Effect of long-term belimumab treatment on b cells in systemic lupus erythematosus: Extension of a phase II, double-blind, placebo-controlled, dose-ranging study


  • Annett M. Jacobi,

    1. University Hospital Munster, Munster, Germany), Weiqing Huang, MD, Tao Wang, BSc, Meggan Mackay, MD, Cynthia Aranow, MD, Betty Diamond, MD, Anne Davidson, MBBS: Feinstein Institute for Medical Research, Manhasset, New York
    Search for more papers by this author
  • Weiqing Huang,

    1. University Hospital Munster, Munster, Germany), Weiqing Huang, MD, Tao Wang, BSc, Meggan Mackay, MD, Cynthia Aranow, MD, Betty Diamond, MD, Anne Davidson, MBBS: Feinstein Institute for Medical Research, Manhasset, New York
    Search for more papers by this author
  • Tao Wang,

    1. University Hospital Munster, Munster, Germany), Weiqing Huang, MD, Tao Wang, BSc, Meggan Mackay, MD, Cynthia Aranow, MD, Betty Diamond, MD, Anne Davidson, MBBS: Feinstein Institute for Medical Research, Manhasset, New York
    Search for more papers by this author
  • William Freimuth,

    1. Human Genome Sciences, Rockville, Maryland
    Search for more papers by this author
    • Dr. Freimuth owns stock or stock options in Human Genome Sciences.

  • Inaki Sanz,

    1. University of Rochester, Rochester, New York
    Search for more papers by this author
  • Richard Furie,

    1. North Shore–Long Island Jewish Health System, Lake Success, New York
    Search for more papers by this author
    • Dr. Furie has received consulting fees, speaking fees, and/or honoraria from Human Genome Sciences (less than $10,000) and has received consulting fees from investment analysts GLG and Guidepoint (less than $10,000).

  • Meggan Mackay,

    1. University Hospital Munster, Munster, Germany), Weiqing Huang, MD, Tao Wang, BSc, Meggan Mackay, MD, Cynthia Aranow, MD, Betty Diamond, MD, Anne Davidson, MBBS: Feinstein Institute for Medical Research, Manhasset, New York
    Search for more papers by this author
  • Cynthia Aranow,

    1. University Hospital Munster, Munster, Germany), Weiqing Huang, MD, Tao Wang, BSc, Meggan Mackay, MD, Cynthia Aranow, MD, Betty Diamond, MD, Anne Davidson, MBBS: Feinstein Institute for Medical Research, Manhasset, New York
    Search for more papers by this author
  • Betty Diamond,

    1. University Hospital Munster, Munster, Germany), Weiqing Huang, MD, Tao Wang, BSc, Meggan Mackay, MD, Cynthia Aranow, MD, Betty Diamond, MD, Anne Davidson, MBBS: Feinstein Institute for Medical Research, Manhasset, New York
    Search for more papers by this author
  • Anne Davidson

    Corresponding author
    1. University Hospital Munster, Munster, Germany), Weiqing Huang, MD, Tao Wang, BSc, Meggan Mackay, MD, Cynthia Aranow, MD, Betty Diamond, MD, Anne Davidson, MBBS: Feinstein Institute for Medical Research, Manhasset, New York
    • Investigator, Feinstein Institute for Medical Research, 350 Community Drive, Manhasset, NY 11030
    Search for more papers by this author

  • ClinicalTrials.gov identifier: NCT00071487.



To understand the effects of long-term BLyS inhibition in human systemic lupus erythematosus (SLE).


Seventeen patients with SLE who were enrolled in a clinical trial of belimumab, a BLyS-specific inhibitor, plus standard of care therapy were studied. Phenotypic analysis of lymphocytes was performed using flow cytometry. Circulating antibody-secreting cells were enumerated using enzyme-linked immunospot assay. Serum was analyzed by enzyme-linked immunosorbent assay using an antibody that recognizes products of the VH4–34 gene. Lymphocyte counts, Ig levels, and anti–double-stranded DNA antibody levels were available as part of the clinical trial analyses.


Samples were collected on days 0, 84, 168, 365, and 532 and after day 730. The total number of B cells started to decrease from baseline between days 84 and 168. This was due to a decrease in naive and transitional B cells. CD27+IgD+ memory B cells and plasmablasts decreased only after 532 days, whereas CD27+IgD− memory B cells were not affected, and there were no changes in T cells. Serum IgM levels began to decline between days 84 and 168, but there were no changes in serum levels of IgG, IgG anti-DNA antibodies, or VH4–34 antibodies during the study. SLE patients had more IgM-, IgG-, and autoantibody-producing B cells than did normal controls on day 0. There was only a modest decrease in the frequency of total IgM-producing, but not IgG-producing, cells on days 365 and 532, consistent with the phenotypic and serologic data.


Our data confirm the dependence of newly formed B cells on BLyS for survival in humans. In contrast, memory B cells and plasma cells are less susceptible to selective BLyS inhibition.

Systemic lupus erythematosus (SLE) is a disorder in which loss of tolerance to nucleic acid antigens is associated with the development of pathogenic autoantibodies that damage target organs. B cells contribute to lupus pathogenesis not only because they produce pathogenic autoantibodies, but also because they have multiple effector functions in the immune system, including antigen presentation to T cells, production of cytokines, and migration to sites of inflammation (1). There has, therefore, been considerable interest in B cell depletion or modulation as a treatment strategy for SLE.

Therapeutic antagonism of the B cell survival molecule B lymphocyte stimulator (BLyS; trademark of Human Genome Sciences, Rockville, MD) in SLE is based on the discoveries that BLyS provides a homeostatic signal for B cell survival (2) and selection (3–5) and that soluble BLyS and its homolog APRIL are expressed at high levels in the serum of individuals with SLE (6) and in the target organs of SLE-prone mice (7, 8). We and others have extensively studied the mechanism of action of BLyS and APRIL blockade in murine lupus (9). Selective BLyS blockade reduces transitional type 2 (T2), follicular, and marginal-zone B cells with modest or no reduction of T1 B cells or T lymphocytes. Although the magnitude of the germinal center response decreases due to B cell reduction (10), high-affinity pathogenic autoantibodies are still generated by somatic mutation (11). Memory B cells do not require BLyS for survival or reactivation (11, 12), and plasma cells are maintained by APRIL when BLyS is absent (13). However, B cell depletion significantly attenuates immune activation, thus decreasing the inflammatory burden and limiting tissue damage (10).

Belimumab, a human monoclonal antibody to BLyS, prevents the binding of soluble BLyS to its receptors (14). In this study, we examined the effect on B cells in a subset of 17 patients who were enrolled in a 52-week study phase II of belimumab in patients with moderately active SLE (15), followed by an extension period and a continuation study. Some patients had received belimumab for >3 years. Initial clinical results of the parent study have been described previously (15). Briefly, belimumab was well tolerated, but the study failed to meet its primary end points at 24 weeks. A number of post hoc analyses, however, suggested that by week 52 belimumab-treated patients with serologically active disease responded better and had fewer disease flares than placebo-treated patients (15). For this reason, 2 larger global phase III trials were initiated to evaluate whether patients receiving 1 mg/kg or 10 mg/kg of belimumab plus standard of care have an improved clinical response as compared with patients receiving placebo plus standard of care.

We used a combination of flow cytometry, enzyme-linked immunospot (ELISpot) assay and serology to enumerate B cell subsets and autoreactive antibodies in treated patients. Our data suggest that in humans, as in mice, BLyS-specific inhibition targets the transitional and naive B cell populations. Effects on memory B cells, plasmablasts, or plasma cells are most likely secondary to B cell reduction, are modest in degree, and require long-term treatment to become evident.



The parent clinical trial was a 52-week, phase II, double-blind, placebo-controlled, dose-ranging study of 1 mg/kg, 4 mg/kg, or 10 mg/kg of belimumab or placebo plus standard of care therapy administered to 449 patients with moderately active SLE (15). On day 392, patients could enter an extension phase in which they continued taking their previous dose or could receive the maximum dose of 10 mg/kg. Patients who were taking placebo received 10 mg/kg of belimumab during the extension phase. Patients received treatment every 28 days. At 80 weeks, patients could roll over into a continuation study in which all patients received 10 mg/kg of belimumab. Patients were enrolled in this mechanistic substudy from 2 centers in the New York area, and 35 cc of blood was collected on days 0, 84, and 168 and every 6 months thereafter. The study blind was maintained until the clinical trial was completed.

Of 21 participating patients, 4 discontinued before day 168 or had insufficient samples collected and were excluded from the mechanistic substudy, leaving 17 patients. Of the 13 belimumab-treated patients, 3 received 1 mg/kg, 5 received 4 mg/kg, and 5 received 10 mg/kg. Ten belimumab-treated patients were rolled over to the extension portion of the phase II study; of these, 7 received 10 mg/kg of belimumab, whereas the other 3 continued their original doses of 1 mg/kg (1 patient) and 4 mg/kg (2 patients) until week 80, when all patients were switched to the 10 mg/kg dose. Four patients received placebo; all transitioned to 10 mg/kg of belimumab on day 392, but only 3 took belimumab for more than 168 days thereafter (Figure 1). Day 392 was counted as day 0 for these 3 patients, with all previous visits referred to as placebo. ELISpot data were obtained on day 0 in 17 patients, day 84 in 12 patients, day 168 in 10 patients, day 365 in 14 patients, day 532 in 12 patients, and after day 730 in 10 patients. Flow cytometry data were obtained on day 0 in 15 patients, day 84 in 8 patients, day 168 in 9 patients, day 365 in 12 patients, day 532 in 12 patients, and on or after day 730 in 11 patients.

Figure 1.

Distribution of the patients with systemic lupus erythematosus (SLE) treated with belimumab or placebo in the primary study and after rollover into the extension study on day 392. At week 80, all patients were rolled over to 10 mg/kg belimumab. Symbols represent individual patients. The table shows the demographic characteristics of the patients. Values for age are the mean ± SD years, values for number of American College of Rheumatology (ACR) criteria and baseline SLE Disease Activity Index (SLEDAI) are the mean ± SD, and values for average prednisone dosage are the mean ± SD mg/day. ANA = antinuclear antibody; RBP = RNA binding protein; aCL = anticardiolipin antibody; AZA = azathioprine; MMF = mycophenolate mofetil.

Flow cytometric analysis.

White blood cells isolated from whole blood by red blood cell lysis (BD PharmLyse; BD Biosciences, San Diego, CA) were incubated with antibodies to CD19 (SJ25-C1), IgD (IA6-2), CD44 (G44-26), CD4 (SK3) (BD Biosciences), CD27 (CLB-27/1), CD10 (MEM-78), CD38 (HIT2), CD3 (UCHT-1), CD8 (3B5), and CD62L (Dreg-56) (Invitrogen, Carlsbad, CA), in phosphate buffered saline/0.2% bovine serum albumin (BSA)/5 mM EDTA at 4°C for 45 minutes. Cells were fixed in 1.0% formaldehyde, and flow cytometry was performed using a FACSCalibur or LSRII instrument and FACSDiva software (Becton Dickinson, San Jose, CA). Up to 2.5 × 106 events were acquired per analysis. Absolute cell counts were calculated using lymphocyte counts and CD19 and CD3 subset gates. Gating was performed using FlowJo software (Tree Star, Ashland, OR), and statistical analyses were performed using GraphPad Prism software (GraphPad Software, San Diego, CA).

ELISpot assays.

Bovine thymus DNA (Sigma-Aldrich, St. Louis, MO) was passed through a Millipore filter (HAWP 45; Millipore, Billerica, MA) and coated onto ELISA plates at 10 μg/ml. Plates were blocked with 5% fetal calf serum/3% BSA. Two-fold serial dilutions of fresh peripheral blood mononuclear cells were plated in duplicate on DNA-coated plates starting at 2.0 × 105 viable cells/well. Alkaline phosphatase–conjugated anti-IgM or anti-IgG diluted 1:500 (Southern Biotechnology, Birmingham, AL) was added, and the plates were incubated overnight at 37°C. Plates were developed with BCIP (Sigma-Aldrich) 1 mg/ml in AMP buffer (0.75 mM MgCl2/0.01% Triton X/9.58% 2-amino-methyl-1-propanol [pH 10.25]). Spots were counted by a single observer (AD) who was unaware of the clinical status of the patients, and the frequency of spots per 2 × 105 cells was calculated. If no spots were detected, the frequency was given as the cutoff value of 0.5. The frequency of total immunoglobulin-secreting cells was measured in the same way using anti-human IgM or IgG (Southern Biotechnology) to coat the plates. The total number of spots/ml was calculated using the frequency and the total lymphocyte count. Fresh cells from 31 normal individuals were run in parallel with the patient samples.

Detection of serum 9G4 antibodies.

ELISA plates (Nunc, Naperville, IL) were coated overnight with 2 μg/ml of anti-human IgG (Sigma-Aldrich). Plates were blocked for 10 minutes at room temperature with SuperBlock (Pierce, Rockford, IL), and serial dilutions of sera were added and incubated for 90 minutes at room temperature. Plates were washed, incubated with biotinylated VH4–34–specific antiidiotypic monoclonal antibody 9G4 (provided by Professor F. Stevenson, Tenovus Research Laboratories, Southampton, UK) for 1 hour at room temperature, followed by alkaline phosphatase–conjugated streptavidin at a 1:2,000 dilution for 1 hour (Jackson ImmunoResearch, West Grove, PA). Plates were developed using BluePhos phosphatase substrate (KPL, Gaithersburg, MD), and optical density at 650 nm was read on a microplate reader. Serum concentrations were determined using a recombinant 9G4+ IgG monoclonal antibody as a standard. Healthy and SLE control sera were included on each plate.

Other parameters.

Lymphocyte counts and serologic assays for IgG anti–double-stranded DNA (anti-dsDNA) antibodies (Bio-Rad, Hercules, CA) and for total immunoglobulins were performed up to day 532 as part of the parent study by a commercial laboratory.

Institutional review board approval.

Patients enrolled in the parent trial consented separately to the mechanistic studies that were approved by the institutional review boards of both parent institutions.

Statistical analysis.

Because there were no differences in the parent study between patients in each of the dose groups with respect to serologic features or B cell counts (15), all belimumab-treated patients were analyzed together. Comparisons for each time point analyzed were performed using Wilcoxon's matched pairs test. In each case, the value for each patient on the indicated day was compared with the matched value on day 0. For patients receiving placebo for the first year, day 392 of the study was considered to be day 0. P values less than 0.05 were considered significant.


Patient characteristics.

A flow chart of the 15 female and 2 male patients along with demographic data is shown in Figure 1. All patients met at least 4 of the American College of Rheumatology criteria for SLE (16), were antinuclear antibody (ANA) positive at enrollment, and had SLE Disease Activity Index (17) scores of ≥4. Thirteen of the 17 patients had antibodies to dsDNA, and 3 of the remaining 4 patients had antibodies to Sm and RNP. The seventeenth patient had high-titer ANA and anticardiolipin antibodies. At enrollment, 6 patients were taking mycophenolate mofetil and 2 were taking azathioprine; 2 of these 8 were also taking hydroxychloroquine. Seven of the 9 remaining patients were taking hydroxychloroquine, and all patients were taking prednisone. Azathioprine was introduced in 1 patient on day 180, methotrexate in 1 patient on day 84, and mycophenolate mofetil in 1 patient 180 days before switching from placebo to active drug. Several patients received temporary increases in prednisone dose over the course of the study.

Serologic data.

Serum levels of IgM, IgG, IgA, and IgG anti-dsDNA antibodies were measured up to day 532. Serum levels of IgM decreased 21%, from a median of 87 μg/ml on day 0 to 69 μg/ml on day 168 (P = 0.03 by repeated-measures analysis of variance); this decrease was maintained to day 532 (Figure 2A). In 3 patients, IgM levels fell from above to below normal cutoff values (<40 μg/ml) during the study. Significant but modest decreases in serum IgA were also detected (data not shown). In contrast, serum levels of IgG (Figure 2B) and IgG anti-dsDNA (Figure 2C) did not decrease significantly.

Figure 2.

Serologic data over time in the 17 belimumab-treated patients. Levels of IgM (A), IgG (B), IgG anti-DNA (C), and VH4–34 (D) over time are shown. Note the decrease over time in levels of IgM. Colors represent treatment groups, and symbols represent individual patients, as shown in Figure 1. Data for days 84 and 168 are shown together. For patients in whom data for day 168 were unavailable, data for day 84 only are shown. ∗ = P < 0.05; ∗∗ = P < 0.01 versus day 0, by Wilcoxon's signed rank test. P = placebo.

B cells using the autoreactive heavy-chain gene VH4–34 are overrepresented in the memory and plasma cell compartments of SLE patients compared with normal controls (18) and can be detected using an antiidiotypic antibody 9G4. Serum levels of 9G4+ antibodies were elevated on day 0 in the belimumab-treated patients (1.25 ± 0.72 mg/ml versus 0.19 ± 0.06 mg/ml in normal controls; P < 0.0001) but did not change significantly during the study (Figure 2D).

Changes in frequencies of IgM- and IgG-producing cells.

On day 0 there was an increased frequency and total number of IgM-and IgG-producing cells in the peripheral blood of study patients compared with 31 normal individuals (Figure 3A). Consistent with the serologic data, the absolute number of IgM-secreting cells decreased by days 168 and 365 (median decrease of 35% and 56%, respectively) in belimumab-treated patients (Figure 3B) with a trend toward a decrease in the number of IgM anti-dsDNA–producing cells (Figure 3D). In contrast, there was no significant change in the total number of IgG-or IgG anti-dsDNA–producing cells (Figures 3C and E).

Figure 3.

Analysis of antibody-forming cells by enzyme-linked immunospot assay. A, Frequency of total and autoreactive IgM- and IgG-producing cells per 2 × 105 cells in the peripheral blood of 17 patients with systemic lupus erythematosus (SLE) at their first pretreatment visit and 31 normal controls. dsDNA = double-stranded DNA; cardio = cardiolipin. B–E, Absolute number of total IgM (B), total IgG (C), IgM anti-DNA (D), and IgG anti-DNA (E) cells/ml over time in 17 belimumab-treated patients. Colors represent treatment groups, and symbols represent individual patients, as shown in Figure 1. Data for days 84 and 168 are shown together. For patients in whom data for day 168 were unavailable, data for day 84 only are shown. ∗ = P < 0.05; ∗∗ = P < 0.01 versus day 0, by Wilcoxon's signed rank test. In B, P = 0.07 for days 532 and >730. In D, P = 0.09 for days 562 and >730. P = placebo.

Analysis of B cell subsets.

Administration of belimumab resulted in a decrease in total CD19+ B cells in the peripheral blood but no change in the number of CD3+ T cells (Figure 4). After day 532, absolute CD19+ B cell numbers were maintained at a median of 23% of their baseline levels for the duration of the study. There were no significant changes in the CD4:CD8 ratio or in the ratio of CD4+ CD45RA:CD45RO or in the percentage of CD4+ CD44highCD62Llow T cells (not shown).

Figure 4.

Absolute numbers of B cells (left) and T cells (right) over time. The total number of B cells decreased significantly by day 365 (P = 0.008 versus day 0), and the decrease remained significant over months (P = 0.001 on day 532 and P = 0.002 on or after day 730 versus day 0), whereas the T cell count did not change significantly throughout the study. Colors represent treatment groups, and symbols represent individual patients, as shown in Figure 1. P = placebo.

B cell subsets were analyzed as shown in Figures 5A–D. The numbers of transitional and naive B cells decreased significantly by day 84 after starting belimumab treatment and continued to decline to day 532, after which they remained stable (Figure 6). On day 532, the median decrease in naive B and transitional B cells was 88% and 75%, respectively. A temporary increase in circulating memory B cells has been observed in patients with rheumatoid arthritis (RA) and patients with SLE treated with BLyS inhibition (15, 19). We similarly observed an increase in circulating total memory B cells on day 84 that did not reach statistical significance (data not shown). When study patients were followed up for >1 year, decreases were observed in non–class-switched memory B cells (median decrease of 15% on day 365 and 52% on day 532) and plasmablasts (median decrease of 0% on day 365 and 40% on day 532) (Figure 6). In contrast, class-switched memory cells were resistant to BLyS inhibition (Figure 6). The heterogeneous CD27−IgD− subset that is expanded in patients with SLE (20, 21) and contains a variable subset of mutated and Ig class–switched B cells (22, 23) decreased over time (median decrease of 71% on day 365) (Figure 6).

Figure 5.

B cell subset analysis. Live CD19+ gated B cells are shown. A, Representative sample obtained 2 years after initiation of belimumab therapy, showing a predominance of Ig class–switched memory B cells and plasmablasts (CD27+/++IgD−) (62%). Other B cell subsets analyzed were CD27+IgD+ preswitched memory B cells (10%) and double-negative CD27−IgD− B cells (14%). B, Reduced percentage of naive B cells (CD27−IgD+CD10−) and transitional B cells (CD27−IgD+CD10+) (12% and 1%, respectively). Naive cells were calculated by subtracting the percent transitional cells from the percent CD27−IgD+ cells (lower right quadrant in A). C, Differentiation of plasmablasts or plasma cells (CD27++CD38++) from CD27+IgD− post-switched memory B cells. Cells were differentiated using a monoclonal antibody to CD38. D, Phenotypes of all subsets analyzed.

Figure 6.

Absolute numbers of B cell subsets over time. Cells were gated on CD19 and then as shown in Figure 5. Data for days 84 and 168 are shown together. For patients in whom data for day 168 were unavailable, data for day 84 only are shown (P < 0.02 on day 84 and not significant on day 168 for naive B cells; P < 0.01 on day 84 and day 168 for transitional B cells). Colors represent treatment groups, and symbols represent individual patients, as shown in Figure 1. ∗ = P < 0.05; ∗∗ = P < 0.01 versus day 0, by Wilcoxon's signed rank test. P = placebo.


BLyS (BAFF) is a homeostatic cytokine for B cells that is up-regulated during inflammation and links adaptive with innate immunity (24). BLyS binds to 3 receptors, BAFF-R, TACI, and BCMA, that are differentially expressed during B cell ontogeny (25). The binding of BLyS to BAFF-R is required for maturation of transitional cells, which exit the bone marrow, to marginal-zone or mature follicular B cells (26), and competition for BLyS at this stage of development determines the stringency of naive B cell selection (3, 4). In contrast, the homologous molecule APRIL, which does not bind to BAFF-R, has no effect on transitional B cell maturation (27, 28). Germinal centers form in the absence of BLyS and APRIL, and although they are smaller and of shorter duration than normal, they support isotype switching and somatic mutation, allowing the production of attenuated titers of high-affinity antibodies to exogenous antigens (11, 29). Neither BLyS nor APRIL is required for the survival or reactivation of memory B cells in normal mice (11–13), although BLyS helps promote memory B cell reactivation in humans during inflammatory states (30). Both BLyS and APRIL support the survival of plasma cells by binding to TACI and BCMA (31, 32).

Excess BLyS production expands marginal-zone and follicular B cell populations and promotes the development of SLE (33, 34). BLyS inhibition is therefore a candidate therapy for SLE. Selective blockade of BLyS alone is as effective as blockade of both BLyS and APRIL in several murine models of SLE and even induces remission of nephritis in some strains (for review, see ref.9). In mice, BLyS blockade depletes splenic B cells within 2 weeks and specifically depletes T2, marginal zone, and follicular cells while sparing T1 and B1 cells (10, 11). Marginal zone–derived responses to T cell–independent antigens are markedly attenuated by BLyS blockade, and primary humoral immune responses to T cell–dependent antigens are decreased in titer with no decrease in affinity (11). Selective BLyS blockade has no effect on plasma cells because APRIL compensates for BLyS deficiency (10, 13). Nonselective BLyS/APRIL blockade depletes short-lived plasma cells of the IgM isotype (10), but the effect on IgG plasma cells is strain and microenvironment dependent (for review, see refs. 9, 13).

In SLE models, BLyS blockade markedly decreases the size of secondary lymphoid organs and, as a result, the total number of T cells and dendritic cells within these organs is decreased. This effect may be secondary to the loss of essential B cell–derived chemokines and cytokines involved in lymphoid organization (35). After cessation of BLyS blockade, the lymphoid organs recover with a significant delay (10, 11). A complete absence of BLyS in a lupus-prone mouse strain markedly attenuates disease and skews the isotype of the glomerular Ig deposits from the complement-fixing IgG2a isotype to the IgG1 isotype that fixes complement only weakly (36).

Belimumab is a fully human recombinant IgG1λ monoclonal antibody that specifically inhibits soluble BLyS (37). Administration of belimumab for 26 weeks to cynomolgus monkeys at doses ranging from 1 to 50 mg/kg resulted in a 50–60% reduction in B cells starting after 4–8 weeks, a longer lag than in mice; the most significant reduction occurred in the spleen, with decreased size and number of the lymphoid follicles and 50–75% loss of mature B cells. T cells were not depleted. Despite the decrease in B cells, serum levels of immunoglobulins did not decrease significantly, even after 26 weeks of treatment (38).

Data from the 1-year, placebo-controlled parent study of belimumab in 449 patients have recently been reported (15). Given the increased power of the parent study compared with our substudy, several serologic findings were observed only in the parent study. Median serum levels of IgG decreased by 10% in the belimumab- treated group, compared with <5% in the placebo group. In addition, there was a modest decrease in IgG anti-dsDNA antibody titers over time that paralleled the decrease in B cell numbers. Median reductions in IgG anti-dsDNA antibodies of 17.3% and 29.4% were observed in belimumab-treated patients on days 168 and 365, respectively, compared with 7.2% and 8.6% in the placebo group (P < 0.03 and P < 0.002), and a small subset of patients that were positive for anti-dsDNA antibodies on day 0 reverted to negative by day 365 (14.6% versus 3.4% in the placebo group) (15).

Our analysis of B cell subsets confirms and extends the more limited analyses performed in the parent cohort (15). First, selective BLyS inhibition at all doses reduced B cells, with kinetics similar to those observed in primates, with preferential reduction of naive and transitional B cells to <20% of their pretreatment numbers. BLyS is therefore essential for survival of newly formed B cells that emerge from the bone marrow in humans. In contrast, non–class-switched memory cells and antibody-secreting cells decreased only after 18 months of treatment with belimumab. Non–class-switched memory cells are a mixed population that contains circulating marginal-zone B cells that are expected to be BLyS dependent (10, 39). Conventional CD27+IgD− class-switched memory cells were resistant to BLyS inhibition even after several years of treatment. Consistent with these findings, serum levels of IgM decreased slowly over time, but there was only a modest effect on serum levels of IgG. Similarly, serum levels of VH4–34–encoded antibodies did not decrease over time in our patient cohort.

Circulating plasmablasts are often increased in patients with SLE (40) and can be enumerated using ELISpot assays. Consistent with the serologic data, we observed only a slow decline in total IgM-producing plasmablasts over time but not in IgG-producing cells. This may reflect the expected decline in marginal-zone B cells and in naive B cells that give rise to short-lived extrafollicular B cell responses, with sparing of germinal center–derived B cells. In SLE patients, circulating IgG-secreting cells are exquisitely sensitive to anti-CD40L, an agent that dissolves germinal centers, indicating that they are mostly T cell–dependent derived plasmablasts (19, 41). Our study suggests that these IgG-secreting cells are generated independently of BLyS. The resistance of plasma cells to belimumab in most patients is expected since survival of plasmablasts and plasma cells is maintained in the absence of BLyS by circulating APRIL, which is not inhibited by belimumab. Similar findings have been reported in mice. In mice, germinal centers can form even when BLyS is completely absent (29), and reactivation of class-switched memory B cells is independent of both BLyS and APRIL (13). Furthermore, while anti-BLyS therapy attenuates primary immune responses in a manner that depends on the degree of naive B cell depletion, it has no effect on secondary immune responses (11, 12). Immunization studies have not been performed as yet in humans treated with a BLyS-specific inhibitor. It is therefore not possible to determine what proportion of the IgG-secreting plasmablasts remaining in treated patients after long-term belimumab treatment is newly derived from germinal centers and what proportion reflects memory B cell reactivation. This is important to address in future studies because ongoing memory cell reactivation could result in a fixed B cell repertoire that remains autoreactive in the setting of increasing B cell immunodeficiency.

The resistance of conventional memory B cells and plasma cells to anti-BLyS therapy suggests that BLyS inhibition is unlikely to correct defects in the selection of the antigen-activated B cell repertoire. We believe, therefore, that the decrease in the anti-dsDNA response observed in the parent belimumab study is most likely due to B cell reduction with a decrease in newly formed anti-dsDNA–producing B cells, offset by the long half-life of pre-existing memory and plasma cells. Nevertheless, the decrease observed in the CD27−IgD− subset, which is expanded in patients with SLE (20) and contains a distinct population of memory B cells (22, 42), is intriguing and needs further study. In addition, anti-dsDNA autoantibodies disappeared from the serum in a small subset of treated patients (15).

An alternative explanation for the heterogeneous decrease in anti-dsDNA antibodies over time is that a variable subset of these antibodies derives from extrafollicular plasma cells in a BLyS-dependent manner (43). It is also possible that BLyS blockade decreases the drive toward autoimmunity contributed by circulating nucleic acid–containing immune complexes because lower levels of BLyS may decrease the response of B cells to B cell receptor and Toll-like receptor signals (44). This, together with the finding that increased competition for BLyS influences the selection of the naive autoreactive B cell repertoire (3, 4), may gradually render the naive B cell repertoire less autoreactive over time. This mechanism should be addressed in further studies.

A remaining question in SLE is whether there is any advantage of therapeutic blockade of both BLyS and APRIL over blockade of BLyS alone. The receptor fusion protein TACI-Ig (atacicept), which is currently in clinical trials for the treatment of SLE, reduces B cells to a similar degree as belimumab because APRIL does not contribute to the survival of naive B cells. Furthermore, blockade of both BLyS and APRIL has no effect on memory B cells in mice (13), so TACI-Ig is unlikely to decrease this B cell compartment. However, since either BLyS or APRIL can support the survival of plasma cells (10), blockade of both cytokines may decrease plasma cell survival (13) and therefore decrease serum levels of autoantibodies.

The results of several phase I trials of multiple- dose TACI-Ig in humans with a variety of diseases have now been reported. In contrast to belimumab, TACI-Ig induces a rapid reduction of serum immunoglobulin levels, especially serum IgM, with decreases of 50% or more. In patients with RA, serum IgG levels variably decreased by up to 20% and serum autoantibody levels decreased by 25–40% (19). In contrast, levels of protective IgG antibodies to tetanus and diphtheria did not decrease. These findings suggest that in humans, emerging plasmablasts are susceptible to blockade of BLyS and APRIL but that long-lived plasma cells are more resistant, perhaps because other survival factors are provided to these cells (30, 45). Since IgM autoantibodies have been reported to be protective in SLE (46) and constitute a protective barrier to viral and bacterial infections, it remains to be determined whether the decrease in circulating antibodies induced by TACI-Ig has either therapeutic or adverse consequences.

Despite the absence of robust effects on autoreactive plasma cells and serum autoantibodies, selective BLyS blockade is highly effective in the treatment of some forms of murine SLE. We have shown that this is in part due to a rapid and marked decrease in the size of the spleen and lymph nodes that occurs as a result of B cell depletion, with a resultant decrease in the total number of activated T cells and dendritic cells (10). In addition, BLyS is expressed in the target organs of mice with SLE (8) and in humans with Sjögren's syndrome and RA (7) and may directly contribute to local inflammation by mechanisms that are not related to B cells (47).

The clinical effects of BLyS-specific inhibition and the limiting dose that achieves B cell reduction in human SLE remain to be determined in 2 larger global phase III studies. Our studies show that there are many similarities between the effects of BLyS inhibition in mice and humans (9). It is clear from the murine studies that depletion of autoantibodies is not necessary to achieve a therapeutic effect with BLyS blockade as long as the effector function of those antibodies is blunted. A major difference in BLyS inhibition between mice and humans is the slower kinetics of B cell reduction in humans. This, together with the presence of established autoreactive memory cell and plasma cell populations, both of which are relatively resistant to BLyS inhibition, could potentially delay or prevent a therapeutic response in some individuals. We would predict that some manifestations of SLE that are mediated by antibodies alone (for example, immune cytopenias) will only respond to BLyS inhibition if the relevant autoantibodies decrease in titer. Finally, our recent murine studies indicate that high levels of type I interferons render mice resistant to the therapeutic effects of BLyS blockade (Liu and Davidson: unpublished observations), suggesting that the presence of the interferon signature should be taken into account when analyzing the clinical data. Ongoing mechanistic studies in patients enrolled in the phase III trials of the BLyS-specific inhibitors will help to define the role of BLyS in SLE and the best use of BLyS-specific inhibitors.


All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Davidson 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 conception and design. Jacobi, Freimuth, Aranow, Diamond, Davidson.

Acquisition of data. Jacobi, Huang, Wang, Freimuth, Sanz, Furie, Mackay, Aranow, Davidson.

Analysis and interpretation of data. Jacobi, Huang, Wang, Sanz, Diamond, Davidson.