Modern immunosuppressive therapies are effective in controlling disease activity and preventing end-stage renal disease (ESRD) in the majority of patients with proliferative lupus nephritis, but none of them cures lupus. Patients may have a complete or partial response to therapy, and some of them experience relapse, especially after shorter courses of treatment (1–11). Flares pose a significant problem because of the risk of deterioration of renal function due to cumulative damage as well as the additional immunosuppressive therapy that may result in additional toxicity. Reliable information on the frequency, predictors, and outcome of renal flares may facilitate the identification of patients at higher risk for relapse.
Several studies have attempted to address these issues (1, 3, 12–18). The cumulative rate of renal flares varies from study to study depending on the definition of flare, the histologic type of lupus nephritis, treatment, and duration of followup. Flares are less common after cytotoxic drug therapy than after corticosteroid therapy and are least common after longer courses of such therapy. Reported predictors of renal flares include high activity index, higher levels of proteinuria, male sex, younger age, hypertension at presentation, use of corticosteroids alone without cytotoxic therapy, delay in initiating cytotoxic therapy, longer time to response, and short duration of treatment. Several of these reports, however, were based on analyses of modest numbers of patients with heterogeneous disease treated in nonrandomized, noncontrolled trials (12–14). Moreover, in most of these studies, African Americans, who are known to have a more aggressive disease, were not included (12–14, 16).
To further investigate this important aspect of lupus nephritis, we explored the prevalence, outcome, and predictive factors of renal flares in a cohort of 145 patients with lupus nephritis who had participated in 2 randomized controlled studies at the National Institutes of Health. To make the analysis more clinically meaningful, we included patients with a partial response to treatment as well as those who had a complete response.
Our findings suggest that flares are common in lupus patients but do not necessarily result in loss of renal function if treated with additional immunosuppressive agents. We also report that partial response to treatment, low C4 levels, and African American ethnicity are independent risk factors for renal flare. These findings suggest that although significant progress has been accomplished with modern immunosuppressive therapy in lupus nephritis, renal flares are an important aspect of the natural history of the disease that should be taken into consideration in evaluating the efficacy of any new treatment.
- Top of page
- PATIENTS AND METHODS
Response to therapy. One hundred forty-five patients with proliferative lupus nephritis entered 2 randomized controlled clinical trials comparing pulse MP, pulse CYC, or the combination of the 2 between 1981 and 1990. Complete or partial response was eventually achieved in the majority of patients, as shown in Figure 1A.
Figure 1. Kaplan-Meier curves for time to response, time to relapse after response, and comparison of time to renal flare. A, Time to partial or complete response after the initiation of protocol-dictated immunosuppressive therapy in 145 patients. Patients had to be off immunosuppressive therapy for at least 6 months to be considered partial or complete responders. B, Time to relapse after complete (n = 73) or partial (n = 19) response was achieved. C, Comparison of time to renal flare in patients with a complete response (CR) or partial response/stabilization (PR) (P = 0.02 by log-rank test). Numbers across the top are the number of patients at risk in each group at the times indicated.
Download figure to PowerPoint
Seventy-three patients fulfilled the criteria for complete response (49 Caucasian, 16 African American, 5 Hispanic, and 3 Asian), and 19 patients met criteria for partial response/stabilization (12 Caucasian, 5 African American, 1 Hispanic, and 1 Asian). Table 1 shows the characteristics of the patients by response group.
Table 1. Characteristics of the study patients, by response group*
|Complete response (n = 73)||Partial response/stabilization (n = 19)|
|Age at entry, median years||29.8||27.1|
|Female, no. (%)||64 (88)||17 (89)|
|Ethnicity, no. (%)|
| Caucasian||49 (67)||12 (63)|
| African American||16 (22)||5 (26)|
| Hispanic||5 (7)||1 (6)|
| Asian||3 (4)||1 (6)|
|Duration of followup, median months||123||116|
|Duration of lupus nephritis before treatment, median months||9||13|
|Histologic features of renal biopsy, no. (%)|
| WHO class III||24 (33)||1 (6)|
| WHO class IV||47 (64)||15 (83)|
| No biopsy||2 (3)||2 (11)|
|Immunosuppressive therapy, no. (%)|
| Pulse CYC or pulse CYC + pulse MP||51 (70)||11 (58)|
| Pulse MP||22 (30)||8 (42)|
|Chronicity index, mean ± SD||2.76 ± 2.08||3.65 ± 1.91|
|Activity index, mean ± SD||7.96 ± 3.73||10.00 ± 4.64|
Fifty-three patients did not achieve either a complete or a partial response at any time during the followup. Seventeen of these patients continued to have active disease, but did not progress to ESRD. In 26 patients, there was continuous deterioration leading to ESRD. Five patients died during the protocol treatments, and 5 patients were lost to followup either during or immediately after the treatment protocols (Figure 2).
Figure 2. Overview of patient response to immunosuppressive therapy. CR = complete response; PR = partial response; ESRD = end-stage renal disease.
Download figure to PowerPoint
Prevalence of renal flares. Flares were common among patients achieving complete or partial response, with most of the flares occurring within 4 years after meeting the response criteria (Figure 1B). Kaplan-Meier survival analysis showed that the probability of renal flare was significantly higher and occurred earlier in the group with partial response/stabilization compared with those with complete response (P = 0.02 by log-rank test) (Figure 1C).
Complete response group. Twenty-nine (40%) of the 73 patients who had a complete response to treatment (17 Caucasian, 8 African American, 3 Hispanic, and 1 Asian) experienced a renal flare after a median of 41 months from the time of complete response. Six patients had proteinuric flares and 23 had nephritic flares (17 mild/moderate and 6 severe) (Table 2). Although initially a distinction was made between mild and moderate nephritic flares, upon further analysis, it was determined that there was little prognostic value in differentiating between the 2 classes (data not shown), and, therefore, they were grouped together into a single category.
Table 2. Treatment and outcome of renal flares in patients who achieved complete response or partial response/stabilization*
| ||Complete response (n = 73)||Partial response/stabilization (n = 19)|
|Total flares (n = 29)||Proteinuric flare (n = 6)||Nephritic flare||Total flares (n = 12)||Proteinuric flare (n = 2)||Nephritic flare|
|Mild/moderate (n = 17)||Severe (n = 6)||Mild/moderate (n = 3)||Severe (n = 7)|
|Treatment of flare|
| Pulse CYC (with or without pulse MP)||12||1||6||5||6||0||1||5|
| Other IST†||5||1||3||1||4||1||2||1|
| Progression to ESRD||3||1||0||2||8||1||1||6|
Partial response/stabilization group. Patients in the partial response/stabilization group (n = 19) did not meet criteria for a complete response but had a sufficient therapeutic response to allow discontinuation of immunosuppressive agents for at least 6 months without significant deterioration in renal function during this period. These 19 patients achieved stabilization after a median of 35 months. Twelve of the 19 patients (63%) had renal exacerbation after a median of 11.5 months from the time of stabilization. Two of these flares were proteinuric and 10 were nephritic (3 mild/moderate and 7 severe) (Table 2).
Treatment and outcome of renal flares.Complete response group. Details of the treatment and outcome of renal flares are shown in Table 2. Four patients fulfilling criteria for complete response eventually progressed to ESRD, 3 of these 4 patients experienced a renal flare (2 severe nephritic and 1 proteinuric). Both patients with severe nephritic flares received aggressive immunosuppressive therapy: one with bolus CYC and the other with bolus MP. The patient with proteinuric flare was not treated with additional immunosuppressive therapy. The remaining patient progressed to ESRD due to uncontrolled hypertension.
Partial response/stabilization group. All renal flares were treated with additional pulse immunosuppressive therapy, except for those in 2 patients. One of these 2 patients had a severe nephritic flare and rapidly progressed to ESRD before the institution of treatment; the other patient was pregnant at the time and was treated with oral prednisone. A total of 8 patients (42%) progressed to ESRD. Six of these patients had a severe nephritic flare (with steady deterioration to ESRD in 4), 1 had a moderate nephritic flare that progressed to a severe nephritic flare and then steadily deteriorated to ESRD, and 1 had a proteinuric flare (Table 2).
Extrarenal flares. After achieving partial or complete response, 22 patients developed 33 episodes of nonrenal flares. Nonrenal flares were more common in patients with renal flares (13 of 41 patients, compared with 9 of 51 patients without renal flares), but the difference was not statistically significant. Only 1 patient experienced a nonrenal flare that occurred within 3 months preceding a renal flare.
Predictors of renal flares and ESRD.Complete response group. Renal flares were associated with certain demographic and laboratory parameters. In univariate analysis at the time of complete response, undetectable C4 levels (LR 11.973, 95% CI 4.125–34.756) and detectable anti-dsDNA (LR 4.885, 95% CI 1.970–12.118) were significantly associated with renal flares. In a multivariate Cox proportional hazards model that included ethnicity, chronicity and activity indices, and C4 levels at the time of complete response, African American ethnicity, C4 level <11 mg/dl, and baseline chronicity index were significant prognostic factors for renal flares (Table 3). C4 <11 mg/dl was most strongly associated with renal flares, since all 6 patients (4 Caucasian and 2 Hispanic) who had undetectable C4 at the time of response experienced a flare. Anti-dsDNA antibodies were not included in the multivariate analyses because anti-dsDNA data at the time of response were not available for 18 of the patients.
Table 3. Multivariate model predicting flares in patients who achieved complete response and in the entire cohort (complete response and partial response/stabilization)
|Variable||P||Likelihood ratio||95% confidence interval|
|Complete response group|
| C4 <11 mg/dl*||<0.0001||22.82||5.51–94.47|
| African American||0.038||3.87||1.00–7.53|
| Chronicity index†||0.017||1.26||1.04–1.53|
| Activity index†||0.57||1.04||0.90–1.19|
| C4 <11 mg/dl||<0.0001||14.20||4.74–42.52|
| African American||0.049||2.45||1.00–5.98|
| Not achieving complete response||0.827||1.13||0.37–3.45|
| Activity index†||0.700||1.02||0.91–1.14|
| Chronicity index†||0.111||1.16||0.96–1.40|
Prognostic factors for ESRD in complete responders were determined in univariate analysis (Table 4). A serum creatinine level >2.0 mg/dl at the time of complete response, a severe nephritic flare, and a higher chronicity index at baseline were associated with ESRD. Two of the 3 patients with a serum creatinine level >2.0 mg/dl at the time of complete response progressed to ESRD. There was no association, however, between progression to ESRD and the activity index at baseline or the C4 level or the presence of anti-dsDNA antibodies at the time of complete response. Multivariate analysis was not performed for ESRD because of an insufficient number of events.
Table 4. Univariate analysis of factors associated with end-stage renal disease in patients who achieved complete response and in the entire cohort (complete response and partial response/stabilization)
|Variable||P||Likelihood ratio||95% confidence interval|
|Complete response group|
| Serum creatinine >2.0 mg/dl*||0.002||24.32||3.04–194.42|
| Severe nephritic flare||0.027||9.18||1.28–65.76|
| Chronicity index†||0.019||1.72||1.09–2.72|
| Severe nephritic flare||<0.0001||11.82||3.48–40.12|
| Serum creatinine >2.0 mg/dl*||0.0007||10.75||2.73–42.26|
| Not achieving complete response||0.002||7.01||2.02–24.26|
| Chronicity index†||0.005||1.51||1.13–2.02|
| Activity index†||0.023||1.20||1.02–1.40|
| Proteinuria ≥0.5 gm/24 hours*||0.041||5.59||1.07–29.27|
| Hematocrit <33%*||0.014||4.64||1.36–15.91|
| Any flare||0.024||10.62||1.26–82.86|
Both response groups (complete and partial response/stabilization). Multivariate analysis including patients who achieved either a complete response or a partial response/stabilization showed that a C4 level <11 mg/dl at the time of complete or partial response and African American ethnicity were significant prognostic factors for renal flares, after adjusting for histologic characteristics at baseline. A serum creatinine level >2.0 mg/dl, severe nephritic flare, not achieving a complete response, and higher activity and/or chronicity indices at baseline were most strongly associated with ESRD in a univariate analysis that included both complete response and partial response groups (Table 4).
Compared with patients who achieved complete response, those with partial response/stabilization were more likely to have a renal flare (LR 2.1, 95% CI 1.08–4.2), develop a severe nephritic flare (LR 5.7, 95% CI 1.9–7.2), or progress to ESRD (LR 7.0, 95% CI 2.0–24.2).
- Top of page
- PATIENTS AND METHODS
In this study, we analyzed the predictors of renal flares/exacerbations and their impact on long-term renal survival in a cohort of patients with proliferative lupus nephritis who achieved at least a partial response to treatment with immunosuppressive agents. Flares were classified as proteinuric or nephritic. Flares/exacerbations were predicted by a lack of detectable C4 at the time of response and by African American ethnicity.
Progression to ESRD was associated with a serum creatinine level >2 mg/dl at the time of response, a severe nephritic flare, higher chronicity and activity indices at study entry, and not achieving a complete response. As expected, patients who fulfilled criteria for stabilization only, without achieving complete response, had a higher likelihood of severe nephritic flares and ESRD.
In this study, we defined a complete response according to published criteria (8), with slight modifications. Patients were required to have <1 gm/day of proteinuria, inactive urinary sediment, and stable serum creatinine levels. In addition, they had to be off immunosuppressive therapy for at least 6 months. We included patients who took <10 mg of prednisone per day because it is standard practice in our institution to maintain patients on low-dose alternate-day therapy with corticosteroids. The requirement that the criteria must be present for at least 6 months, while arbitrary, allows the identification of patients with subclinical nephritis who may develop overt evidence of nephritis after discontinuation of therapy.
Our criteria for partial response/stabilization were based on 2 requirements: immunosuppressive therapy had to be discontinued for at least 6 months, and patients were required to have a <50% increase in their serum creatinine level. These criteria identify patients who may have initially worsened during immunosuppressive therapy but were thought to have eventually responded to treatment, with stabilization of their renal function. We included this group in the analysis because of unresolved issues regarding the natural history and prognosis of their disease. We excluded patients in whom therapy could not be stopped for >6 months and those who had continuous active disease with progression to ESRD despite therapy. As expected, compared with patients who had a complete response, those who had a partial response/stabilization were more likely to have a renal flare, to have a severe nephritic flare, and to progress to ESRD despite immunosuppressive therapy.
In our cohort of complete and partial responders, a total of 8 renal flares (20%) were purely proteinuric; this is lower than the 35% rate described in the study by Moroni et al (14) and may be due to the larger number of patients with membranous glomerulopathy in their cohort. The use of angiotensin-converting enzyme inhibitors can decrease proteinuria and may confound the analysis of proteinuric flares; however, we found no difference in the rate of flares between those who received angiotensin-converting enzyme inhibitors (n = 21) and those who did not (n = 71) (data not shown). To more accurately describe the nephritic flares, we further categorized them as mild, moderate, or severe, based on changes in proteinuria and renal function. The prognosis of mild and moderate flares did not differ significantly, and we therefore combined them into a single group to simplify the analysis. Because all flares were treated with additional immunosuppressive therapy irrespective of their type, we cannot exclude the possibility that patients with a mild-to-moderate flare may have had a comparable outcome with less-aggressive therapy. This will have to be addressed in a separate study. Similar to the experience of Moroni and colleagues, nephritic flares associated with loss of renal function were clear predictors of ESRD. We do not have sufficient data to determine if early treatment of mild/moderate flares may prevent severe nephritic flares, since most patients received aggressive treatment, and only the first episode of flare in each patient was analyzed.
Univariate analysis of clinical and demographic data revealed a significant association of renal flares with undetectable levels of C4 or with elevated titers of anti-dsDNA antibody. In contrast to previous studies, we did not find that a delay in the initiation of therapy was an independent predictor of subsequent flare. This may be due to the more aggressive therapy in our protocols.
To test the significance of these findings, we performed a multivariate analysis, adjusting for important clinical features of the disease. In this model, we included baseline chronicity and activity indices to adjust for differences in disease severity and activity. Although African American ethnicity was not a significant predictor of flares in univariate models, it was included in the multivariate model in view of previous publications suggesting a worse outcome in this ethnic group. Anti-dsDNA antibodies were strongly associated with flares in univariate analysis, but they were not included in the multivariate analysis because for a large number of patients (n = 18), anti-dsDNA antibody data around the time of response were not available. Interestingly, an undetectable C4 level was the strongest predictor of flares: all 6 patients in whom C4 was undetectable experienced a renal flare. C3 levels were normal in these patients, raising the possibility that the low C4 levels may be due to the presence of the C4-null allele, which has been shown to predispose to lupus (22–24). A multivariate subset analysis of patients for whom anti-dsDNA data were included showed a significant association with both elevated anti-dsDNA antibodies and undetectable C4 levels. Although this finding has to be interpreted with caution in view of the small number of patients, it nevertheless suggests that patients with any serologic abnormalities at the time of response should be monitored more closely.
The finding that African American ethnicity is associated with a higher risk of relapse and ESRD is interesting. The clustering of non–lupus-related ESRD in family members of African Americans who have lupus nephritis suggests that in addition to lupus-predisposing genes (25), these patients may also have genes predisposing to nephropathy (26–29). Moreover, compared with Caucasians, African Americans are more likely to develop severe lupus nephritis and less likely to respond to immunosuppressive therapy (6, 7, 30). The higher rate of relapse in this group has not been previously reported and may be due, at least in part, to the more aggressive nature of lupus nephritis in these patients.
In summary, our data show that patients who have a complete response to immunosuppressive therapy have a lower risk of progressing to ESRD than those who have only a partial response. African Americans and patients with low C4 levels are at high risk for subsequent flare and can be identified at the time of response. Vigilant monitoring after immunosuppressive therapy and early institution of additional immunosuppressive therapy may prevent ESRD in most patients.
These data shed additional light on the limitations of existing immunosuppressive therapies and point to the need for controlled studies addressing the role of maintenance immunosuppressive therapies in patients who achieve a response. The data also suggest the importance of using the prevention of renal flares as an outcome measure in future studies of lupus nephritis.