The British Society for Rheumatology Guideline for the Management of Systemic Lupus Erythematosus in Adults

Caroline Gordon; Maame-Boatemaa Amissah-Arthur; Mary Gayed; Sue Brown; Ian N. Bruce; David D'Cruz; Benjamin Empson; Bridget Griffiths; David Jayne; Munther Khamashta; Liz Lightstone; Peter Norton; Yvonne Norton; Karen Schreiber; David Isenberg


Rheumatology. 2018;57(1):e1-e45. 

In This Article


Overview of the Management of Moderate Lupus

Immunosuppressive cytotoxic agents should be used with CSs, while continuing anti-malarials and avoidance of UV radiation, to reduce disease activity in moderate lupus (Table 7), prevent the risk of further flares and lower the risk of damage accrual due to disease and CSs, because they act as steroid-sparing agents. Despite their widespread use in clinical practice and as background standard of care therapy in clinical trials, there are only a few RCTs demonstrating the efficacy of CSs and other immunosuppressive agents for the management of moderate lupus. Additional drugs should be considered if HCQ is insufficient or not tolerated and can be used in addition to HCQ. The evidence supporting the use of MTX has been discussed above, and the evidence supporting the use of CSs, AZA, MMF, calcineurin inhibitors (ciclosporin and tacrolimus) and LEF are discussed in this section. For patients who do not respond to these drugs, the biologic drugs rituximab and belimumab may be considered. It should be noted that there is a specific NHS England 2013 Interim Clinical Commissioning Policy Statement for rituximab in adult SLE patients,[267] and NICE guidance for the use of belimumab in active autoantibody-positive SLE in adults has been published in 2016.[324] Patients being considered for these drugs should be discussed with and/or seen by a specialist lupus centre with experience in using these drugs. The patients should meet specific criteria and be entered in to the BILAG Biologics Register (see below and Figure 1). For patients not requiring biologics, suggested initial target dosing regimens for active disease (as used in most studies) and lower maintenance dosing regimens to prevent recurrence of disease once patients are stable are shown in Table 7. The actual regimen used for individual patients will depend on the clinical picture and the treatment history. It is important to increase the dose and/or change treatment if patients fail to respond in the expected time frame. The LOEs and GORs for all the drugs used to treat lupus are summarized in Table 2.

Figure 1.

Summary of NICE and NHS England guidance for the use of belimumab and rituximab in patients with SLE Belimumab is licensed and NICE-approved (Belimumab for active autoantibody-positive systemic lupus erythematosus: TA397, published June 2016) and should be considered first [324]. Rituximab is not licensed and should only be used according to the NHS England Interim Clinical Commissioning Policy Statement: rituximab for the treatment of systemic lupus erythematosus in adults: published September 2013 A13/PS/a [267]. All patients receiving either drug must be enrolled in the BILAG Biologics Register and be managed at or in collaboration with a specialized centre.

CSs for Moderate Lupus

Summary. Higher doses of oral CSs are required initially than are required for mild lupus, for example prednisolone at up to 0.5 mg/kg/day, and intermittent treatment with i.m. 80–120 mg MP or even i.v. doses of MP (up to 250 mg) are used as well as, or instead of, oral prednisolone to promote a quicker response with less total CS exposure. Prednisolone dosing should be reduced, as disease activity improves, to the lowest possible maintenance dose and stopped, if possible, as other immunosuppressive agents take effect over several weeks or months.

Evidence. There are no data comparing different oral CS regimens for the treatment of moderate lupus. Two controlled studies have shown that treating patients who are clinically stable but showing serological deterioration with a short course of moderate-dose CSs (e.g. 30 mg/day) can prevent more flares than placebo and lead to improvement in serological markers.[46,60] However, there is a risk of treating patients that will not flare, and this approach is not recommended due to the side effects of CSs.

There are some data supporting the use of 100 mg i.v. MP pulses in non-renal lupus as an alternative to 1000 mg pulses,[143] and for 1000 mg pulses on three occasions in patients with moderate or severe lupus, with very little oral prednisolone.[146] The data supporting the use of i.v. pulses of 500 or 1000 mg are discussed further below in the section on the management of severe lupus.[148,326] There is one open-label RCT[142] comparing triamcinolone 100 mg given as an i.m. injection with a short course of oral MP tapered over 1 week. Overall, there was little difference between the regimens but some improvement was seen more quickly with the triamcinolone injection.

Conclusions. Overall the LOE for CSs by i.m. or i.v. injection in non-renal moderate lupus is 2+ and GOR is C.

AZA for Moderate Lupus (Non-renal Disease)

Summary. AZA is not licensed for the treatment of lupus, but has been used for over 40 years, and it is the most frequently used cytotoxic agent[327] in lupus. AZA treatment (1–2.5 mg/kg/day orally) has been associated with prevention of flares and a reduction in CS dosage (see below and Table 2). It is usually started in patients with moderate lupus activity (Table 7) in conjunction with CSs, as it can take up to 3 months to be effective. It is also used for maintenance therapy after remission or significant response has been achieved with other agents used to treat severe lupus (such as CYC) that are less suitable for long-term therapy, particularly in women desiring pregnancy, or who are pregnant or breast-feeding.[24,25,239,328] Most of the evidence (and the only double-blind RCTs) supporting its use relate to the management of LN.[24,25] Only papers discussing the management of non-renal lupus with AZA are discussed here, although in some cases the studies included renal and non-renal patients. There is no evidence that it prevents atherosclerosis or other forms of damage.[12,329]

Evidence. The first reports of AZA being used for renal and non-renal manifestations of lupus with CSs appeared in the late 1960s and 1970s.[149–151,153,330,331] Reduction in disease activity and flare rate and steroid-sparing effects were demonstrated in most of these open-label, controlled studies and in a case series.[158] AZA 200 mg daily was associated with an increased risk of significant liver dysfunction. There was no increased risk of infection, even starting at 3–4 mg/kg/day, but subsequent studies have used 2–2.5 mg/kg/day.

A prospective longitudinal open-label study[154] involving 17 SLE patients showed that AZA reduced lupus activity and anti-dsDNA antibody levels. Subsequently, in a retrospective study[155] with 61 SLE patients, suppression of anti-dsDNA antibodies by AZA (2 mg/kg/day) and low-dose prednisolone (7–12 mg/day) was associated with efficacy and better long-term outcome. However, the presence of renal disease, persistence of anti-dsDNA antibodies for at least 1 year after the beginning of treatment and reduction in AZA dosage to below 2 mg/kg/day predicted flares and was associated with a higher rate of lupus-related death.

An open-label, multicentre, RCT study of 89 SLE patients requiring 15 mg or more of prednisolone compared AZA (mean dose 2.1 mg/kg/day) with ciclosporin (mean dose 2.2 mg/kg/day) for its steroid-sparing properties.[152] The absolute mean change in prednisolone dose at 12 months, adjusted for baseline prednisolone dose, was not significantly different: 9.0 mg for ciclosporin (95% CI: 7.2, 10.8) and 10.7 mg for AZA (95% CI: 8.8, 12.7). There was no difference between groups in change in disease activity or number of flares, development of new damage, change in quality of life or numbers of patients discontinuing study drugs due to adverse events or lack of efficacy.[152] The conclusion was that both drugs can be used in lupus for their steroid-sparing properties, with appropriate monitoring.

AZA is usually well tolerated.[332] The main adverse events are nausea and vomiting, diarrhoea, flu-like illness with fever, rash, leucopenia and hepatotoxicity.[156,157,332–334] Side effects can occur soon after starting AZA and may require drug withdrawal.[156,335] Hepatic veno-occlusive disease is a rare adverse event, but autoimmune hepatitis can improve on AZA, so this is not a contra-indication to its use.[157] AZA is not excreted by the kidney, and it can be used in patients with renal impairment. Managing patients with lupus-related leucopenia with AZA can be difficult.[332,336] The enzyme thiopurine S-methyltransferase (TPMT) catalyses the inactivation of AZA. It is worth testing patients for TPMT[334] before starting AZA, as the very low level phenotype (homozygous deficiency that occurs in 0.3% Caucasians) is associated with potentially life-threatening bone marrow toxicity; otherwise, weekly full blood counts are required as the dose is increased over several weeks.[337,338] Those patients with intermediate TPMT levels due to a heterozygous state have an increased risk of leucopenia as well, and such testing does not remove the need for monitoring the effects of the drug on the full blood count[156,332] and liver function according to national or local guidelines.[337,338]

AZA does not cause infertility and has not been found to be teratogenic in clinical practice, despite theoretical concerns;[339,340] thus, it can be used in women planning conception and is compatible with pregnancy and breast-feeding.[24,98,239] It may reduce the response to some immunizations,[341–344] but this is not a contra-indication to immunization except with live viruses.[74,292] There is no evidence that AZA increases the risk of malignancy in lupus patients,[301,345] but it may increase the risk of cervical dysplasia.[346]

Conclusions. Although the data for AZA in non-renal lupus are much weaker than the data supporting its use in LN (see below), there are four open-label RCTs, three prospective cohort studies, two retrospective cohort studies and one case series supporting the use of AZA for non-renal lupus: overall LOE 2+, GOR C.

MMF for Moderate Lupus (Non-renal Disease)

Summary. There are increasing data showing that MMF in combination with CSs reduces moderate and severe lupus disease activity, reduces renal and non-renal flares, is associated with CS-sparing properties and is tolerated well (see Table 2 and Table 7 for suggested treatment strategies). However, there are no placebo-controlled double-blind RCTs specifically designed to assess the use of MMF in non-renal lupus. It is teratogenic and is contra-indicated in women trying to conceive, or who are pregnant or breast-feeding.

Evidence. The first systematic review of MMF (2–3 g daily) in non-renal lupus was published by Mok in 2007[170] and reviewed 20 papers in terms of the response of specific clinical features (up to 2006) and steroid-sparing properties. This systematic review included patients mostly refractory to other therapies who were treated with MMF in uncontrolled studies for arthritis, renal, haematological and cutaneous manifestations, and a few with neuropsychiatric manifestations, and also covered the use of MMF in prevention of flare in a small prospective study of patients with rising anti-dsDNA antibody levels.[162–164,347]

A later systematic review[133] with a literature search up to end of October 2011 provided further evidence that MMF treatment is associated with reductions in disease activity, flare rate and prednisone dose and included data from five cohort studies[162–166] and from the Aspreva Lupus Management Study (ALMS) trial in LN that specifically reported on non-renal lupus manifestations (see below).[159] Further supporting evidence for MMF comes from a small case series[169] and a study[348] showing that mycophenolic acid (MPA) levels vary between patients and that higher trough levels were associated with less risk of disease flare. MPA levels were more closely associated with efficacy and safety than the dose of MMF. This test is available in some hospitals, but the target trough level of 3.5–4.5 mg/l was recommended to be tested in a controlled trial before being widely applied.

The beneficial effects of MMF on non-renal disease activity[159] were demonstrated in a 6-month open-label RCT (ALMS) that compared oral MMF (target dose 3 g/day, median exposure 2.6 g/day) with pulses of i.v. CYC (0.5–1.0 g/month) as induction treatment for biopsy-proven LN.[349] All patients received prednisone starting at 60 mg/day that was tapered to 10 mg/day. There was induction of remission in >80% of patients treated with MMF for active disease at baseline in mucocutaneous, musculoskeletal, cardiorespiratory and vasculitis systems in addition to renal response in 56% (the primary end point).[349] There were no flares in the patients on MMF, and complement levels and titres of anti-dsDNA antibodies normalized. Very similar renal and non-renal responses were seen in those given CYC.[159] However, more Black and Hispanic patients responded to MMF than i.v. CYC, and further trials are required to assess the role of race, ethnicity and geographical region on treatment response.[350]

In the maintenance phase of ALMS,[160] 227 patients from the 6-month induction study who met the renal clinical response criteria were randomized again to MMF (2 g/day) or AZA (2 mg/kg/day) in a 36-month, double-blind, double-dummy, phase III RCT.[160] Prednisolone ⩽10 mg/day or its equivalent was allowed and was taken by 90% of the MMF group (n = 116) and 87% of the AZA group (n = 111). Secondary end points included an analysis of non-renal severe flare. Severe non-renal flare rates did not differ between groups: 6.9% for the MMF group and 6.3% for the AZA group. There were no significant differences in the changes in anti-dsDNA antibodies or complement levels between groups. However, MMF was superior to AZA in various renal parameters related to maintaining a renal response and in preventing renal relapse in these LN patients, irrespective of which induction treatment had led to their initial response, race and geographical region.[160] Adverse events were common in both groups (>95%) (mostly minor infections and gastrointestinal disorders). Serious adverse events occurred in 24% of the MMF group and 33% of the AZA group (P = 0.11). The rate of withdrawal due to adverse events was lower with MMF than AZA (25% vs 40%, P = 0.02).

Another randomized open-label controlled trial,[161] in Caucasians predominantly, compared MMF (mean 2 g/day) and AZA (mean 124 mg/day) for maintenance therapy over 36 months, starting at week 12 after induction with a short course of i.v. CYC (6 × 500 mg over 10 weeks) for the management of biopsy-proven proliferative LN. All patients initially received three i.v. pulses of MP and were tapered from 0.5 mg/kg/day prednisone down to 5 mg/day at week 52 and then tapered further and stopped if possible. Both regimens were well tolerated, and there was comparable improvement in renal end points and non-renal parameters, including disease activity indices and C3 levels in both groups. There were less renal flares and less haematological adverse events with MMF than AZA (though this was not statistically significant in this study).

Since the systematic review,[133] further studies reporting reduction in disease activity included a retrospective review of patients treated with MMF that found a significant reduction in mean weekly steroid dosage (from about 12.5 to 3 mg/day prednisone).[167] A single-centre retrospective cohort study[168] involving 135 patients with SLE (50% with renal disease) and 43 patients with systemic vasculitis treated with MMF reported good responses in 46% of patients, and the mean prednisolone dosage was significantly reduced from 22 to 8 mg/day at 12 months. These and other studies have shown that adverse events occur in up to 44% of patients over 5 years: mostly mild gastrointestinal intolerance and infections, with leucopenia and hospitalization rare. In one study most patients tolerated the drug well, with 73% of patients on the drug at 12 months, and there was no relationship between adverse events and dose (250 mg to 3 g daily).[351] However, there have been increasing reports of teratogenicity, and it should be stopped at least 6 weeks before a planned pregnancy, and MMF should not be taken by women who are pregnant or breast-feeding.[239]

Yahya et al.[172] reported on a small open-label prospective study of 14 non-renal lupus patients randomized to mycophenolate sodium (MS) or standard care and showed that MS treatment was safe and was associated with reduced disease activity. A randomized open-label trial[171] of 40 patients with primary systemic vasculitis or SLE compared MMF (2000 mg/day) and enteric-coated MS (1440 mg/day). The composite primary end point was treatment failure and/or drug intolerance over 12 months. MS was anticipated to be tolerated better, but no difference in tolerance was observed. Although MS was associated with slightly better efficacy, this may have been due to imbalance in factors affecting remission and relapse, despite randomization with minimization. This study did not support the use of MS as a better tolerated and efficacious alternative to MMF for routine use, but MS could be considered in patients with gastrointestinal side effects from MMF.

Conclusions. The evidence that MMF reduces disease activity, lupus flare and has steroid-sparing properties in non-renal lupus comes from two systematic reviews, three open-label RCTs in LN and seven cohort studies: LOE 2 ++, GOR B. MPA/sodium (MS) may be considered in patients intolerant of MMF based on two studies (LOE three, GOR D).

Ciclosporin and Tacrolimus for Moderate Lupus (Non-renal Disease)

Summary. Ciclosporin and tacrolimus do not cause myelosuppression and have the ability to reduce moderate disease activity (Tables 2 and 7). There is more evidence for ciclosporin in non-renal lupus, and it has been particularly helpful in the treatment of cytopenias, where there is likely to be difficulty distinguishing cytopenias due to lupus from cytopenias due to drugs such as AZA, MTX and MMF. Both ciclosporin and tacrolimus can be used (at the lowest possible dose) in women planning pregnancy, and in those who are pregnant or breast-feeding.[239]

Evidence. There are two open-label RCTs[152,173] and eight non-renal cohort studies supporting the use of ciclosporin at doses of ⩽2.5 mg/kg/day in patients with normal renal function, although a systematic review[133] that included details of two open-label RCTs and a brief summary of six of the cohort studies reported that there was not much evidence supporting the use of ciclosporin in lupus because there were no double-blind, placebo-controlled RCTs.

Nevertheless, the open-label RCTs suggested that ciclosporin reduced disease activity as well as AZA did[152] and better than CSs alone,[173] and that ciclosporin treatment was associated with significant CS-sparing properties in both RCTs, equivalent to that of AZA in one trial[152] as reported previously by the cohort studies. These included two prospective cohort studies[174,175] that showed significant reduction in disease activity at 6 months, with most benefit in patients with renal and/or haematological manifestations, and response maintained to 24 months in one study.[175] Three retrospective studies[176–178] reported a reduction in disease activity and/or flares (particularly haematological manifestations such as thrombocytopenia), and significant steroid-sparing properties were reported in two of these studies.[175,177]

In the first of two additional studies not mentioned in the systematic review, ciclosporin was shown to treat thrombocytopenia in six patients,[179] three of whom were able to stop CSs. In the second study,[180] a retrospective cohort study, ciclosporin was used to manage 40 refractory lupus patients, including 11 patients with neurological conditions and 7 with overlap syndromes, as well as 18 with LN. The study showed reduction in disease activity and only mild transient adverse events not requiring discontinuation.

Adverse events were the focus of another study[181] with doses up to 5 mg/kg/day, so it was not surprising that adverse events were reported in 63%, but these led to discontinuation in only 16% and were reversible within 3 months of stopping the drug, consistent with many other reports. Ciclosporin treatment can cause hypertrichosis, gum hypertrophy, hypertension, paresthesiae, tremor, gastrointestinal symptoms and impaired renal function, especially at higher doses (>3 mg/kg/day). It is best used at lower doses (⩽2.5 mg/kg/day) as that is more tolerable and rarely causes permanent nephrotoxicity if carefully monitored. In the open-label RCT,[152] there were no unexpected adverse events, and with appropriate monitoring of renal function and blood pressure, it was not discontinued due to adverse events or inefficacy more often than AZA.

There are two reports of tacrolimus in non-renal lupus and they were included in the systematic review.[133] The first was a small retrospective cohort study[182] with 10 non-renal patients showing significant reductions in SLEDAI and prednisolone over 1 year on 1–3 mg daily. The second was an open-label prospective study[183] with 21 mostly non-renal patients showing reduction in SLEDAI score over 6 months and no serious side effects, but 29% withdrew due to inefficacy and 10% due to adverse events.

Conclusions. Overall, the LOE for ciclosporin in non-renal lupus from two open-label RCTs, eight non-renal cohort studies and one systematic review is 2+ and GOR is C.

The LOE for tacrolimus from two studies in non-renal lupus and one systematic review is three and GOR is D.

LEF in Moderate Lupus

Summary. The systematic review[133] and our search found little evidence for efficacy and safety of LEF in lupus patients, with only two small studies in the literature. This drug can be considered in patients refractory to, not suitable for or intolerant of MTX, AZA, MMF and calcineurin inhibitors, for whom CYC, rituximab and belimumab are not suitable or not available. It is not suitable for women considering pregnancy, and a cholestyramine washout is required if pregnancy is desired or occurs while it is being taken.[239]

Evidence. There was a randomized, double-blind, placebo-controlled trial in moderate SLE patients, with only six patients in each group.[184] A significant reduction in SLEDAI and prednisone occurred in both groups over 24 weeks. The LEF group showed significantly greater mean reduction in SLEDAI score, but there was no difference in steroid reduction between the groups. Side effects included transiently abnormal alanine aminotransferase (ALT), leucopenia and hypertension. There was a retrospective analysis of 18 patients who received LEF,[185] but 4 patients withdrew (3 due to adverse events, including 1 with rash), and only 9/14 achieved lower SLEDAI scores after 2–3 months of therapy.

Conclusions. Overall the LOE for LEF for reducing non-renal lupus disease activity from two studies is three and the GOR is D. Caution is advised about its use in those with pre-existing subacute cutaneous lupus, as this may worsen as observed in other non-lupus studies.

Rituximab for Refractory Moderate Lupus

Summary. Rituximab can be prescribed and reimbursed in the UK currently according to the NHS England 2013 Interim Clinical Commissioning Policy Statement for rituximab in adult SLE patients[267] who have two or more systems with BILAG B scores; or have severe BILAG A level disease activity, using the BILAG-2004 index;[268,269] or have a SLEDAI-2 K score[270] >6 if they have failed two or more immunosuppressive agents (due to inefficacy or intolerance), at least one of which must be MMF or CYC; or need unacceptably high doses of steroids to achieve lower level of disease activity.

The patients must be managed in conjunction with a specialist centre for lupus and be entered in to the BILAG Biologics Register for standardized reporting of outcome (see Figure 1 flowchart for eligibility and response criteria). This is essential for providing more open-label data in a prospective study with control patients treated with other immunosuppressive therapies, given the failure of the international double-blind, placebo-controlled lupus trials to meet their primary end points, as discussed below (EXPLORER for active non-renal disease[190,191] and LUNAR for LN[352]). This policy was agreed as a result of the increasing published evidence supporting the efficacy of rituximab in refractory lupus patients, who are likely to differ from those recruited to trials where there was no requirement to have failed conventional therapy. Pregnancy should be avoided for at least 6 months after exposure to rituximab.[239]

Evidence. The current evidence supporting the efficacy and safety of rituximab in non-renal lupus was most recently reported in a systematic review[200] in 2014 by Cobo-Ibanez with a literature search up to June 2013. This included the non-renal RCT EXPLORER[190] and its exploratory analysis,[191] 2 open-label phase I/II trials[192,193] and 22 cohort studies which analysed 1231 patients in total.[200] The 2 open-label trials[192,193] and 5 of the cohort studies had been discussed in a previous systematic review summarizing off-label use in 188 cases (including non-renal and renal patients in 9 cohort studies and 26 case series/reports published up to December 2007).[202]

The non-renal patients discussed in the systematic review by Cobo-Ibáñez et al.[200] were heterogeneous, but in general had active lupus disease unresponsive to steroids and/or immunosuppressants prior to treatment with rituximab. Treatment with rituximab was associated with a reduction in global disease activity over 3–9 months, with 64–91% achieving response, including patients with a reduction in complement and anti-dsDNA antibody levels, arthritis and thrombocytopenia. Evidence for a steroid-sparing effect was based on the 2 open-label trials and 10 of the cohort studies.[200] There were few significant adverse events in the RCT, 2 open-label studies and 20 cohort studies.[200] Relapses/flares did occur at variable times (3.7–18 months), although in the RCT there were numerically fewer severe BILAG A flares and longer time to these flares in the rituximab group compared with the placebo group, and this almost achieved statistical significance (hazard ratio = 0.61, P = 0.052).[191] Better clinical response after a second course was observed in 2 of the cohorts that studied retreatment,[200] and a further report supported this observation and that steroid reduction occurred after each of two courses of rituximab.[199] The evidence for rituximab treating mucocutaneous involvement was deemed weak,[200] and this may be explained by a recent report[353] specifically addressing 26 SLE patients with various subtypes of lupus rash, which observed that acute lupus rash responded whereas chronic cutaneous lupus (such as discoid rash) did not respond to rituximab and that new lesions with typical histology may appear despite confirmed B cell depletion.

Rituximab treatment early in the course of lupus disease, followed by AZA, was tried by Ezeonyeji et al.[194] specifically for its steroid-sparing effect in a pilot study with 8 SLE patients whose results were compared with 23 matched historical control patients treated conventionally.[194] Reduction in disease activity, a fall in anti-dsDNA antibodies and complement, and significant lower cumulative prednisolone at 6 months compared with controls was observed. There is also an open-label LN study suggesting that early rituximab with i.v. MP followed by MMF may avoid the use of oral CSs, and this regimen is currently being tested in a controlled randomized RCT called RITUXILUP.[354]

The Duxbury systematic review and meta-analysis[201] reported response rates for various disease activity measures for patients in the open-label studies of refractory lupus treated with rituximab also reviewed by Cobo-Ibáñez et al..[200] The Duxbury review and meta-analysis did include a section on LN (not discussed here) and included a few non-renal studies not in the Cobo-Ibáñez review, although the latter also included a few not in the Duxbury review. The BILAG index was used in 188 patients treated with rituximab in 8 open-label studies (3 prospective, 4 retrospective and 1 small case–control).[201] The pooled global response in seven of these studies was 83%. The complete response rate was 47% and the partial response rate was 38% in six studies. A significant reduction in anti-dsDNA antibodies was observed in 6 of the 8 studies and a significant rise in complement was observed in 5 of 6 studies. Various versions of the SLEDAI were used in 513 patients treated with rituximab in 12 open-label studies: 5 prospective, 6 retrospective and 1 open-label randomized trial, only 1 of which also analysed BILAG response. With SLEDAI the global response was 77% in 11 studies. In 6 studies the complete response rate was 57% and the partial response rate was 31%. Anti-dsDNA levels fell in 3 of 3 studies and complement rose in 2 of 3 studies.[201]

Publications from cohorts in Germany,[195] Italy[196] and Japan[197] have confirmed similar levels of efficacy with various disease activity measures and provided further safety data in another 264 patients. Long-term follow-up of 98 SLE patients treated with rituximab over a 12-year period has shown in a retrospective analysis that the group with longer duration of depletion (⩾12 months) was associated with a better response (greater decrease in BILAG score at 6 and 12 months) than those with shorter period of B cell depletion.[198]

The results of these open-label studies are much better than the response rates observed in the EXPLORER RCT (for rituximab vs placebo: complete 12% vs 16%, partial 17% vs 13%).[190] However, EXPLORER used more stringent BILAG response criteria than used in any other study,[201] but did observe a reduced rate and time to severe BILAG A flare.[191] High-dose CSs and background immunosuppression were used in both arms of the EXPLORER trial and may have reduced the ability to discriminate benefit from rituximab.[201] Patients on MTX as the background immunosuppressant derived more benefit from rituximab in a post hoc analysis than those in the placebo group,[190] and in contrast to those on background AZA or MMF.[190] Patients of Afro-American or Hispanic origin were also shown to benefit from rituximab in the RCT, in contrast to Caucasians.[190]

However, two case series reports have suggested that repeat courses of rituximab may increase the risk of hypogammaglobulinaemia and infection.[199,293] Progressive multifocal leukoencephalopathy (PML) has been reported in 17 SLE patients, of whom 5 had been treated with rituximab. It seems likely that immunosuppression, however it is achieved, is the key factor in the development of PML. Lupus patients may be at increased risk of developing PML compared with other rheumatic diseases.[355] The risk of rituximab causing PML in rheumatic diseases, including RA and SLE, has been estimated at 5/100 000, which is less than the risk observed with some other immunosuppressants in other diseases.[356]

Conclusions. There is now considerable evidence for the ability of rituximab to reduce disease activity in refractory non-renal SLE of moderate and severe severity, albeit mostly from cohort studies. There have been relatively few concerns in the individual reports and systematic reviews about adverse events, including infections, in lupus patients on rituximab. There is increasing evidence that rituximab has steroid-sparing properties, but further evidence for its use early in the disease course is needed. Overall, the LOE for rituximab from 3 systematic reviews (including a meta-analysis and 30 studies, including 1 RCT and 3 open-label trials for reducing disease activity and for steroid-sparing properties) is 2+ and the GOR is C.

Belimumab for Refractory Moderate Lupus

Summary. There have been two large phase III RCTs[203,204] investigating the use of belimumab in moderate–severe seropositive lupus (mostly musculoskeletal and cutaneous disease; as severe active renal and NPSLE disease were exclusions). All patients received steroids, HCQ and/or immunosuppressive drugs, with specific criteria for dosing changes allowed or contra-indicated in the protocol. Both trials showed a significantly increased proportion of responders to belimumab at a 10 mg/kg dose in addition to standard care. A variety of secondary end points were met, and there were no significant differences in adverse events, leading to the drug being approved and licensed by the US Food and Drug Administration and the European Medicines Agency. NICE guidance for use of belimumab in active autoantibody-positive SLE in adults has been published[324] and is summarized in Figure 1. Patients must have positive anti-dsDNA antibodies, low complement and a SELENA-SLEDAI score ⩾10 despite standard therapy. Patients should be recruited to the BILAG Biologics Register so that outcomes can be recorded, and treatment with belimumab should not be continued for >24 weeks unless the SELENA-SLEDAI score has improved by 4 points or more. Pregnancy should not occur while on belimumab, but first trimester exposure is unlikely to be harmful.[239]

Evidence. In the BLISS52 trial,[203] at week 52 the response rate with placebo was 44%, with belimumab 1 mg/kg it was 51% (P = 0.013) and with 10 mg/kg it was 58% (P = 0.001). In the BLISS76 trial,[204] the placebo response rate at week 52 was 34%, with belimumab 1 mg/kg it was 41% (P = 0.089) and with 10 mg/kg it was 43% (P = 0.017). The response rates at week 76 were a little lower in all groups. A meta-analysis of the response at 52 weeks in the phase II trial of belimumab[205] as well as BLISS 52 and BLISS 76 trials showed benefit for belimumab, with an odds ratio of 1.63 (95% CI: 1.27, 2.09).[209] Safety data from the phase II trial and its open-label extension have not shown any significant concerns and continued benefit for up to 7 years.[207,208] The most common side effects have been upper respiratory tract and urinary tract infections, arthralgia, headaches, fatigue and nausea. Serious infusion reactions and infections have been rare.[207,208] There have been two case reports of progressive multifocal leukoencephalopathy,[357,358] but there is no evidence that belimumab increases the risk more than other immunosuppressive regimens in SLE patients.[356]

Further post hoc analyses[359,360] on the pooled datasets from BLISS 52 and BLISS 76 trials have demonstrated that belimumab therapy was associated with significantly more patients showing improvements than with placebo in the most commonly affected musculoskeletal and mucocutaneous systems, and more immunological abnormalities normalized than with placebo.[359] Improvement was reported less consistently in other systems that were less often affected.[359] There was less worsening in haematological, immunological and renal parameters in those patients on belimumab than in those on placebo,[359] but as with improvement, effects were not always dose related. Serological improvements (reduction in anti-dsDNA antibodies and increase in C3/C4 levels, without reduction in memory T or B cell numbers or levels of anti-pneumococcal or anti-tetanus toxoid antibodies) have been reported.[361] This is consistent with the low rate of serious infections in the long-term open-label study of belimumab.[207,208]

Another pooled analysis of BLISS 52 and BLISS 76 trials identified that belimumab had most therapeutic benefit compared with standard therapy alone in patients with higher disease activity (SELENA-SLEDAI ⩾10), positive anti-dsDNA antibodies, low complement, or CS treatment at baseline.[206] Week 52 response rates in the low complement/anti-dsDNA–positive subgroup were 32% for placebo, 42% for belimumab 1 mg/kg (P = 0.002) and 52% for belimumab 10 mg/kg groups (P < 0.001). For the SELENA-SLEDAI ⩾10 subgroup, the response rates were 44%, 58% (P < 0.001) and 63% (P < 0.001), respectively. Belimumab was also shown to reduce severe flares and CS use and to improve health-related quality of life most in these more severe subgroups.[206] These analyses contributed to the decision by the European Medicines Agency to limit the market authorization for belimumab (Benlysta) to add-on therapy in adult patients with active autoantibody-positive SLE with a high degree of disease activity (e.g. positive anti-dsDNA and low complement) despite standard therapy.[362]

Conclusions. Treatment with belimumab in addition to standard therapy in autoantibody-positive SLE patients was associated with some improvements in clinical, laboratory and patient-reported outcome measures (compared with placebo in addition to standard therapy) and had a low risk of serious side effects. Based on the results of the two RCTs and the post hoc analyses, belimumab is considered by NICE to be cost-effective in the UK only for patients who meet the specific criteria[324] (see summary above and Figure 1), so availability is limited. The drug is being used in other countries, particularly in the USA, where the licence covers patients with moderate disease activity and only specifies that patients must have active, autoantibody-positive lupus and be receiving standard therapy (such as CSs, antimalarials, immunosuppressives and NSAIDs).[363] Overall, the LOE for belimumab in non-renal lupus from a meta-analysis, one phase II study, two phase III RCTs, their open-label extension study and post hoc analyses combining the data from the two RCTs is 1+ and the GOR is B.