Associations of B cell‐activating factor (BAFF) and anti‐BAFF autoantibodies with disease activity in multi‐ethnic Asian systemic lupus erythematosus patients in Singapore

Summary

To measure the levels of B cell‐activating factor (BAFF) and endogenous anti‐BAFF autoantibodies in a cohort of multi‐ethnic Asian systemic lupus erythematosus (SLE) patients in Singapore, to determine their correlation with disease activity. Serum samples from 121 SLE patients and 24 age‐ and sex‐matched healthy controls were assayed for BAFF and anti‐BAFF immunoglobulin (Ig)G antibody levels by enzyme‐linked immunosorbent assay (ELISA). The lowest reliable detection limit for anti‐BAFF‐IgG antibody levels was defined as 2 standard deviations (s.d.) from blank. Correlation of serum BAFF and anti‐BAFF IgG levels with disease activity [scored by SLE Activity Measure revised (SLAM‐R)], and disease manifestations were determined in these 121 patients. SLE patients had elevated BAFF levels compared to controls; mean 820 ± 40 pg/ml and 152 pg ± 45/ml, respectively [mean ± standard error of the mean (s.e.m.), P < 0·01], which were correlated positively with anti‐dsDNA antibody levels (r = 0·253, P < 0·03), and SLAM‐R scores (r = 0·627, P < 0·01). In addition, SLE patients had significantly higher levels of anti‐BAFF IgG, which were correlated negatively with disease activity (r = –0·436, P < 0·01), levels of anti‐dsDNA antibody (r = –0·347, P < 0·02) and BAFF (r = –0·459, P < 0·01). The majority of patients in this multi‐ethnic Asian SLE cohort had elevated levels of BAFF and anti‐BAFF antibodies. Anti‐BAFF autoantibody levels correlated negatively with clinical disease activity, anti‐dsDNA and BAFF levels, suggesting that they may be disease‐modifying. Our results provide further information about the complexity of BAFF pathophysiology in different SLE disease populations and phenotypes, and suggest that studies of the influence of anti‐cytokine antibodies in different SLE populations will be required when selecting patients for trials using targeted anti‐cytokine therapies.

Introduction

B cell‐activating factor (BAFF), or Blys, a tumour necrosis factor (TNF) family ligand involved in the selection, differentiation and survival of B cells, is implicated in the pathogenesis of systemic lupus erythematosus (SLE), promoting the escape of autoreactive B cells from negative selection and influencing the interactions of T cells and dendritic cells 1. Increased circulating levels of BAFF are found in SLE patients, but correlation with disease activity varies between studies. The inconsistent results have been attributed to differences in disease activity scoring, BAFF isoforms and population and phenotype subsets between studies 2, 3. Endogenous autoantibodies against pathogenic cytokines have been reported in systemic autoimmune disease, where they may modulate the bioactivity of the targeted cytokine, thereby influencing the disease course 4. In one study, endogenous autoantibodies to BAFF have been found in 10% of SLE patients 3. Another recent study employing multiplex protein microarray observed that these autoantibodies were associated with the high interferon (IFN) signature metric (ISM) that characterizes SLE patients with more severe disease 5. This was an unexpected finding, as BAFF signalling was blocked more in their patients with high anti‐BAFF immunoglobulin (Ig)G compared to patients and healthy controls with low anti‐BAFF IgG. As the number of SLE patients (n = 30) in the latter study was small, and SLE disease phenotypes in different populations differ, we sought to examine levels of BAFF and anti‐BAFF antibodies in a larger cohort of our multi‐ethnic Asian SLE patients, in order to determine the prevalence of serum BAFF and endogenous anti‐BAFF autoantibodies and their relationship with disease activity. As current management of SLE may utilize targeted cytokine therapy, the findings obtained could influence the choice of therapy.

Materials and methods

Serum BAFF and anti‐BAFF IgG antibody levels were determined in 121 SLE patients and 24 healthy controls by enzyme‐linked immunosorbent assay (ELISA). The 24 healthy controls were age‐ and sex‐matched, and comprised 22 females and two males, mean age 37·6 years ± 13·2. Correlation analysis of serum BAFF and anti‐BAFF IgG with SLE patient demographics, current and past disease manifestations, laboratory variables, anti‐dsDNA antibody levels and therapy were performed. All subjects were recruited with their written informed consent in accordance with the Declaration of Helsinki and local institutional review board (National Healthcare Group Domain Specific Review Board) guidelines. SLE patients all fulfilled 1997 revised American College of Rheumatology (ACR) classification criteria 6. Clinical and laboratory manifestations, disease activity scored with the SLE Activity Measure‐revised (SLAM‐R) 7, damage scored with the ACR/Systemic Lupus International Collaborating Clinic (SLICC) SLE damage index (SDI) 8 and therapy were recorded at the time of sample collection. None of the SLE patients had received treatment with anti‐BAFF monoclonal antibody or any other biologics.

BAFF, anti‐BAFF and anti‐dsDNA enzyme‐linked immunosorbent assay (ELISA)

Serum BAFF was assayed by ELISA, according to the manufacturer's recommendations (R&D Systems, Abingdon, UK), with 25 pg/ml being the lowest detection limit. Serum anti‐BAFF levels were measured by ELISA. Briefly, 96‐well plates (Nunc‐Immuno Maxisorp; ThermoFisher, Fremont, CA, USA) were coated with 4 μg/ml recombinant BAFF protein (eBioscience, San Diego, CA, USA) in 0·1 M NaH₂CO₃ overnight at 4°C, blocked in sample buffer which was comprised of 1% bovine serum albumin (BSA) in phosphate‐buffered saline (PBS) (Sigma‐Aldrich, St Louis, MO, USA), and twofold serial dilutions (N/50 onwards) of sera were performed in sample buffer. Bound, total IgG was detected with biotin‐conjugated anti‐human IgG, followed by avidin peroxidase (both diluted in sample buffer) and developed with tetramethylbenzidine substrate (all Sigma‐Aldrich). The lowest reliable detection limit was 2 standard deviations (s.d.) from blank. To verify that the specificity of anti‐BAFF IgG in our ELISA was not affected by potential interference from SLE autoantibodies and other serum factors, we performed the following BAFF preclearing experiment. Serum samples from 16 SLE patients with various anti‐BAFF IgG titres (N/3200, N/1600, N/400, N/200, four per group) and four normal controls with undetectable anti‐BAFF IgG were diluted in sample buffer (1% BSA PBS) containing 100 μg/ml of recombinant BAFF protein and incubated overnight at 4°C, then assayed by ELISA for anti‐BAFF IgG. All SLE and normal control BAFF preincubated serum samples were found to be undetectable for anti‐BAFF IgG, with optical density (OD) within 2 s.d. from blank. Serum anti‐dsDNA was assayed by a commercial ELISA, according to the manufacturer's recommendations (EliA dsDNA; Phadia GmbH, Freiburg, Germany). All ELISA samples were assayed in duplicate.

Statistics

Comparison of BAFF and anti‐BAFF IgG reactivity with SLE disease activity and clinical manifestations was performed using Fisher's exact test. SLE disease activity by system and organ involvement was assessed with SLAM‐R, which has a score ranging from 0 to 81, where a higher score signifies more severe disease 7. The nine organs and systems are mucocutaneous (oral/nasal ulcers, peri‐ungual erythema, malar rash, photosensitive rash, erythematous maculo or papular rash, discoid lupus, lupus profundus, bullous lesions, urticarial rash, livedo reticularis, leucocytoclastic vasculitis, palpable purpura, panniculitis, vasculitic ulcer, nailfold infarct or alopecia), musculoskeletal (arthritis), pulmonary (pleurisy/pleural effusion, pneumonitis), cardiac (hypertension, carditis, Raynaud's), gastrointestinal (serositis, pancreatitis, ischaemic bowel or other causes of abdominal pain because of SLE), nervous system (headache, stroke syndromes, cortical dysfunction or seizures), eye (cytoid bodies, papillitis, pseudotumour cerebri, episcleritis, retinal or choroidal haemorrhages), renal (active urinary sediments, proteinuria or renal impairment) and haematological (cytopenias or haemolytic anaemia). The anti‐BAFF IgG levels and SLAM‐R scores were found to be non‐normally distributed. Non‐parametric χ² and Mann–Whitney U‐tests were performed for univariate comparisons among subject groups. Correlation between disease activity score, anti‐dsDNA, BAFF and anti‐BAFF IgG levels were assessed using Spearman's correlation. All statistical analyses were performed using the Intercooled STATA (Stata Corporation, College Station, TX, USA). P‐values of less than 0·05 were considered significant.

Results

Demographics and disease characteristics

As shown in Table 1, the 121 SLE patients were predominantly female (85·9%), with a mean age of 38·7 ± 12·4 years, mean disease duration of 102 ± 89 months; and comprised 80·9% (98) Chinese, 11·6% (14) Malay and 7·4% (9) Indian and other ethnicity, approximating the population ethnic composition (74·3, 13·3 and 12·3%, respectively). The mean scores of SLAM‐R and SDI were 2·8 ± 2·2 and 0·6 ± 0·97, respectively. SLE disease manifestations at the time of sample collection included mucocutaneous in seven (5·8%), fatigue in four (3·3%) and active urine sediment in 33 (27·3%). Eighty‐eight (72·7%) patients had hypocomplementaemia and 80 (66·1%) had raised titres of anti‐dsDNA antibody. The majority were on corticosteroids (72.7%) and hydroxychloroquine (67.8%). Immunosuppressive drugs included azathioprine in 35·5%, mycophenolate in 5·8% and intravenous pulse cyclophosphamide in 3·3%. Only one patient was on cyclosporin.

Table 1.

Patient characteristics

	SLE patients (n = 121)
Age (years), mean ± s.d.	38·7 ± 12·4
Female, %	85·9
Disease duration (months), ± s.d.	102 ± 89
SLAM‐R, mean ± s.d.	2·8 ± 2·2
SDI, mean ± s.d.	0·6 ± 0·97
SLE manifestations*
Mucocutaneous involvement, %	5·8
Fatigue, %	3·3
Active urine sediment, %	27·3
Hypocomplementaemia, %	72·7
Elevated anti‐dsDNA antibody titre, %	66·1
Therapy*
Corticosteroids, %	72·7
Hydroxychloroquine, %	67·8
Azathioprine, %	35·5
Mycophenolate, %	5·8
i.v. cyclophosphamide, %	3.3
Cyclosporin, %	0.8

SLAM‐R = Systemic Lupus Activity Measure revised; SDI = American College of Rheumatology/Systemic Lupus International Collaborating Clinic (ACR/SLICC) systemic lupus erythematosus (SLE) damage index. *SLE manifestations and therapy at the time of sample collection; s.d. = standard deviation; i.v. = intravenous.

Serum BAFF levels in SLE patients were associated with anti‐dsDNA antibody levels and disease activity

SLE patients had significantly higher serum BAFF compared to controls; mean 820 ± 40 pg/ml versus 151 ± 10 pg/ml, respectively [mean ± standard error of the mean (s.e.m.), P < 0·01, Fig. 1a), levels of which correlated positively with levels of anti‐dsDNA antibody (r = 0·253, P < 0·03, Fig. 1b). Serum BAFF levels were also correlated positively with clinical disease activity scored with SLAM‐R in the SLE patients (r = 0·627, P < 0·01, Fig. 1c).

Figure 1.

Systemic lupus erythematosus (SLE) patients had elevated serum B cell‐activating factor (BAFF) levels which correlated with anti‐dsDNA antibody levels and disease activity. P‐values (two‐tailed) represent significance of serum BAFF levels in SLE (n = 121) versus healthy controls (Fig. 1a), and correlation with anti‐dsDNA levels (Fig. 1b) and with Systemic Lupus Activity Measure revised (SLAM‐R) scores (Fig. 1c). Data are mean ± standard error of the mean (s.e.m.).

Anti‐BAFF IgG autoantibody levels were elevated in SLE patients

The lowest reliable detection limit of anti‐BAFF IgG in our ELISA assays, defined as 2 s.d. from the blank control in the normal control group, was N/100 dilution. Based on this, all 121 SLE patients were found to have anti‐BAFF IgG levels of at least N/200 or higher, with N/3200 the highest level detected, and a mean detection level of N/1256, s.e.m. ± 71 (Fig. 2a). As the levels of anti‐BAFF IgG detected in serum of SLE patients could be affected by non‐specific binding of autoantibodies and other immune factors associated with SLE, we have performed additional BAFF preclearing assays by preincubating SLE and normal control samples with recombinant BAFF protein, as described in Materials and methods, prior to anti‐BAFF IgG estimation by ELISA. Upon clearing, all samples (16 SLE and four healthy controls) were found to be undetectable for anti‐BAFF IgG at OD within 2 s.d. from blank, thus ruling out any potential interference in our anti‐BAFF IgG assays.

Figure 2.

Systemic lupus erythematosus (SLE) patients had elevated serum anti‐B cell‐activating factor (BAFF) immunoglobulin (Ig)G levels which correlated negatively with anti‐dsDNA levels and disease activity. P‐values (two‐tailed) represent significance of serum anti‐BAFF IgG levels in SLE (n = 121) versus healthy controls (Fig. 2a), and of the correlation with anti‐dsDNA levels (Fig. 2b) and with Systemic Lupus Activity Measure revised (SLAM‐R) scores (Fig. 2c). Data are mean ± standard error of the mean (s.e.m.).

Serum anti‐BAFF antibody levels correlated negatively with disease activity as well as levels of anti‐dsDNA antibody and serum BAFF in SLE patients

We observed a negative correlation between levels of circulating anti‐BAFF IgG and anti‐dsDNA antibody in SLE patients (r = –0·347, P < 0·02, Fig. 2b). In addition, levels of serum anti‐BAFF IgG were correlated negatively (r = –0·436, P < 0·01, Fig. 2c) with clinical disease activity of SLE as measured by SLAM‐R. Finally, a significant negative correlation (r = –0·459, P < 0·01), between serum BAFF and circulating anti‐BAFF IgG levels in our patients was also observed, data not shown. No correlation of serum BAFF or anti‐BAFF IgG levels was found with sociodemographic characteristics, disease or organ manifestations, disease duration, treatment or damage (P > 0·05).

Discussion

Our study demonstrated a high prevalence of endogenous antibodies to BAFF in our multi‐ethnic Asian SLE cohort. Importantly, while levels of serum BAFF correlated positively with disease activity in our patients, levels of anti‐BAFF antibody were correlated negatively with levels of its target cytokine, anti‐dsDNA antibody and clinical disease activity. This suggests that the presence of these autoantibodies does not merely reflect an autoimmune response to high levels of circulating BAFF, but is demonstrative of the known phenomenon of anti‐pathogenic cytokine autoantibodies influencing the course of autoimmune disease 4. Endogenous anti‐BAFF autoantibodies may blunt the effects of overproduced BAFF through forming complexes with BAFF, thereby enhancing its clearance.

The variation in prevalence of endogenous anti‐BAFF and other anti‐cytokine antibodies between different SLE populations has potential clinical implications. The efficacy of belimumab, the therapeutic antibody against BAFF, has been modest, possibly because of pre‐existent endogenous autoantibodies already modulating BAFF‐mediated pathogenic mechanisms. Indeed, Price et al. showed that BAFF‐binding IgG autoantibodies in their SLE patients neutralized the activity of BAFF via its interaction with the BAFF receptor 5. Interestingly, they also observed that these neutralizing anti‐BAFF autoantibodies were more prevalent in patients possessing the high IFN signature metric (ISM), which has been associated with more severe disease. The explanations put forward to account for the unexpected association of these endogenous anti‐BAFF antibodies with the ISM included paradoxical stimulation of BAFF production by anti‐BAFF autoantibodies, preferential binding of these autoantibodies to ΔBAFF, the natural regulator of BAFF, allowing the effects of BAFF to go unchecked and the possibility that these BAFF‐binding antibodies were an autoimmune response to the high amounts of circulating BAFF 9. However, the ISM has been shown recently to be associated with serological manifestations of elevated levels of antibodies to dsDNA, extractable nuclear antigens (ENA), serum BAFF and hypocomplementaemia, but not clinical disease activity 10. Their study also found no difference in serum BAFF levels between the anti‐BAFF antibody‐high and ‐low samples. Contrary to this, we found that serum BAFF levels were correlated negatively with anti‐BAFF IgG levels in our multi‐ethnic Asian SLE patients. Although we did not measure the IFN signature in our 121 SLE patient cohort, we have observed similar type 1 IFN gene expression profiles in previous studies of our SLE patients 11. It would have been useful to determine if similar IFN signature profiles exist in these 121 SLE patients; however, this is beyond the scope of the current study.

We believe that the conflicting results between the studies may also be due to differences not only between SLE population disease phenotypes, but also in anti‐cytokine autoantibody subsets in individual patients, as shown in the recent reports on antibodies to anti‐IFN‐α. Neutralizing anti‐IFN‐α antibodies detected in approximately 25% of lupus patients by Morimoto et al. were associated with decreased bioactivity of IFN‐α and lower disease activity and severity 12. Contrary to this, the recent study by Gupta et al. 13 found that of their 12 SLE patients (of a total of 199) who had anti‐IFN‐α antibodies, only 50% of these samples demonstrated functional activity. Moreover, there were no significant differences in clinical disease activity, hypocomplementaemia or anti‐dsDNA antibody titres between those with and without anti‐IFN‐α antibodies.

In common with other non‐Caucasian SLE populations, our multi‐ethnic Asian SLE patients have more severe disease 14, 15, and the higher prevalence of anti‐BAFF autoantibodies may be a reflection of this. A limitation of our study is that we did not determine if the anti‐BAFF antibodies in our patients were capable of neutralizing BAFF, although the negative correlation of their levels with serum BAFF, clinical disease activity and anti‐dsDNA antibody titres would suggest that they are disease‐modifying. In addition, prospective follow‐up studies of anti‐BAFF IgG levels in tandem with clinical disease activity over the longer term is needed for further elucidation of these preliminary findings.

Our study demonstrates clearly the significant differences between different ethnic populations of SLE patients in not only the prevalence of endogenous anti‐cytokine antibodies, but also their correlation with clinical and serological disease activity. Our observations on the relationship of serum BAFF and anti‐BAFF autoantibodies in our multi‐ethnic Asian cohort provides further information about the complexity of the biology of pathogenic cytokines in SLE, and highlights the need for further studies into the different clinical and immunological phenotypes. The efficacy of belimumab, the therapeutic antibody against BAFF, has been modest, possibly because of pre‐existent endogenous autoantibodies already modulating BAFF‐mediated pathogenic mechanisms 16. Evaluation for the presence and function of these endogenous anti‐cytokine autoantibodies would not only add to the understanding of SLE disease pathogenesis but also inform the design of new targeted cytokine therapies and the individualization of patient selection.

Disclosure

The authors declare no conflicts of interest.