To determine the association of plasma B lymphocyte stimulator (BLyS) levels, immunosuppressive therapy, and other clinical parameters with disease activity in systemic lupus erythematosus (SLE).
To determine the association of plasma B lymphocyte stimulator (BLyS) levels, immunosuppressive therapy, and other clinical parameters with disease activity in systemic lupus erythematosus (SLE).
Two hundred forty-five SLE patients were evaluated prospectively over a 2-year period at 4 centers. Assessments were performed every 3–6 months. Univariate analysis was used to determine the association among the Safety of Estrogens in Lupus Erythematosus: National Assessment (SELENA) version of the Systemic Lupus Erythematosus Disease Activity Index (SLEDAI) score, serum anti–double-stranded DNA (anti-dsDNA), and plasma BLyS levels. A multivariate repeated-measures model incorporating immunosuppressive therapy was utilized.
Ninety-two percent of the patients were female. Sixty-seven percent were white, 31% African American, and 2% Asian (all of these groups may include Hispanic). Mean values at baseline were as follows: age 41.5 years, disease duration 8.1 years, SELENA–SLEDAI 3.3 (median 2, range 0–18), BLyS 5.57 ng/ml, IgG 1,439 mg/dl, C3 104.4 mg/dl, and C4 21.3 mg/dl; among those positive for anti-dsDNA, the median titer was 1:40 (range 1:10–1:1,280). Univariate analysis showed that plasma BLyS levels were associated with anti-dsDNA titers (P = 0.0465) and SELENA–SLEDAI scores (P = 0.0002). In multivariate analyses, a greater increase in the SELENA–SLEDAI score from the previous visit was associated with higher BLyS levels at the previous visit (P = 0.0042) and with a greater increase in the BLyS level from the previous visit (P = 0.0007).
The findings of association between a greater increase in the BLyS level from the previous visit and a greater increase in the SELENA–SLEDAI score at the subsequent visit, and between an elevated BLyS level at the previous visit and a greater SELENA–SLEDAI score at the subsequent visit, demonstrate a relationship between circulating BLyS levels and SLE disease activity. These results lend support to the notion that BLyS is a candidate for therapeutic targeting in SLE.
Blymphocyte stimulator (BLyS; trademark of Human Genome Sciences, Rockville, MD), also known as BAFF (B cell–activating factor from the tumor necrosis factor [TNF] family), is a type II transmembrane protein (1). BLyS exists in both membrane-bound and soluble forms and is expressed on monocytes, macrophages, and monocyte-derived dendritic cells. BLyS gene expression and levels of membrane-associated soluble BLyS are regulated by cytokines, in particular interferon-γ and to a lesser extent interleukin-10 (2). The biologically active soluble form of BLyS binds to 3 receptors expressed primarily on B lymphocytes: TACI (TNF transmembrane activator and calcium modulator and cyclophilin ligand interactor), BCMA (B lymphocyte maturation antigen), and BR3 (BAFF/BLyS receptor 3) (3, 4). Activation of the BLyS receptors leads to B cell and plasma cell proliferation, differentiation, and survival and IgG class switching (1, 5, 6). A related TNF family member, APRIL (a proliferation-inducing ligand), can also bind TACI and BCMA and can mediate effects similar to those of BLyS, but its biologic action may be primarily on memory plasma cells. BLyS–APRIL heterotrimers have also been characterized (7), but their function in vivo is unclear.
Circulating BLyS levels are elevated in patients with systemic lupus erythematosus (SLE), rheumatoid arthritis (RA), and other systemic immune-based rheumatic diseases. A correlation between plasma BLyS protein levels and rheumatoid factor titers in patients with RA has been reported (8, 9). In addition, elevated levels of BLyS have been detected in the synovial fluid of patients with RA and have been correlated with levels of autoantibodies associated with Sjögren's syndrome (10). In cross-sectional studies, 20–40% of patients with SLE appear to have significantly higher levels of circulating BLyS protein than are found in normal controls, and plasma BLyS protein levels appear to correlate with IgG levels and anti–double-stranded DNA (anti-dsDNA) titers (8, 9). In SLE and RA patients treated with rituximab, BLyS levels increase with B cell depletion and return to near-baseline levels upon B cell repopulation (11, 12).
Although the findings of elevated levels of BLyS protein in autoimmune diseases suggest that BLyS may be a marker of current or impending autoimmune disease activity and may be a potential target for therapeutic intervention, the relationship between circulating BLyS levels and SLE disease activity has not been adequately characterized. Accordingly, this 2-year observational study was designed to assess the relationships between plasma BLyS concentrations and disease activity, autoantibody and immunoglobulin levels, and treatments received for SLE.
This 2-year multicenter observational study was undertaken to assess disease activity, therapies, and multiple biomarkers including plasma BLyS levels in patients with SLE as defined by the American College of Rheumatology (ACR) classification criteria (13, 14). Patients received standard-of-care therapy for their disease. To be eligible for the study, patients had to be at least 18 years old and could not have been treated with an investigational agent within 28 days prior to the start of the study. None of the study patients were found to have received an investigational agent at any time. Originally designed as a single-year natural history investigation, the study was extended to measure BLyS and evaluate disease activity through a second year. Patients were assessed clinically, and blood and urine samples were collected every 3 months during the first year of the study and every 6 months during the second year. Patients were also assessed during any unscheduled clinical visits that were necessary due to acute exacerbation of SLE disease activity. Written informed consent was obtained from all study patients prior to enrollment.
At baseline, the medical history was reviewed and information on current medications and on the components of the ACR SLE classification criteria that the patient fulfilled was recorded. To assess disease activity, the Safety of Estrogens in Lupus Erythematosus: National Assessment (SELENA) version of the Systemic Lupus Erythematosus Disease Activity Index (SLEDAI) (15, 16) was administered, and plasma BLyS, C3/C4, and anti-dsDNA autoantibodies measured, at baseline, at 3, 6, 9, 12, 18, and 24 months, and at any unscheduled visits. Total immunoglobulins and subclasses (IgM, IgA, IgG, and IgE) were measured at baseline and at the 6-, 12-, 18-, and 24-month scheduled visits. A complete blood cell count and urinalysis were performed at baseline, at 3, 6, 9, 12, 18, and 24 months, and at any unscheduled visit.
To measure plasma BLyS levels, ∼10 ml of blood was collected in Vacutainer tubes containing EDTA, centrifuged for 15 minutes at 1,500 revolutions per minute, aliquoted (0.5 ml) into cryotubes, and immediately frozen at –70°C. Plasma BLyS levels were measured by enzyme-linked immunosorbent assay (ELISA) with mouse anti-human BLyS monoclonal antibody 3D4 as the capture antibody and biotinylated rabbit anti-human BLyS polyclonal antibody as the detector antibody. Recombinant soluble BLyS was used to standardize the assay (interassay variability <15%) (17). The normal plasma BLyS level measured using this assay ranged from 1 ng/ml to 5 ng/ml in healthy adults. The BLyS assay was performed at Human Genome Sciences Laboratory. Serum levels of biologic markers, including anti-dsDNA, total immunoglobulin, and immunoglobulin subclasses, were also measured by ELISA.
Disease history, use of immunosuppressive agents, and baseline demographic characteristics were compared among study sites (see below). These comparisons were performed using the likelihood ratio test for discrete variables and analysis of variance (ANOVA) or Wilcoxon's rank sum test, as appropriate, for continuous variables.
Descriptive statistics were used to summarize demographic and baseline disease characteristics, as well as changes in BLyS levels over time. Univariate association analysis was used to analyze BLyS levels and other variables, such as SELENA–SLEDAI scores and immunoglobulin levels. The significance of differences in BLyS levels between patients with and those without detectable anti-dsDNA antibodies was assessed by Student's t-test.
A repeated-measures analysis of covariance model was used to examine the relationship between disease activity and BLyS level, immunosuppressive drug use, and steroid dosage. In this model, the dependent variable was the change in SELENA–SLEDAI score from the previous to the current visit, and the explanatory variables were 1) BLyS level at the previous visit, 2) change in BLyS level from the previous to the current visit, 3) whether the patient was receiving prednisone in the last 3 months, 4) change in the prednisone dosage from the previous to the current visit, 5) whether the patient was receiving immunosuppressive treatment in the last 3 months, and 6) SELENA–SLEDAI score at the previous visit. The model was also adjusted for study site, the patient's race (white versus other), the interaction effect between study site and race, and age. A spatial covariance structure was assumed to account for the correlation between repeated observations within each patient.
A total of 245 patients at 4 clinical centers (Johns Hopkins University, University of Southern California, University of Alabama at Birmingham, and University of Michigan) were enrolled in this observational study. Patient demographic characteristics, baseline disease characteristics, and ACR classification criteria by clinical center are summarized in Table 1. Approximately 90% of the patients at each clinical center were female. The mean age was 41.5 years. The majority of the patients were white (ethnicity was self-reported, with 67% white, 31% African American, and 2% Asian [all of these ethnicity categories could include subjects of Hispanic origin]).
|JHU (n = 100)||USC (n = 69)||UAB (n = 50)||UM (n = 26)|
|Female||90 (90)||65 (94)||47 (94)||24 (92)|
|Male||10 (10)||4 (6)||3 (6)||2 (8)|
|White||51 (51)||62 (90)§||27 (54)||23 (88)|
|African American||47 (47)||4 (6)||23 (46)||3 (12)|
|Asian||2 (2)||3 (4)||0||0|
|Age, mean ± SD years||43 ± 12||41 ± 13||39 ± 11||43 ± 15||0.372|
|Duration of SLE, mean ± SD years||8.8 ± 7.7||8.7 ± 9.3||4.6 ± 4.7||11.0 ± 9.7||0.003|
|Mean ± SD||2.0 ± 2.4||4.7 ± 4.5||4.3 ± 3.6||3.1 ± 3.0|
|Prednisone use||53 (53)||58 (84)||35 (70)||19 (73)||0.0002|
|Prednisone dosage, median (range) mg/day||7.5 (1–140)||10 (1–40)||10 (2.5–20)||5 (1–15)||0.0200|
|Immunosuppressive drug use¶|
|Azathioprine||7 (7)||36 (52.2)||14 (28)||2 (7.7)||<0.0001|
|Mycophenolate mofetil||15 (15)||3 (4.3)||0||12 (46.2)||<0.0001|
|Cyclophosphamide||13 (13)||4 (5.8)||2 (4)||0||0.1013|
|Methotrexate||9 (9)||5 (7.2)||11 (22)||4 (15.4)||0.0787|
|Hydroxychloroquine||63 (63)||59 (86)||37 (74)||19 (73)||0.0149|
|Components of SLE|
|ANA positive||97 (97)||63 (91)||49 (98)||25 (96)||0.290|
|Arthritis||62 (62)||50 (72)||41 (82)||21 (81)||0.039|
|Malar rash||50 (50)||30 (43)||28 (56)||12 (46)||0.583|
|Photosensitivity||48 (48)||30 (43)||24 (48)||14 (54)||0.830|
|Immunologic||39 (39)||40 (58)||19 (38)||14 (54)||0.048|
|Hematologic||49 (49)||30 (43)||16 (32)||11 (42)||0.261|
|Oral ulcers||42 (42)||17 (25)||25 (50)||6 (23)||0.009|
|Renal||28 (28)||31 (45)||11 (22)||9 (35)||0.040|
|Serositis||26 (26)||19 (28)||8 (16)||5 (19)||0.404|
|Discoid rash||18 (18)||5 (7)||5 (10)||5 (19)||0.137|
|Neurologic||7 (7)||2 (3)||0||2 (8)||0.077|
The mean duration of SLE was 8.1 years, and the mean baseline SELENA–SLEDAI score was 3.3 (median 2, range 0–18). The SLE study population was observed to be heterogeneous as indicated by statistically significant differences in race distribution, duration of SLE, and baseline SELENA–SLEDAI scores across the 4 clinical centers. Based on historical record review or assessment at the baseline study visit, nearly all of the patients (96%) were antinuclear antibody positive, and 71% had arthritis. Overall, malar rash, photosensitivity, immunologic disorders, and hematologic disorders were each reported in approximately half of the patients. The incidences of arthritis, immunologic disorders, oral ulcers, and renal disorders were significantly different among the 4 study sites (P < 0.05). The use of corticosteroids and immunosuppressive drugs also differed significantly across sites (Table 1).
Levels of BLyS, immunoglobulin subgroups, and anti-dsDNA were assessed at baseline (Table 2). The mean baseline plasma level of BLyS in all patients combined was 5.57 ng/ml (range 1.5–19.1). At baseline, BLyS levels were positively correlated with SELENA–SLEDAI scores (P = 0.057) and IgG levels (P = 0.008). In addition, BLyS levels were significantly higher among patients with detectable anti-dsDNA antibodies (P = 0.017) (Figure 1).
|BLyS, ng/ml (normal 1–5)||5.57||4.85||1.5–9.1|
|IgG, mg/dl (normal 694–1,618)||1,439||1,325||461–5,180|
|IgA, mg/dl (normal 84–463)||303||279||0–1,110|
|IgM, mg/dl (normal 48–291)||132||116||3–708|
|Anti-dsDNA, titer (by IF) (normal <1:30)‡||NA||1:40||1:10–1:1,280|
Eighty-eight percent of the patients exhibited a >25% change from baseline in the plasma BLyS level, on at least 1 clinical visit. Plasma BLyS levels varied over time during the 2 years of observation, with >55% of the patients exhibiting a >50% increase or decrease at any 1 (or more) clinical visit compared with baseline.
Association analyses (Figure 2) showed that anti-dsDNA levels were positively associated with BLyS concentrations (P = 0.0127) and with the SELENA–SLEDAI score (P = 0.0003) across all visits. BLyS levels at the current visit correlated with the SELENA–SLEDAI score (P = 0.0002), and an elevated BLyS level at the previous visit predicted anti-dsDNA positivity (P = 0.0465). On average, patients with BLyS levels of >10 ng/ml had ∼2-point higher SELENA–SLEDAI scores than those with BLyS levels of ≤10 ng/ml.
A multivariate analysis was performed to investigate the relationship between the change in the SELENA–SLEDAI scores from the previous visit and BLyS levels, with adjustment for immunosuppressive drug treatment and changes in prednisone use since the last visit, SELENA–SLEDAI score at the previous visit, and baseline demographic characteristics (Table 3). The model showed that changes in the SELENA–SLEDAI score from the previous visit were significantly and positively associated with the BLyS level at the previous visit and with the change in the BLyS level from the previous to the current visit. There was a trend toward greater increases in the SELENA–SLEDAI score at the current visit among patients who had higher BLyS levels at the previous visit and among those with greater increases in BLyS levels from the previous to the current visit.
|Independent variable†||Relationship to increase in SELENA–SLEDAI||Parameter estimate (SE)‡||P|
|BLyS level at previous visit||Positive||0.096 (0.033)||0.0042|
|Change in BLyS level from previous to current visit||Positive||0.121 (0.035)||0.0007|
|SELENA–SLEDAI score at previous visit||Negative||−0.639 (0.028)||<0.0001|
|Race effect (nonwhite), within site|
|Prednisone use in last 3 months||Positive||0.405 (0.200)||0.0520|
|Prednisone dosage change from last visit||Negative||−0.002 (0.010)||0.8168|
|Immunosuppressive drug use in last 3 months||Positive||0.358 (0.194)||0.0793|
Analysis using a multivariate repeated-measures ANOVA model with adjustment for the effects of arthritis, immunologic disorders, oral ulcers, and renal disorders showed the same significant effects of both BLyS level at the previous visit (P = 0.004) and change in BLyS level from the previous visit to the current visit (P = 0.0007). The study center effect also remained similar. The results also showed that patients with oral ulcers at the time of diagnosis were significantly more likely to have an increase in disease activity (P = 0.0075).
To assess the effect of the differences in frequency of data collection between year 1 and year 2, an analysis using the multivariate repeated-measures model was performed. The results showed the same significant effects of both BLyS level at the previous visit (P = 0.0034) and changes in BLyS level from the previous visit to the current visit (P = 0.0006). The effect of the different schedules for data collection was not statistically significant (P = 0.4619).
To evaluate the relationship between changes in BLyS level and development of a mild-to-moderate SLE flare (18) or improvement in SLE disease activity (19), BLyS levels were correlated with the percentage of patients who had an increase of ≥3 points in the SELENA–SLEDAI score from the previous visit to the current visit, and with the percentage who had a reduction of ≥4 points in the SELENA–SLEDAI score from the previous visit to the current visit. Univariate analysis showed that the increase in BLyS level from the previous visit to the current visit appeared to be associated with the occurrence of mild-to-moderate flares (P = 0.0512). In the subgroup of patients with SELENA–SLEDAI scores of ≥4, improvement of ≥4 points in the SELENA–SLEDAI score from the previous visit to the current visit was more likely to be observed among those who had a ≥1 ng/ml reduction in the BLyS level from the previous visit to the current visit with a BLyS level that was within the normal range (1–5 ng/ml) at the current visit. Twenty percent of these patients exhibited improvement of ≥4 points in the SELENA–SLEDAI score, whereas only 10% of the patients with significant increases in the BLyS level (increases of ≥1 ng/ml, from within the normal range to >5 ng/ml) showed improvement of ≥4 points in the SELENA–SLEDAI score (P = 0.0495).
Several reports have suggested a central role of BLyS in the pathogenesis of SLE (3, 5, 20). Transgenic mice that overexpress BLyS have an increased number of B cells, enlarged lymphoid organs, and hypergammaglobulinemia, and express symptoms that resemble SLE (5, 20). In addition, the constitutive overexpression of BLyS in these animals is associated with antinuclear and anti-dsDNA antibodies. Elevated levels of BLyS have been found in other murine models of SLE, including MRL-lpr/lpr and (NZB × NZW)F1 strains. It has been shown that treatment of mice overexpressing BLyS with a BLyS receptor fusion protein ameliorates the progression of SLE and improves survival (3). In SLE patients with B cell depletion, BLyS levels increase until B cell repopulation occurs (11).
Given the apparent link between BLyS and autoimmune disease pathogenesis in animal studies, in this prospective study we tracked BLyS levels, clinical disease activity, and biologic markers, and accounted for changes in therapy. Our study cohort should be broadly generalizable, in that it incorporates observations at 4 clinical centers with differences in patient ethnicity and medications. BLyS levels predicted increases in SLE disease activity across all study sites, supporting the notion that elevated BLyS has a role as a biomarker for current and/or future SLE disease activity. This observation was consistent when differences in disease manifestations, immunosuppressive drug treatment, study site, and race distribution were addressed in the model.
The results of this study should be contrasted with findings of past investigations. In one longitudinal cohort, SLE patients exhibited heterogeneity, with 50% manifesting persistently or intermittently elevated plasma BLyS levels and 61% manifesting persistently or intermittently elevated blood BLyS messenger RNA (mRNA) phenotypes (17). In another study, changes in SLEDAI scores were more closely associated with changes in BLyS mRNA levels than with changes in BLyS protein levels (21).
It is important to note that the findings in the present study are based on measurement of BLyS levels in plasma. Glomerulonephritis may increase BLyS excretion in the urine, thereby resulting in lower plasma BLyS levels when renal disease contributes to the SELENA–SLEDAI score. Future analyses of urine BLyS protein and local production of BLyS (e.g., in lymphatic tissues and joints) may prove to be highly informative.
The apparent role of increased BLyS in SLE disease activity suggests that an anti-BLyS therapeutic approach may prove useful in controlling disease activity and progression. If disease activity is driven in part by a failure of dysregulated B cells to undergo apoptosis, targeting of BLyS signaling may afford the opportunity to specifically reduce pathogenic B cell responses while maintaining sufficient B cell activity to preserve normal immune function (22).
Multivariate analyses in this observational study have shown that use of immunosuppressive therapy is positively associated with disease activity and with BLyS level at the current and previous visits. Decreases in BLyS levels were associated with improvement in SLE disease activity (19), whereas increases in BLyS levels were associated with worsening of SLE disease activity indicative of mild-to-moderate flare (18). BLyS concentrations appear to be associated with current and future disease activity in patients with SLE, as measured by the SELENA–SLEDAI, and this finding supports the rationale for investigation of BLyS antagonism as a potential therapeutic approach in SLE.
Dr. Petri had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study design. Petri, Stohl, Zhong, Freimuth.
Acquisition of data. Petri, Stohl, Chatham, McCune, Ryel, Zhong, Freimuth.
Analysis and interpretation of data. Petri, Chatham, McCune, Chevrier, Recta, Zhong, Freimuth.
Manuscript preparation. Petri, Stohl, McCune, Chevrier, Recta, Zhong, Freimuth, Dr. Lilia Pineda (nonauthor; Human Genome Sciences, Vivian Fernandez (nonauthor; Human Genome Sciences).
Statistical analysis. Chevrier, Recta, Zhong.
Medical monitoring. Freimuth.