What do we know about lupus membranous nephropathy? An analytic review



The management of lupus nephritis has undergone a major revolution since the introduction of percutaneous renal biopsy (1), which resulted in the recognition that certain histologic subtypes are associated with very different natural histories and responses to therapy (2). Renal lesions due to systemic lupus erythematosus (SLE) initially were classified into 3 subtypes (focal, diffuse, and membranous glomerulonephritis) (3). This classification gave way to the World Health Organization's (WHO) more elaborate scheme using 6 different types and multiple subtypes (4). After more than a decade of use, the WHO system was revised, and the more recent WHO classification (4) reflects the prevailing view that the presence of cellular proliferation on renal biopsy relegates the renal specimen with coincident membranous pathology to either the focal proliferative or diffuse proliferative categories. Unfortunately, much of the currently available literature uses the 1982 WHO classification system.

The use of the 1982 WHO classification system has been supplemented by activity and chronicity indices that quantify reversible and irreversible lesions (5). Generally in lupus nephritis, the WHO classification parallels these indices with regard to outcome (6). However, the activity and chronicity indices may not independently predict outcome, in part because there is inconsistency in scoring from one center to another (7). In general, very inflammatory biopsy specimens have a high activity index and a poor prognosis if left untreated. However, these lesions are thought to be relatively responsive to immunosuppressive therapy with corticosteroids and cytotoxic agents, particularly cyclophosphamide (8). Biopsy specimens with little inflammation and a lower activity index can be associated with good outcomes with more modest intervention, as long as the kidneys are not scarred. Finally, patients with scarred kidneys with a high chronicity index do poorly no matter what intervention is chosen (6, 9).

Diffuse subepithelial deposits of immune complexes histologically characterize the membranous lesion of lupus nephritis. It is similar to idiopathic membranous nephritis (10), but has several distinct pathologic characteristics, including mesangial and subendothelial electron dense deposits and intense C1q deposition (11). The pure membranous lesion in both the idiopathic and lupus forms has little cellular proliferation and usually results in the nephrotic syndrome.

In an effort to elucidate the prognosis and optimal management of lupus membranous nephropathy (LMN), we analyzed the literature on natural history and treatment.

Materials and methods

We identified references that provided information about outcome, either related to natural history or to treatment, of biopsy-proven LMN. To accomplish this, we performed a MEDLINE search using the keywords “membranous nephropathy” and “nephrotic syndrome” with date limits of 1965–1998. Once reviewed, articles were chosen for inclusion in this analysis if they provided information that was specific to patients with biopsy-proven LMN for at least 2 points in time.

Aggregation of data across studies is necessary to apply meta-analytic techniques (12). We have utilized quantitative analysis to the extent permissible by the study designs and the data included. Furthermore, several large studies (10, 13–16) that provide important information about LMN were not included in the analysis because the data were not in a form that could be aggregated with those outlined in the results section.


Proteinuria. Six studies provided information concerning the degree of proteinuria at presentation and followup in 117 biopsy-proven LMN patients (Table 1). Reported followup periods ranged from less than 1 year to 23 years. The majority (64.1%) of patients with LMN presented with nephrotic-range proteinuria. Only 23.1% had complete resolution of proteinuria at followup. Another 12% remained nephrotic. These data suggest that the remaining 65% had some level of persistent proteinuria.

Table 1. Outcome of patients with lupus membranous nephropathy: proteinuria and renal function
ReferenceFollowupProteinuriaRenal function
Banfi, 1998 (25)51–177 months19/19 nephrotic1/19 nephrotic19/19 normal3/19 doubling of creatinine
   9/19 partial remission  
   9/19 complete remission 1/19 end-stage renal disease
Chan, 1998 (19)25–123 months0/20 nephrotic1/20 nephrotic18/20 normal20/20 stable creatinine clearance
   9/20 partial improvement2/20 abnormal creatinine 
   10/20 complete resolution  
Radhakrishnan, 1994 (24)23–43 months9/10 nephrotic3/10 nephrotic9/10 normal8/10 stable creatinine
   7/10 improved1/10 creatinine >1.5 mg/dl1/10 normalized creatinine
     1/10 end-stage renal disease
Pasquali, 1993 (17)6–128 months29/42 nephrotic9/42 nephrotic35/42 normal4/42 creatinine >1.5 mg/dl
   25/42 proteinuria7/42 creatinine >1.5 mg/dl 
   8/42 complete remission 6/42 end-stage renal disease
Moroni, 1992 (26)132–276 months4/6 nephrotic0/6 nephrotic6/6 normal1/6 elevated creatinine
Adler, 1990 (18)59–89 months18/18 normal6/18 creatinine <1.5 mg/dl
     3/18 creatinine <1.5 mg/dl
Leaker, 1987 (23)64 months14/20 nephrotic4/17 proteinuria >1 g/24 hour20/20 normal0/20 end-stage renal disease
   13/17 proteinuria <1 g/24 hour  
Schwartz, 1984 (13)60 months17/22 normal3/22 creatinine improved by 20% (2 with Vd)
    5/22 abnormal creatinine 
     7/22 creatinine worsened by 20% (1 Vb, 5 Vd)
Total 75/117 (64.1%)14/117 (12.0%)  
   nephrotic nephrotic15/157 (9.6%) abnormal creatinine23/157 (14.6%) abnormal creatinine
   76/117 (65.0%) persistent proteinuria 8/157 (5.1%) end stage renal disease at followup
   27/117 (23.1%) complete resolution of proteinuria  

Renal function

Eight studies provided information about renal function in 157 patients with LMN (Table 1). Followup in these studies ranged from 6 months to 23 years. The vast majority (90.4%) of patients with LMN presented with normal renal function as measured by serum creatinine. Most (85.4%) continued to have normal renal function. However, 14.6% had abnormal renal function at followup, and 5.1% developed end-stage renal disease.

Renal function by WHO classification

Three studies presented data according to the 1982 WHO classification for LMN. This classification divides membranous nephropathy into 4 subclasses based on histologic features. Subclass Va includes those with only changes of membranous nephropathy; subclass Vb includes mesangial widening or mesangial hypercellularity; subclass Vc includes focal segmental changes with mild to moderate mesangial change; subclass Vd includes diffuse proliferative changes (4). The more recent WHO classification has been updated to eliminate the Vc and Vd designations. These have been subsumed under the mesangial and diffuse proliferative categories (4). The literature available for analysis, however, uses Vc and Vd.

Data on the outcome of patients with regard to proteinuria could not be abstracted from 2 of the 3 studies with information by WHO subclassification. However, Pasquali et al (17) were able to show that patients with more proliferative change on renal biopsy presented with more proteinuria, were more likely to have persistent nonnephrotic proteinuria, and were less likely to have complete resolution of their proteinuria.

Data on renal function as measured by serum creatinine were available in all 3 studies that divided patients by WHO subclassification (17–19). Patients in these studies did worse (Table 2) than patients in our analysis as a whole (Table 1). This could be due to the smaller number of patients available for analysis and to a larger proportion with proliferative changes. WHO classification does not appear to be an independent predictor of renal failure when adjusted for impaired renal function, proteinuria, anemia, or age (10).

Table 2. Renal function of patients with LMN by WHO subclassification*
ReferenceWHO subclasses Va and VbWHO subclasses Vc and Vd
Stable creatinineIncreased creatinineEnd-stage renal diseaseStable creatinineIncreased creatinineEnd-stage renal disease
  • *

    LMN = lupus membranous nephritis; WHO = World Health Organization.

Sloan, 1996 (16)15/3613/365/3610/4314/4314/43
Pasquali, 1993 (17)21/263/262/2611/161/164/16
Adler, 1990 (18)4/71/71/72/112/117/11
Total40/69 (58.0%)17/69 (24.6%)8/69 (11.6%)23/70 (32.9%)17/70 (24.3%)25/70 (35.7%)

Most of the studies reviewed provided information about mortality by WHO class. The number of deaths among patients with LMN and the reported causes of mortality are listed in Table 3. Deaths directly attributable to end-stage renal disease are in the minority. However, the causes of many of the deaths were not reported.

Table 3. Mortality of patients with lupus membranous nephritis
ReferenceNumber of deathsFollowupCauses of death
Banfi, 1998 (25)0/1951–177 months
Chan, 1998 (19)0/2025–123 months
Sloan, 1996 (16)8/7932–133 monthsNot reported
Radhakrishnan, 1994 (24)0/1022–43 months
Pasquali, 1993 (17)2/266–128 monthsCerebrovascular accidents
Moroni, 1992 (26)0/6132–276 months
Adler, 1990 (18)4/1859–89 monthsNot reported (3 with end-stage renal disease)
Leaker, 1987 (23)2/2064 monthsPulmonary embolus Ischemic heart disease
Schwartz, 1984 (13)2/2260 monthsAspiration during seizure Fungal pneumonia
Total deaths 18/220 (8.2%)   

From 25%–50% of lupus patients with renal involvement will switch histologic classification on second or subsequent renal biopsies (2, 9). In one study (20), 5 of 42 patients had class V histology on initial biopsy. This number increased to 14 of 42 on repeat biopsy. Nine patients with class IV and one with class II lesions on initial biopsy subsequently converted to class V. One of the original class V patients converted to class III. Thus, the conversions to a different histologic class were predominantly to class V. It is unclear whether patients who convert to class V are similar in natural history to those who present initially with class V. It may be that this does not represent conversion to a more benign lesion, but the residua that aggressive therapy is unable to reverse (21).

Clinical characteristics

A number of the reports included in our analysis assessed whether or not clinical variables correlated with outcome measures. Others have previously shown that renal morphology does not always correlate with clinical features of patients with renal disease (22). Several of the studies we analyzed drew similar conclusions. In general, for patients with LMN, no significant correlation with renal outcome could be demonstrated for age (16, 23), sex (23), the presence of hypertension (16, 18, 23), complement levels (16, 18, 19), anti–double-stranded DNA antibody levels (18, 19), or serum albumin levels (16, 19). However, 2 studies did suggest that serum creatinine on presentation was a significant clinical marker for worse outcome (16, 23), as previously reported for lupus nephritis patients as a whole (6, 10).


Moroni and colleagues (14) presented considerable information on cardiovascular and cerebrovascular complications in their series of patients who had lupus nephritis for at least 10 years. They point out that although most patients with lupus nephritis may expect kidney survival, they are at increased risk of subsequent vascular complications such as angina, myocardial infarction, cerebral thrombosis, and cerebral hemorrhage. Unfortunately, they did not present control data on vascular complications in lupus patients without nephritis for comparison. In their publication detailing outcomes by WHO classification in a group of patients with shorter followup (17), these same authors note that, in addition to 2 deaths due to cerebrovascular accidents, a number of nonfatal vascular events were seen, 75% of which were associated with antiphospholipid antibodies. These included 1 inferior vena cava thrombosis, 3 renal vein thromboses, 2 superficial iliac artery thromboses, and 1 axillary vein thrombosis.


No prospective, randomized, double-blind placebo-controlled trials exist of treatment options for LMN. Reports that address treatment have been retrospective reviews, usually from a single institution. Treatment protocols have varied widely. All of the studies included in this analysis report using corticosteroids either intravenously or orally, with and without pulse dosing (14, 16, 17, 19, 23–25). Additional reported treatments, usually used in combination with corticosteroids, have included azathioprine (14, 16–19, 23, 24), chlorambucil (14, 17, 25, 26), cyclophosphamide (14, 16–19, 23, 24), cyclosporine (24), and plasmapheresis (15, 16, 18). Dosing and duration of immunosuppressive therapy have generally not been reported. Outcome measures have generally not been reported by choice of treatment. Because of the variability in treatment regimens and the lack of prospective data, no conclusions can be made regarding the optimal treatment for LMN from the studies analyzed here.

One study included in our analysis (24) presents the only prospective data on treatment of LMN. This pilot trial was not randomized, double-blinded, or placebo-controlled and included only 10 patients, 9 of whom had nephrotic-range proteinuria. Cyclosporine (4–6 mg/kg/day) was given alone or with steroids for up to 43 months. Although no patient had a complete resolution of proteinuria, all patients had an initial decline in proteinuria, most out of the nephrotic range. Three patients subsequently experienced a flare and had increasing proteinuria while still taking cyclosporine. Interestingly, given the known nephrotoxicity of cyclosporine, no patient had a significant increase in creatinine during the study. One patient went on to develop renal failure after cyclosporine was discontinued. The role of cyclosporine remains unclear (27, 28).


Lupus membranous nephropathy remains a therapeutic enigma, in part because membranous lesions represent the minority of biopsy specimens. For example, of the 659 patients reported in the 1996 GISNEL study, only 14% had lupus membranous nephritis (29). Earlier studies quoted similar figures (10, 15, 23, 30). Thus, conclusions must be drawn from relatively few patients. The lack of uniformity in the nomenclature makes analysis, particularly of the literature prior to the 1980s, problematic because information specific to LMN cannot be obtained from early studies that did not separate data by WHO subtype. Furthermore, there are no randomized, controlled trials that specifically target membranous lupus nephritis or stratify all lupus nephritis by histologic subtype, either in the entry criteria or in the analysis, to aid clinical decision making about therapy.

Arguments in favor of treating patients with LMN rest on the belief that treatment will affect outcome and that the complications of nephrotic-range proteinuria will be reduced. Complications include hypercholesterolemia and hypertension, both of which can result in accelerated atherosclerosis, infectious complications, and hypercoagulability. In fact, studies have documented hypertension as a variable that predicts poor prognosis in patients with lupus nephritis (20, 31) and vascular complications as a frequent cause of poor outcome (10, 17).

In the absence of evidence-based guidelines regarding natural history and treatment, one might look to idiopathic membranous nephropathy as a model for decision making about LMN. If LMN is similar to the idiopathic form, then the outcome generally is favorable (32). However, the inference that lupus nephritis patients should do well based on experience with idiopathic membranous nephritis is hazardous for a number of reasons. These include the fact that the natural history of untreated membranous disease in SLE is unknown and that SLE patients are probably at greater risk for comorbidity. Furthermore, the offending agent, such as a drug (e.g., penicillamine), malignancy, or viral disease (e.g., hepatitis B), in idiopathic membranous nephritis can occasionally be eliminated (33), but not in SLE.

Although the literature on the idiopathic form is large, some disagreement exists with regard to treatment. Many authors have questioned whether idiopathic membranous glomerulonephritis is clearly improved by aggressive immunotherapy (33–35) and studies can lack methodologic rigor (36). Other trials suggest that aggressive therapy is useful (37, 38). Studies using azathioprine (39, 40), cyclophosphamide (41, 42), and chlorambucil (41, 42) reported favorable results, although 1 study suggested azathioprine did not affect prognosis (40). Several reviews suggested immunosuppressives are useful, especially if high-grade proteinuria is persistent (43, 44), and two meta-analyses showed a cumulative beneficial effect for cyclophosphamide and chlorambucil (45, 46). Other trials indicate that alkylating therapy may be effective when administered on a daily basis, but not on an intermittent, less toxic, schedule (47, 48). A cautionary note was introduced by the recent observation that bone marrow failure occurred in Chinese patients treated with chlorambucil for LMN (49). Mycophenolate mofetil appears to be promising for both idiopathic (50) and lupus membranous glomerulonephritis (51), but long-term followup is unavailable.

Our analysis shows that most patients who present with LMN improve in their degree of proteinuria and maintain normal renal function. However, given the limitations outlined above, it is difficult to say how much of this is attributable to the natural history of the disease and how much to treatment. At initial presentation, virtually all patients have significant proteinuria, and two-thirds are frankly nephrotic. In time, 25% of patients will have complete resolution of their proteinuria. Most of the rest will have diminished proteinuria, although some of these may be a consequence of worsening glomerular filtration rate. Only about 1 in 10 individuals is left with persistent nephrotic syndrome.

Renal function is impaired in only a small number of patients at presentation, about 10%. An additional 5% develop renal insufficiency during followup. About 33% of those with renal insufficiency will progress to renal failure, but this represents only 1 in 20 patients with membranous disease. Serum creatinine at presentation emerges as the most consistent predictor of renal failure.

Some additional insight is obtained by subdividing the group by the degree of inflammation on histologic analysis. About 3 times as many patients developed end-stage renal disease with the Vc and Vd classification as those with Va or Vb. This is in keeping with the conclusions of the Lupus Nephritis Collaborative Study Group that proliferative changes in the presence of membranous nephropathy confer a worse prognosis, similar to that of class IV histology (13), and supports the recent change in WHO classification.

Mortality is about 8% in patients with LMN. This figure is difficult to interpret because the data do not permit reconstruction into life tables. The causes of death agree with the frequently cited causes of death in SLE generally, including infections, renal failure, and cardiovascular events.

Unfortunately, little can be concluded regarding therapy for membranous disease. It would be valuable for future randomized, controlled trials of therapeutic agents for lupus nephritis to include patients with membranous disease. However, because membranous nephropathy appears to be relatively uncommon, it would be most appropriate to stratify patients by histologic type prior to randomization to ensure that control and treated groups are balanced with regard to this important variable.