To assess the overexpression of B lymphocyte stimulator (BLyS) over time in patients with systemic lupus erythematosus (SLE).
To assess the overexpression of B lymphocyte stimulator (BLyS) over time in patients with systemic lupus erythematosus (SLE).
Sixty-eight SLE patients were followed up longitudinally for a median 369 days. At each physician encounter, disease activity was assessed by the Systemic Lupus Erythematosus Disease Activity Index, and blood was collected for determination of the serum BLyS level, blood BLyS messenger RNA (mRNA) level, and cell surface BLyS expression. Twenty normal control subjects underwent similar laboratory evaluations.
In contrast to the uniformly normal serum BLyS and blood BLyS mRNA phenotypes in control subjects, SLE patients displayed marked heterogeneity, with 50% and 61% of patients manifesting persistently or intermittently elevated serum BLyS and blood BLyS mRNA phenotypes, respectively. Surface BLyS expression by SLE peripheral blood mononuclear cells was also often increased. Treatment of patients who had elevated serum BLyS levels with intensive courses of high-dose corticosteroids resulted in marked reductions in serum BLyS levels, and tapering of the corticosteroid dosage often resulted in increases in serum BLyS levels. Serum BLyS levels generally correlated with anti–double-stranded DNA (anti-dsDNA) titers (in those with detectable anti-dsDNA titers), but changes in serum BLyS levels did not correlate with changes in disease activity in individual patients. Serum BLyS phenotype did not associate with specific organ system involvement.
Dysregulation of BLyS over extended periods of time is common in patients with SLE. Neutralization of BLyS activity with an appropriate BLyS antagonist may be therapeutically beneficial.
B lymphocyte stimulator (BLyS; trademark of Human Genome Sciences, Rockville, MD) protein (also known as BAFF, TALL-1, THANK, TNFSF13B, and zTNF4) is a 285–amino acid member of the tumor necrosis factor (TNF) ligand superfamily (1–6). It is expressed as a type II transmembrane protein, which is cleaved from the cell surface by a furin protease to release a biologically active soluble 17-kd protein (1–5).
BLyS is a potent B cell survival factor (7–10). The numbers of mature B cells in secondary lymphoid organs, as well as baseline serum Ig levels and Ig responses to T cell–dependent and T cell–independent antigens, are markedly reduced in mice rendered genetically deficient in BLyS (11, 12). Conversely, in vivo administration of exogenous BLyS to mice induces B cell expansion and polyclonal hypergammaglobulinemia (1). Although 3 distinct BLyS receptors are known (BCMA, TACI, and BAFF-R) (6, 7, 13–18), the agonist effects of BLyS on B cells are mediated predominantly (if not solely) via BAFF-R (17–20).
A clear relationship between BLyS overexpression and systemic lupus erythematosus (SLE) has been established through 3 sets of seminal in vivo observations in mice. First, constitutive overproduction of BLyS in some (albeit not all) mice transgenic for the human or murine blys gene leads not just to polyclonal hypergammaglobulinemia, but to elevated titers of multiple autoantibodies (including anti–double-stranded DNA [anti-dsDNA]), circulating immune complexes, and renal Ig deposits as well (6, 21, 22). Second, mice genetically prone to spontaneous development of SLE ([NZB × NZW]F1 and MRL-lpr/lpr mice) harbor elevated circulating levels of BLyS by the time of disease onset (6). Third, treatment of these SLE mice with a BLyS antagonist (a soluble fusion protein between one of the BLyS receptors and IgG Fc) ameliorates the progression of disease and improves survival (6, 19).
BLyS overexpression is a feature of human SLE as well. Cross-sectional studies have documented elevated circulating levels of BLyS in ∼30% of SLE patients (23, 24). A weak (but significant) correlation was observed between circulating levels of BLyS and total IgG, and a stronger correlation was observed between circulating levels of BLyS and anti-dsDNA autoantibodies. This is consistent with observations in blys-transgenic mice that elevations in autoantibody titers are out of proportion to elevations in total Ig levels (6, 21, 22).
Based on the likely contributory role of BLyS to SLE, there is considerable interest in the development and testing of BLyS antagonists as therapeutic agents. A fully human anti-BLyS monoclonal antibody (mAb) (25) has undergone phase I testing in SLE patients and has been shown to be both safe and biologically active (26). It remains to be determined, however, whether all SLE patients would be candidates for BLyS-antagonist therapy or whether only certain patients might be suitable. Suitability may depend upon the duration and persistence of BLyS dysregulation in a given patient. To better delineate such dysregulation in SLE, we conducted a longitudinal observational study of SLE patients in which we assessed BLyS expression over time.
Eighty patients with SLE (27) who had been admitted to the Los Angeles County + University of Southern California Medical Center (LAC+USC MC) and seen in consultation by the Rheumatology Service or seen as outpatients at the Rheumatology Clinics of LAC+USC MC were recruited for this study. In order to be enrolled, the patient had to have active clinical disease at the time of enrollment, as assessed subjectively by the patient's physician (prior to complete formal assessment of disease activity). In addition, the patient had to indicate his or her intention to receive ongoing rheumatologic care at the LAC+USC MC Rheumatology Clinics to permit observational followup. The frequency of patient encounters and the medical management (e.g., adjustment of medication dosages, diagnostic studies/procedures, referral to consultant physicians, etc.) were dictated solely by clinical need. Each patient received state-of-the-art care that was not influenced by his or her participation in this study.
At each encounter, clinical disease activity was assessed by the Systemic Lupus Erythematosus Disease Activity Index (SLEDAI) (28), which is sensitive to disease flares (29), and is therefore useful in monitoring changes in disease activity for any given SLE patient. Medications being taken and ongoing organ involvement were recorded, and nonheparinized and heparinized samples of venous blood were collected for laboratory-based studies (see below). Dosages of corticosteroids other than prednisone were converted to, and reported as, daily prednisone-equivalent dosages. Each patient's medical record was reviewed retrospectively to capture prior organ involvement that may have been quiescent during the longitudinal observation period.
Twelve patients had insufficient followup at LAC+USC MC to permit meaningful longitudinal observation (1 patient died; 5 patients moved and/or changed their medical insurance status and chose to receive their rheumatologic care elsewhere; 1 patient chose to withdraw from the study but continued to receive rheumatologic care at LAC+USC MC; 5 patients were lost to followup). There were 3–12 temporally distinct encounters (median 8) for each of the remaining 68 enrolled SLE patients, who were longitudinally studied for 147–420 days (median 369 days), and the results presented herein are for these 68 patients.
Twenty healthy control subjects were recruited from personnel of the LAC+USC MC and USC Keck School of Medicine. The characteristics of the 68 patients and 20 controls at the time of enrollment are summarized in Table 1. Day 0 was defined as the day that a given recruited patient or control underwent his or her first clinical and/or laboratory evaluation.
|Control subjects (n = 20)||SLE patients (n = 68)|
|No. taking/not taking prednisone (or equivalent)||–||67/1|
|Median dosage, mg/day||–||15.0|
|Dosage range, mg/day||–||0–60|
|No. taking/not taking DMARDs|
Serum BLyS levels were determined by enzyme-linked immunosorbent assay as previously described (24, 30). The lower limit of detection in this assay is 0.1 ng/ml. To eliminate day-to-day interassay variability, all serum samples from a given subject were stored at –70°C until the subject's entire observation period was complete, and then all of that subject's serum samples were assayed on the same day.
The normal range for serum BLyS levels was defined as the geometric mean (±2 SD) of all the values obtained over time from the 20 control subjects (n = 107). Each SLE patient was assigned a serum BLyS phenotype based on his or her serum BLyS levels over time. A “persistently elevated” phenotype was assigned when all or all but 1 of the patient's serum BLyS levels were greater than the upper limit of normal. An “intermittently elevated” phenotype was assigned when at least 2 of the patient's serum BLyS levels were greater than the upper limit of normal and at least 2 were in the normal range. A “persistently normal” phenotype was assigned when no more than 1 of the patient's serum BLyS levels was greater than the upper limit of normal. A “persistently low” phenotype was assigned when the patient's serum BLyS levels were consistently at or below the lower limits of detection.
Levels of BLyS mRNA were determined by real-time quantitative polymerase chain reaction (PCR) using an ABI 7700 TaqMan Sequence Detector (Applied Biosystems, Foster City, CA). Total RNA was prepared from heparinized whole blood by TRIzol (Invitrogen, Carlsbad, CA) extraction and was cleaned using RNeasy resin (Qiagen, Valencia, CA). As for the serum BLyS determinations, all RNA samples from a given subject were stored at –70°C until the subject's entire observation period was complete, at which time all the collected samples for that subject were processed in parallel. BLyS mRNA was detected by a 1-step reverse transcription–PCR procedure, using the oligonucleotide 5′-CACCAGCTCCAGGAGAAGGCAACTCC-3′ as probe and the primers 5′-CGCGGGACTGAAAATCTTTG-3′ and 5′-CACGCTTATTTCTGCTGTTCTGA-3′. The BLyS amplicon was designed to span the region containing nucleotides 327–400 of the BLyS coding sequence. The comparative delta Ct method (Perkin Elmer User Bulletin no. 2, 1997; Perkin Elmer, Emeryville, CA) was used for quantitation of mRNA using an 18S ribosomal RNA probe as the endogenous reference. Results are presented as ratios of BLyS mRNA to 18S mRNA.
Peripheral blood mononuclear cells (PBMCs) were isolated from heparinized venous blood by Ficoll density-gradient centrifugation and were double-stained with fluorescein isothiocyanate–conjugated anti-CD14 monoclonal antibody (mAb; PharMingen, San Diego, CA) plus biotinylated anti-BLyS mAb 9B6 followed by phycoerythrin-conjugated streptavidin (Dako, Carpinteria, CA). The specificity of mAb 9B6 for BLyS has been previously documented (31). Cell surface staining observed with mAb 9B6 is not observed with control IgG1 mAb (30). Stained cells were analyzed by flow cytometry, with at least 10,000 events analyzed for each sample. Cell debris, as determined by forward- and side-scatter characteristics, was electronically excluded from the analysis.
All analyses were performed using SigmaStat software (SPSS, Chicago, IL). Nominal data were analyzed by chi-square analysis-of-contingency tables. Correlations were determined by Pearson's product-moment correlation for interval data and by Spearman's rank order correlation for ordinal data or for interval data that did not follow a normal distribution.
Among normal controls, serum BLyS levels fell within a narrow range and remained remarkably stable over time in any given subject (Figure 1A). Among the 20 control subjects, only 2 harbored single elevated serum BLyS levels (4.09 and 3.70 ng/ml, respectively; upper limit of normal 3.37 ng/ml) during their 12-month observation periods. Other than these single modest elevations, serum BLyS levels were normal at all additional time points (data not shown). Thus, all control subjects exhibited a persistently normal serum BLyS phenotype (Table 2).
|Persistently normal||Persistently elevated||Intermittently elevated||Persistently low|
|Control subjects||20 (100)||0 (0)||0 (0)||0 (0)|
|SLE patients||33 (49)||16 (24)||18 (26)||1 (1)|
|Control subjects||20 (100)||0 (0)||0 (0)||0 (0)|
|SLE patients||26 (38)||20 (29)||22 (32)||0 (0)|
In contrast, there was considerable heterogeneity in the serum BLyS phenotype among SLE patients (Table 2). Approximately one-half of the patients exhibited a persistently normal serum BLyS phenotype (Figures 1B and C), approximately one-fourth exhibited a persistently elevated phenotype (Figures 1D and E), approximately one-fourth exhibited an intermittently elevated phenotype (Figures 1F and G), and 1 patient exhibited a persistently low phenotype (data not shown). This distribution of serum BLyS phenotypes in SLE patients was significantly different from that in normal control subjects (P < 0.001). In aggregate, BLyS levels were elevated in 194 of the 532 individual sera collected from the 68 SLE patients, in comparison to 2 of the 107 individual sera collected from the 20 control subjects (P < 0.001).
The higher serum BLyS levels in SLE patients pointed to increased in vivo BLyS production and release into the circulation. Although we had no technical means of directly measuring these parameters, we did measure 2 surrogate markers that likely reflect in vivo BLyS production: peripheral blood BLyS mRNA levels and cell surface BLyS expression.
Blood levels of BLyS mRNA in normal controls showed somewhat greater variations over time, with a broader range of values in comparison to those observed for serum BLyS levels (compare Figure 2A with Figure 1A). As with the serum BLyS levels, only 2 of the 107 BLyS mRNA determinations in control blood samples were elevated (BLyS mRNA ratios of 0.00113 and 0.00106, respectively; upper limit of normal 0.00103). Thus, all control subjects exhibited a persistently normal blood BLyS mRNA phenotype (Table 2).
As with the serum BLyS phenotype, there was considerable heterogeneity in the blood BLyS mRNA phenotype among SLE patients, with the patients almost equally divided among the persistently normal, persistently elevated, and intermittently elevated groups (Table 2). This distribution of blood BLyS mRNA phenotypes in SLE patients was significantly different from that in the control subjects (P < 0.001). Moreover, 241 of the 526 individual SLE blood samples assayed had elevated BLyS mRNA levels. This was also significantly different from the results in the normal controls (P < 0.001).
In most patients, serum BLyS and blood BLyS mRNA levels closely paralleled each other (Figures 2B and C), consistent with increased blys gene expression leading to increased levels of BLyS mRNA, and that in turn, leading to increased BLyS production, and in turn, leading to increased circulating BLyS levels. However, in 14 patients, changes in serum BLyS levels were, at times, antiparallel with blood BLyS mRNA levels (Figures 2D and E). This raises the possibility of there being some feedback regulatory mechanism through which changes in circulating levels of the final product (BLyS protein) affect blys gene transcription, BLyS mRNA stability, BLyS translation, and/or BLyS posttranslational processing (e.g., secretion or release).
Also of note, the serum BLyS phenotype did not match the blood BLyS mRNA phenotype in 29 patients (Figures 2F and G). In some of these patients, the blood BLyS mRNA phenotype was the more “abnormal” one (e.g., a persistently elevated blood BLyS mRNA phenotype in conjunction with an intermittently elevated serum BLyS phenotype). The relatively depressed serum BLyS levels may have indicated increased utilization of BLyS by B cells and/or decreased cleavage of membrane BLyS to soluble BLyS. In other patients, the serum BLyS phenotype was the more “abnormal” one (e.g., a persistently elevated serum BLyS phenotype in conjunction with an intermittently elevated blood BLyS mRNA phenotype). The relatively augmented serum BLyS levels may have indicated more efficient BLyS protein translation and/or cleavage of membrane BLyS to soluble BLyS in these patients.
Of the PBMCs from normal controls, only monocytes (CD14+ cells) expressed surface BLyS, and this expression tended to be very modest. In contrast, the level of BLyS expression was frequently increased in SLE PBMCs and involved some CD14– cells as well as CD14+ cells. As an illustrative example, PBMCs from 3 control subjects and 3 SLE patients were isolated on the same day and stained in parallel for cell surface expression of BLyS (Figure 3, top). For each of the control subjects, BLyS staining of PBMCs was limited. Among the 3 SLE patients, PBMCs from 1 displayed a staining pattern similar to that of the normal controls, PBMCs from another displayed a moderate increase in surface BLyS expression (involving both CD14– and CD14+ cells), and PBMCs from the other displayed a marked increase in surface BLyS expression (involving both CD14– and CD14+ cells).
For technical reasons, PBMCs had to be stained on the day of phlebotomy. Accordingly, we could not save all the PBMC samples collected over time from a given subject and process them and analyze them on the same day. This introduced a certain degree of day-to-day variability in the staining results, compromising our ability to definitively assign a cell surface BLyS phenotype to the individual subjects. Nevertheless, longitudinal analysis demonstrated that cell surface expression of BLyS in SLE patients was often greater than that in normal controls (Figure 3, bottom). No significant correlation between serum BLyS levels and surface BLyS expression by PBMCs was noted (data not shown). This is consistent with the premise that serum BLyS levels are regulated through a composite of multiple factors, only one of which is cell surface expression of BLyS.
The wide swings in serum BLyS levels in many SLE patients were in striking contrast to the stability of serum BLyS levels among normal controls (Figure 1). One possible contributing factor to the wide swings in SLE patients is that in addition to their inherent immune dysregulation, SLE patients are also treated with immunosuppressive medications that may profoundly affect serum BLyS levels. Changes in the dosages of such medications (especially abrupt changes) could have dramatic effects on serum BLyS levels.
In response to clinical need, 8 patients with elevated serum BLyS levels were treated with short intense courses of high-dose corticosteroids (Table 3, patients A–H). All experienced marked reductions in serum BLyS levels 15–35 days later, and in all but 1 patient, serum BLyS levels declined to the normal range. An additional patient (patient I) with a normal serum BLyS level was treated with an intense course of high-dose corticosteroids, and this patient also experienced a reduction in the serum BLyS level 43 days later. Treatment with cytotoxics (e.g., cyclophosphamide, azathioprine) had no discernible effect on serum BLyS levels (data not shown).
|A||0||6.5||Prednisone 10 mg/day orally (6 doses); methylprednisolone 16 mg q8h IV (9 doses); prednisone 60 mg/day orally (10 doses)||19||2.2|
|B||0||11.2||Methylprednisolone 16 mg q6h IV (20 doses); methylprednisolone 24 mg q12h IV (2 doses); prednisone 60 mg/day orally (14 doses)||20||1.5|
|C||0||7.4||Methylprednisolone 60 mg q8h IV (3 doses); methylprednisolone 16 mg q8h IV (15 doses); methylprednisolone 16 mg q12h IV (2 doses); prednisone 60 mg/day orally (5 doses); prednisone 50 mg/day orally (3 doses); prednisone 40 mg/day orally (3 doses); prednisone 30 mg/day orally (10 doses)||28||2.9|
|D||0||4.6||Methylprednisolone 16 mg q8h IV (33 doses); prednisone 60 mg/day orally (4 doses)||15||2.6|
|E||0||8.3||Hydrocortisone 100 mg q8h IV (3 doses); methylprednisolone 16 mg q8h IV (21 doses); prednisone 60 mg/day orally (14 doses)||25||1.9|
|F||0||10.4||Methylprednisolone 12 mg q12h IV (6 doses); methylprednisolone 16 mg q8h IV (3 doses); methylprednisolone 125 mg q12h IV (2 doses); methylprednisolone 32 mg q12h IV (2 doses); methylprednisolone 24 mg q12h IV (8 doses); prednisone 70 mg/day orally (2 doses); methylprednisolone 32 mg q12h IV (14 doses); prednisone 75 mg/day orally (4 doses); prednisone 70 mg/day orally (10 doses)||33||1.8|
|G||63||5.0||Prednisone 10 mg/day (27 doses); methylprednisolone 24 mg q12h IV (10 doses); prednisone 60 mg/day orally (3 doses)||98||2.7|
|H||146||11.6||Prednisone 10 mg/day orally (6 doses); methylprednisolone 100 mg IV (1 dose); methylprednisolone 20 mg q6h IV (6 doses); prednisone 60 mg/day orally (2 doses); prednisone 20 mg/day orally (4 doses)||160||4.7|
|I||19||3.0||Prednisone 40 mg/day orally (35 doses); methylprednisolone 24 mg q12h IV (14 doses)||62||1.7|
Just as treatment with high-dose corticosteroids led to reductions in serum BLyS levels, tapering of the corticosteroid dosage routinely led to rebound increases in serum BLyS levels (Figure 4). Four representative patients illustrate this point clearly.
Serum BLyS levels in the patient whose data are shown in Figure 4A fell in response to treatment with high-dose corticosteroids. As the corticosteroid dosage was tapered, serum BLyS levels rose. When the corticosteroid dosage was increased because of a disease flare, the serum BLyS levels again fell. The second patient (Figure 4B) also experienced a dramatic decline in serum BLyS levels consequent to treatment with high-dose corticosteroids. As her corticosteroid dosage was tapered, the serum BLyS levels gradually and steadily crept upward. The third patient (Figure 4C) was already taking high-dose corticosteroids (60 mg/day prednisone-equivalent dosage) at the time of enrollment. Serum BLyS levels at that time were actually lower than normal and remained so until the corticosteroid dosage was tapered, at which point the serum BLyS levels dramatically increased. In the fourth patient (Figure 4D), the initially elevated serum BLyS level remained suppressed for a protracted period of time. This corresponded to a daily prednisone dosage maintained at ≥30 mg. When the daily prednisone dosage was finally tapered below this level, serum BLyS levels appeared to resume their ascent.
Regardless of the patient's serum BLyS phenotype or blood BLyS mRNA phenotype, clinical disease activity (as measured by the SLEDAI) for any given patient at any point in time was not reliably paralleled by the serum BLyS level (Figures 1B–G) or the blood BLyS mRNA level (data not shown). This is not surprising, since BLyS has no known direct or immediate proinflammatory properties. Thus, increases or decreases in serum BLyS levels at any point in time should not be expected to acutely promote increases or decreases in systemic or organ-specific inflammation (which would be reflected in the SLEDAI score).
In addition, there was no association between serum BLyS phenotype and organ system involvement (Table 4). This included the subset of 18 SLE patients who excreted substantial amounts (≥2.0 gm) of protein per day at some point during their observational periods (Figure 5) and might have been expected to have lower serum BLyS levels secondary to excretion of large amounts of BLyS in their urine (32). One of this latter group of patients exhibited the persistently low serum BLyS phenotype, so we were unable to accurately assess any relationship between the serum BLyS level and the degree of proteinuria. Among the remaining 17 patients, there was a suggestion that serum BLyS levels correlated inversely with the degree of proteinuria in only 3 of them (Figures 5A–C). In the other 14 patients, no obvious association was appreciated (Figures 5D–F).
|Persistently normal (n = 33)||28 (85)||26 (79)||15 (45)||14 (42)||9 (27)||7 (21)||6 (18)|
|Intermittently elevated (n = 18)||17 (94)||13 (72)||12 (67)||13 (72)||4 (22)||2 (11)||4 (22)|
|Persistently elevated (n = 16)||15 (94)||12 (75)||9 (56)||7 (44)||5 (31)||4 (25)||1 (6)|
The lack of detectable effect of serum BLyS levels on disease activity and organ involvement notwithstanding, the serum BLyS levels usually correlated well with serum anti-dsDNA autoantibody titers over time. Thirty-four SLE patients had a serum anti-dsDNA titer of ≥1:20 at ≥2 observational time points. In the 1 patient with the persistently low serum BLyS phenotype, the relationship between serum levels of BLyS and anti-dsDNA autoantibodies was indeterminate, since serum BLyS could not be reliably detected.
In 24 of the remaining 33 patients, changes in the anti-dsDNA titer were largely paralleled by similar qualitative changes in the serum BLyS level (Figures 6A–D). In the other 9 patients, changes in anti-dsDNA titers were, at times, antiparallel with changes in serum BLyS levels (Figures 6E and F), indicating that factors other than BLyS affect circulating levels of anti-dsDNA autoantibodies. Nevertheless, in the 269 individual serum samples from all 33 of these patients, serum BLyS levels correlated modestly, but significantly, with serum titers of anti-dsDNA (Figure 6G). These findings extend previous cross-sectional observations that circulating BLyS levels correlate positively with circulating levels of anti-dsDNA autoantibody (23, 24).
Cross-sectional studies have documented elevated circulating levels of BLyS in variable percentages of patients with SLE, rheumatoid arthritis, Sjögren's syndrome, or human immunodeficiency virus infection (23, 24, 33–35). The entry of at least 1 BLyS antagonist (anti-BLyS mAb) into clinical trials in SLE (26) underscores the need to fully delineate the “normal” fluctuations in BLyS expression among both healthy individuals and patients with SLE. Previous cross-sectional studies indicated that ∼30% of SLE patients have elevated circulating levels of BLyS at a single point in time (23, 24). Since cross-sectional studies are silent with regard to the duration of an abnormality, it remained unknown whether the ∼30% of SLE patients have abnormally elevated circulating BLyS levels 100% of the time, whether 100% of SLE patients have abnormally elevated circulating BLyS levels ∼30% of the time, or whether some intermediate percentage of SLE patients have abnormally elevated circulating BLyS levels some intermediate percentage of time. To address this unresolved issue, we longitudinally monitored serum BLyS protein levels, blood BLyS mRNA ratios, and PBMC surface expression of BLyS along with clinical laboratory and disease features in a cohort of SLE patients.
Our findings permit us to draw several conclusions. First, there is marked heterogeneity in the serum BLyS phenotype among SLE patients that is in dramatic contrast to the almost monolithic response observed among normal controls (Figure 1 and Table 2). Approximately 50% of SLE patients displayed abnormal serum BLyS phenotypes (intermittently elevated or persistently elevated). The patients we studied may not be completely representative of the entire SLE population, since the presence of active disease was a criterion for study enrollment. The broader SLE population may therefore manifest a somewhat lower percentage of abnormal serum BLyS phenotypes. On the other hand, our empirically determined percentage is based on a median observation period of 369 days. The true frequency of abnormal serum BLyS phenotypes in SLE may actually be greater than ∼50% and may only become apparent with longer followup.
In any case, it is likely that the elevated serum BLyS levels reflect dysregulated (increased) BLyS production in vivo. Indeed, blood BLyS mRNA levels and PBMC surface expression of BLyS, each of which likely reflects in vivo BLyS production, were also substantially increased in SLE (Figures 2 and 3 and Table 2). It is noteworthy that a subset of CD14– cells from several SLE patients expressed surface BLyS, an observation not made for corresponding cells from control subjects (Figure 3). Whether these CD14–,BLyS+ cells are monocytes that have down-regulated their surface CD14 consequent to differentiation into dendritic cells (36), other myeloid lineage cells that are inherently CD14–, or CD14– non–myeloid-lineage cells that “aberrantly” express surface BLyS in SLE patients remains to be established.
Discordance between serum BLyS levels and blood BLyS mRNA levels or PBMC surface BLyS expression was not uncommon (data not shown), pointing to complex regulation of serum BLyS levels. For example, surface expression of BLyS by monocytes in synovial fluid from patients with rheumatoid arthritis is decreased despite the presence of increased levels of soluble BLyS in these same synovial fluid samples (31). This raises the possibility that the rate of cleavage of membrane BLyS to soluble BLyS may be an important regulatory factor in the joints of rheumatoid arthritis patients, and it may play a similar important regulatory role in SLE patients. In addition, elaboration of BLyS by activated neutrophils can occur in the absence of surface BLyS expression (37). Either of these two phenomena may contribute to a dissociation between circulating BLyS levels and cell surface BLyS expression.
Furthermore, surface expression of BLyS receptors on B cells varies with the activation/differentiation state of the B cells (38). Since the vast majority of B cells reside in extravascular tissues (e.g., spleen, lymph nodes, bone marrow), unmeasured variations in BLyS receptor expression by these B cells may affect consumption of soluble BLyS and lead to a selective decrease in circulating BLyS levels without concomitant changes in blood BLyS mRNA levels or in cell surface BLyS expression. Moreover, the bulk of endogenous BLyS production likely occurs extravascularly (e.g., in spleen and lymph nodes that are rich in myeloid-lineage cells), so BLyS mRNA levels of circulating BLyS-producing cells may not faithfully reflect BLyS mRNA levels of tissue BLyS-producing cells. Indeed, elevated levels of BLyS mRNA are present in the kidneys of MRL-lpr/lpr mice but not in their PBMCs (Baker KP, Wu Y: unpublished observations). This disparity between BLyS expression in tissues and BLyS expression in the circulation notwithstanding, it is clear that chronic persistent or intermittent overexpression of BLyS is highly prevalent among SLE patients.
A second conclusion from our findings is that in patients with elevated serum BLyS levels, intense courses of high-dose corticosteroids profoundly reduce these levels, oftentimes to normal (Table 3). It may be that corticosteroids can “mask” BLyS dysregulation, leading to an underestimation of the true percentage of SLE patients with persistently elevated or intermittently elevated serum BLyS phenotypes. Although this might represent a direct effect of corticosteroids on blys gene transcription and/or on BLyS protein translation, it may actually be an indirect consequence of corticosteroid-mediated inhibition of interferon (IFN). IFNγ and IFNα each up-regulate BLyS production (30, 39), and PBMCs from many (but not all) SLE patients bear an “IFN signature” indicative of IFN up-regulation (40, 41). High-dose corticosteroid treatment extinguishes the IFN signature (41), raising the possibility that IFN is a major inducer of BLyS overproduction in SLE patients. Experiments are currently in progress to address this issue.
In any case, it is tempting to speculate that part of the salutary clinical effect of high-dose corticosteroids is related to the decrease in serum BLyS levels. Indeed, serum BLyS levels often inversely correlated with the corticosteroid dosage (Figure 4). It may be that the requirement for corticosteroids in many patients can be reduced (eliminated) by neutralization of BLyS. Clinical trials will be necessary to address this issue.
A third conclusion is that the serum BLyS phenotype does not define a specific constellation of affected organ systems in SLE and, for any given SLE patient, serum BLyS levels do not correlate with disease activity (Table 4 and Figure 1). This was predictable, given that BLyS has no known direct or immediate proinflammatory properties. Changes in serum BLyS levels at any point in time should not be expected to acutely promote changes in systemic or organ-specific inflammation (which would be reflected in the SLEDAI score). Even when we focused on the subset of patients with “substantial” proteinuria (≥2.0 gm/day) who presumably had increased renal excretion of BLyS (32), we failed to find a correlation between the degree of proteinuria and serum BLyS levels in most patients. In only 3 of the 17 informative patients was there any suggestion of an inverse relationship between these 2 parameters (Figure 5). It may be that our previous finding of an inverse correlation between serum BLyS levels and the presence of nephrotic-range proteinuria (24) was not solely a consequence of the proteinuria per se but was also influenced by the high doses of corticosteroids that these patients were taking.
The failure to document a correlation between serum BLyS levels and disease activity for any individual patient should not be interpreted to mean that elevated serum BLyS levels have no effect on disease activity. Of note, the positive correlation between serum BLyS levels and anti-dsDNA titers previously documented in cross-sectional studies (23, 24) was confirmed and extended by the current longitudinal study (Figure 6). It may be that increases in BLyS levels, by virtue of the effects on autoantibody production, do increase the likelihood of aggravating and/or exacerbating disease. Such an association might be appreciated in a large SLE population when analyzed in aggregate. Indeed, preliminary findings in a larger cohort of SLE patients suggest that plasma BLyS levels correlate significantly with SLEDAI scores across all patient visits in aggregate over time (42). However, analysis of single individual patients would fail to demonstrate such association because of the numerous factors other than BLyS that contribute to fluctuations in disease activity.
Based on our results, then, are all SLE patients candidates for BLyS-antagonist therapy? The answer to this question depends upon how one views the role of BLyS in SLE. One model has BLyS functioning as a contributor to development of SLE. BLyS per se does not cause loss of tolerance to self antigens, but once such tolerance is broken, the ever-present nature of the autoantigen permits it to repetitively stimulate the host immune system. In the presence of increasing amounts of BLyS, the autoimmune response is exaggerated. In a susceptible host, this exaggerated autoimmune response can trigger/maintain frank clinical disease (SLE).
According to this model, reducing SLE-contributory BLyS levels to normal should ameliorate disease by suppressing the BLyS-driven acceleration or exaggeration of the autoimmune response. Thus, SLE patients with the most elevated circulating BLyS levels should be the ones who are most responsive to BLyS-antagonist therapy. Those patients with persistently normal circulating BLyS levels might be relatively resistant to BLyS-antagonist therapy, since “excess” BLyS is not driving the clinical autoimmunity in these patients.
An alternative model has BLyS functioning as a passive facilitator in the development of SLE. In this model, development of the pathogenic anti-self response is inherently BLyS-independent. Regardless of whether BLyS levels are normal or elevated, the pathogenic component of the autoimmune response is similar. That is, the trigger of autoimmunity elicits a response so robust that its pathogenicity is not further amplified by elevated levels of BLyS. Accordingly, the critical genetic and/or environmental factors that lead to the autoimmune response do not directly require BLyS. Indeed, the fact that many SLE patients have normal circulating BLyS levels strongly suggests that BLyS overexpression is not absolutely essential to development of SLE. Nevertheless, given the indispensable role of BLyS in B cell development (11, 12), a certain threshold level of BLyS is required to permit any antibody responses (including autoantibody responses). When BLyS levels are reduced below this critical threshold level, the ability to fully mount the pathogenic autoimmune response (along with other B cell and humoral responses) is impaired. According to this second model, the SLE patients who should be most responsive to BLyS-antagonist therapy are those with normal, rather than elevated, circulating levels of BLyS, since in such patients, less neutralization of BLyS would be required to reach the critical threshold level.
These two models are not necessarily mutually exclusive. Within the population of humans with SLE, there may be individuals in whom BLyS plays more of a contributor role, and there may be others in whom BLyS plays more of a facilitator role. Indeed, from a therapeutic perspective, the models may operationally be viewed as a continuum, with some patients requiring more neutralization of BLyS than is required by others before salutary clinical effects can be appreciated. Thus, all SLE patients may benefit from BLyS-antagonist therapy. Appropriate clinical trials should be able to resolve these issues.
The authors thank Hal Soucier for performing the flow cytometry and all the subjects for their participation in the study.