Systemic lupus erythematosus (SLE) is a chronic autoimmune disease with an incidence rate of 3.9 cases per 100 000 Caucasian females and 0.4 cases per 100 000 Caucasian males.1 Despite improvements in prognosis and the establishment of corticosteroid and immunosuppressive drug-based treatments, specific therapies are still needed for patients with SLE refractory to conventional drugs. Treatments that target B cells appear particularly promising, since SLE pathogenesis has been linked to B cell hyperproliferation in patients.1 These B cells generate large quantities of IgG autoantibodies that can form immune complexes, ultimately leading to lupus nephritis and renal failure.1

B lymphocyte stimulator (BLyS, also termed B cell activating factor (BAFF), THANK, TALL-1 and zTNF4) and a proliferation-inducing ligand (APRIL) form a subfamily of tumour necrosis factor (TNF)-like ligands that regulate B cell development and immune responses, and are associated with SLE pathology and its treatment.2 Although cells of myeloid and lymphoid lineages were originally reported to produce BLyS and APRIL,3 both cytokines can be also expressed by non-haematopoietic cells (eg, BLyS by osteoclasts and synovial fibroblasts and APRIL by osteoclasts, tumour cells and epithelial cells).2

TNF ligand family members are synthesised as type II transmembrane proteins, but cleavage is commonly observed in the stalk region between the transmembrane and receptor-binding domains. Accordingly, BLyS is released from the cell surface, where it appears as a membrane-anchored protein. In contrast, APRIL has a unique maturation pathway among TNF ligand family members, as it is processed intracellularly prior to secretion.4

BLyS is a fundamental survival factor for transitional and mature B cells, whereas APRIL mainly affects B1 cell activity, humoral responses and class switching of immunoglobulins.2 BLyS and APRIL share the ability to bind to two receptors, transmembrane activator and calcium modulator cyclophilin ligand interactor (TACI) and B cell maturation antigen (BCMA).2 Both of these receptor are expressed on B cells, whereas TACI was also detected on a subset of activated T cells.2 A BLyS-specific receptor termed BAFF receptor 3 (BR3) has been identified on B cells and appears to mediate most BLyS-elicited B cell survival signals.3 Heparan sulfate proteoglycans (HSPG) were shown to mediate APRIL binding to haematopoietic and tumour cells.2

Constitutive overexpression of BLyS in mice can lead to a SLE-like syndrome.3 Mice that develop SLE have elevated circulating levels of BLyS; treatment of these mice with soluble TACI–Ig fusion protein ameliorates disease progression and enhances survival.5 High serum BLyS levels have been detected in patients with SLE6 and several companies are presently testing the therapeutic capacity of BLyS antagonists.2

The role of APRIL in SLE is less well understood, and reports on APRIL serum levels in patients with SLE are debated. One publication reported increased APRIL levels in serum from patients with SLE, suggesting a contribution of APRIL to SLE pathogenesis.7 A second study described an inverse correlation of serum APRIL levels and disease activity, however, indicating a negative regulatory role for APRIL.8 To clarify this issue, we performed a study of 43 patients with SLE and compared serum APRIL levels with disease activity and autoantibodies. The median APRIL levels were elevated in patients with SLE compared to healthy controls, but did not correlate with SLEDAI scores. Serum APRIL levels correlated inversely with serum BLyS levels, and weakly with anti-double-stranded DNA (dsDNA) titres and tended to associate with a lack of renal manifestations. A follow-up of 27 patients showed an inverse correlation between serum APRIL levels and SLE Disease Activity Index (SLEDAI) score.

PATIENTS AND METHODS

Subjects

The study included 43 patients who met the 1982 American College of Rheumatology (ACR) criteria for the classification of SLE. In addition, positivity for anti-dsDNA antibodies in the patient history was an obligatory criterion because of the high specificity of these autoantibodies for SLE.9 As controls, we used 27 healthy volunteers and 28 patients with osteoarthritis. Patient and control serum samples were stored at −20°C until use. The Ethics Committee of the CHU Montpellier approved the study protocol, and informed consent was obtained from each participant.

Disease activity was measured by the SLEDAI, which contains 24 descriptors (clinical and biological parameters) with preassigned severity weighting.10 The score ranges from 0 (no disease activity) to 105 (maximum disease activity); a score of 6 is considered clinically important.11 Biological parameters included lactate dehydrogenase (LDH), as well as immunoglobulins and β2 microglobulin to estimate B cell activation. In addition, the SLEDAI measures rheumatoid factor, anti-nuclear antibodies (ANA), anti-dsDNA antibodies, soluble anti-nuclear extract antibodies (ENA) and anti-phospholipid antibodies, as well as levels of the complement components C3 and C4, the erythrocyte sedimentation rate (ESR), C-reactive protein (CRP), creatininaemia and proteinuria. All patients were proposed for a follow-up 4 months after their first visit and 27 patients agreed to be re-evaluated for all baseline clinical and biological parameters. Antinuclear antibodies were detected by indirect immunofluorescence on Hep-2 cells by applying serial serum dilutions from 1:80 to 1:5000, with 1:160 as the cut-off for positivity (Bio-Rad, Marnes-la-Coquette, France). Anti-dsDNA and anti-ENA antibodies were measured using the FIDIS connective 100 MX006 kit (BioMedical Diagnostics, Marne la Vallée, France) with a normal value ⩽40 arbitrary units (AU)/litre. An ELISA was used to measure anticardiolipin antibodies (Bio-Rad) and was considered positive if IgG was >23 units of IgG antiphospholipid (UGPL)/ml and IgM >11 units of IgM antiphospholipid (UMPL)/ml. Anti-β2GP1 was also measured by ELISA (Bio-Rad) with a positive threshold >20 AU/ml for IgG or IgM. Rheumatoid factor, CRP, C3 and C4 levels were evaluated using a Nephelometer analyser II (Dade-Boehring, Paris, France). The range of normality for β2 microglobulin was 0.8–2.4 mg/litre, 220–450 IU/litre for LDH, 6.90–14 g/litre for IgG, 0.7–4.1 g/litre for IgA and 0.34–2.40 g/litre for IgM.

Determination of serum BLyS and APRIL levels

Patient and control sera were assayed in parallel for BLyS and APRIL using ELISA as described.12 13 APRIL was captured using B-cell maturation antigen (BCMA)-Fc (1 μg/ml; Sigma-Aldrich, St Louis, Missouri, USA) followed by anti-APRIL antibody (1 μg/ml) and human IgG-absorbed anti-rabbit peroxidase secondary antibody (Jackson Laboratory, Bar Harbor, Maine, USA); specificity was controlled by coating wells with an irrelevant Fc-hybrid protein. Addition of up to 6.3 ng/ml recombinant BLyS, the highest concentration detected in our cohort of patients with SLE, did not interfere with APRIL detection in the sera (data not shown). Serum dilutions used for quantification were within the linear absorbance range at 492 nm, and recombinant APRIL (R&D Systems, Minneapolis, Minnesota, USA) was used for standardisation. Assay sensitivity was 750 pg/ml. To detect soluble BLyS protein, we used the Human BLyS Quantikine ELISA Kit (R&D Systems).

Statistical analysis

All statistical analyses were performed using GraphPad Prism software version 4.00 and results were given as median and range (GraphPad Software, San Diego, CA, USA). Correlations were determined by Spearman rank order correlation for ordinal data. Non-parametric testing was performed using the Mann–Whitney U test for non-ordinal data and for comparison of APRIL and BLyS levels between two groups. The Kruskal–Wallis test was used to compare APRIL and BLyS levels between different activity groups. A p value <0.05 was considered statistically significant. For comparison of paired samples during the follow-up of the 27 patients, a Wilcoxon matched pair test was used.

RESULTS

Patient characteristics

A total of 43 patients with SLE (40 females, 3 males), with a median age of 36 years (range 19 to 63), were included in this study. The clinical features of these patients are depicted in table 1. The median delay between diagnosis and enrolment in the study was 5 years (range 1 to 27). Articular symptoms were common (95.4%) and renal involvement was observed in 37.2% of patients. According to the World Health Organization (WHO) classification, one patient had class II nephritis, five patients had class III, five had class IV and two had class V, while the nephritis class could not be determined for four patients. At baseline, the median SLEDAI was 6 (range 0 to 30) despite treatment with corticosteroids (53.5%) and immunomodulators or an immunosuppressor (83.7%). Biological parameters observed at baseline are shown in table 2. ESR and CRP values were elevated in 20 and 12 patients, respectively. In all, 10 patients showed high creatinine levels and proteinuria (>0.5 g/24 h). We found detectable ANA levels at a dilution of at least 1/640 for 39 patients and anti-dsDNA antibodies in 25 patients, with a median titre of 48 AU/litre (range 0 to 1077). A total of 15 patients had positive anti-ENA antibodies (Sjögren syndrome antigen A (SSA), SSB, Sm (named for Smith, the patient the antigen was first discovered in), or ribonucleoprotein (RNP)) and 12 had anti-phospholipid antibodies. Only four patients were positive for rheumatoid factor. Decreased C3 and C4 levels were measured in 10 and 7 patients, respectively, and elevated LDH and β2 microglobulin levels in 6 and 10 patients, respectively. A total of 7 patients (16%) had a monoclonal gammopathy, a rather large proportion considering a previous study that reported monoclonal gammopathy in 59 (5.3%) of 1083 patients.14 Patients in our study with monoclonal gammopathy had median APRIL and BLyS values similar to those found for all patients (data not shown).

Elevated median serum APRIL and BLyS levels correlate inversely in patients with SLE

To evaluate their implication in SLE, we measured circulating APRIL and BLyS concentrations by ELISA. The median APRIL serum level was 4 ng/ml in healthy donors (range 0 to 24 ng/ml; n = 27) and 24 ng/ml in patients with SLE (range 0 to 631 ng/ml; n = 43; p<0.001; fig 1A). No significant differences were found in APRIL levels in serum from patients with osteoarthritis (median = 1.5 ng/ml; n = 28) compared to healthy donors. APRIL was undetectable in 7 of the 43 SLE sera, and was <24 ng/ml in another 14 samples from patients with SLE (ie, lower than the highest control serum value). Approximately 53% of all patients with SLE thus showed increased circulating APRIL protein compared to controls (fig 1A). Sera from patients with SLE also showed high circulating BLyS levels compared to healthy controls (fig 1B, p<0.001). The median BLyS serum level was 1.7 ng/ml in sera from patients with SLE (n = 43), 1.3 in patients with osteoarthritis (n = 28) and 0.9 ng/ml in healthy donors (n = 27). Serum APRIL and BLyS levels correlated inversely in patients with SLE (r = −0.339; p = 0.026; n = 43), but not in patients with osteoarthritis or healthy donors (fig 2). Patients that had positive anti-dsDNA titres at their first visit manifested a more pronounced negative correlation between APRIL and BLyS serum levels (r = −0.465; p = 0.022; n = 25).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Serum a proliferation-inducing ligand (APRIL) and B lymphocyte stimulator (BLyS) levels in patients with systemic lupus erythematosus (SLE) and correlation with disease parameters. Serum levels of (A) APRIL and (B) BLyS in our cohort of patients with SLE (n = 43), patients with osteoarthritis (OA, n = 28) and healthy donors (controls, n = 27). Median value is marked (solid line).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Correlation of serum a proliferation-inducing ligand (APRIL) and B lymphocyte stimulator (BLyS) levels in patients with systemic lupus erythematosus (SLE) (n = 43).

Relationship between serum APRIL levels and disease parameters

Serum APRIL levels showed weak inverse correlation with anti-dsDNA antibody titres (r = −0.261; p = 0.093; n = 42). When only patients with positive anti-dsDNA antibody titres (>40 AU/ml) were considered, the inverse correlation between APRIL and anti-dsDNA levels was significant (r = −0.411; p = 0.046; n = 25) (fig 3A). Patients with high serum APRIL levels tended to have a lower incidence of renal involvement (p = 0.103). The median APRIL serum level was 35 ng/ml (range 0 to 83 ng/ml; n = 17) in patients with SLE with no kidney manifestations and 11 ng/ml (range 0 to 631 ng/ml; n = 28) in those with renal involvement. Serum APRIL correlated weakly with LDH (r = 0.315; p = 0.05; n = 39) but did not correlate with age, sex, disease duration, SLEDAI, erythrocyte sedimentation rate, CRP, creatininaemia, C3 and C4 levels, proteinuria or haematuria, ANA titres, anti-ENA, anti-phospholipid antibodies, presence of rheumatoid factor or articular involvement (data not shown).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

Correlation of serum a proliferation-inducing ligand (APRIL) and B lymphocyte stimulator (BLyS) levels with disease parameters in patients with systemic lupus erythematosus (SLE). A. Correlation between APRIL and anti-double-stranded DNA (dsDNA) antibodies in patients with positive anti-dsDNA antibody titres (>40 arbitrary units (AU)/ml) (n = 25). B. Correlation between BLyS and anti-nuclear antibodies (ANA; n = 42).

Relationship between serum levels of BLyS and disease parameters

In comparison with APRIL, serum BLyS levels correlated with ANA (p = 0.028; r = 0.338; n = 42) (fig 3B), but we found no correlation with anti-dsDNA titres (data not shown). Moreover, BLyS did not correlate with the SLEDAI for the 43 patients (p = 0.885). There was no correlation between serum BLyS levels and other immunological markers (anti-ENA, anti-phospholipid antibodies, rheumatoid factor) or clinical parameters (data not shown). Serum BLyS correlated weakly with β2 microglobulin (p = 0.106; r = 0.269; n = 37).

Patient follow-up

A total of 27 patients with SLE were re-evaluated 4 months after the first visit. Median age was 36 years (range 21 to 60). Of these patients, 24 were treated prior to the second visit with corticosteroids (median 6 mg/day; range 2 to 60), of which 18 also received immunosuppressive drugs (azathioprine, n = 3; methotrexate, n = 5; hydroxychloroquine, n = 5; cyclophosphamide, n = 2; ciclosporine with mycophenolate mofetil, n = 2; mycophenolate mofetil alone, n = 1). For 11 patients, treatment was modified between the 2 visits. Median SLEDAI was 6 (range 0 to 30) at the first visit and 1.5 (range 0 to 22) at the second visit.

Between the first and second visits, only one patient showed a moderate disease flare (ΔSLEDAI = 10), seven patients improved to remission (SLEDAI>3 at baseline and SLEDAI = 0 at re-evaluation), five were in remission (SLEDAI = 0 for both), eight had improved (ΔSLEDAI >3) and six were stable (ΔSLEDAI <3). Median patient serum levels of APRIL and BLyS remained mostly unchanged between the first and second visits, with respective median serum APRIL and BLyS levels of 26 ng/ml (range 0 to 229) and 2.083 ng/ml (range 0 to 4.767) at the first visit and 24 ng/ml (range 0 to 237) and 2.033 ng/ml (range 0.717 to 10.55) at the second. Serum BLyS levels at the follow-up correlated with the re-evaluated ANA levels (r = 0.420; p = 0.032). Serum APRIL in the follow-up patients correlated inversely with the re-evaluated SLEDAI (r = −0.464; p = 0.01) (fig 4A), with anti-dsDNA antibody titres (r = −0.408; p = 0.03) (fig 4B) and BLyS levels measured at the second visit (r = −0.398; p = 0.03) (fig 4C).

• Download figure
• Open in new tab
• Download powerpoint

Figure 4

Correlation of serum a proliferation-inducing ligand (APRIL) levels with disease parameters in patients with systemic lupus erythematosus (SLE) evaluated at their second visit (n = 27). Inverse correlation of serum APRIL levels with SLE Disease Activity Index (SLEDAI) (A), anti-double-stranded DNA (dsDNA) antibody titres (B) and serum B lymphocyte stimulator (BLyS) levels (C).

DISCUSSION

SLE is a disease with a heterogeneous pattern of symptoms.1 We selected only those patients who met the ACR criteria and had tested positive for anti-dsDNA antibodies at least once in their medical record.

Concurring with previous reports, we observed increased serum BLyS levels in patients with SLE compared to control subjects, and no correlation between BLyS levels and disease activity.6 15 16 Nevertheless, we did not detect correlation between serum BLyS and anti-dsDNA levels, as observed in other studies,6 15 16 but with ANA.

While there is a consensus that increased serum BLyS levels in SLE reflect its contribution to disease, previous reports of serum APRIL levels in SLE are contradictory. Stohl et al described a modest inverse association of APRIL levels and disease activity,8 Koyama et al detected elevated serum APRIL levels that correlated slightly with anti-dsDNA antibodies,7 and Vallerskog et al found no significant difference in circulating APRIL between patients with SLE and controls.17 Based on these observations, distinct therapeutic strategies have been suggested, ranging from the application of APRIL antagonists7 to the use of specific BLyS-antagonising agents rather than of soluble TACI or BCMA receptors, which antagonise BLyS as well as APRIL.8

In our cohort of patients with SLE, we observed elevated serum APRIL levels in 53% of the sera tested. Koyama et al found elevated serum APRIL levels in the majority of patients with SLE and 42% had levels above 100 ng/ml,7 whereas only 5% of the sera displayed elevated serum APRIL levels in the study of Stohl et al.8 We can currently only speculate about the reason for these discrepancies; one explanation could be the differences between the cohorts of patients with SLE analysed. A comparison of the cohorts is not possible, however, given the lack of the clinical information in the studies mentioned.

APRIL levels in patients with SLE correlated inversely with levels of BLyS, although median concentrations of both cytokines were elevated. This finding might be related to the >10-fold higher median concentration of APRIL compared to that of BLyS, as also observed by Stohl et al.8 This suggests that small differences in circulating APRIL could trigger its biological action in patients with SLE, and that increased APRIL serum levels alone cannot be used as an indicator.

Stohl et al analysed in parallel serum and mRNA levels of APRIL and BLyS in blood cells of patients with SLE,8 although both cytokines are now known to be expressed by non-haematopoietic cells as well (summarised in Dillon et al).2 Stohl et al found a correlation between APRIL and BLyS mRNA levels in blood cells, but not in serum levels of these cytokines; to explain this discrepancy, they discuss non-haematopoietic cells as important APRIL producers.8 These observations led us to measure the cytokines in serum.

The weak inverse correlation observed between serum APRIL levels and anti-dsDNA antibodies concurs with the description by Stohl et al of an inverse correlation between serum APRIL and anti-dsDNA antibodies (r = −0.144; p = 0.017; n = 526).8 The authors found no correlation with disease activity for individual patients over time, but a modest inverse correlation when considering the 526 samples derived from 68 patients. Anti-dsDNA antibodies are detected in renal lesions of patients with SLE18 19 and anti-dsDNA antibody titres are reported to predict a renal flare.20 We observed that high serum APRIL levels tended to associate with a lack of renal involvement in patients with SLE, supporting the link between anti-dsDNA antibody titres and kidney involvement.

The follow-up of 27 patients in our study revealed an inverse correlation between re-evaluated serum APRIL levels and circulating BLyS, but also with the re-evaluated anti-dsDNA antibody titres and the re-evaluated SLEDAI. Moreover, patients had similar serum APRIL levels at their first and second visits, and baseline APRIL levels correlated with the re-evaluated SLEDAI (r = −0.417; p = 0.031, n = 27). APRIL therefore appears to modulate disease activity, particularly during treatment with immunosuppressive drugs. The similarity of APRIL levels in patients between visits indicates that the cells that regulate APRIL production in patients with SLE are not affected by immunosuppressive drug treatment in our study.

It is tempting to speculate that APRIL has a negative regulatory role or could be protective factor in SLE. Analysis of APRIL deficient mice revealed a negative regulatory role for APRIL in T cell-dependent (TD) humoral responses, whereas BLyS is an established promoter of TD humoral responses.21 It is thus possible that APRIL and BLyS production can inversely regulate each other in SLE, perhaps in concert with other cytokines.

Our analysis of a cohort of 43 patients with SLE suggests opposing roles for APRIL and BLyS in SLE disease. BLyS-antagonising agents, including a monoclonal antibody and a soluble TACI-Fc construct, are currently being evaluated in clinical trials for their potential to treat patients with SLE, and both appear to be effective.22 23 In light of our results, modification of the TACI-Fc construct such that it antagonises BLyS, but not APRIL, could increase its efficiency.