As for the other biological agents, live vaccines should not be given while patients are receiving TCZ (category A/D evidence174 ,199 ,216 ,233 ,234).
Unusual adverse events
Non-rheumatological arthralgia after TCZ has been noted (category C evidence18).
TNF blocking agents (TNFi)
TNFi differ in composition, precise mechanism of action, pharmacokinetics and biopharmaceutical properties, but this document emphasises areas of commonality. Studies that have clearly differentiated among compounds will be discussed where appropriate.
In most patients, TNFi are used in conjunction with another DMARD, usually MTX. TNFi have also been used successfully with other DMARDs, including sulfasalazine and leflunomide. They are effective for the treatment of RA in MTX-naive patients (category A, D evidence155 ,235–253) and can be used as the first DMARD in some patients (category A/B evidence51 ,155 ,182 ,235–244 ,246 ,247 ,251 ,254–258). They have been used successfully in combination with MTX for early RA. Adalimumab, certolizumab and etanercept are approved as monotherapy for RA. Infliximab and golimumab are only approved for use with MTX in RA; however, observational data indicate that they have been used as monotherapy (category C evidence259–261).
Adalimumab, etanercept, golimumab and infliximab are widely approved for the treatment of patients with PsA with inadequate response to conventional treatments. Efficacy has been demonstrated both as monotherapy and with background MTX (category A/B/D evidence198 ,233 ,262–279).
Adalimumab, etanercept, golimumab and infliximab are widely approved for the treatment of active AS that is refractory to conventional treatments. In clinical trials, the efficacy of these TNFi improved signs and symptoms, function and QoL as monotherapy as well as with concomitant second-line agents, including sulfasalazine or MTX (category A/B evidence257 ,280–288). There is no evidence that combination treatment with conventional DMARDs is better than monotherapy.
Juvenile idiopathic arthritis
Etanercept and adalimumab are approved for JIA with a polyarticular course (FDA and EMA: ≥2 years of age for etanercept and ≥4 years of age for adalimumab; EMA: age 13–17 years for both) (category A/B evidence289–296). Infliximab was beneficial at 6 mg/kg in polyarticular JIA (category B evidence289 ,290 ,295 ,296).
Increasing the dose or reducing the dosing intervals of infliximab and adalimumab may provide additional clinical benefit in RA, whereas increased doses of etanercept or golimumab confer no increased clinical benefit at a group level (category A/B/D evidence145 ,251 ,289 ,297–301). Likewise, no effect of increasing certolizumab dosing from 200 to 400 mg every 2 weeks has been shown (category D evidence301). The addition or substitution of other DMARDs may increase efficacy in some patients. In clinical practices one purely observational study showed that 12–36% of patients with RA have their TNFi dose increased (category C evidence302).
Data are conflicting as to whether a triple combination of traditional DMARDs is clinically as effective as a combination of MTX plus TNFi. Triple combination treatment appears to be clinically as effective as TNFi plus MTX but the latter may have better radiographic effects (category A/D evidence305–307).
For patients in remission or with low disease activity for a prolonged period, some studies indicate that lowering or even stopping the TNFi may be successful (category C/D evidence155 ,238 ,239 ,243 ,245 ,251 ,259–261 308–311).
The combination of a TNFi and MTX yields better results both in established and early RA for clinical responses, radiological outcomes, tolerability and immunogenicity. If there are contraindications to MTX use, other DMARDs (eg, leflunomide) are recommended, although there is some controversy (category A/B evidence155 ,235–245 247–249 254–256 259–261 ,308 ,309 ,312–314).
TNFi are effective for controlling pain in RA and other conditions. Data indicate that this may be associated with changes in the central nervous system (category D evidence317).
In recent studies, patients with RA, particularly those with early RA as compared with established disease, successfully discontinued or reduced their dose of TNFi if they had achieved low disease activity or remission for a prolonged period (category B evidence318 ,319). In one study, when the dose of the drug was lowered, clinical efficacy continued but radiographic benefit did not (category C evidence320–322). Longer-term data on discontinuation are needed.
Patients with a higher body mass index respond less well than those with a lower BMI to infliximab or etanercept (category C evidence323). Cigarette smoking attenuates the clinical response to TNF inhibitors (category C evidence320 ,321).
Time to response/predictors
TNFi, when given up to the maximum approved dosing regimens for RA and polyarticular JIA, may elicit a response in 2–4 weeks in some patients. They usually lead to significant, documentable improvement in symptoms, signs and/or laboratory variables within 12–24 weeks (category A/B evidence155 ,235–249 ,251 ,254–257 ,261 ,280–282 ,288 ,289 ,296 ,299 ,308 ,324–333). Clinically significant important responses should include patient-oriented measures (eg, HAQ-DI, patient's global VAS, Medical Outcome Survey and SF-36) and/or physical measures (eg, joint counts) (category A evidence155 ,236–239 241–247 ,249 ,254–257 ,261 ,280–282 ,285 ,308 ,313 ,325–328 ,331 ,332).
If improvement occurs, treatment should be continued. If patients show no response to these agents, their continued use should be re-evaluated.
Etanercept weekly dosing in children (0.8 mg/kg up to 50 mg weekly) also improves health-related QoL and results in reduced disease activity (category A evidence334).
For certolizumab the rapidity and degree of clinical response at 6–12 weeks predicts the 1-year response. Some patients require longer than 12 weeks to respond or achieve their target disease state, with 35–40% of patients requiring up to 24 weeks (category C evidence 335).
Comparing TNF inhibitors
There is no evidence that any one TNFi should be used before another one can be tried. There is also no evidence that any TNFi is more effective than any other in RA (category A/B evidence28 ,34 ,40 ,57 ,145 ,152 ,155 ,236 ,248 ,251–253 ,296 ,329 ,336). One recent meta-analysis of data from RCTs demonstrated that etanercept, infliximab and adalimumab were comparable and all were more effective than anakinra for RA (category A evidence164). A meta-analysis also contended that etanercept was safer than anakinra, adalimumab or infliximab (category A evidence145).
A number of meta-analyses and other assessments (including mixed-treatment comparisons) generally concluded that the lowest rate and degree of response was to infliximab in comparison with etanercept and adalimumab (category A/C evidence34 ,337–340).
Because channelling bias may occur in these registry and observational studies, their conclusions will require well-done, prospective, corroborative, comparative studies.
In long-term observational studies, some patients continue to respond for up to 15 years (category C/D evidence152 ,341–344). Loss of response to a TNFi can occur. Patients have been switched successfully from one TNFi to another. Several retrospective and observational studies suggest the efficacy of such switches. One recent RCT supports this regimen (category B/D evidence27 ,345–347).
Data from some, but not all, observational data suggest that primary non-responding patients may be less likely to respond to a second TNFi. Patients who have not tolerated one TNFi may respond to a second but are also less likely to tolerate a second TNFi (category B/C/D evidence251 ,253 ,296 ,329 ,348–352). Patients who have responded to a TNFi but have lost response may respond to a second TNFi. The optimal treatment of patients not responding to TNFi remains to be determined (category C evidence181 ,236 ,239 ,244 ,261 ,281 ,289 ,315).
Patients with high or moderate disease at baseline can respond well to TNFi (category C/D evidence315 ,316). The concomitant use of DMARDs can aid in maintaining TNFi treatment (category C/D evidence315 ,316).
TNFi inhibit or reduce radiographic progression in RA, even in some patients who do not experience clinical response (category A evidence155 ,240 ,242 ,245 ,247 ,258 ,328 ,332 ,353–357). Better clinical and radiological outcomes are achieved when TNFi are used in combination with a traditional DMARD, particularly MTX) category D evidence358). TNFi slow down radiographic progression in early and established RA (category A evidence356 ,357).
TNFi may be cost-effective from a societal perspective and improve QoL. The pharmacoeconomics evaluations are highly dependent on the specific circumstances of the analysis and the society in which the analysis is done (category B/C/D evidence237 ,359–365).
Juvenile idiopathic arthritis
TNFi, when given up to the maximum approved dosing regimens for polyarticular JIA, usually lead to an early, significant, documentable improvement in symptoms, signs and/or laboratory variables.
Etanercept weekly dosing in children (0.8 mg/kg up to 50 mg weekly) also improves health-related QoL and reduces disease activity (category B evidence334).
MTX or the combination of etanercept, MTX and prednisolone effectively induced clinically inactive disease. The trial also showed that the earlier in disease that treatment occurred, the more likely it was that clinically inactive disease could be achieved (category A evidence369).
Time to response/productors: see TNFi in RA above.
Comparing TNF inhibitors in JIA
Etanercept appears less effective in patients with systemic-onset JIA than in patients with other forms of JIA (category C evidence291–294). All tested TNFi demonstrate comparable efficacy and safety in non-systemic-onset JIA and can be tried (category C/D evidence291 ,370).
In JIA-associated uveitis, adalimumab and infliximab are effective more often than etanercept (category C/D evidence371 ,372). In an open trial, infliximab, 5 mg/kg up to every 6 weeks was not as effective as adalimumab 24 mg/m2 used every 14 days for uveitis (category C evidence373). However, others have suggested that a clinically effective dose of infliximab for uveitis may require up to 10–20 mg/kg as often as every 4 weeks (category D evidence371).
Switching TNF inhibitors
Anecdotal studies indicate that TNFi can be successfully switched in JIA (category D evidence294–296).
TNFi contribute to restoration of growth velocity in children whose JIA-associated inflammation is controlled (category B evidence377). Bone density improves after treatment with TNFi, even in patients who have incomplete disease control (category C/D evidence291–295).
The Group for Research and Assessment of Psoriasis and Psoriatic Arthritis (GRAPPA), and subsequently EULAR, have developed treatment recommendations for PsA based on a systematic evidence-based review of the efficacy of TNFi and other agents (category D evidence270).
In addition to efficacy in joints and skin, TNFi treatment has been shown to be efficacious for enthesitis, dactylitis, function, QoL fatigue, productivity, work disability and inhibition of structural damage (category A/B/D evidence 233 ,266–269 271–280 ,324 ,378–388). The GRAPPA group has established a quantitative composite measure of minimal disease activity as a target for treatment (category D evidence379 ,389). In studies using the 1981 ACR remission criteria and the RA DAS28 remission criteria, respectively, remission was more frequent and enduring in patients with PsA than in those with RA (category B evidence390).
As with other TNFi, golimumab inhibits structural damage in PsA (category A evidence391).
Dose and time to response
Improvement of signs and symptoms, function and QoL usually occur within 12 weeks. Some patients continue to improve to week 24. For etanercept, 100 mg a week for 12 weeks, followed by 50 mg a week, was more effective than 50 mg a week for skin but not arthritis, enthesitis or dactylitis (category D evidence385 ,386). Clinical response has been demonstrated in the axial disease of PsA (category C evidence392).
Use of TNFi in patients with early PsA has increased effectiveness when used early in disease (category C evidence393).
In children, maximal response to etanercept may take longer than 3 months (category C evidence292).
Raised CRP at baseline was predictive of good treatment responses and continued treatment (category B evidence378).
Comparing TNF inhibitors in PsA
A meta-analysis of randomised trials suggests that the efficacy of TNFi antibodies may be better than that of soluble receptor for improvement of skin manifestations (category A evidence378).
Switching between TNF inhibitors in PsA
Preliminary data suggest that PsA-related joint and skin signs and symptoms may be improved by switching to a different TNFi, even if efficacy from a previous TNFi was never achieved (category C/D evidence382 ,394).
In open-label studies of up to 2 years in PsA, the durability of clinical efficacy and improvement in radiographic damage have been demonstrated with etanercept, infliximab, adalimumab and golimumab (category A/B/C evidence233 ,272 ,274 ,277 ,278 ,324). In the Danish registry median drug survival was 2.9 years, and 1- and 2-year drug survival rates were 70% and 57%, respectively.
Structural changes: Both golimumab 50 mg and 100 mg monthly decreased radiographic progression using the Sharp–van der Heijde score (category A evidence391). This is a difference from RA, where only the 100 mg dose was effective for decreasing radiographic progression.
In the British Registry, patients with PsA treated with TNFi had a similar safety profile to that of a control cohort of patients with seronegative arthritis receiving DMARD treatment (category B evidence384).
The approved doses of TNFi for treatment of AS are 5 mg/kg infliximab intravenously every 6–8 weeks after induction; subcutaneous etanercept, 25 mg twice a week or 50 mg once a week; 50 mg subcutaneous golimumab monthly and 40 mg adalimumab subcutaneously every other week (category A/B evidence23 ,281–288 ,322 ,398 ,399). No major differences in efficacy and safety between 50 mg and 100 mg golimumab doses were seen when treating AS (category B evidence272).
In clinical trials in patients fulfilling the modified New York criteria for AS, improvement in signs and symptoms were seen after TNFi as measured by patient-reported outcomes (BASDAI, BASFI, patient global VAS, SF-36), spinal mobility measures, peripheral arthritis, enthesitis and acute phase reactants (category A/B/D evidence257 ,280–288 ,395 ,396 ,398 ,400–402).
Two placebo-controlled trials showed significant improvement in signs and symptoms in patients with non-radiographic axial spondyloarthritis (SpA) (category A/D evidence398 ,403) according to the ASAS axial criteria for axial SpA (category A evidence396). Two RCTs failed to demonstrate the superiority of a combination of MTX with infliximab versus infliximab alone in the treatment of active AS over 1 year (category B evidence102 ,257 ,280). Regular treatment with infliximab was more effective than ‘on-demand’ treatment for treating AS (category A evidence 405 , 405a).
In a head-to-head comparison trial of a conventional DMARD (sulfasalazine) with a TNFi (etanercept), the latter was more effective, including for heel enthesitis (category A evidence327 ,405 ,406). Enthesitis scores have also been shown to improve in studies with other TNFi.
There is evidence that the incidence of uveitis flares is reduced and anaemia improves when patients are treated with TNFi. TNFi antibodies produce a greater decrease in the frequency of uveitis episodes than etanercept (category A evidence371 ,372 ,407).
The importance of adding regular physical therapy to TNFi use has been highlighted in an observational trial (category C evidence408).
There is some evidence that the presence and/or history of colitis associated with Crohn's disease or ulcerative colitis should influence the choice of the TNFi, since TNF antibodies such as adalimumab, certolizumab, infliximab are approved for the treatment of Crohn's disease, and antibodies directed against adalimumab or infliximab correlate with decreased clinical response in some patients with AS. This was not found for etanercept. Acute phase reactions correlate with response (category B/C/D evidence207 ,210 ,211 ,214 ,410–412).
Young patients with active AS and raised CRP levels responded better to TNFi than older patients without such markers (category A evidence257 ,402 ,410 ,413 ,414). However, even in patients with advanced and severe AS, there is evidence that TNFi can be efficacious (category D evidence414 ,415).
An observational study indicates that switching to a second TNFi may be effective in a small percentage of patients (15% in a study of 514 patients with AS), although the efficacy may be a little less with the second TNFi (category B evidence305).
In two head-to-head comparison trials of a conventional DMARD (sulfasalazine) with a TNFi (etanercept), the latter was more effective in patients with established AS and non-radiographic axial SpA (category A evidence416 ,417).
A 108 patient double-blind, randomised trial in AS showed no difference between 100 mg etanercept and 50 mg etanercept weekly over 12 weeks for any clinical parameters, erythrocyte sedimentation rate or CRP (category A evidence418).
Time to response
Although improvement may be seen more rapidly, a reduction in signs and symptoms, and improvement in function and QoL will usually be seen by 6–12 weeks in response to treatment with a TNFi (category A evidence288 ,405).
Comparing TNFi in AS
TNFi (adalimumab, etanercept, infliximab) remain efficacious for 2–7 years in open-label studies (category A/B/D evidence396 ,401 ,403 ,405 ,413). The disease usually flares after discontinuation of TNFi category C evidence396 ,401 ,403 ,413 ,414). When TNFi are restarted, treatment response re-occurs in over 70% of patients (category C evidence414 ,421).
Several studies have shown that active inflammation of the sacroiliac joints and spine, as shown by MRI, is significantly reduced for up to 3 years by adalimumab, etanercept, infliximab and golimumab (category A/B/C evidence399 ,401 ,403 ,422).
One open, RCT in 76 patients with AS showed that etanercept was clinically more efficacious and MRI evidence of decreased inflammation was stronger than with 2–3 g daily sulfasalazine (category B evidence417).
TNFi reduce spinal inflammation (category A/C evidence423 ,424) but they did not inhibit new bone formation. The effect was particularly evident with continuous or high-dose concomitant NSAID (category B evidence425 ,426).
Pharmacoeconomic data in AS
Safety (arranged alphabetically)-across indications
Two systematic reviews of adalimumab and etanercept use across disease indications suggest that, in general, there is a numerically lower frequency of adverse events, in particular infection, in PsA, AS, JIA and psoriasis than in RA. This may be due to background factors such as less corticosteroid and concomitant immunosuppressive drug use, as well as possible disease state factors (category A evidence431 ,432).
In the British Registry, patients with PsA treated with TNFi had a similar safety profile to that of a control cohort of patients with seronegative arthritis receiving DMARD treatment (category B evidence384).
Antiphospholipid and lupus-like syndromes have occurred in both adult and paediatric patients during treatment with TNFi. Autoantibody formation is common after TNFi treatment (eg, antinuclear antibodies), but clinical syndromes associated with these antibodies are rare (category C/D evidence404 ,433–436). In some cases, development of anti-drug antibodies has been associated with decreased efficacy (category C evidence436a).
In patients with SpA treated with infliximab alone, 25.5% patients developed anti-drug antibodies. The presence of these antibodies was associated with a poorer response, more infusion reactions and increased discontinuation rates (category D evidence412).
A case of dermatomyositis associated with the use of adalimumab has been reported (category D evidence439).
Reviews (category C evidence349 ,440 ,440a) and multiple open studies of TNFi have been published (category C/D evidence435 ,439–451). Reports on TNFi and lipids are somewhat conflicting, but it has been reported that patients receiving infliximab, etanercept and golimumab have improved lipid and arthrogenic profiles, less arterial stiffness and reduced insulin resistance in comparison with controls (category C/D evidence348 ,349 ,422 ,435 ,440–458). Two long-term studies demonstrated a reduction in myocardial infarction (category C evidence446 ,447), but no effect on overall mortality (category C evidence455).
In an analysis of about 10 000 TNFi-treated patients in a North American Registry, the risk of CV events was reduced (HR=0.39) in patients treated with a TNFi but not in patients treated with MTX. In contrast, prednisone use was associated with a dose-dependent increased risk. The risk reduction associated with TNFi was seen both for fatal and non-fatal CV events (category C evidence456).
To date these profiles seem to reflect the degree to which inflammation is controlled. Better disease control was reflected in either unchanged or improved lipid profiles, while incomplete control was associated with worsening profiles. The clinical significance of these changes in CV symptoms is unknown, but recent studies suggest that the risk of CV events may be decreased in patients using these agents (category C evidence446 ,447 ,456).
There continue to be inconsistent results about the CV risk during TNFi treatment, all from observational data. In one large multisite observational study, a benefit was only found in those aged ≥65 years (HR=0.51; 95% CI 0.33 to 0.78), but not for those aged <65 years (category D evidence457). A cohort of 6000 patients with RA from the Swedish Rheumatology Registry from 2003 found no TNFi effect on acute coronary syndrome, including no effect of disease activity and no difference among TNFi (category B evidence458). A meta-analysis of RCTs in chronic plaque psoriasis found no TNFi effect versus placebo on CV events (category A evidence).
A meta-analysis of the effect of TNFi on lipid levels (category A evidence459) identified a significant increase of total cholesterol (maximum 10%) and HDL (maximum 7%) within the first 2–6 weeks of treatment, with no substantial changes of the atherogenic index. A review of RA and lipids showed that decreasing inflammation with all biological agents increased lipids (category B evidence460). A systematic review and meta-analysis of subclinical atherosclerosis and lipid profiles showed that, despite significant changes of inflammatory markers, TNFi did not lead to significant changes in intima-media thickness, endothelial function or lipid profiles over 52 weeks (category D evidence451 ,461).
The incidence of non-cardiac vascular disease in patients with RA has been analysed recently (category B evidence462). This inception cohort study noted no change in the rate of cerebrovascular or peripheral arterial events between the time before TNFi became available (1980–1994) and after their introduction (1995–2007).
Rare instances of pancytopenia and aplastic anaemia have been reported (category C evidence327). If haematological adverse events occur, TNFi should be stopped and patients evaluated for evidence of other underlying diseases or association with concomitant drugs.
Raised liver function tests (LFTs) have been seen in patients treated with adalimumab, etanercept, certolizumab, golimumab and infliximab, with alanine transaminase/aspartate transaminase increased in 0.1–17.6%, often transiently and usually decreasing with continued use. Mild increases in LFTs were more consistently seen with infliximab, less commonly with adalimumab and etanercept in comparison with DMARDs. Golimumab toxicity in a separate study was shown to be associated with LFT increases (category A/C/D evidence463–468). The use of concomitant drugs and other clinical conditions confound the interpretation of these observations (FDA; category B/C evidence350 ,404 ,469–474). Follow-up and monitoring for increases in LFT should be governed by the patient's concomitant drugs, conditions and patient-related risk factors. Worsening of alcoholic hepatitis has been seen in patients receiving TNFi (category C evidence350).
An increased susceptibility to TB or reactivation of latent TB has been reported for all TNFi (category A/B/C evidence463 ,475–496), although one report of 354 patients with AS did not find this (category B evidence497). The risk of TB is also increased by the use of corticosteroids. There appears to be a higher incidence of TB in patients using the monoclonal antibodies, infliximab and adalimumab, than in those using etanercept (category B/C/D evidence478 ,480–482 ,498).
TB risk data for golimumab and certolizumab are limited. Trials of golimumab excluded patients with active or latent TB, and cases of TB were uncommon (category B evidence463). In trials of certolizumab there was an increased incidence of TB in patients relative to controls, but TB screening procedures were not standardised among sites (category C evidence152).
While differences may be due, in part, to differences in mechanism of action, biology or kinetics among drugs (category C/D evidence463 ,478–493), it may also be, in part, due to the fact that populations treated with the various TNFi differ (eg, higher background rates of TB in some countries) and the data come from registries and voluntary reporting systems.
The clinical manifestations of active TB may be atypical in patients treated with TNFi (eg, miliary or extrapulmonary presentations), as has been seen with other immunocompromised patients (category C evidence486–488).
In the USA, an area with low TB prevalence, most mycobacterial infections among TNFi users were caused by non-TB mycobacteria, with only 35% caused by Mycobacterium tuberculosis. Mycobacterium avium was found as often as M tuberculosis, and multiple other non-tuberculous mycobacterial infections accounted for the rest of the mycobacterial infections (category C evidence489).
Screening of patients about to start TNFi treatment has reduced the risk of reactivating latent TB for patients treated with these agents (category B evidence490 ,491). Every patient should be evaluated for the possibility of latent TB, including a history of previous exposure, previous or active drug addiction, HIV infection, birth or extended living in a region of high TB prevalence, and working or living in TB high-risk settings such as jails, homeless shelters and drug rehabilitation centres (category B evidence475 ,486).
In addition, physical examination and screening tests such as TST and/or interferon γ release assays (IFNγ release), and chest radiographs should be carried out before TNFi treatment is started, according to local recommendations. In areas of high TB prevalence (ie, high-risk populations or in the event of potential TB exposure), repeat screening should be considered (category C evidence477 ,492 ,493).
The TST is a diagnostic aid, and false-negative results can occur when immune suppression is present (eg, HIV, renal dialysis, corticosteroid use and RA) (category C evidence 477 ,494 ,495). The TST can also be falsely positive due to prior BCG vaccination. New blood-based diagnostic assays (IFNγ release) have been developed using TB-specific antigens. These tests (eg, Quantiferon-Gold and T-Spot TB) have greater specificity for latent TB infection than the TST and therefore provide a useful tool (particularly for those with a history of BCG vaccination). It should be noted that false-negative results and indeterminate results also occur with the IFNγ release assays (category C evidence494 ,495). The background rate of TB in the population should be considered in the interpretation of these tests, as it influences their positive predictive value.
The precise role of these tests in diagnosing latent TB in patients with rheumatoid disease remains under study (category C evidence489).
Repeat screening should be performed in the event of TB exposure, and should be considered in patients who are at continuing risk for TB exposure (eg, living or extended travel to endemic areas) (category C evidence477). Local screening guidelines should be followed.
Continued vigilance is required to detect reactivation of latent TB or acquisition of new cases.
The optimal time frame between starting preventive treatment for latent TB and starting TNFi treatment is unknown. Given the low numbers of bacilli present in latent TB infection, it is likely that leaving a long time between starting preventive treatment and TNF blockade is unnecessary. Although there are no prospective trials assessing this question, observational data from Spain suggest that starting isoniazid treatment 1 month before TNFi substantially reduces the risk of latent TB reactivation (category C evidence476 ,490 ,491). Before starting preventive anti-TB treatment in accordance with local guidelines, consultation with an infectious disease specialist should be considered. There are case reports of restarting TNFi treatment after successful completion of a full course of anti-TB treatment (category C evidence496).
Other opportunistic infections
Other opportunistic infections have been reported in patients treated with TNFi (category C/D evidence155 ,254 ,316 ,498–502). Particular vigilance is needed when considering infections whose containment is macrophage/granuloma dependent, such as listeriosis, non-tuberculous mycobacteria (category D evidence489 ,502), coccidiomycosis or histoplasmosis (including reactivation of latent histoplasmosis) (category C/D evidence155 ,254 ,498 ,500 ,501).
A British registry study found that the rate of intracellular infections among patients with RA treated with TNFi was 200/100 000 and significantly higher than in similar patients treated with DMARDs or corticosteroids (category C/D evidence485 ,499). A French registry analysed age- and sex-adjusted rates of opportunistic infections (excluding TB) in 152 of 100 000 TNFi users. In case–control analysis, the risk of these infections was significantly higher in those patients using monoclonal antibodies or steroids (>10 mg/day) than in those using etanercept (category C evidence503).
Serious bacterial infections (usually defined as bacterial infections requiring intravenous antibiotics or admission to hospital) have also been seen in patients receiving TNFi. They occur at rates of between 0.04 and 0.09/patient-year compared with 0.01–0.06/patient-year in controls using other DMARDs (category C evidence50 ,345–347 ,355 ,504). Risk ratios of 1–3 were documented (category B/C, evidence355). TNFi should not be given in the presence of active serious infections and/or opportunistic infections, including septic arthritis, infected prostheses, acute abscess, osteomyelitis, sepsis, systemic fungal infections and listeriosis (category C/D evidence50 ,355 ,358).
Some studies indicate that serious infections in certain sites are more common when TNFi inhibitors are used, such as the skin, soft tissues and joints, and the risk may be highest during the first 6 months of treatment (category C evidence347 ,355).
A recent US, population-based, observational study suggested that the risk of serious infection in new users of TNFi was similar to that seen in patients who begin treatment with a new non-biological DMARD. These data may be influenced by new clinical practices, coinciding with changes in screening and prevention practices associated with starting TNFi (category B evidence505). Also, decreased RA disease activity and a reduction or elimination of concomitant steroid use may decrease infection risk over time. A large registry study from Germany evaluated the influence of concomitant treatment, cormorbidities and disease activity after starting a biological agent and documented a small, independent increased relative risk for serious infection of 1.8 (1.2 to 2.7) associated with TNFi treatment (category C evidence506).
Three meta-analyses of the safety of TNFi across indications demonstrated the highest rates of serious infections in patients with RA compared with those with PsA, AS and JIA. In these studies, patients with RA were more likely to be receiving concomitant DMARD and corticosteroid treatment (category A evidence441 ,479 ,507 ,508).
Among patients with JIA in an open study, the rate of serious infections did not differ among MTX, etanercept and etanercept plus MTX groups in clinical trials (category C evidence25), and data were inconclusive from the Dutch ABC registry 22056397. Further, in a large cohort study in the UK, discontinuations for infections in patients with JIA were only about 1% (category B evidence509).
Biological agents and high-dose corticosteroids affect acute phase reactions (eg, erythrocyte sedimentation rate < CRP), irrespective of the cause of the inflammation. Therefore care needs to be exercised when these measures are used to help diagnose infection in the presence of these agents (category B/C evidence346 ,347 ,510–512).
The incidence of serious infections is approximately doubled when IL-1 receptor antagonist or abatacept is used with any of the TNFi in combination (category A evidence; FDA51 ,52 ,260 ,420). The use of TNFi plus IL-1 blocking agents or abatacept in combination is not recommended.
Patients should be screened for viral hepatitis before starting TNFi, as the long-term safety of TNFi in patients with chronic active viral hepatitis (hepatitis B and C) is not known.
In patients with hepatitis C and RA, several observational studies and one systematic review in 153 hepatitis C-infected patients treated with IFNα co-therapy showed no increased incidence of toxicity (eg, raised LFTs or viral load) associated with TNFi treatment. Only one definite case and five probable cases of worsening HCV disease were found, while 120 patients remained stable and 29 patients improved (category B/C/D evidence469 ,472 ,513 ,514).
The lack of HCV reactivation was supported by small case studies of patients with psoriasis treated with etanercept (n=3) (category D evidence515).
Patients with hepatitis B treated with adalimumab, etanercept and infliximab have experienced worsening of their symptoms, increased viral load and, in some cases, hepatic failure (especially after stopping the TNFi) (category C/D evidence404 ,469 ,471–474 ,505 ,513 ,516–518).
Although specific warnings about hepatitis B reactivation have been added to the US label by the FDA, TNFi have been used in patients with known persistent hepatitis B infection, with concomitant hepatitis B treatment (category B evidence518). If hepatitis B infection is discovered during use of TNFi, prophylactic antiviral treatment can be employed (category C evidence516).
Small cases series have been reported, in which TNFi were used in patients with evidence of previous hepatitis B (hepatitis B surface antibody positive, hepatitis B surface antigen and HBV DNA negative) with only transient increases in transaminases and no change in viral load (category D evidence472).
In contrast, another case series documented an increased risk of hepatitis B (category C/D evidence519 ,520). The hepatitis B viral load should be carefully monitored if TNFi are used in patients with previous hepatitis B.
Hepatitis: Two small studies in patients with PsA and SpA with inactive hepatitis B and normal liver enzymes and viral load were treated with TNFi (adalimumab, etanercept and infliximab) plus antiviral agents without reactivation (category C/D evidence521).
Infliximab did not increase aspartate aminotransferase/alanine aminotransferase or reactivate viraemia over 26 weeks in 41 patients with RA with normal baseline liver function, negative hepatitis B surface antigen and some form of anti-hepatitis B antibody (category D evidence522). However, a comprehensive analysis of potential HBV reactivation in patients using TNFi (n=257 cases) reported it as occurring in 39%, with a higher risk in patients with prior immunosuppressive treatment (96% vs 70%) and lower risk in those using antiviral prophylaxis (23% vs 62%). Five of 257 patients developed acute liver failure and four died. The increased risk of HBV reactivation, including liver failure, when using TNFi requires strict examination of viral status in patients with RA. Antibody response to hepatitis B vaccination was inadequate in 96% (RA) and 89% (SpA) of infliximab-treated patients (category D evidence523). Hepatitis B viral status should be checked before starting TNFi.
Herpes zoster: Recent observational studies and series reported either no increased risk (in patients with AS) or a small increased risk of herpes zoster (in patients with RA) with monoclonal antibodies, while another observational study found no increase in risk with TNFi treatment as a whole, and reported significantly lower risks for etanercept and adalimumab than for infliximab (category B/D evidence505 ,510 ,524 ,525). In a large observational cohort of over 20 000 patients, no increased risk of herpes zoster was seen with TNFi initiation (category C evidence524). Vaccination for herpes zoster before starting a TNF blocking agent is still advised (category D evidence525).
Injection site/infusion reactions
In placebo-controlled trials, injection site reactions, most of which were mild to moderate (but some of which resulted in drug discontinuation), were more common with subcutaneously administered TNFi than with placebo (category A/B/C/D evidence155 ,236 ,254 ,258 ,316 ,419 ,430). A survey indicates that administration reactions after adalimumab and etanercept may be more common than previously thought, occurring in >50% of patients, although the great majority did not interrupt treatment; 13% of the reactions were moderate to severe (category C evidence526).
One study indicates that human anti-chimeric antibodies against infliximab were associated with decreased response and increased infusion reactions (category C evidence527).
Acute reactions after adalimumab or golimumab administration are uncommon and are usually mild to moderate, but may, rarely, be serious (category A/B evidence155 ,243 ,254 ,283 ,528). In most instances, infusion reactions can be treated by the use of corticosteroids or antihistamines or by slowing the infusion rate (category B/C evidence511 ,527 ,529).
Hypersensitivity reactions (eg, urticaria/angioedema) occurred in 54 (8%) of 671 patients among all TNFi. Of the group who had infusion reactions, 60.8% were receiving infliximab, 25% etanercept and 11% adalimumab (category C evidence530).
Rare instances of acute, severe and even fatal interstitial lung disease (ILD) have been reported in patients using TNFi (category C evidence531–536). However, in an analysis from a British registry, the mortality of patients with RA with ILD after treatment with TNFi was no higher than with traditional DMARD treatment (23% vs 21%), despite probable channelling bias toward more use of TNFi in those with underlying ILD or organising pneumonia (category C evidence455 ,534).
More than 24 instances of ILD have been documented associated with TNFi (category C evidence438 ,535 ,536 ,538). An investigational observational cohort indicated improvement in pulmonary function and high-resolution CT (HRCT) of the lung in patients with RA or AS with baseline HRCT abnormalities, using a recently published HRCT scoring method (category C evidence539–541).
The incidence of lymphoma is increased in chronic inflammatory diseases such as RA. This increase is associated with high disease activity (category C evidence542 ,543). In most studies the risk for lymphoma (especially NHL) is increased two- to fivefold in patients with RA compared with the general population (category B/D evidence432 ,544 ,545). A similar risk is seen in patients with RA who have received TNFi treatment (category A/B/C/D evidence325 ,430 ,432 ,542–544 546–551). It is unclear if the risk of lymphoma is increased.
In patients with COPD, there may be an increased risk of lung cancers associated with TNFi treatment. In a trial of patients with COPD assigned to infliximab versus placebo, nine developed lung cancers during the trial, and another four lung cancers were found during open-label follow-up (category A evidence551 ,552). Lung cancer appears to be increased in RA, although whether this is due to disease activity or confounding factors is not known (category C evidence551 ,552). In a study of Wegener's granulomatosis, the use of etanercept with cyclophosphamide was associated with six solid malignancies compared with none in the cyclophosphamide placebo group (category A evidence553). After starting TNFi, cancer risk does not increase with time (category B evidence554).
The concomitant use of azathioprine with infliximab in adolescents has been associated with the occurrence of rare hepatosplenic lymphomas. Malignancies in JIA have been reported, but data did not show an increased frequency of malignancy in patients with JIA treated with TNFi compared with other patients with JIA (category C evidence, FDA letter293 ,555–557).
Although two meta-analyses (with infliximab and adalimumab) reported a higher rate of solid malignancies, including skin (category A evidence558 ,559), numerous other large observational databases, three meta-analyses and other studies did not show an increased incidence of solid tumours in patients receiving TNFi (category A/B/C/D evidence18–21 ,432 ,542 ,545–549 551–554 560–566).
Open-label extension studies (>5 years) of adalimumab demonstrated a decrease in overall malignancies (standardised incidence ratio=0.64 (0. 53–0.76), decreased overall mortality (standardised mortality ratio=0.71 (0.57–0.81) and increased lymphomas (standardised incidence ratio=1.99 (1.11–3.28) versus the general population (SEER database) (category A/D evidence566).
Despite many methodological problems associated with observational data, the totality of data support the notion that TNFi (from data on adalimumab, etanercept or infliximab) are not associated with cancer overall (category C evidence566).
Vigilance for the occurrence of lymphomas and other malignancies (including recurrence of solid tumours) remains appropriate in patients treated with TNFi.
Skin cancers: A literature review of skin malignancies in patients with RA, PsA and AS suggests an increased frequency of non-melanotic skin cancers (category A/B evidence526 ,560 ,565 ,566). This was supported by another cohort of 20 648 patients with RA in a systematic review and meta-analysis, where there was a small increase in non-melanotic skin cancers (HR=1.45; 95% CI >1.1 to 1.76) (category C evidence567) but not by another meta-analysis (category A evidence568). Among patients with rheumatic disease using certolizumab or golimumab (comparing RA), there was no increased incidence of non-melanotic skin cancers (category A evidence569).
Consideration should be given to screening patients for cutaneous malignancy before starting biological agents and monitoring during treatment.
Rare instances of central and peripheral demyelinating syndromes, including multiple sclerosis, optic neuritis and Guillan–Barré syndrome, have been reported in patients using TNFi (category C/D evidence570–579). In some cases, but not all, these syndromes improved after withdrawal of TNFi treatment and steroids were given. Accordingly, TNFi treatment should not be given to patients with a history of demyelinating disease, multiple sclerosis or optic neuritis (category C/D evidence570–579).
Rare reports of unusual adverse events
Unusual adverse events have been seen in conjunction with the administration of TNFi, although a causal connection has not been proved. There have been case reports of membranous glomerulonephritis (category D evidence580), new onset uveitis (category D evidence407 ,581), polychondritis (category D evidence582) and inflammatory bowel disease (category D evidence582a).
Risks during pregnancy
The safety of anti-TNF treatment during pregnancy is unknown. Experts disagree about whether TNFi should be stopped when pregnancy is being considered or whether they can be continued throughout pregnancy. Some studies found no increased fetal loss or spontaneous abortion during the use of TNFi (category C/D evidence583–586).
A rare combination of congenital abnormalities (VACTERL—vertebral abnormalities, anal atresia, cardiac defect, tracheoesophageal, renal and limb abnormalities) and partial VACTERL defect have been reported rarely, although the risk and causality is unclear (category C evidence585).
According to the US FDA, this drug class is considered category B, meaning no evidence of risk in humans (if no adequate human studies have been done, no animal studies have been done or animal studies show risk but human studies do not). A systematic review of 667 pregnancies came to the conclusion that, to date, no definite harm to pregnancy can be ascribed to TNFi (FDA category B evidence130).
A single-patient study examined maternal serum, placenta, breast milk and infant etanercept levels, finding about 3% transfer of etanercept from serum to placenta and no transfer of etanercept in breast milk (category D evidence586).
A recent study indicated that there was no increase in VACTERL syndrome among TNFi users compared with the general population (category C evidence587).
Eighty-eight live births in a total of 130 pregnancies were reported in patients with RA with high disease activity treated with TNFi from a British register of biological agents. The rate of spontaneous abortion—24%—was highest among patients who were exposed to TNFi at the time of conception. This was higher than the 14% spontaneous abortions in those with prior exposure to TNFi and the 10% in the control group (category C evidence588).
Male sexual function
Limited data show that sperm volume and function do not differ from normal. In a comparison of 26 men using TNFi with those not using the drug, the TNFi-treated men were equally able to father normal children and sperm seemed either unaffected or improved (category B evidence589).
More than 200 cases of psoriasis, psoriaform lesions or exacerbation of psoriasis have been reported with the use of all TNFi and some have noted more frequent psoriatic lesions after the use of TNF antibodies than after using etanercept. In some, but not all, cases, switching TNFi allowed continuation of treatment without recrudescence of skin lesions (category D evidence590–594).
In addition, rare cases of Henoch–Schönlein purpura, Stevens–Johnson syndrome, digital vasculitis, erythema multiforme, toxic epidermal necrolysis, granulomatous reactions in skin and lungs, bullous pemphigus were reported (category D evidence529 ,595).
Appropriate vaccinations should be carried out before initiation of TNFi treatment, according to national guidelines. This includes herpes zoster (category D evidence525).
TNFi do not usually adversely affect the development of protective antibodies after vaccination with flu or polysaccharide pneumococcal vaccine, although there is a small decrease in the prevalence of adequate protection and a decrease in the titre of response, especially in combination with MTX (category A/B/D evidence598–601). Vaccination with live attenuated vaccines (eg, nasal flu vaccine, BCG, yellow fever, herpes zoster) is not recommended, although the incidence of dissemination (especially herpes zoster) is very low (category D evidence525).
Measles, mumps and rubella vaccination with appropriate secondary response has been reported in patients with JIA treated with etanercept and MTX, even though in clinical practice live attenuated vaccines are not generally recommended in children with JIA treated with MTX (category D602). In some patients whose flu titres do not rise with an initial vaccination, repeat vaccination may be effective (category D evidence603).
As expected, the degree of flu and pneumococcal vaccination responses decreases after TNFi, although serological protection is adequate in most cases (category C/D evidence604–606).
Use of the inactivated vaccine against hepatitis A was effective in 43 of 47 children (91%), while none of four children with systemic-onset JIA receiving TNFi developed an immune response (category D evidence607).
Other biological agents
Alefacept (approved in the USA for psoriasis but not PsA) is a fully human fusion protein that blocks interaction between lymphocyte function-associated antigen-3 on antigen-presenting cells and CD2 on T cells, leading to decreased T cell activation and deletion of certain T cell clones. It is approved for the treatment of psoriasis in the USA. A phase 2 trial in PsA demonstrated only modest efficacy in joints and skin at 24 weeks (category B evidence608). A second course (each course is 12 weekly intramuscular injections followed by 12 weeks without injections) during an open-label extension demonstrated sustained articular efficacy (category A evidence608).
Ustekinumab is an inhibitor of IL-12 and IL-23 which acts in both the TH17 and TH1 pathways of inflammation and is approved for the treatment of psoriasis, dosed at 0, 4 and then every 12 weeks subcutaneously (category B evidence609). For skin, it was slightly more cost-effective than etanercept in one study, although local conditions can drastically alter cost-effectiveness estimates (category C evidence610). Modest efficacy has been demonstrated in a phase 2 trial in PsA.611 ,612
The treatment of RA and other rheumatic diseases and conditions of altered immunoreactivity has changed dramatically for the better since the introduction of biological agents into the armamentarium of the treating physician. It is hoped that this consensus statement will provide guidance to the clinician in his/her efforts to improve the QoL of patients with these conditions. In addition, this consensus statement should provide evidence-based support for the selection of agents and justification for their use.
Appendix: categories of evidence
Category A evidence: based on evidence from at least one RCT or meta-analyses of RCTs. Also includes reviews if these contain category A references.
Category B evidence: based on evidence from at least one controlled trial without randomisation or at least one other type of experimental study, or on extrapolated recommendations from RCTs or meta-analyses.
Category C evidence: based on non-experimental descriptive studies such as comparative studies, correlational studies and case–control studies which are extrapolated from RCTs, non-RCTs or other experimental studies.
Category D evidence: based on expert committee reports or opinions or clinical experience of respected authorities or both, or extrapolated recommendations from RCTs, meta-analyses, non-RCTs, experimental studies or non-experimental descriptive studies. Also includes all abstracts.