Outcome and prognostic indicators of diffuse proliferative lupus glomerulonephritis treated with sequential oral cyclophosphamide and azathioprine

Authors


Abstract

Objective

To study the outcome and prognostic indicators of diffuse proliferative glomerulonephritis (DPGN) in patients with systemic lupus erythematosus (SLE) treated with sequential oral cyclophosphamide (CYC) and azathioprine (AZA).

Methods

SLE patients with biopsy-proven DPGN treated with sequential oral CYC and AZA were studied. Those who achieved renal remission at 12 months were identified, and the clinical predictors of complete remission were evaluated by regression analysis. All patients were followed up until a relapse of the nephritis or a doubling of the serum creatinine level occurred. The timing and risk factors for flares and creatinine doubling were evaluated by Kaplan-Meier analysis and with the Cox proportional hazards model.

Results

We studied 55 patients (47 women, 8 men; mean ± SD age at renal biopsy 31.1 ± 10.4 years); 25 (46%) had a serum creatinine level >106 μmoles/liter, and 29 (53%) had nephrotic syndrome. At 12 months posttreatment, 37 (67%) had complete remission and 12 (22%) had partial remission. The initial serum creatinine level was an independent predictor of complete remission. Excluding the 4 patients who were treatment- resistant or died, 21 patients (41%) had renal flares during a median followup of 4 years. The cumulative risk of renal flare was 6% at 1 year, 21% at 3 years, and 32% at 5 years. The median time to relapse was 43 months. The histologic activity score and the mean daily dose of CYC were multivariate predictors of renal flare, by Cox regression. At the last followup visit, 9 of 54 patients (17%) had a doubling of the creatinine level, 6 of whom (11%) underwent dialysis. The cumulative risk of creatinine doubling was 8.4% at 5 years and 18.2% at 10 years. An increasing chronicity index at the time of initial renal biopsy was an independent predictor of deterioration in renal function.

Conclusion

Sequential therapy with oral CYC followed by AZA appears to be an effective treatment regimen for DPGN in patients with SLE, with 89% of patients achieving complete or partial remission at 12 months, 62.8% remaining in remission after 5 years, and 81.8% having stable renal function after 10 years. Predictors of treatment resistance and relapse include increasing serum creatinine level, higher histologic activity scores, and a lower dose of CYC. Increasing chronicity indices predict a deterioration of renal function.

Nephritis is a major complication of systemic lupus erythematosus (SLE) and is a strong determinant of morbidity and mortality (1– 3). Up to two-thirds of patients develop renal disease at some stage of their illness (4). Renal involvement in SLE ranges from mild proteinuria to rapidly deteriorating renal function. Among the various World Health Organization histologic classes of lupus nephritis, diffuse proliferative glomerulonephritis (DPGN) is associated with the worst prognosis. Various studies have reported a 5-year cumulative probability of end-stage renal disease of 11–48% (5–8).

Because of the unfavorable prognosis of DPGN in SLE patients, aggressive treatment must be instituted. Information from the available literature suggests that cytotoxic agents used in conjunction with oral prednisone are superior to prednisone alone in the preservation of renal function and in patient survival (9– 11). In studies by the National Institutes of Health (NIH), regimens consisting of steroid and cyclophosphamide (CYC) were found to be significantly better than steroid alone in the long-term preservation of renal function (9, 10). Among the various CYC-containing regimens, intermittent intravenous (IV) pulse CYC tended to be the most effective and was associated with less toxicity than the oral regimens (12).

Although IV pulse CYC was widely used in the treatment of SLE patients with DPGN, the reported response rates varied according to the characteristics of the study patients, the severity of the nephritis at study entry, race, and dosage and frequency of CYC pulses (13–25). Moreover, since 30–50% of patients might still develop either a doubling of the serum creatinine level or end-stage renal failure within 5 years (16– 23), more-aggressive regimens are necessary.

At Queen Mary Hospital, a therapeutic regimen consisting of high-dose steroid and sequential administration of oral CYC followed by azathioprine (AZA) is being adopted. Complete remission has been achieved in 77% of patients after a mean observation of 33 months (26). In our recent study comparing the sequential oral regimen with IV pulse CYC in 2 historical cohorts of patients with SLE and DPGN, we demonstrated that the sequential oral regimen was associated with a tendency toward better renal outcome at 24 months posttreatment (27). However, data regarding renal flares and deterioration of renal function after this sequential regimen are largely unreported. The current study was undertaken to describe the long-term outcome and prognostic indicators of DPGN in SLE patients treated with sequential oral CYC and AZA in a moderately large patient cohort.

PATIENTS AND METHODS

Patient population

We studied 55 SLE patients with histologically confirmed DPGN who were treated with sequential oral CYC and AZA between 1978 and 1998. All patients were ethnic Chinese whose families originated in southern China. All fulfilled at least 4 of the American College of Rheumatology criteria for the classification of SLE (28). The treatment regimen was as follows: prednisone (1 mg/kg/day) for 8–10 weeks, gradually tapering to a maintenance dosage of 5–10 mg/day, plus oral CYC (1–2 mg/kg/day) for 6–9 months followed by AZA (50–100 mg/day). The exact daily dose and duration of the CYC treatment were determined by individual specialists- in-charge of the patients and were adjusted according to the patient's body weight, white blood cell counts, the presence of severe intercurrent infection, and the tolerability and toxicity of the drug.

All patients were evaluated at 12 months posttreatment. Those who achieved complete renal remission were identified, and the predictors of remission were studied. All patients were followed up until a relapse of the nephritis or a doubling of the serum creatinine level occurred. For those who did not meet the end points, data were censored at the time of the latest followup visit. Timing and risk factors for renal flares and creatinine doubling were evaluated by Kaplan-Meier analysis and Cox proportional hazards model.

Definitions of renal remission and relapse

Complete, partial, and no response of nephritis after immunosuppressive treatment were defined as suggested by Boumpas and Balow (29), with slight modifications. Complete remission was defined as stabilization or improvement in renal function with reduction of proteinuria to ≤1 gm/day and persistent normalization of the C3 level for at least 6 months. Partial remission was defined as stabilization or improvement in renal function with persistent reduction of proteinuria (if nephrotic range at baseline, a ≥50% decrease in proteinuria but <3 gm/day; if non-nephrotic at baseline, a decrease to ≤50% of the pretreatment value but >1 gm/day) for at least 6 months. Nonresponse was defined as a deterioration in renal function exclusive of other causes (such as sepsis, nephrotoxic agents, overdiuresis, and renal vein thrombosis), an increase in proteinuria, or a reduction in proteinuria but not to the extent of complete or partial remission.

A proteinuric renal flare was defined as an increase in proteinuria to >2 gm/ day, with or without deterioration of renal function, in patients who achieved complete remission and a doubling of proteinuria, with or without deterioration in renal function, in patients who achieved partial remission. A nephritic flare was defined as an increase or recurrence of active urinary sediments, with or without a concomitant increase in proteinuria or a deterioration of renal function.

Renal biopsy was performed within 2 weeks after patients presented with signs and symptoms of renal disease. Biopsy specimens were examined by light, immunofluorescence, and electron microscopy. Activity (maximum score 24 points) and chronicity (maximum score 12 points) indices were determined according to the histologic criteria described previously (30, 31). The pathologist (KWC) was blinded to the clinical data.

Laboratory evaluation

Anti–double-stranded DNA (anti-dsDNA) antibody was determined by enzyme-linked immunosorbent assay. Anti–extractable nuclear antigen antibodies (Ro, La, nuclear RNP, and Sm) were studied by counterimmunoelectrophoresis. Serum C3 levels were measured by nephelometry. Urinary protein was measured by a dye method using pyrogallol red. Assays for serum and urinary creatinine were performed according to the Jaffe method. Creatinine clearance (in ml/minute) was calculated based on the serum creatinine level and a 24- hour urinary creatinine assay using the formula UV/P, where U = urinary creatinine concentration, V = urine flow rate, and P = plasma creatinine level.

Data collection

The following data were collected: age and duration of SLE at the time of the renal biopsy; sex; history of nephritis; history of treatment with prednisone, AZA, or CYC prior to the current episode of nephritis; presence of various extrarenal manifestations of SLE, including mucocutaneous and musculoskeletal involvement, serositis, central nervous system disease (psychosis, seizure, transverse myelitis, and aseptic meningitis), and hematologic complications (hemolytic anemia, thrombocytopenia with a platelet count <100 × 109/liter, leukopenia with a white blood cell count <4 × 109/liter); autoantibodies (anti-dsDNA, anti-Ro, anti-La, anti-Sm, anti–nuclear RNP); several renal and disease activity parameters at the time of renal biopsy, such as histologic activity and chronicity scores, 24-hour urinary protein, serum creatinine level, creatinine clearance, serum albumin level, presence of hypertension and nephrotic syndrome, and serum C3 level; time between the diagnosis of nephritis and the initiation of CYC, duration of CYC, and the mean daily dose and cumulative dose of CYC; mean daily dose and cumulative dose of prednisone during the first 4 months of treatment; and the use of hydroxychloroquine at the time of sequential immunosuppression. These data were considered to be predictor variables for complete remission at 12 months in univariate and multivariate analyses.

The same set of variables was used for univariate and multivariate analyses of the risk factors for renal relapse or doubling of the serum creatinine level in the entire cohort. The effect of the dosage of AZA on remission and relapse could not be evaluated, because this drug was continued indefinitely. The cumulative dose of AZA was affected by the duration of followup and whether the end points were reached.

Statistical analysis

Unless specified otherwise, values are expressed as the mean ± SD. Comparison of categorical data between 2 groups was performed with the chi-square test; Yates' continuity correction was performed when the frequency was small. Comparison of continuous data between 2 groups was performed with Student's t- test. When normal distribution or equal variance could not be assumed, the Mann- Whitney rank sum test was used instead.

The clinical predictors of complete renal remission at 12 months after renal biopsy were analyzed by logistic regression using the above-mentioned clinical variables as covariates and remission as the outcome. The Kaplan-Meier method was used to evaluate the cumulative risk of renal relapse or doubling of the serum creatinine level over time in the cohort of patients. Time zero was the time when the course of oral CYC was finished. The log-rank test and Cox proportional hazards model were used for univariate and multivariate analyses, respectively, of the risk factors for events; the same set of covariates was used. A stepwise backward elimination procedure was adopted for the multivariate models based on a likelihood ratio test, with P > 0.10 for removal of variables and P < 0.05 for entry of variables. To avoid entering too many variables into the multivariate models given the limited number of observations and events, only variables with P values that were less than 0.18 in univariate analysis were included for study.

Statistical significance was defined as a P value less than 0.05, 2 tailed. All statistical analyses were performed on a computer using the SPSS program, version 8.0 (SPSS, Chicago, IL) for Windows 98.

RESULTS

We studied 55 patients (47 women and 8 men) with SLE and DPGN who were treated with prednisone and with oral CYC followed by AZA. Their mean ± SD age at the time of renal biopsy was 31.1 ± 10.4 years, and their mean ± SD serum creatinine level was 107 ± 40 μmoles/liter (range 54–276). Twenty-five patients (46%) had a serum creatinine level >106 μmoles/liter (upper limit of normal range) and 29 (53%) had nephrotic syndrome. The mean ± SD histologic activity and chronicity scores were 10.5 ± 4.1 and 2.7 ± 2.6, respectively. At 12 months after treatment, 37 patients (67%) and 12 patients (22%) achieved complete remission and partial remission, respectively. Three patients were treatment-resistant, and 2 had early relapse of nephritis soon after cessation of oral CYC treatment. One patient died of subarachnoid hemorrhage 2 months after CYC was initiated.

Table 1 shows the clinical characteristics of the study patients, grouped according to renal status, at 12 months posttreatment. Patients with complete remission had significantly lower serum creatinine levels and chronicity scores and were less likely to have hypertension at the time of renal biopsy compared with patients without complete remission. Other unfavorable, although not statistically significant, factors for complete remission were a history of steroid and AZA treatment, a lower creatinine clearance rate, nephrotic syndrome at renal biopsy, male sex, increasing age, and a lower cumulative dose of CYC.

Table 1. Clinical characteristics of patients who did or did not achieve complete renal remission after sequential oral treatment with CYC followed by AZA, at 12 months after immunosuppressive treatment*
 Patients with complete remission (n = 37)Patients without complete remission (n = 18)P
  • *

    Except where indicated otherwise, values are the mean ± SD. CYC = cyclophosphamide; AZA = azathioprine; HCQ = hydroxychloroquine; BP = blood pressure.

Age at the time of renal biopsy, years29.8 ± 10.333.8 ± 10.20.18
Female, no. (%)34 (92)13 (72)0.13
Disease duration at the time of nephritis, months41.3 ± 5256.8 ± 450.26
History of nephritis, no. (%)9 (24)7 (39)0.26
History of corticosteroid treatment, no. (%)25 (68)17 (94)0.06
History of AZA treatment, no. (%)9 (24)9 (50)0.06
History of CYC treatment, no. (%)4 (11)1 (6)0.90
Current use of HCQ, no. (%)11 (30)7 (39)0.50
CYC treatment
 Interval between nephritis and CYC initiation, weeks9.6 ± 155.7 ± 4.00.30
 Mean daily dose, mg84.4 ± 2087.3 ± 170.62
 Mean duration of treatment, months8.2 ± 6.56.2 ± 2.70.23
 Mean cumulative dose, gm20.4 ± 1315.7 ± 6.90.17
Prednisone treatment in first 4 months
 Mean cumulative dose, gm3.76 ± 0.533.83 ± 0.800.71
 Mean daily dose, mg33.6 ± 5.233.0 ± 7.10.70
Prevalence of extrarenal features
 Musculoskeletal, no. (%)37 (100)17 (94)0.71
 Mucocutaneous, no. (%)31 (84)12 (67)0.15
 Central nervous system, no. (%)5 (14)2 (11)1.00
 Hematologic, no. (%)23 (62)7 (39)0.10
 Serositis, no. (%)7 (19)4 (22)1.00
Clinical parameters at initial renal biopsy
 Body weight, kg51.4 ± 8.155.1 ± 8.90.19
 Activity score, range 0–2410.3 ± 4.210.9 ± 3.90.67
 Chronicity score, range 0–122.0 ± 2.14.0 ± 2.90.02
 Serum creatinine, μmoles/liter (normal 60–106)94.7 ± 27132 ± 500.001
 Creatinine clearance, ml/minute69.6 ± 2954.1 ± 230.05
 Proteinuria in 24 hours, gm4.7 ± 4.36.8 ± 7.00.18
 Serum albumin, gm/dl (normal 44– 56)27.7 ± 7.226.2 ± 5.40.43
 Nephrotic syndrome, no. (%)17 (46)12 (67)0.15
 Serum C3, mg/dl (normal 60– 130)45.0 ± 1748.3 ± 210.54
 Hypertension (diastolic BP ≥100 mm Hg), no. (%)6 (16)9 (50)0.008
Autoantibodies
 Double-stranded DNA, no. (%)34 (92)13 (72)0.30
 Ro, no. (%)19 (51)7 (39)0.39
 La, no. (%)3 (8)1 (6)1.00
 Sm, no. (%)4 (11)1 (6)0.89
 Nuclear RNP, no. (%)4 (11)1 (6)0.89

Multivariate analysis revealed that the serum creatinine level at the time of renal biopsy was the only independent predictor of complete remission at 12 months (Table 2). The multivariate adjusted odds ratio for renal remission in patients with a serum creatinine level >106 μmoles/liter at the time of renal biopsy was 0.05 (95% confidence interval [95% CI] 0.01–0.35; P = 0.002) compared with those with a creatinine level ≤106 μmoles/liter.

Table 2. Univariate and multivariate predictors of complete remission after sequential oral treatment with CYC followed by AZA in patients with systemic lupus erythematosus and diffuse proliferative glomerulonephritis*
Clinical predictorsUnivariate analysisMultivariate analysis
RR (95% CI)PRR (95% CI)P
  • *

    CYC = cyclophosphamide; AZA = azathioprine; RR = relative risk; 95% CI = 95% confidence interval.

  • Data obtained at time of renal biopsy.

Serum creatinine, per μmole/liter0.97 (0.95–0.99)0.0050.96 (0.93–0.99)0.004
Hypertension0.19 (0.05–0.69)0.01
Chronicity score, per point0.72 (0.55–0.94)0.02
History of corticosteroid treatment0.12 (0.15–1.03)0.05
Creatinine clearance, per ml/minute1.02 (1.00–1.05)0.06
History of AZA treatment0.32 (0.98–1.06)0.06
Female4.40 (0.91–20.9)0.07
Hematologic involvement2.58 (0.81–8.21)0.11
Nephrotic syndrome0.43 (0.13–1.38)0.15
Mucocutaneous manifestation2.58 (0.70–9.61)0.16
Cumulative dose of CYC1.05 (0.98–1.12)0.16
Age at renal biopsy0.96 (0.91–1.02)0.18

Patients with complete or partial remission after steroid and oral CYC therapy were maintained on low-dose prednisone (5–10 mg/day) and AZA (50–100 mg/day). Excluding those who were treatment-resistant (n = 3) or died (n = 1), 21 patients (41%) had renal flares (17 nephrotic and 4 nephritic) during a median followup period of 48 months (interquartile range 129) after cessation of oral CYC. Sixteen of these 21 patients (76%) had a relapse of DPGN, and 5 (24%) had a relapse of focal proliferative glomerulonephritis documented by a repeat renal biopsy. Figure 1 shows the cumulative probability of remaining free of renal relapse over time in our patients who achieved remission (partial or complete) after sequential CYC and AZA treatment. The cumulative hazard of renal relapse was 6% at 12 months, 21% at 36 months, and 32% at 60 months. The median nephritis-free interval for those who experienced relapse was 43 months. A subanalysis of the patients who achieved complete remission after treatment (n = 37) was performed and revealed cumulative renal relapse rates of 3%, 19%, and 33% at 12, 36, and 60 months, respectively.

Figure 1.

Cumulative probability of remaining free of a relapse of lupus nephritis in the systemic lupus erythematosus study patients who achieved remission (partial or complete) after sequential oral cyclophosphamide followed by azathioprine.

The clinical features of the patients who experienced relapse or remained in remission at the time of analysis are shown in Table 3. Patients who experienced renal relapse had received a significantly lower daily dose of oral CYC. Moreover, they tended to be younger, they had a higher histologic activity score, and they were more likely to have hypertension, serositis, and hematologic complications. The duration of oral CYC treatment and the interval between renal biopsy and the institution of CYC also tended to be longer in those who experienced relapse, although statistical significance was not achieved. Cox regression analysis demonstrated that the histologic activity score (hazard ratio [HR] 1.13 per 1 point [95% CI 1.01–1.27]; P = 0.04] and the mean daily dose of CYC (HR 0.95 per mg [95% CI 0.92–0.98]; P = 0.001) were multivariate predictors of renal flare.

Table 3. Clinical characteristics of patients who did or did not experience a relapse after sequential oral treatment with CYC followed by AZA*
Patients with remission (n = 30)Patients with relapse (n = 21)P
  • *

    Except where indicated otherwise, values are the mean ± SD. CYC = cyclophosphamide; AZA = azathioprine; HCQ = hydroxychloroquine; BP = blood pressure.

Age at the time of renal biopsy, years31.7 ± 9.827.9 ± 8.90.16
Female, no. (%)24 (80)19 (90)0.53
Disease duration at the time of nephritis, months44.5 ± 4841.0 ± 470.79
History of nephritis, no. (%)7 (23)7 (33)0.43
History of corticosteroid treatment, no. (%)22 (73)16 (76)0.82
History of AZA treatment, no. (%)9 (30)6 (29)0.91
History of CYC treatment, no. (%)2 (7)3 (14)0.67
Current use of HCQ, no. (%)10 (33)7 (33)1.00
CYC treatment
 Interval between nephritis and CYC initiation, weeks6.2 ± 10.111.8 ± 170.15
 Mean daily dose, mg90.6 ± 1876.4 ± 190.01
 Mean duration of treatment, months6.6 ± 2.59.2 ± 8.20.11
 Mean cumulative dose, gm18.2 ± 7.120.8 ± 170.44
Prednisone treatment in first 4 months
 Mean cumulative dose, gm3.85 ± 0.73.30 ± 5.90.48
 Mean daily dose, mg34.0 ± 5.937.2 ± 0.60.57
Prevalence of extrarenal features
 Musculoskeletal, no. (%)29 (97)21 (100)1.00
 Mucocutaneous, no. (%)23 (77)18 (86)0.66
 Central nervous system, no. (%)4 (13)3 (14)1.00
 Hematologic, no. (%)14 (47)14 (67)0.16
 Serositis, no. (%)3 (10)6 (29)0.18
Clinical parameters at initial renal biopsy
 Body weight, kg52.4 ± 7.952.2 ± 9.70.93
 Activity score, range 0–249.9 ± 3.912.1 ± 4.00.08
 Chronicity score, range 0–122.3 ± 2.32.6 ± 2.50.70
 Serum creatinine, μmoles/liter (normal 60–106)110 ± 45101 ± 330.45
 Creatinine clearance, ml/minute62.2 ± 2970.2 ± 270.31
 Proteinuria in 24 hours, gm5.4 ± 6.05.6 ± 5.00.93
 Serum albumin, gm/dl (normal 44– 56)27.1 ± 7.527.4 ± 5.40.85
 Nephrotic syndrome, no. (%)16 (53)11 (52)0.95
 Serum C3, mg/dl (normal 60– 130)46.5 ± 1944.5 ± 190.72
 Hypertension (diastolic BP ≥100 mm Hg), no. (%)6 (20)8 (38)0.15
Autoantibodies
 Double-stranded DNA, no. (%)28 (93)18 (86)0.67
 Ro, no. (%)15 (50)10 (48)0.87
 La, no. (%)2 (7)1 (5)1.00
 Sm, no. (%)1 (3)3 (14)0.37
 Nuclear RNP, no. (%)3 (10)3 (14)0.98

Considering all the patients in our cohort, the cumulative probability of remaining in renal remission (complete or partial) was 62.8% at 5 years and 47.9% at 10 years. In patients with a relapse of DPGN, the prednisone dosage was increased (1 mg/kg/day) for 6–8 weeks, and another 6-month course of oral CYC (1–2 mg/kg/day) was initiated, followed by resumption of AZA (1–2 mg/kg/day). In those with renal relapses other than DPGN, the prednisone dosage was increased for 6–8 weeks, and the AZA was continued. At the latest followup, 9 of 54 patients (17%) had a doubling of the serum creatinine level, 6 of whom (11%) underwent dialysis. The cumulative risk of creatinine doubling was 6% at 3 years, 8.4% at 5 years, and 18.2% at 10 years (Figure 2).

Figure 2.

Cumulative probability of not having a doubling of the serum creatinine level in all study patients with diffuse proliferative glomerulonephritis and systemic lupus erythematosus.

Univariate analysis revealed that an increasing creatinine level, history of AZA treatment, longer disease duration, higher chronicity indices, failure of complete remission at 12 months after treatment, hypertension at the time of renal biopsy, and a history of serositis were associated with creatinine doubling. However, after multivariate adjustment, only the chronicity index at initial renal biopsy was found to be the independent predictor of deterioration of renal function (adjusted HR 2.1 per 1 point [95% CI 1.2–3.6]; P = 0.009).

Table 4 shows the drug-related adverse events experienced by our patients at the time of analysis. The most common side effect was menstrual disturbances, followed by herpes zoster infection, major infection, and hemorrhagic cystitis.

Table 4. Occurrence of adverse events in 55 patients during the study period
ToxicityFrequency (%)
  • *

    One patient died of subarachnoid hemorrhage.

Amenorrhea/oligomenorrhea (n = 47)20 (43)
Herpes zoster10 (18)
Major infection6 (11)
Hemorrhagic cystitis2 (4)
Death*1 (2)
Malignancies0 (0)

DISCUSSION

The optimum treatment for DPGN in SLE patients is unclear, because large prospective randomized trials are lacking. Previous studies have shown that regimens containing CYC and steroid are associated with a better renal outcome compared with those containing steroid alone. Although CYC is being widely used for the treatment of this condition, the optimum route of administration, dosage, and duration of treatment are unclear. Moreover, little information regarding the rate of renal flares after discontinuation of CYC is available in the literature. This issue is important because recurrence of nephritis is associated with new immunologic and inflammatory insults to the kidney, with the subsequent occurrence of new renal lesions. Incomplete resolution of these renal lesions may lead to glomerular sclerosis and interstitial fibrosis, and consequently, a relentless progression to renal failure (32, 33).

Studies at the NIH showed that CYC-containing regimens were superior to those with steroid alone. Although no significant differences among the different routes of CYC administration could be demonstrated, IV pulse CYC showed a tendency toward better efficacy than the oral route and was associated with less toxicity (9, 10, 12). The lower incidence of CYC-related toxicities was probably because of the lower cumulative dose of CYC administered. It is noteworthy that in the oral treatment arm, the median duration of treatment was 48 months, with a daily dose of 1–4 mg/kg. Although a lower daily dose of oral CYC (1 mg/kg) was used in the combination treatment arm (CYC + AZA), the median duration of treatment was 4.3 years. An average patient would definitely have received a cumulative dose of CYC that was >70 gm.

For the sequential regimen evaluated in the present study, a lower daily dose and shorter duration of oral CYC (maximum 9 months) was used, after which AZA was substituted. The mean ± SD cumulative dose of CYC (18.8 ± 11.8 gm) in the current study was much lower than that used in the oral treatment arms of the NIH studies. This may account for the low incidence of serious drug-related toxicities, such as major infection and hemorrhagic cystitis, identified in our patients, although herpes zoster infection and menstrual disturbances were still common (26, 27, 34, 35).

Table 5 summarizes the rates of remission, relapse, and doubling of the serum creatinine level in lupus nephritis patients treated with CYC that have been reported in major international series (10, 13, 23, 25, 36–41). Direct comparisons among these studies are difficult because of the discrepancies in the proportion of patients with DPGN (World Health Organization class IV) who were included, the severity of renal disease at biopsy, the criteria for remission/relapse, the ethnicity of the patients, and the great variation in treatment protocols in terms of steroid dosage, route and duration of CYC, and the use of plasmapheresis.

Table 5. Rates of renal remission, flares, and serum creatinine doubling after immunosuppressive treatment of severe lupus nephritis*
Author, year (ref.) [no. of patients]Renal statusTreatmentRemission rateRelapse rateDoubling of serum Cr or occurrence of ESFR
  • *

    ESRF = end-stage renal failure; WHO = World Health Organization (class IV represents diffuse proliferative glomerulonephritis); Cr = creatinine; Pred. = prednisolone/prednisone; CYC = cyclophosphamide; AZA = azathioprine; IV = intravenous; NS = not stated; MP = methylprednisolone.

Donadio et al, 1978 (36) [n = 50]WHO class IV; mean Cr clearance 53 ml/ minutePred. (n = 26) or Pred. + oral CYC for 6 months (n = 24)21 patients in each treatment arm were improved at 6 months78% for Pred. alone, 30% for Pred. + oral CYC at 50 monthsESRF or death at 4 years: 27% for Pred. alone, 29% for Pred. + oral CYC
Steinberg et al, 1991 (10) [n = 111]59% WHO class IV; mean serum Cr 90 μ moles/literPred. alone, Pred. + oral CYC, oral CYC + AZA, IV pulse CYC, or AZANSNSCumulative probability of ESRF at 5 years < 25% for all treatment arms; at 10 years, IV pulse CYC had lowest rate (10%)
Boumpas et al, 1992 (13) [n = 65]88% WHO class IV; 64% with serum Cr > 114 μmoles/literPred. + IV pulse MP, Pred. + short- term (6 pulses) IV pulse CYC, or Pred. + long-term (12 pulses) IV pulse CYCNS10% for long-term IV pulse CYC, 55% for short-term IV pulse CYC at 60 months, for those with stable disease after treatmentDoubling of serum Cr at 5 years: 50% for Pred. alone, 30% for short-term IV pulse CYC, 20% for long-term IV pulse CYC
Ciruelo et al, 1996 (38) [n = 48]89% WHO class IV; mean serum Cr 88 μ moles/literPred. + either IV or oral CYC (median treatment duration 31 months)NS25% at 5 years and 46% at 10 yearsNS
Gourley et al, 1996 (37) [n = 82]75% WHO class IV; mean serum Cr 102 μ moles/literMonthly IV pulse MP for ≥1 year, IV pulse CYC (12 pulses), or IV pulse MP + pulse CYCCumulative remission rate at 60 months: 30% for IV pulse MP alone, 65% for IV pulse CYC, and 75% for IV pulse MP + IV pulse CYCProportion with relapse after 5 years among those who achieved remission: 36% for IV pulse MP alone, 7% for IV pulse CYC, 0% for IV pulse MP + IV pulse CYCCumulative risk of serum Cr doubling at 60 months: 25% for IV pulse MP alone, 5% for IV pulse CYC, 0% for IV pulse MP + IV pulse CYC
Moroni et al, 1996 (39) [n = 70]60% WHO class IV; mean serum Cr 159 μ moles/literPred./IV pulse MP + either oral AZA, oral CYC, or chlorambucilNS66% relapsed after a median observation of 127 monthsCumulative risk of serum Cr doubling: 15% at 10 years and 28% at 20 years
Dooley et al, 1997 (23) [n = 89]WHO class IV; mean serum Cr 124 μmoles/ literPred. + IV pulse CYC (monthly for 6 pulses ± 3 monthly pulses)NSNSRenal survival rate at 5 years: 71% overall; 57% for blacks, 95% for non-blacks
Ioannidis et al, 2000 (25) [n = 85]61% WHO class IV; mean serum Cr 115 μ moles/literPred. + IV pulse CYC (18 doses)Cumulative probability of remission at 2 years: 78%50% at 80 months for those who initially achieved remission8 (9%) developed ESRF after a median followup of 31 months
Korbet et al, 2000 (40) [n = 86]42% WHO class IV; mean serum Cr 180 μ moles/literPred. + oral CYC (2–3 months), with or without plasmapheresis43% remitted after 5 yearsNSRenal survival at 5 and 10 years: 94% and 94% in remission group; 46% and 31% in nonremission group, respectively
Mosca et al, 2001 (41) [n = 33]WHO class IV; 33% with abnormal serum CrIV pulse MP + IV pulse CYC (6–9 pulses)39% with complete remission, 61% with partial remission after treatment45% had renal flare after a median observation of 48 months (24% had early flare within 12 months of cessation of CYC)27% had serum Cr doubling at 51 months
Present study [n = 55]WHO class IV; mean serum Cr 107 μmoles/ literPred. + oral CYC (6–9 months) followed by AZA indefinitely67% with complete remission, 22% with partial remission after treatment; cumulative probability of remission at 5 and 10 years: 62.8% and 47.9%, respectively32% at 60 months after cessation of CYC among those who initially achieved remissionCumulative risk of serum Cr doubling at 5 and 10 years: 8.4% and 18.2%, respectively

In the present study, 89% of patients with SLE and DPGN experienced complete or partial remission at 12 months after oral CYC and AZA treatment, and 62.8% remained in renal remission 5 years after cessation of CYC. The remission rate appears to be higher than that described by Korbet et al (40), who reported a 43% remission rate at 5 years in their patients treated with steroid and oral CYC, with and without plasmapheresis. The mean serum creatinine level in their patients was much higher than that in our patients, and the duration of CYC treatment was much shorter (2 months if no flares). Moreover, the stricter remission criteria they used (proteinuria <0.33 gm/day) might contribute to a lower number of remissions. Nevertheless, our results are consistent with their finding in that an increasing creatinine level is a multivariate predictor of clinical nonremission.

The NIH group (37) reported a 5-year cumulative renal remission rate of 75% after pulse methylprednisolone (monthly) and pulse CYC (12 pulses in 24 months) for severe lupus nephritis. Seventy-five percent of their patients had DPGN, and the mean serum creatinine level at study entry was very close to that of the patients in our study. The inclusion of patients with milder forms of lupus nephritis and the allowance of an increasing creatinine level (as long as the value did not double the value at study entry) as a remission criterion might contribute to a slightly higher remission rate compared with our results.

The rate of relapse of nephritis after CYC treatment of proliferative lupus nephritis ranges from 10% to 66% in various studies (13, 25, 36, 38, 39, 41). The wide range of relapse rates is, again, probably related to the differences in the characteristics of the study patients, the treatment regimens, the definition of relapse, and the duration of observation.

Donadio et al (36) reported that 13 (31%) of 42 of their patients with SLE and DPGN who improved initially after immunosuppressive treatment had renal flares during a mean followup of 43 months. The probability of flare at 50 months was 78% for patients treated with prednisone alone and 30% for those treated with prednisone and oral CYC (mean daily dose 107 mg for 6 months). Ciruelo et al (38) reported a cumulative rate of renal relapse of 46% at 10 years in 48 patients treated with corticosteroids and CYC (IV or oral) for a median of 31 months and followed up for 33 months after cessation of CYC. In the study by Boumpas et al (13), it was found that 11 (27%) of 40 patients who were treated with oral prednisone and IV pulse CYC had a renal flare within 5 years of therapy. The probability of flare was 55% in those treated with 6 pulses of CYC and only 10% in those treated with 12 pulses of CYC.

Moroni et al (39) observed that 46 (66%) of 70 patients treated with corticosteroids and cytotoxic agents had relapse of nephritis over a median followup period of 127 months. A recent report by Ioannidis et al (25) described a cumulative relapse rate of 50% in 63 patients with either focal or diffuse proliferative lupus glomerulonephritis and SLE who remitted after IV pulse CYC treatment, over a followup period of 80 months after remission. In our study, the cumulative risk of nephritis relapse after an initial remission was 32% at 5 years after cessation of CYC. This is similar to the values reported by Donadio et al (36) and Ciruelo et al (38), but appears to be lower than those reported by Ioannidis et al (25) and Mosca et al (41), who adopted a pulse CYC regimen for the treatment of severe lupus nephritis.

Few studies have described the clinical predictors of renal flare after initial improvement of severe lupus nephritis with immunosuppressive therapy. A high activity index at initial renal biopsy (41, 42), younger age (38, 41), delay of initiation of CYC treatment (38), treatment with steroid alone (36), short duration of CYC treatment (13), male sex and hypertension at onset of nephritis (39), and history of central nervous system involvement and longer time to remission (25) have been found to be risk factors for renal relapse. The rate of renal relapse also depends on the duration of observation. Moroni et al (43) observed a significant reduction of renal flares after the tenth year from the onset of nephritis, as compared with the period between 0 and 10 years. This observation is consistent with the findings by Ioannidis et al (25) and with our findings, in that the cumulative risk of relapse of nephritis tended to decrease over time, and the proportion of patients who remained in remission became relatively static after 7 years. Whether this reflects a spontaneous quenching of the disease over time or the efficacy of intense and prolonged immunosuppressive treatment remains to be determined.

In the present study, we were unable to demonstrate that male sex and age at onset of SLE were significant risk factors for renal flare after initial remission with sequential oral CYC and AZA treatment. The former may be related to the small number of male patients included in our study. Moreover, since the interval between the onset of lupus nephritis and the administration of CYC was <11 weeks in the majority of our patients, delay in the initiation of CYC was not shown to be a risk factor for renal relapse. Instead, we observed that treatment with a lower daily dose of oral CYC and a higher activity score at the onset of nephritis were multivariate predictors of renal flare after achieving an initial complete or partial remission. The latter finding is consistent with the findings by Mosca et al (41) and Leaker et al (42).

Apart from remission and relapse, another very important outcome of lupus nephritis is deterioration in renal function. In the first NIH study (9, 10), no differences in the risk of renal failure among patients in the different treatment arms could be demonstrated in the first 5 years. Differences were apparent only after 10 years, and IV pulse CYC was associated with the least risk (10%) at that time point. In the subsequent NIH studies, the 5-year cumulative risk of creatinine doubling was reported to be 5–20% after long-term pulse CYC (12 pulses in 24 months) treatment for severe lupus nephritis (13, 37). This is similar to our finding that the cumulative rates of creatinine doubling after sequential oral CYC and AZA were 8.4% and 18.2% at 5 years and 10 years, respectively, after cessation of CYC. Since patients with milder form of lupus nephritis were included in the NIH studies (12–41%), an underestimation of the risk of renal failure might be expected.

Ethnicity is an important determining factor for the prognosis of lupus nephritis. Proliferative lupus nephritis in African Americans is associated with a poor prognosis, with a 5- and 10-year renal survival rate of 52–58% and 0%, respectively (8, 23). The overall 5- and 10-year renal survival of southern Chinese patients with lupus nephritis was reported to be 92% and 81%, respectively, a rate which is comparable to that in most western series (5). A more recent prospective study revealed a 93% 5-year survival rate for our local southern Chinese SLE patients (44). This value is similar to the values reported in most major series of lupus patients during the 1990s (45). Thus, the prognosis of SLE in Chinese is not significantly different from that in Caucasians, and ethnicity per se is unlikely to be a major contributing factor to the renal outcome reported in the current study.

AZA, in conjunction with prednisone, has been shown in meta-analyses to be more effective than prednisone alone in reducing mortality and end-stage renal disease in patients with severe lupus nephritis (11, 46). Although there was a tendency toward lower efficacy of AZA when compared with CYC-containing regimens in the NIH studies (9, 10), the difference was not statistically significant, and there are no prospective trials directly comparing these 2 cytotoxic agents. A recent retrospective study described a 5-year renal survival rate of 92% in 26 patients with SLE and DPGN treated with prednisolone and AZA (up to 2.5 mg/kg/day for a median of 53 months) (47). Thus, AZA is likely to have therapeutic value of its own in lupus nephritis, and the efficacy of our sequential regimen in inducing and maintaining renal remission, as well as in preserving renal function, is likely the result of both CYC and AZA.

Finally, the initial dosage, route, and duration of steroid used in different treatment protocols may also influence renal outcome. The initial dosage (1 mg/kg/ day) and duration (8–10 weeks) of prednisone used in our study are slightly higher than those used in other studies in which oral steroid treatment was given (0.5–1 mg/kg of prednisone per day for 4–8 weeks) (9, 10, 40). This more-intensive steroid regimen may, to some extent, contribute to a better remission rate.

In summary, our study has shown that a sequential regimen of oral CYC followed by AZA appears to be effective in the treatment of DPGN in patients with SLE, with 89% of patients achieving renal remission after initial treatment. Two-thirds of the patients remained in renal remission after 5 years, and four-fifths had stable renal function after 10 years. Although the reported rates of renal remission and end-stage renal failure were comparable to those described in the NIH studies, the lack of a control arm in our study makes direct comparison impossible and interpretation difficult. Moreover, the initial use of a slightly higher dosage and a longer duration of oral steroid may also influence the renal outcome to some extent. Nevertheless, with the use of a lower daily dose and shorter duration of CYC, the occurrence of serious toxicities such as major infection was not excessively high.

Ancillary