1. Top of page
  2. Abstract


Serial measurements of anti–double-stranded DNA (anti-dsDNA) and complement are routine in the management of systemic lupus erythematosus (SLE), but their utility as biomarkers in preemptive treatment to prevent flares remains a subject of controversy. We hypothesized that concomitant elevation of anti-dsDNA and C3a can predict SLE activity in patients with stable or inactive disease and that short-term treatment with corticosteroids can avert flares.


In this prospective, randomized, double-blind, placebo-controlled trial, 154 patients were evaluated monthly for up to 18 months, with measurements of C3a, C3, C4, CH50, and anti-dsDNA levels. Patients who remained clinically stable but showed serologic evidence of an SLE flare (elevation of both the anti-dsDNA level by 25% and the C3a level by 50% over the previous 1–2 monthly visits) were randomized to receive either prednisone or placebo therapy at a dosage of 30 mg/day for 2 weeks, 20 mg/day for 1 week, and 10 mg/day for 1 week.


Forty-one patients (21 randomized to prednisone and 20 randomized to placebo) experienced a serologic flare. Analysis of severe flares occurring ≤90 days from randomization revealed that 6 occurred in patients taking placebo and none occurred in patients taking prednisone (P = 0.007). Severe flares resulted in an increase in the prednisone dosage to >40 mg/day and/or the addition of an immunosuppressive agent. Furthermore, improvement in scores on the Systemic Lupus Erythematosus Disease Activity Index, decreased levels of anti-dsDNA antibodies, and increased levels of C4 occurred 1 month after initiation of prednisone treatment.


These preliminary data support our hypothesis that in a subset of clinically stable SLE patients with a combination of elevated C3a and anti-dsDNA levels, short-term corticosteroid therapy may avert a severe flare.

Systemic lupus erythematosus (SLE) is a chronic disease that is frequently characterized by an undulating course of exacerbations and remissions. Since a major determinant of long-term outcome is organ damage consequent to tissue injury that accompanies disease activity and toxicity of therapy (1, 2), identification of biomarkers that accurately predict activity would be a major advance. Serial measurements of anti–double-stranded DNA (anti-dsDNA) antibodies and complement are routinely performed in clinically quiescent patients to track activity and predict flare, but their utility in decision-making and preemptive treatment remains a subject of controversy (3–5).

Bootsma et al (6) reported that treatment with prednisone reduced the likelihood of flare without increasing the cumulative dose of corticosteroids in asymptomatic patients whose sera demonstrated rising titers of anti-dsDNA antibodies. In that study, patients randomized to the treatment group required less immunosuppressive therapy and experienced fewer severe flares, but these differences did not reach statistical significance. Therefore, despite the prevailing clinical practice management of continued surveillance by serial serologic evaluations, there is insufficient evidence-based data to justify therapeutic efforts to normalize the levels of complement components or anti-dsDNA antibodies in clinically asymptomatic or stable patients.

In the present study, we assessed serial changes in 2 serologic biomarkers, anti-dsDNA antibodies and plasma C3a, as predictors of disease flare, and we sought to determine whether intervention with moderate-dose corticosteroid therapy would avert a clinical disease flare. We used a trial design that included a combination of rising C3a and anti-dsDNA antibody levels, since pathogenically, immune complexes consisting of anti-dsDNA/DNA are complement-fixing and are more likely to incite tissue injury. The primary objective in this prospective, randomized, double-blind, placebo-controlled trial was to evaluate the effectiveness of short-term, moderate-dose corticosteroid treatment in preventing severe flares when elevations of C3a levels by 50% were accompanied by a 25% increase in the anti-dsDNA titer in patients with inactive or stable/active SLE.


  1. Top of page
  2. Abstract


Patients were enrolled between 1997 and 2002. The study design is presented in Figure 1. Institutional Review Boards at all sites approved the protocol and consent forms. The study protocol was given to clinics or private practices, and eligible patients who had a history of anti-dsDNA antibody positivity and stable laboratory values/stable disease were referred to the study physicians. Each of the study visits described in the protocol were visits to the study physicians. After written consent was obtained, 180 SLE patients were screened at the following sites: the lupus clinics at the Hospital for Joint Diseases (83 enrolled; 26 randomized)/Bellevue Hospital (26 enrolled; 5 randomized) at New York University School of Medicine; Bronx Municipal Hospital Center/Montefiore Medical Center at Albert Einstein School of Medicine (35 enrolled; 7 randomized); Lenox Hill Hospital (4 enrolled; 1 randomized); Long Island Jewish Health Center (6 enrolled; 2 randomized) and affiliated private practices. Bellevue Clinic pediatric patients (ages 12–18 years) were included.

thumbnail image

Figure 1. Summary flow chart showing the outcomes in the systemic lupus erythematosus study patients from initial screening to end of study.

Download figure to PowerPoint

At study entry, all patients fulfilled at least 4 of the American College of Rheumatology criteria for SLE (7) and had a history of anti-dsDNA antibody positivity. Only patients with inactive disease or with stable/active disease (as defined by the Safety of Estrogens in Lupus Erythematosus: National Assessment [SELENA] version of the Systemic Lupus Erythematosus Disease Activity Index [SLEDAI] [8]) who were receiving no more than 15 mg of prednisone per day were eligible for the study. Inactive disease was defined as a SLEDAI score ≤4 and stable/active disease was defined as a SLEDAI score of 5–12. Patients in both these groups were considered to be clinically stable and may have been taking immunosuppressive medications, such as cyclophosphamide or mycophenolate mofetil, but remained at risk of breakthrough serologic activity. Patients had stable disease for at least 2 months, as reflected by increases in the SLEDAI scores of ≤2 points during this time interval. Dosages of cytotoxic drugs had to be stable for at least 2 months. Exclusion criteria were active infection, uncontrolled diabetes mellitus or hypertension, and pregnancy. All patients underwent skin testing for tuberculosis within 2 months of study entry. Patients with purified protein derivative reactions measuring >10 mm underwent chest radiography to exclude the presence of active tuberculosis.

At each monthly visit, levels of C3a, C3, C4, CH50, and anti-dsDNA antibodies by enzyme-linked immunosorbent assay (ELISA), a complete blood cell count, analyses of renal and hepatic function, and urinalysis (performed by certified technicians) were obtained. A 24-urine collection for determination of the protein:creatinine ratio was also obtained at baseline or if >2+ protein was noted on routine urinalysis. While the Farr assay is acknowledged to be the most specific test for measuring anti-dsDNA antibodies and to have high specificity for high-affinity antibodies, the ELISA has the advantages of not requiring radioactivity and of being used in almost all commercial laboratories, and is thereby more relevant for most rheumatologists. Disease activity was measured by the SELENA–SLEDAI and by physician's global assessment (0–3 visual analog scale). All physicians who assessed disease activity underwent training on the use of the SLEDAI conducted by 1 physician at the Hospital for Joint Diseases (JPB). The Systemic Lupus International Collaborating Clinics (SLICC) Damage Index was used for assessment of damage (2).

Randomization and treatment.

Patients in the observational phase who remained clinically stable but experienced an elevation in serologic values were considered to have had a serologic flare. Specifically, a serologic flare was defined as a concomitant elevation of anti-dsDNA antibody levels by 25% (to the abnormal range) and an elevation of C3a levels by 50% (reaching an absolute level of ≥500 ng/ml) as compared with the previous 1–2 monthly visits. Patients who had a serologic flare during a period of clinical stability (defined as a change in the SLEDAI score of ≤2 points) were randomized to receive either prednisone or placebo treatment. A stratified randomization scheme with variable block size was used, where the strata were defined according to disease activity (SLEDAI scores of 0–4 or 4–12) and cyclophosphamide infusion within 1 month (yes/no), which resulted in 4 strata. Stratification for the use of cyclophosphamide controlled for any influence of this drug on a subsequent flare when given within 30 days of randomization. Each patient was randomly assigned to receive placebo or active drug, which consisted of daily prednisone dosages of 30 mg for 2 weeks, 20 mg for 1 week, and 10 mg for 1 week.

Followup visits.

All patients in the observational phase were followed up monthly for up to 18 months until 1 of the 6 prespecified end points was reached: 1) serologic flare while remaining clinically stable, thereby becoming a candidate for randomization; 2) completion of the trial for 18 months without serologic or clinical activity; 3) clinical flare, with or without serologic activity, prior to reaching randomization criteria; 4) voluntary dropout; 5) protocol violation, such as self-adjustment of the steroid dosage or refusal to be randomized; or 6) development of 1 of the exclusion criteria (active infection, uncontrolled diabetes mellitus or hypertension, or pregnancy). Tapering of corticosteroid dosages or cytotoxic medications was allowed, but an increase in the dosage was considered to indicate a flare.

After randomization, the patients were followed up for at least 3 monthly visits to assess primary and secondary outcomes. To further minimize the possibility of observer and patient bias in outcome ascertainment, the SLEDAI was completed by 2 different investigators (C-ET, MM, GM, and/or PR), one performed the history and physical examinations and was blinded to the serologic findings, and the other performed the SLEDAI scoring, which included the serologic assessments. This measure was taken since serologic improvement could potentially unblind the investigator who was taking the history and scoring the physical examination findings.

Definition of clinical flares.

Mild-to-moderate flares were defined as one or more of the following 5 features: 1) a >3-point change in the SELENA–SLEDAI score, with a total score of ≤12; 2) new or worsening discoid, photosensitivity, or other rash attributable to lupus (including lupus profundus, cutaneous vasculitis, or bullous lupus), nasopharyngeal ulcers, pleuritis, pericarditis, arthritis, or fever not attributable to infection; 3) an increase in the prednisone dosage, but not to >0.5 mg/kg of body weight per day; 4) initiation of therapy with either hydroxychloroquine or nonsteroidal antiinflammatory drugs, without an increase in the prednisone dosage; and 5) a change of ≥1.0 in the physician's global assessment score, but with the score remaining ≤2.5.

Severe flares were defined as one or more of the following 5 features: 1) a SELENA–SLEDAI score >12; 2) new or worsening central nervous system (CNS) involvement, vasculitis, glomerulonephritis, myositis, thrombocytopenia (platelet count <60 × 109 cells/liter), or hemolytic anemia (hemoglobin level <70 gm/liter or a decrease in the hemoglobin level of >30 gm/liter over a 2-week period), each of which required doubling of the corticosteroid dosage to a final dosage of >0.5 mg/kg/day or hospitalization; 3) any manifestation requiring an increase in the dosage of prednisone or equivalent drug to >0.5 mg/kg/day, or initiation of therapy with cyclophosphamide, azathioprine, mycophenolate mofetil, or methotrexate; 4) hospitalization because of lupus activity; or 5) a change from baseline in the physician's global assessment score to >2.5.

Laboratory studies.

All assays for C3, C4, CH50, plasma C3a, and anti-dsDNA antibodies were performed at the Clinical Immunology Laboratory at the Hospital for Joint Diseases. C3 and C4 levels were determined by nephelometry (Dade-Behring, Marburg, Germany), and plasma C3a levels were determined by capture enzyme immunoassay (Quidel, San Diego, CA). CH50 levels were determined by hemolytic assay of sheep erythrocytes and anti-dsDNA antibodies by ELISA (Bio-Rad, Hercules, CA).

Statistical analysis.

All analyses were based on an intent-to-treat approach. The proportions of subjects in the 2 treatment groups who experienced a severe flare within 3 months of the time of randomization were compared using Fisher's exact test. The log-rank test was used to compare the between-group distributions of time to first severe flare and time to first flare of any type. The Cox proportional hazards model was used to estimate relative risks. Analyses adjusting for randomization strata were also performed, but since the results did not differ from the unstratified results, only the latter are reported. Data on multiple flares were analyzed using the approach of Andersen and Gill (9), which is an extension of the Cox proportional hazards model for recurrent event data. Changes in the C3, C4, C3a, CH50, and anti-dsDNA levels and in the SLEDAI scores from baseline were compared at individual time points between groups using Wilcoxon's rank sum test. Mixed effects models were fit to the repeated measurements to evaluate the treatment effect over all time points. The mixed effects models included a random subject effect and fixed effects for time, treatment, and baseline levels. Results that were non-normally distributed were log-transformed prior to model fitting. Analyses were conducted using SAS version 8.2 software (SAS Institute, Cary, NC). All reported P values are 2-sided. P values less than 0.05 were considered significant.


  1. Top of page
  2. Abstract

Enrollment and randomization of the study patients.

One hundred eighty patients with inactive or stable/active SLE were screened; 154 of them qualified for study and were enrolled into the observational phase. Figure 1 illustrates the course of the 154 patients who were followed up monthly for 12–18 months (total of 1,514 patient visits). While in the observational phase, 41 of the 154 patients developed a serologic flare (i.e., prespecified elevations of C3a and anti-dsDNA levels), yet remained clinically stable. These patients were randomized to receive either prednisone (n = 20) or placebo (n = 21). The distributions of the demographic and laboratory features and scores on the SLEDAI, physician's global assessment, and SLICC Damage Index were not statistically different between the treatment groups (Table 1).

Table 1. Characteristics of the SLE patients at randomization, by treatment group*
CharacteristicPrednisone group (n = 21)Placebo group (n = 20)P
  • *

    Of 154 systemic lupus erythematosus (SLE) patients entered into the observational phase of study, those who remained clinically stable but showed serologic evidence of an SLE flare (elevation of both the anti–double-stranded DNA [anti-dsDNA] level by 25% and the C3a level by 50% over the previous 1–2 monthly visits) were randomized to receive either prednisone or placebo therapy. IV = intravenous; SLEDAI = Systemic Lupus Erythematosus Disease Activity Index; SLICC = Systemic Lupus International Collaborating Clinics.

Age, mean ± SD years5.7 ± 13.534.1 ± 13.30.68
No. (%) female19 (90)18 (90)1.00
Race, no. (%)   
 Hispanic11 (52)8 (40)0.70
 African American3 (14)6 (30) 
 Asian4 (19)3 (15) 
 Caucasian3 (14)3 (15) 
Duration of disease, mean ± SD days245 ± 2,1162,878 ± 2,7490.48
Prednisone dosage, mean ± SD mg/day4.1 ± 4.92.7 ± 3.40.47
Cytotoxic medications, mean ± SD % of patients   
 Any cytotoxic medications5 ± 246 ± 300.73
 IV cyclophosphamide2 ± 102 ± 101.00
Hydroxychloroquine16 ± 7614 ± 700.73
No. (%) with stable/active disease5 (24)5 (25)1.00
Serologic findings, mean ± SD   
 C3a, ng/ml (normal 0–380)713.8 ± 146714 ± 1461.00
 Anti-dsDNA, IU/ml (normal 0–50)391.1 ± 417.8970.5 ± 1,892.60.19
 C3, mg/dl (normal 60–175)90.7 ± 38.476.7 ± 27.20.27
 C4, mg/dl (normal 16–37)15.1 ± 8.713.9 ± 7.40.74
 CH50, units/ml (normal 150–250)151.5 ± 73.6134.7 ± 62.10.51
SLEDAI score, mean ± SD4.2 ± 2.54.2 ± 1.30.42
Physician's global assessment, mean ± SD score0.61 ± 0.440.49 ± 0.380.41
SLICC Damage Index, mean ± SD1.05 ± 1.50.55 ± 0.830.42

Clinical outcomes in the groups randomized to prednisone or placebo treatment.

Following randomization, 11 patients experienced 13 flares within the 90-day followup period: 6 were mild/moderate and 7 were severe (Table 2). One patient (patient 42) had a mild/moderate flare as well as a severe flare, whereas another (patient 26) had 2 severe flares, over a 2-month period. In the placebo group, 6 of the 20 patients (30%) experienced a severe flare, but in the prednisone group, none of the 21 patients (0%) experienced a severe flare (P = 0.0086 by Fisher's exact test and P = 0.0071 by log-rank test). Since all severe flares occurred in the placebo group, we were not able to estimate the relative risk using the Cox proportional hazards model.

Table 2. Patients who experienced a disease flare after randomization, by severity of the flare*
 SLEDAI score at randomizationType of flareTreatmentTime of flare (days since randomization)Treatment group
  • *

    SLEDAI = Systemic Lupus Erythematosus Disease Activity Index; HCQ = hydroxychloroquine; NSAIDs = nonsteroidal antiinflammatory drugs; pred. = prednisone; ACE = angiotensin-converting enzyme; MMF = mycophenolate mofetil; CNS = central nervous system; methylpred. = methylprednisolone; IV = intravenously; CYC = cyclophosphamide; DPGN = diffuse proliferative glomerulonephritis; AZA = azathioprine.

  • Patient 42 had a mild/moderate flare as well as a severe flare over a 2-month period, and patient 26 had 2 severe flares over a 2-month period.

Mild/moderate flare     
 Patient 14ArthritisNone89Prednisone
 Patient 612ArthritisNone63Prednisone
 Patient 418Arthritis, rashNone32Prednisone
 Patient 424ArthritisNone39Placebo
 Patient 1084AlopeciaNone40Placebo
 Patient 1677Discoid rashHCQ added81Prednisone
Severe flare     
 Patient 36Arthritis, renal (membranous nephropathy)NSAIDs; pred. (20 mg/day taper); ACE inhibitor17Placebo
 Patient 266Oral ulcer, malar rash, arthritis, pleurisyPred. (15 mg/day taper)29Placebo
 Patient 266Renal (worsening of membranous nephropathy)MMF dosage increased64Placebo
 Patient 424CNS lupus with hospitalizationMethylpred. (80 mg/day IV for 4 days), then pred. (60 mg/day taper); continued monthly IV CYC62Placebo
 Patient 624Arthritis, fever, renal (DPGN)Pred. (60 mg/day taper); MMF added22Placebo
 Patient 1744Pyoderma gangrenosum, pancytopeniaPred. (40 mg/day taper); AZA added42Placebo
 Patient 1756Serositis, pleural effusion, fever with hospitalizationMethylpred. (40 mg/day IV for 5 days), then pred. (60 mg/day taper)82Placebo

The 7 severe flares, which occurred in 6 patients taking placebo, consisted of 3 nephritis (1 de novo in patient 62), 1 mucocutaneous with arthritis and pleurisy (later active nephritis), 1 vasculitic pyoderma gangrenosum with pancytopenia, 1 CNS lupus, and 1 serositis associated with fever and dyspnea. Hospitalizations were required for the latter 2 patients. All patients required an increase in the prednisone dosage to >40 mg/day and/or the addition of an immunosuppressive agent.

Analysis of flares of any type (mild/moderate or severe) that occurred ≤90 days from the randomization date indicated that 4 of 21 patients (19%) in the prednisone group and 7 of 20 patients (35%) in the placebo group experienced at least 1 flare (P = 0.31 by Fisher's exact test and P = 0.20 by log-rank test). The relative risk (placebo versus prednisone) was estimated to be 2.2 (95% confidence interval 0.64–7.47). Two subjects in the placebo arm experienced multiple flares. One patient had a mild/moderate and a severe flare; the other patient had 2 severe flares. When these multiple flares were included in the analysis using the approach described by Anderson and Gill (9), the relative risk increased to 2.36 (95% confidence interval 0.73–7.66), but remained nonsignificant (P = 0.15).

Although differences in the baseline characteristics were not statistically significant, the median anti-dsDNA antibody level at the time of randomization was higher in the placebo group than in the prednisone group. The Cox proportional hazards model could not be used to adjust for baseline anti-dsDNA levels in the analysis of severe flares because all events occurred in the placebo group. Therefore, to determine whether the higher level of anti-dsDNA antibodies at baseline could account for the increase in severe flares in the placebo group, the study population was subgrouped according to the median anti-dsDNA antibody level in all study patients at baseline: ≤416 IU/ml or >416 IU/ml. In the subgroup with higher levels of anti-dsDNA antibodies, none of the 7 patients taking prednisone (0%) had a severe flare, but 5 of the 13 patients taking placebo (38%) had a severe flare (P = 0.07 by log-rank test), indicating that preventive prednisone treatment was also effective in this subgroup. In the subgroup with lower levels of anti-dsDNA antibodies, 0 of 14 patients taking prednisone (0%) had a severe flare, as compared with 1 of 7 patients taking placebo (14%) (P = 0.16 by log-rank test).

Effects of treatment on serologic findings and SLEDAI scores.

Outcome measures that showed improvement at 1 month in the prednisone group included the SLEDAI score (P = 0.02), the anti-dsDNA antibody level (P = 0.04), and the C4 level (P = 0.04). A trend toward decreased C3a levels after 1 month of prednisone treatment was also observed, but this did not reach statistical significance. Prednisone was significantly associated with an overall decrease in anti-dsDNA levels (P = 0.02) and SLEDAI scores (P = 0.04) after adjusting for baseline values and time; treatment effects on other serologic measures were not significant. When a treatment-by-time interaction term was added to each model, the interaction terms were significant for the anti-dsDNA antibody level (P = 0.01), SLEDAI score (P = 0.003), and C4 level (P = 0.01), indicating that the effect of treatment on these outcomes was not constant over time.

Cumulative dose of corticosteroids and initiation of new immunosuppressive therapy.

Among the patients who experienced a flare, only those with severe flares in the placebo group required initiation or increased dosages of corticosteroids or required the addition of immunosuppressive therapy during the study period. None of the patients who had mild/moderate flares needed additional medication, with the exception of 1 patient in the placebo group, who was given hydroxychloroquine for discoid rash (Table 2).

The mean daily dose of prednisone was calculated for each patient in both groups at randomization and for 3 subsequent monthly visits (Figure 2). The mean daily dose of prednisone was higher in the prednisone-treated cohort during the first month following randomization, reflecting the study treatment of 30 mg of prednisone per day superimposed on the baseline dosage. In the 6 patients who experienced a flare while taking placebo, the mean daily dose of prednisone sharply increased at month 2 and remained higher for the duration of followup. The median dose of prednisone in the placebo and prednisone groups was 350 mg and 135 mg, respectively, but the difference was not significant (P = 0.43). However, the median dose in the 6 patients who experienced a flare while taking placebo was 837 mg, and in 2 of these patients, immunosuppressive therapy was initiated (mycophenolate mofetil in 1 and azathioprine in 1). Further analysis with regard to the risk/benefit of preventive prednisone therapy indicated that the number needed to treat in order to prevent 1 severe flare was 3.33 patients.

thumbnail image

Figure 2. Mean daily dose of prednisone in the systemic lupus erythematosus patients randomized to receive prednisone or placebo treatment. The mean daily prednisone dose was calculated for each patient in the prednisone group (n = 21) (•) and the placebo group (n = 20) (▪) at randomization (visit 0) and at 3 subsequent monthly visits. The 6 patients in the placebo group who experienced a flare are also shown (▴). Their mean daily dose of prednisone sharply increased at month 2 and remained higher for the duration of followup.

Download figure to PowerPoint

Prednisone was well tolerated (no hypertension or glucose intolerance), with only 1 patient in the active treatment group discontinuing after 2 weeks because of dyspepsia. The numbers and severity of adverse events occurring ≤90 days following randomization did not differ between the prednisone and placebo groups. Twelve patients taking prednisone and 11 patients taking placebo experienced an adverse event.

Outcomes in the nonrandomized patients.

One hundred thirteen clinically stable patients who entered the observational phase of the study did not meet our criteria for randomization (Figure 1). Fifty-seven of these patients completed the trial without experiencing a serologic flare or clinical event. The other 56 patients reached prespecified stopping points, as follows: 8 had protocol violations, which included self adjustment of the prednisone dosage without physician consent and refusal to be randomized; 8 were excluded because of pregnancy or active infection; 19 dropped out; and 21 had clinical flares, which occurred without the serial serologic abnormalities required to meet the randomization criteria.

Of the 21 clinical flares, 8 were renal flares and the others were rash, arthritis/arthralgia, or serositis consistent with mild/moderate flares. In 6 of the 21 flares, increased serologic activity was not noted prior to or at the time of the flare. Among the 8 patients with new renal flare, 2 had elevations of both the C3a and the anti-dsDNA antibody levels prior to the flare, but the values did not meet the criteria for randomization. Four patients had elevations of either the C3a or the anti-dsDNA antibody levels prior to clinical renal flares, and the 2 remaining patients had normal serologic findings despite clinical activity. Of the 2 patients who had renal flares in the absence of serologic activity, 1 had a history of biopsy-proven membranous nephropathy, and the other (who has a history of diffuse proliferative glomerulonephritis) did not have an active urinary sediment despite an increase in proteinuria of >2 gm in a 24-hour collection. No additional biopsy was performed at the time, but the patient was started on prednisone 1 mg/kg/day and mycophenolate mofetil.


  1. Top of page
  2. Abstract

Discordance between serologic tests and clinical features in SLE has been previously described, and its significance is unclear. In this study, we used a prospective, randomized, double-blind, placebo-controlled study design to assess interval changes of a combination of 2 biomarkers, C3a and anti-dsDNA antibodies, as predictors of impending disease flare. We focused on a limited but important population of patients who were clinically stable, but had serologic activity, a group first described by Gladman et al (10). Our findings indicate that short-term, moderate-dose corticosteroid treatment can avert severe flares of SLE in patients with serologically active disease who demonstrate increasing levels of C3a and anti-dsDNA antibodies over the previous 4–8 weeks. Proof of concept was further provided by the normalization of anti-dsDNA antibody, C3a, and C4 levels and the improvement in SLEDAI scores during the month that followed prednisone treatment.

Although the followup period was limited to 3 months, the treatment appeared to have a sparing effect on the use of immunosuppressive agents, since no additional cytotoxic agents were started in the group receiving prednisone. Moreover, there was no serious recurrent or new organ involvement in a group of preselected anti-dsDNA–positive patients who were likely to be at greater risk of life-threatening lupus, as evidenced by the 1 CNS flare and the 3 renal flares (1 de novo) that occurred only in the placebo group.

In this study, we intentionally set the bar for serologic activity requiring prednisone intervention high (requiring both a 50% elevation of the C3a level and a 25% elevation of the anti-dsDNA antibody level) in order to minimize the use of corticosteroids in patients who would otherwise develop only mild-to-moderate flares. Minor fluctuations in these levels without any correlation with clinical activity have also been observed. A consequence of these stringent requirements was that 21 of the 113 patients followed up during the observational phase of the study developed a flare without meeting the stringent protocol definition of serologic flare. Of these 21 clinical flares, 8 were considered severe, and all of them were renal flares. All but 2 of the 8 patients exhibited deterioration of the C3a and/or anti-DNA antibody levels that were not sufficient to meet our prespecified criteria for randomization. Future trials using less stringent criteria for randomization could help determine the optimal serologic biomarker for detecting mild/moderate versus severe flares.

Certain limitations regarding the generalizability of the study findings should be noted. First, this was a randomized trial in which serologic tests were performed monthly, a frequency not typically achieved in clinical practice. Second, the observations pertain to a specific subset of SLE patients: those with a history of anti-DNA antibodies and a subsequent serologic flare as defined in Patients and Methods. The numbers of patients who met the strict criteria of serologic flare represented ∼27% of the 154 patients who were enrolled in the observational phase of the study. Indeed, most patients who were enrolled exhibited neither a serologic flare nor a clinical flare during the observation period, which is consistent with previous observations of SLE patients at our institution and others (8). Third, patients were considered for the formal “screening” phase of the study only if they had a history of anti-dsDNA antibodies and had been clinically stable for at least 2 months. Although we did not record the number of patients who were initially excluded at prescreening, we estimate that the percentage of such patients in our clinics was 50%. Finally, as detailed above, 21 patients experienced clinical flares but did not meet our serologic criteria for randomization to study treatment. Thus, our data indicate that corticosteroids reduced the incidence of severe flare in the subset of patients, followed up monthly, who met our stringent criteria for serologic flare.

Several previous studies have established the reliability of using serial anti-dsDNA antibody measurements to predict lupus exacerbations (4, 11–18). Bootsma et al (6) studied 156 SLE patients in whom anti-dsDNA antibodies were measured monthly. Based on increases in this biomarker by 25% from baseline, 22 patients were randomized to preemptive treatment with prednisone 30 mg/day, which was then tapered to baseline over the next 18 weeks, and 24 patients were randomized to conventional treatment, with adjustment of the prednisone dosage and cytotoxic drugs only in cases of clinical relapse. Consistent with our findings, the flare rate was greater in the conventional treatment group than in the prednisone treatment group (20 versus 2; P < 0.0001). Furthermore, 7 major relapses in the conventional treatment group required cytotoxic therapy, as compared with 2 major relapses in the preemptive prednisone treatment group.

Notwithstanding the observations by Bootsma et al (6), and although activation of complement plays a central role in the pathogenesis of tissue injury in SLE (12, 19–25), the use of surrogate biomarkers, including anti-dsDNA antibodies, or conventional measures of C3, C4, or CH50 to predict disease activity in patients with clinically quiescent or stable disease has fallen short of expectations because of heterogeneity of the study populations, irregularity or infrequency of serologic testing, variations in the synthesis of complement components, hereditary deficiencies, extravascular distribution of complement components, and ethnic differences (26–29). Circulating levels of complement activation products (C3a, C4d, erythrocyte-bound C4d, Ba, Bb, and SC5b–9) have been shown to be superior discriminants of disease activity as compared with C3 or C4 (30–40), and increased levels of C3a may precede other serologic or clinical evidence of an impending flare (31).

Therefore, this study, which showed the benefit of corticosteroid treatment initiated on the basis of serologic evidence of exacerbation, was designed to address several of the weaknesses inherent in previous studies which have dampened enthusiasm for biomarker-driven treatment strategies in the past, as follows. First, the study population was limited to patients at higher risk of serious visceral involvement because of a known history of complement-fixing anti-dsDNA antibodies. Second, monitoring of the sensitive complement analyte C3a was performed at monthly intervals. Third, the criterion for preemptive treatment required concomitant abnormalities of 2 biomarkers (C3a and anti-dsDNA antibody levels), and was based on presumed pathogenic mechanisms. Using this targeted treatment strategy, we demonstrated that the likelihood of a severe flare in such clinically stable patients exceeds 30% over a period of 12 weeks and that corticosteroids can effectively avert those flares.

These findings support the premise that consideration should be given to the use of corticosteroids in clinically stable SLE patients who develop evidence of serologic (biomarker) activity. The risk of any preemptive treatment must be weighed against several factors, including the toxicity of the intervention, the number needed to treat in order to prevent a severe flare, and the consequences of a serious flare that might otherwise have been averted. In our series, the number and severity of adverse events reported in both groups were comparable. The patients continue to be monitored for the development of late toxicity, such as avascular necrosis.

The number needed to treat in order to prevent 1 severe flare was 3.3 patients, indicating that preventive treatment would require the exposure of patients to corticosteroids who might otherwise not have required such treatment. However, the risk of developing a serious clinical flare within 90 days in the untreated patients was substantial at 30%, and the consequences of such flares were serious, perhaps life-threatening. Moreover, serious flares required not only higher corticosteroid dosages, as evidenced by the conservative estimate of the cumulative dose of corticosteroids among the 6 patients with severe flares, but also increased dosages or the addition of immunosuppressive agents that would likely be maintained for at least 12 months.

In summary, we evaluated a cohort of serologically active, yet clinically stable SLE patients and demonstrated the benefit of moderate-dose corticosteroid therapy to avert severe flares. The data indicate that concurrent elevations of pathogenic anti-dsDNA autoantibody and C3a levels in SLE patients should be considered with prognostic gravity, similar to that of silent hypertension or persistent uncontrolled hyperglycemia in diabetic patients (41). The preliminary results of this study should therefore cause us to reevaluate the current “watch-and-wait” treatment strategies and consider the use of carefully monitored, biomarker-determined intervention with moderate-dose corticosteroids or other mechanism-based therapies in the management of high-risk SLE patients. The finding by Roman et al (42) of a negative correlation between atherosclerosis and aggressive therapy in patients with SLE is an additional rationale for this strategy, in which more vigorous treatment can decrease the likelihood and burden of atherosclerosis. This study provides a model for the application of serologic assays in patients to serve as surrogate end points in the development of innovative treatments for SLE. Accordingly, the monitoring of serologic markers as an indicator of when to initiate treatment with corticosteroids or newer agents is an attractive strategy.


  1. Top of page
  2. Abstract